Abstract: Cement is one of the most consumed materials in the world. The cement industry is responsible for a large portion of global carbon
Downloaded from ascelibrary.org by 106.222.187.25 on 12/18/19. Copyright ASCE. For personal use only; all rights reserved.
dioxide emissions. Cement production and therefore carbon dioxide emissions can be decreased by increasing the durability and enhancing
the mechanical properties of cement-based materials. On the other hand, an important weakness of concrete is its weak tensile properties,
which are the main reasons for its failure and low durability. Therefore, over the past 30 years, many studies have focused on improving
tensile properties using a variety of physical and chemical methods. One of the most successful attempts is to use polymer fibers in the
structure of concrete to obtain a composite with high tensile strength and ductility. However, a thorough understanding of the mechanical
behavior of fiber-reinforced concrete requires knowledge of fiber cement interfaces at the nano scale. In this study, a combination of ex-
perimental and molecular dynamics (MD) techniques is used to study the nanostructure of fiber cement interfaces. Scanning electron micros-
copy (SEM) and energy-dispersive X-ray spectroscopy (EDX) analysis are used to obtain a better understanding of the C-S-H fiber (in cement
chemistry notation, C = CaO, S ¼ SiO2 , and H ¼ H2 O) interfaces. The results show that the C/S ratio changes in the interface
of cement and polymeric fibers and is largely affected by the functional group of the polymers. The results are then used to propose a
more realistic molecular dynamics model for C-S-H in the vicinity of the three most used polymeric fibers: polypropylene, polyvinyl alcohol,
and nylon-6. The full atomistic simulations show that the molecular structure of C-S-H at the interface depends on the properties of the
polymer functional group. The adhesion energy between the polymeric fibers and the relevant C-S-H structure is then computed using
atomistic simulations. The adhesion energy between C-S-H and polymers increases with the polarity of the fiber. The mechanical response
of cement paste with added polymeric fibers is then experimentally studied using the split-cylinder test. The experimental results further
show that the adhesion energies between the fibers and cement increase as a function of the polarity of the fibers. DOI: 10.1061/(ASCE)
NM.2153-5477.0000119. © 2017 American Society of Civil Engineers.
Author keywords: C-S-H; Polymeric fibers; Interface; Energy-dispersive X-ray spectroscopy (EDX) mapping.
Introduction would decrease the need for cement production and therefore
decrease carbon dioxide emissions (Van Vliet et al. 2012).
The cement-based matrix of concrete has a complex structure To improve the mechanical performance of concrete and other
that makes concrete a weak material in terms of material properties cement-based composites, cement material and chemical composi-
such as tensile strength and toughness. Therefore, many studies tions need to be characterized, well understood, and then improved.
have focused on improving the tensile properties of concrete using In portland cement, the raw materials are limestone and clay.
a variety of physical and chemical methods (Balaguru and Shah Hence, a typical composition of portland cement is approximately
1992). One of the most successful attempts is to use polymeric ma- 67% calcium oxide (CaO) and 22% silicon dioxide (SiO2 ), with
terials in the structure of concrete to obtain a composite with high mainly alite and belite molecules. The products of the cement hy-
tensile strength and ductility (Abbasnia et al. 2013; Akihama et al. dration reaction are mainly portlandite and calcium silicate hydrate,
1985; Betterman et al. 1995; Garcia et al. 1997; Li et al. 1994). or C-S-H gel. The C-S-H gel is up to 70% of the final volume.
Polypropylene, polyvinyl alcohol (PVA), and nylon-6 are some The produced C-S-H gel from the hydration reaction is respon-
of the most used polymeric fibers in this application. They also sible for the cohesion and strong mechanical properties of cement
represent a range of fibers with polar to nonpolar functional groups. pastes (Dolado et al. 2007). An understanding of the nature of chemi-
Improving the performance of polymer fiber-reinforced cement cal reactions occurring in C-S-H is also necessary in modeling the
composites can also increase the durability of cement-based struc- mechanical properties of concrete at the nano scale. To understand
tures. Producing stronger and more durable cement-based materials the chemical reactions in C-S-H, many researchers have previously
tried to characterize the structure of C-S-H gel at the atomistic
1
Dept. of Mechanical Engineering, Worcester Polytechnic Institute, level (Hajilar and Shafei 2015; Hamid 1981; Richardson 2008;
Worcester, MA 01609. Shahsavari et al. 2011; Shalchy and Rahbar 2015b; Taylor 1993).
2
Associate Professor, Dept. of Civil and Environmental Engineering, In that regard, different strategies were used to model the atomistic
Worcester Polytechnic Institute, Worcester, MA 01609 (corresponding structure of C-S-H to understand its nanostructure properties
author). E-mail: nrahbar@wpi.edu
(Richardson 2004; Wu et al. 2011). Bauchy et al. (2014) constructed
Note. This manuscript was submitted on February 17, 2016; approved
on October 31, 2016; published online on February 7, 2017. Discussion
a tobermorite crystalline structure, a more realistic C-S-H, and an
period open until July 7, 2017; separate discussions must be submitted artificial ideal glass. They showed that although C-S-H retains some
for individual papers. This paper is part of the Journal of Nanomechanics signatures of a tobermoritelike layered structure, hydrated species
and Micromechanics, © ASCE, ISSN 2153-5434. are completely amorphous. Earlier, Pellenq et al. (2009) proposed
important step in performing atomistic simulations. In general, the were added to the structure for the following reasons: (1) to satisfy
potential energy consists of valence, cross-term, and nonbonded the C/S ratio in the C-S-H structure at the interface with different
energies (Sun 1998; Sun et al. 1998) polymer fibers, (2) to satisfy the overall charge balance in the
system, and (3) to provide the necessary defects in the structure of
Etotal ¼ Evalence þ Ecross-term þ Enonbonded ð1Þ C-S-H. This approach was applied to construct different C-S-H
models with different C/S ratios using the EDX analysis results.
In Eq. (1), a bond-stretching term, a bending energy term, and The layered structure of the C-S-H surface was modeled and the
four body terms, including a dihedral bond-torsion angle term and potential energy was optimized using the steepest descent approach
an inversion (out-of-plane interaction) term, contribute to the va- followed by the conjugate gradient method. The dimensions of the
lence energy. The cross-term energy Ecross-term accounts for the en- system were 40 × 40 × 45 Å. The charge distribution was calcu-
ergy induced by the changes in the bond length and the bond angle lated using the QEq method (Rappe and Goddard III 1991). The
with the surrounding atoms. Enon-bonded , the nonbonded term, con- nonbonded summations were calculated using Ewald for electro-
sists of intermolecular and intramolecular interaction. The non- static interaction with an accuracy of 0.001 kcal=mol. Based on
bonded terms include hydrogen bonds (H-bonds) and van der each atom charge with the truncation of atoms further than the
Waals (vdW) interactions that are induced dipole-dipole interac- cut-off distance of 15.5 Å, the charge distribution for van der Waals
tions (also known as London forces). Moreover, the Coulomb interaction was computed (Fig. 3).
interaction accounts for electrostatic interaction. To model the polymer fiber macromolecules, monomers of the
The condensed-phase optimized molecular potentials for chosen polymers (vinyl alcohol, propylene, and hexano-6-lactam)
atomistic simulation studies (COMPASS) (Sun 1998) force field were initially constructed. Chains of PVA, polypropylene, and ny-
was used for this calculation. It is a powerful force field for atom- lon-6 were then created by assembling 100-monomer units of vinyl
istic simulation of condensed materials and one of the first ab alcohol, propylene, and hexano-6-lactam, respectively (Gujrati and
Leonov 2010).
Five possible structures of each prescribed polymer molecule
with the desired density of 1.2 g=cm3 at 298 K were created
and energetically optimized. Each configuration was then subjected
Fig. 2. Schematic of the initial molecular model of C-S-H Fig. 3. Molecular model of C-S-H after optimization
be calculated as
Mechanical Testing
2P
There are different approaches to predict, analyze, and improve the σ¼ ð2Þ
πDL
tensile and compressive strength of cement-based composites
(Misra and Poorsolhjouy 2015). The tensile strength of concrete where P = compression force; D = diameter of the cylinder; and L
is an important parameter in the design of civil engineering struc- = length of cylinder.
tures. To determine the tensile strength of concrete for existing
structures, different experiments are essential. Compression tests
are very common to investigate the compressive strength of con- Results and Discussion
crete and fiber-reinforced concrete samples. Because of the com-
plex nature of uniaxial tension tests, splitting tension tests are
SEM and EDX Interface Characterization
usually carried out on cylindrical specimens or cores. In this study,
a split testing setup was chosen because of the fiber-pullout effect Typical SEM images of the polymer fiber, matrix, and polymer ma-
on the polymeric fibers, which is influenced by the polymers’ trix interfacial transition zones for all three types of polymer fibers
molecular structures. are shown in Fig. 4. The SEM images of the samples clearly show
To compare the mechanical properties of fiber-reinforced ce- there is a transition zone between the polymer fibers and cement
ment paste and investigate the effect of different polymeric fibers paste. These figures show the interfacial layer between fiber and
in these composites and their molecular-level interactions, numer- matrix has a thickness of approximately 1–5 μm around the fiber.
ous samples were prepared. Nine samples each of the control, ce- Regions of fiber, cement matrix, unhydrated cements, and fiber ma-
ment and PVA, cement and nylon-6, and cement and polypropylene trix interface are also indicated in the figure. The EDX results and
batches were prepared. Three samples from each batch were tested analysis for three different fiber cement interfaces are shown in
at 3, 7, and 28 days. The fibers shape was identical. Nineteen- Fig. 5. X-ray mapping is used to investigate the distribution and
mm-long and 175-μm-thick cylindrical polymeric fibers were used. density of existing elements such as Si and Ca in the area around
each fiber. The X-ray mappings of the Si and Ca elements for ny-
lon-6 and PVA fibers clearly show the accumulation of calcium at
30 µm the fiber matrix interface, which is represented by a circle in the
figures. However, this phenomenon is not observed in the samples
with nonpolar polypropylene fibers. Moreover, the EDX is per-
formed on 10 spectra of the fiber matrix interface and 10 of regular
C-S-H gel. The numerical EDX results show that the ratio of
C/S in the interfacial zone significantly changes for PVA and ny-
Fiber/ Matrix lon-6 fibers, whereas it remains almost constant for polypropyl-
Interface ene fibers, shown in Fig. 6. The ratio of C/S in the interfacial
Nylon-6 fiber transition zone between PVA and nylon-6 fibers and cement in-
(a) creases in comparison to the regular C-S-H in fiber-reinforced
50 µm 30 µm composite matrices. Furthermore, this ratio does not change for
polypropylene fibers. This can be due to the polarity of different
fibers caused by the functional groups on their molecular
structure.
Fig. 5. EDX X-ray map results for three polymeric fibers in the C-S-H matrix; the results show that the Ca2þ ions are absorbed on the interface of
nylon-6 and PVA fibers, but the interface of polypropylene fiber has not significantly changed
and 2.64 ðg=cm3 Þ, respectively. The modulus of elasticity is esti- stress is commonly used to relate the computed stress in molecular
mated to be 57.02, 24.52, and 22.20 GPa in the x, y, and z-direc- dynamics to continuum stresses. The Young’s moduli is then com-
tions (shown in Fig. 3), respectively. Two different methods are puted as the average of the three initial stress-strain slopes after
used to calculate the elastic moduli. In the first method, the struc- relaxation. In the second method, Young’s moduli are indirectly
ture is stretched in all three directions to the maximum strain am-
plitude of 0.01 in 10 steps, and virial stresses are computed. Virial
Table 1. Adhesion Energy (J=m2 ) between Polymers and the Relevant
C-S-H Gel
Polymer Adhesion energy
Polypropylene 64.18
PVA 89.91
Nylon-6 95.48
30,000 30,000
25,000 25,000
20,000 20,000
Load (N)
Load (N)
15,000 15,000
10,000 10,000
5,000 5,000
Polypropylene PVA
0 0
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5 0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5
30,000
25,000
20,000
Load (N)
15,000
10,000
5,000
Nylon-6
0
0 1 2 3 4 5
(c) Displacement (mm)
Fig. 9. Load-displacement curve for cement paste with added fibers cured for 3 days
rium; EC-S-H = energy of the surface without the polymer; and fiber detachment or debonding, and fiber pull-out continuation. The
Epolymer = energy of the polymer without the surface, both separated first occurs in the failure of the tensile strength of the cement paste;
in vacuum in equilibrium (Kisin et al. 2007). As shown in the EDX the second is when the stress in the sample exceeds the strength of
analysis, the C/S ratio in the C-S-H gel increases for polymer ce- the interface between the fibers and cement paste, which is affected
ment interfaces with polar fibers. The polarity of polymer fibers by the polarity of the polymers’ functional groups; and the third
enhances the absorption of water molecules and positive ions, such occurs due to the friction between fibers and cement paste, which
as Ca2þ , during hardening of the fiber molecules. Hence, a more is affected by fiber roughness.
realistic adhesion energy for the polymer cement interface in each The failure strength of fiber-reinforced cement paste can be ob-
type of fiber-reinforced concrete can be calculated using the actual tained from testing control samples. A typical load-displacement
C/S ratio in the C-S-H gel. The final interaction energies and the graph of a cement paste sample with no additive and a sample
computed adhesion energies between PVA, polypropylene, and with polymeric fibers is shown in Fig. 8. In the samples with fibers,
nylon-6 and the relevant C-S-H gel model are presented in the last step of failure strongly depends on the roughness of the
35,000
35,000
30,000
30,000
25,000
25,000
Load (N)
Load (N)
20,000
20,000
15,000
15,000
10,000 10,000
5,000 5,000
Polypropylene PVA
0 0
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5 0 1 2 3 4 5
(a) Displacement (mm) (b) Displacement (mm)
35,000
30,000
25,000
Load (N)
20,000
15,000
10,000
5,000
Nylon-6
0
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5
(c) Displacement (mm)
Fig. 10. Load-displacement curve for cement paste with added fibers cured for 7 days
35,000 35,000
30,000 30,000
25,000 25,000
Load (N)
Load (N)
20,000 20,000
15,000 15,000
10,000 10,000
5,000 5,000
PVA
Polypropylene
0 0
Downloaded from ascelibrary.org by 106.222.187.25 on 12/18/19. Copyright ASCE. For personal use only; all rights reserved.
40,000
35,000
30,000
25,000
Load (N)
20,000
15,000
10,000
5,000
Nylon-6
0
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5
(c) Displacement (mm)
Fig. 11. Load-displacement curve for cement paste with added fibers cured for 28 days
fibers at the nano and micro scales. However, there is a correlation is always challenging and quite important. The results of this study
between the second step of failure and the adhesion energies show there is a correlation between the yield load of fiber-
between the polymers and the relevant C-S-H model. The reinforced cement paste and the polarity of functional groups in
force corresponding to this point is called the yield load in the chemical composition of the polymeric fibers.
this study.
Fig. 9 shows the load-displacement behavior of tested samples
on the third day. The circles in the graphs show the second step of
failure. The yield load is higher for cement and nylon-6 samples
than for cement and PVA samples, and it is higher for cement
and PVA samples than for cement and polypropylene samples.
Figs. 10 and 11 show the tested samples at 7 and 28 days. The same
trend for the yield load can be observed in the samples.
The yield loads for samples with PVA and polypropylene are
close; however, the load for samples with nylon-6 fibers is mark-
edly higher. To conclude, the graphs can be summarized in Fig. 12.
These results agree well with the experiments done by Song
et al. (2005).
The implications of these results are quite significant. First, they
show that the polymers’ chemical composition and functional
groups change the structure of C-S-H gel in the vicinity of the pol-
ymers. This result is important first because changing the chemical
structure and C/S ratio of C-S-H gel affects the overall mechanical
response of fiber-reinforced cement paste, and second, the ion
transfers during the hydration reaction may affect the corrosion
in the composite, especially if there are other reinforcing materials
in the composite. The correlation between the atomic-scale and
Fig. 12. Summary of the split-cylinder test results
nanoscale properties with the macroscale properties of materials
that in the polypropylene and C-S-H interface. This is mainly a Lu, Z., and Dunn, M. L. (2010). “Van der Waals adhesion of graphene
result of the high polarity of the hydroxyl and amide functional membranes.” J. Appl. Phys., 107(4), 044301.
groups. To further investigate this observation, atomistic simula- Manzano, H., Dolado, J., and Ayuela, A. (2009). “Elastic properties of the
main species present in portland cement pastes.” Acta Materialia.,
tions are performed to study the structure of C-S-H in the vicinity
57(5), 1666–1674.
of the three different polymers. The adhesion energies between the Manzano, H., Dolado, J., Guerrero, A., and Ayuela, A. (2007). “Mechanical
polymers and the relevant C-S-H model were also found to study properties of crystalline calcium-silicate-hydrates: Comparison with
the strength of polymer cement interfaces. Additionally, split-cyl- cementitious C-S-H gels.” Phys. Status Solidi A., 204(6), 1775–1780.
inder testing was performed on samples and a correlation between McNaught, A. D., and McNaught, A. D. (1997). Compendium of chemical
the polarity of the polymers’ functional groups (adhesion energies) terminology, Royal Society of Chemistry, Cambridge, U.K.
and load-displacement curves were found. This result was used Misra, A., and Poorsolhjouy, P. (2015). “Granular micromechanics model
to investigate the macroscopic behavior of samples and was com- for damage and plasticity of cementitious materials based upon
pared to the molecular interaction between C-S-H and polymers. thermomechanics.” Math. Mech. Solids., 227(5), 1393–1413.
The properties of the interface between fibers and cement paste Pellenq, R. J. M., et al. (2009). “A realistic molecular model of cement
can be correlated with the fiber detachment in load-displacement hydrates.” Proc. Natl. Acad. Sci., 106(38), 16102–16107.
curves. The implications of these results are significant in further Pellenq, R. M., Lequeux, N., and Van Damme, H. (2008). “Engineering the
bonding scheme in CSH: The iono-covalent framework.” Cem. Concr.
development of robust fiber-reinforced cement composites.
Res., 38(2), 159–174.
Qomi, M. A., et al. (2014). “Combinatorial molecular optimization of ce-
ment hydrates.” Nat. Commun., 5, 4960.
References
Rappe, A. K., and Goddard, W. A., III (1991). “Charge equilibration for
Abbasnia, R., Hosseinpour, F., Rostamian, M., and Ziaadiny, H. (2013). molecular dynamics simulations.” J. Phys. Chem., 95(8), 3358–3363.
“Cyclic and monotonic behavior of FRP confined concrete rectangular Richardson, I. (2004). “Tobermorite/jennite-and tobermorite/calcium
prisms with different aspect ratios.” Constr. Build. Mater., 40, 118–125. hydroxide-based models for the structure of CSH: Applicability to hard-
Akihama, S., Suenaga, T., Nakagawa, H., and Suzuki, K. (1985). “Exper- ened pastes of tricalcium silicate, -dicalcium silicate, portland cement,
imental study of vinylon fiber reinforced cement composites: VFRC and blends of portland cement with blast-furnace slag, metakaolin, or
(Part I).” Kajima Annu. Tech. Rep., 33, 123–128. silica fume.” Cem. Concr. Res., 34(9), 1733–1777.
Allen, A. J., Thomas, J. J., and Jennings, H. M. (2007). “Composition and Richardson, I. (2008). “The calcium silicate hydrates.” Cem. Concr. Res.,
density of nanoscale calciumsilicatehydrate in cement.” Nat. Mater., 38(2), 137–158.
6(4), 311–316. Sakulich, A. R., Li, V. C. (2011). “Nanoscale characterization of engi-
Balaguru, P. N., and Shah, S. P. (1992). Fiber-reinforced cement neered cementitious composites (ECC).” Cem. Concr. Res., 41(2),
composites, McGraw-Hill, New York. 169–175.
Bauchy, M., Qomi, M. A., Ulm, F. J., and Pellenq, R. M. (2014). “Order Salahshoor, H., and Rahbar, N. (2012). “Nano-scale fracture toughness and
and disorder in calciumsilicatehydrate.” J. Chem. Phys., 140(21), behavior of graphene/epoxy interface.” J. Appl. Phys., 112(2), 023510.
214503. Shahsavari, R., Pellenq, R. J. M., and Ulm, F. J. (2011). “Empirical force
Bentur, A., and Mindess, S. (2006). Fibre reinforced cementitious fields for complex hydrated calcio-silicate layered materials.” PCCP,
composites, Taylor & Francis, New York. 13(3), 1002–1011.
Betterman, L., Ouyang, C., and Shah, S. (1995). “Fiber-matrix interaction Shalchy, F., and Rahbar, N. (2015a). “Nanostructural characteristics and
in microfiber-reinforced mortar.” Adv. Cem. Based Mater., 2(2), 53–61. interfacial properties of polymer fibers in cement matrix.” ACS Appl.
Constantinides, G., and Ulm, F. J. (2004). “The effect of two types of CSH Mater. Interfaces, 7(31), 17278–17286.
on the elasticity of cement-based materials: Results from nanoindenta- Shalchy, F., and Rahbar, N. (2015b). “Nanostructure of cement/polymer
tion and micromechanical modeling.” Cem. Concr. Res., 34(1), 67–80. fiber interfaces.” Proc., 10th Int. Conf. on Mechanics and Physics of
Dolado, J. S., Griebel, M., and Hamaekers, J. (2007). “A molecular Creep, Shrinkage, and Durability of Concrete and Concrete Structures,
dynamic study of cementitious calcium silicate hydrate (CSH) gels.” ASCE, Reston, VA, 872–876.
J. Am. Ceram. Soc., 90(12), 3938–3942. Song, P. S., Hwang, S., and Sheu, B. C. (2005). “Strength properties of
Garcia, S., Naaman, A. E., and Pera, J. (1997). “Experimental investigation nylon- and polypropylene-fiber-reinforced concretes.” Cem. Concr.
on the potential use of poly (vinyl alcohol) short fibers in fiber- Res., 35(8), 1546–1550.
reinforced cement-based composites.” Mater. Struct., 30(1), 43–52. Soyer-Uzun, S., Chae, S. R., Benmore, C. J., Wenk, H. R., and Monteiro,
Grujicic, M., Sun, Y. P., and Koudela, K. (2007). “The effect of covalent P. J. (2012). “Compositional evolution of calcium silicate hydrate
functionalization of carbon nanotube reinforcements on the atomic- (CSH) structures by total X-ray scattering.” J. Am. Ceram. Soc., 95(2),
level mechanical properties of poly-vinyl-ester-epoxy.” Appl. Surf. 793–798.
Sci., 253(6), 3009–3021. Sun, H. (1998). “COMPASS: An ab initio force-field optimized for
Gujrati, P. D., and Leonov, A. I. (2010). Modeling and simulation in condensed-phase applications overview with details on alkane and
polymers, Wiley, Germany. benzene compounds.” J. Phys. Chem. B., 102(38), 7338–7364.