You are on page 1of 6

1611

NOTE
Tensile strength of sands treated with microbially induced
carbonate precipitation
Ashkan Nafisi, Douglas Mocelin, Brina M. Montoya, and Shane Underwood

Abstract: During large earthquake events where bending moments within soil cements are induced, the tensile strength of
Can. Geotech. J. Downloaded from cdnsciencepub.com by TU DELFT LIBRARY on 03/28/24

cemented soil may govern the deformational behavior of improved ground. Several studies have been conducted to assess the
tensile strength of artificially cemented sands that use Portland cement or gypsum; however, the tensile strength of microbially
induced carbonate precipitation (MICP)-treated sands with various particle sizes measured through direct tension tests has not
been evaluated. MICP is a biomediated improvement technique that binds soil particles through carbonate precipitation. In this
study, the tensile strength of nine specimens were measured by conducting direct tension tests. Three types of sand (coarse,
medium, and fine) were cemented to reach a heavy level of cementation (e.g., shear wave velocity of ⬃900 m/s or higher). The
results show that the tensile strength varies between 210 and 710 kPa depending on sand type and mass of carbonate. Unconfined
compressive strength (UCS) tests were performed for each sand type to assess the ratio between tensile strength and UCS in
MICP-treated sands. Scanning electron microscopy (SEM) images and surface energy measurements were used to determine the
predominant failure mode at particle contacts under tensile loading condition.

Key words: bio-cementation, MICP, tensile strength, unconfined compressive strength (UCS).

Résumé : Pendant les tremblements de terre de grande ampleur où lors de moments de flexion dans de ciments de sol induits,
la résistance à la traction du sol cimenté peut régir le comportement de déformation d’un sol amélioré. Plusieurs études ont été
For personal use only.

réalisées pour évaluer la résistance à la traction des sables cimentés artificiellement utilisant du ciment Portland ou du gypse;
cependant, la résistance à la traction de sables traités par précipitation carbonatée induite par des microorganismes (MICP) dont
la taille de particules varie, mesurée par essai de tension directe n’a pas été évaluée. Le MICP est une technique d’amélioration
biomédiée reliant les particules du sol par précipitation carbonatée. Dans cette étude, la résistance à la traction de neuf
éprouvettes a été mesurée par des essais de traction directe. Trois types de sable (grossier, moyen et fin) ont été cimentés afin
d’atteindre un niveau élevé de scellement (p. ex. vitesse de l’onde de cisaillement de ⬃900 m/s ou plus). Les résultats démontrent
que la résistance à la traction varie entre 210 et 710 kPa selon le type de sable et selon la masse de carbonate. Des essais de
résistance à la compression sans confinement ont été réalisés pour chaque type de sable dans le but d’évaluer le rapport entre
la résistance à la traction et la résistance à la compression sans confinement dans les sables traités au MICP. Des images la
microscopie électronique à balayage (MEB) et des mesures de l’énergie de surface ont été utilisées pour déterminer le mode de
défaillance prédominant aux contacts de particules soumis à la charge de traction. [Traduit par la Rédaction]

Mots-clés : biocimentation, MICP, résistance à la traction, résistance non confinée à la compression (UCS).

1. Introduction Direct tension, bending, and Brazilian (splitting tensile) tests


are used to estimate the tensile strength of cemented soils (Das
Tensile strength is an important parameter that governs the
et al. 1995; Namikawa and Koseki 2007). In the direct tension test,
behavior of cemented sands in some special cases. As an example,
tensile loading is applied to a cylindrical specimen along the axis
Namikawa et al. (2007) evaluated the dynamic behavior of lattice-
of symmetry until it fails, and then the tensile strength is calcu-
shaped ground improvement in artificially cemented sand through
lated based on the measured load and the sectional area of the
a finite element analysis and concluded that tensile failure hap- specimen. Tensile strength can also be measured by applying a
pens at the corners of the cemented zone. Consoli et al. (2003) point load at the center of a simple supported beam (i.e., in bend-
showed that improved layers of soils start to fail under tensile ing). The Brazilian test measures the tensile strength indirectly by
stresses in soil–cement mixtures. Tensile cracks and bending ten- applying a compressional force along the entire length of the
sile failure are reported in soils cemented using deep mixing specimen, thus inducing tensile stresses in the diametrical direc-
method (Kitazume and Maruyama 2007). Boulanger et al. (2018) tion. Namikawa and Koseki (2007) used analytical simulations of
numerically simulated centrifuge model tests of an embankment the three tensile loading scenarios and compared the simulations
on a liquefiable sand layer treated by soil–cement walls and found with experimental results. They found direct tension is the most
that tension cracks were fully developed in soil–cement walls accurate method among the three approaches to assess the tensile
during large shaking events. These examples of soil–cement walls strength, while the bending test overestimates and the Brazilian
illustrate tensile strength of cemented sand performance. test underestimates the tensile strength. Dass et al. (1994) com-

Received 3 April 2019. Accepted 23 October 2019.


A. Nafisi. Terracon Consultants Inc., Germantown, MD, USA.
D. Mocelin, B.M. Montoya, and S. Underwood. Department of Civil, Construction, and Environmental Engineering, North Carolina State University,
Raleigh, NC, USA.
Corresponding author: Ashkan Nafisi (email: ashkan.nafisi@terracon.com).
Copyright remains with the author(s) or their institution(s). Permission for reuse (free in most cases) can be obtained from copyright.com.

Can. Geotech. J. 57: 1611–1616 (2020) dx.doi.org/10.1139/cgj-2019-0230 Published at www.nrcresearchpress.com/cgj on 1 November 2019.
1612 Can. Geotech. J. Vol. 57, 2020

Table 1. Physical properties of sands. Table 2. Chemical recipes.


Sand Ottawa 20-30 Ottawa 50-70 Nevada Biological Cementation
solution solution
D50 (mm) 0.72 0.22 0.12
Cu 1.17 1.40 1.70 Chemical (mmol/L) (mmol/L)
Cc 1.02 0.90 1.24 Urea 333 333
Gs 2.65 2.65 2.65 NH4Cl 374 374
emin 0.50 0.55 0.56 CaCl2 — 100
emax 0.74 0.87 0.86
Note: D50, mean particle size; Cu, coefficient of uniformity;
Cc, coefficient of curvature; Gs, specific gravity of solid soil particles;
emin and emax, minimum and maximum void ratios, respectively. three specimens to assess the ratio between tensile strength and
UCS.
pared the tensile strength of Portland cemented sands with ce-
2. Materials and methods
Can. Geotech. J. Downloaded from cdnsciencepub.com by TU DELFT LIBRARY on 03/28/24

ment content varying from 4% to 8% through direct tensile and


Brazilian tests. They also found that the Brazilian test underesti- 2.1. Sand properties and sample preparation
mates the tensile strength. Properties of the sand types used in this study are tabulated in
Several studies have evaluated the tensile strength of artificially Table 1. The specimens were prepared and treated in 50 mL vials
and naturally cemented sands. Das et al. (1995) measured the ten- without applying any surcharge or vertical loads during the treat-
sile strength of a fine-grained sand artificially cemented with Type ment. The height of vials (from tip to top) was ⬃110 mm and the
I Portland cement. Brazilian tension tests were conducted to mea- diameter was 25 mm. Two lines were connected to the top and
sure tensile strengths of about 140 to 300 kPa, depending upon bottom of the vials to inject and collect the solutions. Two filters
the cement content, which ranged from 4% to 8% by weight. (Porex) were also put at the top and bottom of the vials to keep the
Namikawa et al. (2017) conducted a series of direct tension tests on sand particles in place. All specimens were prepared using air
artificially cemented sand under confinements, which ranged pluviation at relative density of about 40%. The diameter and
from 30 to 500 kPa. They concluded that tensile failure happens at height of the specimens were 25 and 75 mm, respectively. After
low confinements, while shear failure occurs at high confine- treatment, the specimens were flushed with tap water to remove
ments. Tensile strength as high as 350 kPa was reported for ce-
residual salts. The tips of vials were then cut with a band saw, and
mented sand with more than 10% of cement content (Consoli et al.
the specimens were extruded carefully by applying uniform load
For personal use only.

2010). Tensile strength of naturally cemented carbonate sand was


to the bottom of the specimens. To minimize any potential dam-
assessed varying from 50 to 500 kPa, based on the cement content
age, the vials were cut from a place that did not have any contact
and density of the soil (Airey 1993). Choi et al. (2016) conducted the
with the cemented sand (just below the bottom filter). The speci-
Brazilian test on fiber reinforced, bio-cemented sands and found
mens were extruded carefully and smoothly to avoid cracking
that tensile and unconfined strengths increase as the fiber con-
during extrusion. The specimens were oven-dried 24 h before the
tent increases. The tensile strength of sands treated with micro-
bially induced carbonate precipitation (MICP), measured through tensile strength and UCS measurements.
direct tension test in sand types with different particle sizes, has 2.2. Bio-cementation process
not been evaluated. Knowing that the behavior of MICP-treated Biological treatment was performed by using the Sporosarcina
sands and other types of cemented sands (naturally and artifi-
pasteurii bacterium, prepared by following the method described
cially) are not necessarily similar (Ismail et al. 2002), it is essential
in Mortensen et al. (2011). The final optical density (OD600) of the
to measure the tensile strength of MICP-treated sands to have a
bacterial suspension was between 0.8 and 1.2, which corresponds
reliable assessment.
to about 1 × 106 cells/mL based on a lab-specific OD to total cells
MICP is a promising improvement technique, which has the
correlation. A two-phase injection method was used for treat-
potential to improve shear strength, stiffness, volumetric behav-
ments. The treatment process was initiated by injecting a biolog-
ior, compressibility, and erodibility (DeJong et al. 2006; Whiffin
ical solution (containing the bacteria) with a retention time of 4 to
et al. 2007; Van Paassen et al. 2010; Lin et al. 2015; Montoya and
DeJong 2015; Montoya et al. 2018; Terzis and Laloui 2019; Zamani 6 h followed by injecting two pore volumes of cementation solu-
and Montoya 2019). In addition, MICP can improve the leaching tion. Table 2 shows the chemical recipes for biological and cemen-
behavior of contaminated soils or coal combustion residual in a tation solutions. Before injecting the cementation solution at
more eco-friendly and economic manner than conventional tech- each step, 0.4 pore volumes of a solution containing 333 mmol/L
niques (Safavizadeh 2017). In this technique, urea is hydrolyzed to of urea was injected to prevent immediate precipitation of car-
ammonia and carbonate by ureolytic bacteria. In the presence of bonate (CaCO3) near the injection points. Cementation treat-
calcium ions, the generated carbonate reacts with calcium, and ments were repeated in 6–6–12 h intervals. The direction of
calcium carbonate precipitates. In some cases, the concentration injection was alternately changed from top to bottom and bottom
of urea is selected to be higher than the concentration of calcium to top to improve the uniformity of carbonate distribution. The
in the injecting solution to improve the efficiency of the MICP solutions were injected using a peristaltic pump with the flow rate
process (Martinez 2012). Under this circumstance, the excessive of 10 mL/min.
produced carbonate in the system increases the pH of the effluent Since a pH of about 9 could indicate that bacteria are actively
solution. Therefore, measuring the pH of the effluent solution hydrolyzing urea, and a pH of about 7 might be an indication that
may be used as a tool to monitor the performance of the bacteria the ureolytic activity is less active in the soil (for the chemical
in the soil media and cementation process when an unbalanced recipe used here in this study), the pH was measured after each
recipe is used for the injecting solution. Precipitation of carbonate treatment. For cases in which the pH dropped from 9 to 7, a small
leads to the binding and roughening of sand particles (DeJong amount of bacteria was injected in the specimens with cementa-
et al. 2010). MICP improves the tensile strength of the sands, but tion solution to increase the activity of bacteria inside the speci-
the amount of improvement is still unknown. In the study pre- men (Feng and Montoya 2015; Martinez 2012). During this process,
sented herein, the tensile behavior of MICP-treated sand is evalu- a low density of ureolytic microbes with concentration of about
ated by conducting nine direct tension tests on three types of 4 × 104 cells/mL was added to the cementation solution to increase
sand. Unconfined compressive strength (UCS) was measured for the bacterial activity inside the specimen.

Published by NRC Research Press


Nafisi et al. 1613

Table 3. Characteristics of specimens.


Loading Test No. of CaCO3 content (%) Failure Strength Failure
mode No. Sand injections (top, bottom, mean) location (kPa) strain (%)
Tension 1 Ottawa 20-30 25 (4.8, 6.1, 5.5) Top third 640 0.017
2 Ottawa 20-30 25 (4.9, 5.6, 5.2) Top third 600 0.026
3 Ottawa 20-30 25 (3.2, 4.4, 3.8) Top third 310 0.029
4 Ottawa 50-70 30 (9.1, 11.9, 10.5) Middle third 710 0.027
5 Ottawa 50-70 30 (8.4, 7.7, 8.0) Middle third 480 0.026
6 Ottawa 50-70 30 (7.9, 6.4, 7.1) Bottom third 370 0.030
7 Nevada 40 (12.9, 14.1, 13.5) Middle third 490 0.027
8 Nevada 40 (15.2, 13.6, 14.4) Bottom third 480 0.039
9 Nevada 40 (12.7, 15.5, 14.1) Top third 210 0.037
Compression 10 Ottawa 20-30 25 (NA, 6.5, 6.5)* Inclined 2500 0.21
11 Ottawa 50-70 30 (10.2, 11.5, 10.8) Inclined 3040 0.36
Can. Geotech. J. Downloaded from cdnsciencepub.com by TU DELFT LIBRARY on 03/28/24

12 Nevada 40 (13.7, 14.1, 13.9) Inclined 2600 0.28


*NA, not available.

2.3. Cementation degree 2.5. Scanning electron microscopy (SEM) images


All of the specimens were treated to reach a heavy level of SEM images were taken from the failure surfaces to find the
cementation, which resembles rock behavior with normalized failure mode under tensile load condition using a Hitachi S3200N
shear wave velocity (Vs1) of 900 m/s or higher (Montoya and DeJong Variable Pressure Scanning Electron Microscope. Aggregates of
2015; Nafisi et al. 2020). The specimens were treated to a number cemented sand from the exposed failure surface were carefully
of treatments depending on the particle size. Since the shear wave cut and removed from the specimen to be used in the SEM images.
velocity was not measured in this study, the number of treat- The SEM samples were cut from the center of the specimen using
ments were chosen based on the authors’ experience in monitor- a sharp blade, then placed on carbide tape on the sample stub for
ing the shear wave velocity during the course of treatment in SEM images with the failure surface exposed for imaging. Owing
triaxial specimens for the sand types (Nafisi 2019; Nafisi et al. to the level of cementation, the aggregates used for imaging were
2020). Table 3 shows the treatment numbers for each specimen. workable and not friable during SEM sample preparation. The
Since a Vs1 above 900 m/s was intended for all specimens, a higher samples were coated with gold–palladium alloy to decrease charg-
For personal use only.

number of treatments (and mass of carbonate) was needed as the ing during imaging.
sand particle size reduces (Nafisi et al. 2020). Therefore, the Ne-
vada sand specimens were treated 40 times, whereas the Ottawa 2.6. Mass of carbonate
20-30 sand specimens were treated 25 times to reach the heavy Carbonate content was determined using a gravimetric acid-
level of cementation. washing method. After the completion of each test, the specimens
were divided into two sections along the length. Each section was
2.4. Direct tension and unconfined compressive strength
washed with 1 mol/L HCl. The mass of carbonate was calculated by
(UCS) tests
measuring the difference between oven-dried mass (at 110 °C) of
Both direct tension and UCS tests were conducted using an MTS
the soil sample before and after acid-washing.
810 machine that was programmed to apply a constant actuator
rate displacement. During the test, the applied force and actuator 2.7. Surface energy measurements
displacements were recorded for later analysis. The rate of dis-
Surface energy measurements are currently used to determine
placement was 0.015 mm/min (equivalent to mean specimen strain of
the cohesive and adhesive characteristics of asphalt-binder and
0.02%/min) and 0.15 mm/min (equivalent to mean specimen strain
asphalt-aggregate systems (Bhasin and Little 2007). Generally, the
of 0.2%/min) for tension and UCS tests, respectively. The rate of
surface energy of a material is defined as the amount of work
displacement for tension was selected based on Japanese Geotech-
needed to create a unit area of the material and is calculated by
nical Society standard JGS2552-2015 (Japanese Geotechnical Society
Standards 2015). The same rate is used to measure tensile strength of knowing the contact angle between the material surface and a
Portland cemented sands through direct tensile test in the litera- drop of probe fluids. Indeed, works of cohesion and adhesion
ture (Namikawa et al. 2017). The rate of applying load in UCS in represent the energy required for microcracks to propagate
similar studies is about 0.1 to 1%/min (Van Paassen et al. 2010; within calcium carbonate bonds and at the interface of silica–
Qabany and Soga 2013). To account for the machine compliance, a calcium carbonate, respectively, regardless of the particle geom-
pilot test was carried out by first instrumenting a specimen with etry and the localized stress at particle contacts. Indeed, these
surface-mounted LVDTs (linear variable differential transform- analyses assume equal stresses would be present at all locations.
ers). Measurements of the force, on-specimen displacement, and To measure the contact angle, the sessile drop method was used
actuator displacements were used to calibrate the machine com- (Koc and Bulut 2014). Three different fluids (water, glycerol, and
pliance factors, which were then used on the subsequent tests. For ethylene glycol) were dropped on a silica aggregate plate and a
the subsequent tests, displacements were measured by the actua- sheet of MICP–calcium carbonate (three drops per each fluid and
tor LVDTs installed on the MTS machine. each material). To prepare MICP–calcium carbonate sheets, the
To perform the tensile tests, the specimens were attached to the mixture of bacterial cultures and cementation solution (shown in
machine actuator by gluing the specimens ends to metal plates Table 2) was poured into a weighing dish. The solution inside the
and then attaching the plates to the machine actuator. A gluing jig weighing dish was replaced in 6–6–12 h intervals. The thickness of
was used to ensure the proper alignment of the specimens and precipitated calcium carbonate on the bottom of the dish gradu-
also ensure parallelism of the plates. The glue used was a plastic ally increased so that calcium carbonate sheets were formed. The
steel putty composite that allowed a fast (5 min) curing period. formed sheets were detached from the plastic dish and used for
This procedure was also used for the compression tests to ensure the tests in the study. Geologic silica sheet was used for the mea-
homogeneous load distribution on the specimen–plate contact. surements on silica.
All the specimens were oven-dried for 24 h before gluing and The total surface energy (␥total) of any material consists of three
testing. component: (1) the van der Waals or the dispersive component

Published by NRC Research Press


1614 Can. Geotech. J. Vol. 57, 2020

Fig. 1. Tensile strength of (a) Ottawa 20-30; (b) Ottawa 50-70; and (c) Nevada sands. [Colour online.]
Can. Geotech. J. Downloaded from cdnsciencepub.com by TU DELFT LIBRARY on 03/28/24

(␥LW); (2) the Lewis acid component (␥+); and (3) the Lewis base Fig. 2. Variation of tensile strength versus average carbonate content.
component (␥−). [Colour online.]

(1) ␥total ⫽ ␥LW ⫹ ␥AB ⫽ ␥LW ⫹ 2兹␥⫹␥⫺

These components were calculated for MICP–calcium carbon-


ate and silica based on measured contact angles, known surface
tension parameters of the fluids, and Young–Dupré equation
(Carré 2007; Van Oss et al. 1988).
Knowing these components, work of cohesion within calcium
C
carbonate bonds 共WCaCO 3
兲 and adhesion between silica and cal-
A
cium carbonate 共WSiCaCO3兲 were obtained through the following
equations:
For personal use only.

(2) C
WCaCO 3
⫽ 2␥CaCO
total
3

兹␥ 兹␥ 兹␥
⫹ ⫺ ⫺ +
Si ␥CaCO3 Si␥CaCO3 Si␥CaCO3
A LW LW
(3) WSiCaCO 3
⫽2 ⫹2 ⫹2 cannot comment on the sensitivity of the tensile strength to car-
bonate content in the sand types, considering the limited data
herein. Any decisive conclusion about the sensitivity of tensile
In these equations, calcium carbonate and silica are denoted by strength to carbonate content needs further investigation. How-
CaCO3 and Si, respectively. ever, it is hypothesized that the higher number of particle con-
tacts in the Nevada sand specimens (due to the finer size and the
3. Results and discussion higher angularity of Nevada sand particles) may influence the
Nine direct tension tests were conducted to assess the tensile sensitivity. If it is assumed that the mass of carbonate is distrib-
strength of MICP-treated sand, the results of which are illustrated uted to each particle contact, a difference in carbonate content in
in Fig. 1 and tabulated in Table 3. The results show that the tensile the Nevada sand specimen may result in less change in carbonate
strength improves substantially in all three sand types, and it at each contact compared with coarser sands, such as Ottawa
increases as the carbonate content increases. The measured ten- 20-30 sand. The precipitated carbonate in the top and bottom
sile strengths were within the range of 210 to 710 kPa. The axial sections are also presented in Table 3. The mean carbonate con-
strain at failure (␧f) ranged from 0.02% to 0.04% for the specimens, tent varied from 3.8% to 14.4%. Nevada sand specimens have
which is in the same range of ␧f for Portland cemented sands higher carbonate content than the other two sand types due to
previously reported (Das et al. 1995; Namikawa et al. 2017). The the higher number of treatments; however, the average tensile
stress–strain response of the specimens demonstrates softening strength is almost similar for all three sand types. This is consis-
behavior. Based on visual observations during testing, small tent with the reported peak shear strength under isotropically
cracks were initiated approximately at the peak strength and then consolidated drained compression loading of these three sand
propagated with continued load until the specimen was sepa- types with similar cementation levels (Nafisi et al. 2020).
rated. The Nevada-3 specimen showed a lower tensile strength Although the number of treatments for a given sand type was
than other specimens. Since the carbonate content in this speci- the same, the precipitated carbonate is different, which might
men is not remarkably lower than that in other specimens stem from varying bacteria activity from one batch to another or
(Table 3), it is speculated that the specimen was damaged either the presence of preferential paths during the course of treatment.
during extrusion or gluing processes. In general, the observed These variations highlight the imperative of monitoring the ce-
variation of tensile strength is dependent on the amount of car- mentation process in real time in a nondestructive manner, such
bonate mineral precipitated in the specimens over the range of as measuring the shear wave velocity of the material (e.g., DeJong
carbonate contents in this study, as demonstrated in Fig. 2. The et al. 2010; Nafisi et al. 2020).
Ottawa 20-30-3 specimen showed about 300 kPa lower of the ten- To determine the predominant mode of failure at particle con-
sile strength compared with the two other Ottawa 20-30 speci- tacts, surface energy measurements and SEM were used. Adhesive
mens, which can be attributed to its lower carbonate content (failure at the interface of carbonate and sand particle) and cohe-
(about 1.5% lower as shown in Table 3). The same trend was ob- sive (failure within the carbonate bonds) modes of failure are two
served for the Ottawa 50-70 sand specimens in which 3.4% differ- possibilities at particle contacts (DeJong et al. 2010; Montoya and
ence in the average carbonate content resulted in ⬃340 kPa Feng 2015). The predominant failure mode under tension loads
difference in the tensile strength (test Nos. 4 and 6). The authors was first intended to be investigated visually with the aid of SEM

Published by NRC Research Press


Nafisi et al. 1615

Table 4. Work of cohesion (WC) and adhesion (WA) of calcium carbonate and silica.
Material ␥AB (mJ/m2) ␥LW (mJ/m2) ␥total (mJ/m2) WC (mJ/m2) WA (mJ/m2)
Calcium carbonate 3.1 38.3 41.4 82.8 84.2
Silica 6.2 37.0 43.2 86.4 84.2
Note: Results for silica are from Feng (2015).

Fig. 3. Unconfined compressive strength (UCS) measurements. Fig. 4. Typical mode of failure under (a) compressive and (b) tensile
[Colour online.] loading. [Colour online.]
Can. Geotech. J. Downloaded from cdnsciencepub.com by TU DELFT LIBRARY on 03/28/24
For personal use only.

Terzis and Laloui 2018). The failure strain for UCS tests were 0.21%,
0.36%, and 0.28% in Ottawa 20-30, Ottawa 50-70, and Nevada
sands, respectively. Comparing tensile and compressive behaviors
indicates that failure occurs in lower axial strains under tensile
loads compared with compressive loads (i.e., about an order of
magnitude lower). To assess the magnitude of the ratio of tensile
images; however, SEM images did not provide any reliable infor-
strength to UCS, tests with similar carbonate contents (Nos. 1 and
mation, so they are not presented in this paper. Additionally, any
10, Nos. 4 and 11, and Nos. 7 and 12) were considered. According to
conclusion regarding the predominant mode of failure based on
the test results, this ratio was 0.25, 0.23, and 0.19 for Ottawa 20-30,
SEM images requires statistical studies to analyze an extensive
Ottawa 50-70, and Nevada sands, respectively. The ratio of the
number of SEM images, which is beyond the scope of this study.
Moreover, determining the mode of failure visually based on im- tensile strength to UCS for Portland cemented sand varied be-
ages and the shape of calcium carbonate at contacts can be sub- tween 0.1 and 0.3, depending on the cement content and the type
jective. Therefore, the work of cohesion and adhesion of the MICP of tensile test (Das et al. 1995; Namikawa and Koseki 2007; Consoli
calcite–silica system was calculated through surface energy mea- et al. 2010).
surements and Young–Dupré equation (Van Oss et al. 1988) so that Figure 4 illustrates the typical failure mode of specimens under
the predominant mode of failure can be determined quantita- compressive and tensile loads. The location of failure for each
tively. Although any decisive conclusion requires a more in-depth specimen under tensile loading condition is presented in Table 3.
study (numerically and experimentally), surface energy measure- The failure surface was almost perpendicular to the loading direc-
ments give an insight on the mode of failure under tensile loads. tion in all tension tests (similar to Fig. 4b).
Table 4 presents the adhesion and cohesion work for the MICP
calcium carbonate–silica system. WC and WA show the work of
4. Conclusion
cohesion and adhesion, respectively. According to the values, co- Direct tension and UCS tests were conducted on 12 specimens to
hesive failure is more likely to happen at the particle contact since evaluate the tensile behavior of three types of sands (coarse, me-
C C A C dium, and fine sands) cemented through MICP. The results con-
WCaCO 3
has the least value among WSi , WSiCaCO 3
, and WCaCO 3
, which
means that microcracks initiate and propagate within the cal- firm that MICP is an appropriate technique in increasing the
cium carbonate bonds before they initiate at the interface be- tensile strength of sands. The fine sand (Nevada sand) required a
tween calcium carbonate and silica. higher carbonate content than medium and coarse sands to reach
Note that the lower work of cohesion does not indicate that a similar range of tensile strength. The magnitude of the ratio
cohesive failure happens at every single contact, since other pa- between tensile strength and UCS ranged from 0.19 to 0.25 de-
rameters such as roughness of sand particles, the presence of pending upon the particle size. The ratio of the tensile strain at
impurity, and congestion of bacteria on sand particles may affect failure to that of UCS is ⬃0.1. Moreover, the specimens after ten-
the mode of failure. sile failure showed that the failure surface is perpendicular to the
The UCS test was performed to assess the ratio between the applied tensile load direction. Surface energy measurements were
tensile strength and UCS for MICP-treated sands. The UCS values performed on carbonate and silica sheets. Based on surface energy
were approximately 2.5, 3.0, and 2.6 MPa for Ottawa 20-30, Ottawa measurements, cohesive failure within calcium carbonate bonds
50-70, and Nevada sands, respectively (Fig. 3). Similar UCS values is more likely to occur than adhesive failure at particle contacts
have been reported for MICP-treated sands with the similar ranges under tensile loads. This conclusion assumes that stresses at all
of carbonate content (Gomez et al. 2017; Van Paassen et al. 2010; locations within a bond are equal.

Published by NRC Research Press


1616 Can. Geotech. J. Vol. 57, 2020

Acknowledgements Japanese Geotechnical Society Standards. 2015. Laboratory Testing Standards of


Geomaterial, 2015, Vol. 3, Chapter 2552.
Funding from the National Science Foundation (CMMI Nos. Kitazume, M., and Maruyama, K. 2007. Internal stability of group column type
1537007 and 1554056) is appreciated. Any opinions, findings, and deep mixing improved ground under embankment loading. Soils and Foun-
conclusions or recommendations expressed are those of the au- dations, 47(3): 437–455. doi:10.3208/sandf.47.437.
thors and do not necessarily reflect the views of the National Koc, M., and Bulut, R. 2014. Assessment of a sessile drop device and a new testing
approach measuring contact angles on aggregates and asphalt binders. Jour-
Science Foundation. This work was performed in part at the Ana- nal. Material Civil Engineering, 26(3): 391–398. doi:10.1061/(ASCE)MT.1943-
lytical Instrumentation Facility (AIF) at North Carolina State Uni- 5533.0000852.
versity, which is supported by the State of North Carolina and the Lin, H., Suleiman, M.T., Brown, D.G., and Kavazanjian, E. 2015. Mechanical be-
havior of sands treated by microbially induced carbonate precipitation. Jour-
National Science Foundation (ECCS No. 1542015). The AIF is a nal of Geotechnical and Geoenvironmental Engineering, 142(2). doi:10.1061/
member of the North Carolina Research Triangle Nanotechnology %28ASCE%29GT.1943-5606.0001383.
Network (RTNN), a site in the National Nanotechnology Coordi- Martinez, B.C. 2012. Up-Scaling of microbial induced calcite precipitation in
nated Infrastructure (NNCI). sands for gotechnical ground improvement. University of California, Davis,
Calif.
Montoya, B.M., and DeJong, J.T. 2015. Stress-strain behavior of sands cemented
References by microbially induced calcite precipitation. Journal of Geotechnical and
Can. Geotech. J. Downloaded from cdnsciencepub.com by TU DELFT LIBRARY on 03/28/24

Airey, D.W. 1993. Triaxial testing of naturally cemented carbonate soil. Journal Geoenvironmental Engineering, 141(6): 4015019. doi:10.1061/(ASCE)GT.1943-
of Geotechnical Engineering, 119(9): 1379–1395. doi:10.1061/(ASCE)0733- 5606.0001302.
9410(1993)119:9(1379). Montoya, B.M., Do, J., and Gabr, M.M. 2018. Erodibility of microbial induced
Bhasin, A., and Little, D.N. 2007. Characterization of aggregate surface energy carbonate precipitation-stabilized sand under submerged impinging jet.
using the universal sorption device. Journal of Materials in Civil Engineering, International Foundation Congress and Equipment Expo 2018, American
19(8): 634–641. doi:10.1061/(ASCE)0899-1561(2007)19:8(634). Society of Civil Engineers, Orlando, Fla. pp. 19–28.
Boulanger, R.W., Khosravi, M., Khosravi, A., and Wilson, D.W. 2018. Remediation Mortensen, B.M., Haber, M.J., Dejong, J.T., Caslake, L.F., and Nelson, D.C. 2011.
of liquefaction effects for an embankment using soil–cement walls: centri- Effects of environmental factors on microbial induced calcium carbonate
fuge and numerical modeling. Soil Dynamics and Earthquake Engineering, precipitation. Journal of Applied Microbiology, 111(2): 338–349. doi:10.1111/j.
114: 38–50. doi:10.1016/j.soildyn.2018.07.001. 1365-2672.2011.05065.x. PMID:21624021.
Carré, A. 2007. Polar interactions at liquid/polymer interfaces. Journal of Adhesion Nafisi, A. 2019. Elucidating the failure behavior and bond mechanics of bio-
Science and Technology, 21(10): 961–981. doi:10.1163/156856107781393875. cemented sands. North Carolina State University. Raleigh, N.C.
Choi, S.-G., Wang, K., and Chu, J. 2016. Properties of biocemented, fiber rein- Nafisi, A., Montoya, B.M., and Evans, T.M. 2020. Shear strength envelopes of
bio-cemented sands with varying particle size and cementation level. Journal
forced sand. Construction and Building Materials, 120: 623–629. doi:10.1016/
of Geotechnical and Geoenvironmental Engineering, 146(3): 04020002. doi:
j.conbuildmat.2016.05.124.
10.1061/(ASCE)GT.1943-5606.0002201.
Consoli, N.C., Vendruscolo, A., and Prietto, P.D. 2003. Behavior of plate load tests
Namikawa, T., and Koseki, J. 2007. Evaluation of tensile strength of cement-
on soil layers improved with cement and fiber. Journal of Geotechnical and
treated sand based on several types of laboratory tests. Soils and Founda-
Geoenvironmental Engeering, 129(1): 96–101. doi:10.1061/(ASCE)1090-0241(2003)
For personal use only.

tions, 47(4): 657–674. doi:10.3208/sandf.47.657.


129:1(96).
Namikawa, T., Koseki, J., and Suzuki, Y. 2007. Finite element analysis of lattice-
Consoli, N.C., Cruz, R.C., Floss, F., and Festugato, L. 2010. Parameters controlling shaped ground improvement by cement-mixing for liquefaction mitigation.
tensile and compressive strength of artificially cemented sand. Journal of Soils and Foundations, The Japanese Geotechnical Society, 47(3): 559–576.
Geotechnical and Geoenvironmental Engeering, 136(5): 759–763. doi:10.1061/ doi:10.3208/sandf.47.559.
(ASCE)GT.1943-5606.0000278. Namikawa, T., Hiyama, S., Ando, Y., and Shibata, T. 2017. Failure behavior of
Das, B., Yen, S., and Dass, R. 1995. Brazilian tensile strength test of lightly cement-treated soil under triaxial tension conditions. Soils and Foundations,
cemented sand. Canadian Geotechnical Journal, 32: 166–171. doi:10.1139/t95- Japanese Geotechnical Society, 57(5): 815–827. doi:10.1016/j.sandf.2017.08.011.
013. Qabany, A., and Soga, A. 2013. Effect of chemical treatment used in MICP on
Dass, R.N., Yen, S.-C., Das, B.M., Puri, V.K., and Wright, M.A. 1994. Tensile stress– engineering properties of cemented soils. Geotechnique, 63(4): 331–339. doi:
strain characteristics of lightly cemented sand. Geotechnical Testing Journal, 10.1680/geot.SIP13.P.022.
17(3): 305–314. doi:10.1520/GTJ10105J. Safavizadeh, S. 2017. Evaluating the effect of MICP treatment on the structural
DeJong, J.T., Fritzges, M.B., and Nüsslein, K. 2006. Microbially induced cementa- and chemical stabilization of coal ash impoundments. North Carolina State
tion to control sand response to undrained shear. Journal of Geotechnical University, Raleigh, N.C.
and Geoenvironmental Engeering, 132(11): 1381–1392. doi:10.1061/(ASCE)1090- Terzis, D., and Laloui, L. 2018. 3-D micro-architecture and mechanical response
0241(2006)132:11(1381). of soil cemented via microbial-induced calcite precipitation. Scientific Re-
DeJong, J.T., Mortensen, B.M., Martinez, B.C., and Nelson, D.C. 2010. Bio- ports, 8(1): 1416.
mediated soil improvement. Ecological Engeering, 36: 197–210. doi:10.1016/j. Terzis, D., and Laloui, L. 2019. Cell-free soil bio-cementation with strength, dila-
ecoleng.2008.12.029. tancy and fabric characterization. Acta Geotechnica, 14(3): 639–656. doi:10.
Feng, K. 2015. Constitutive response of microbial induced calcite precipitation 1007/s11440-019-00764-3.
cemented sands. North Carolina State University, Raleigh, N.C. Van Oss, C.J., Chaudhury, M.K., and Good, R.J. 1988. Interfacial Lifshitz–van der
Feng, K., and Montoya, B.M. 2015. Influence of confinement and cementation Waals and polar interactions in macroscopic systems. Chemical reviews,
level on the behavior of microbial-induced calcite precipitated Sands under 88(6). doi:10.1021/cr00088a006.
monotonic drained loading. Journal of Geotechtechnical and Geoenviron- Van Paassen, L.A., Ghose, R., Van Der Linden, T.J.M., Van Der Star, R.L.W., and
mental Engineering, 142(1). doi:10.1061/(ASCE)GT.1943-5606.0001379. Van Loosdrecht, M.C.M. 2010. Quantifying biomediated ground improve-
Gomez, M.G., Graddy, C.M.R., Dejong, J.T., Nelson, D.C., and Tsesarsky, M. 2017. ment by ureolysis: large-scale biogrout experiment. Journal of Geotechnical
Stimulation of native microorganisms for biocementation in samples recov- and Geoenvironmental Engineering, 136(12): 1721–1728. doi:10.1061/(ASCE)GT.
ered from field-scale treatment depths. Journal of Geotechtechnical and Geo- 1943-5606.0000382.
environmental Engineering, 144(1). doi:10.1061/%28ASCE%29GT.1943-5606. Whiffin, V.S., Van Paassen, L.A., and Harkes, M.P. 2007. Microbial carbonate
0001804. precipitation as a soil improvement technique. Geomicrobiology Journal, 24:
Ismail, M.A., Joer, H.A., Sim, W.H., and Randolph, M.F. 2002. Effect of cement 417–423. doi:10.1080/01490450701436505.
type on shear behavior of cemented calcareous soil. Journal of Geotechnical Zamani, A., and Montoya, B.M. 2019. Undrained cyclic response of silty sands
and Geoenvironmental Engineering, 128(6): 520–529. doi:10.1061/(ASCE)1090- with microbial induced calcium carbonate precipitation. Soil Dynamic and
0241(2002)128:6(520). Earthquake Engineering, 120: 436–448. doi:10.1016/j.soildyn.2019.01.010.

Published by NRC Research Press

You might also like