molecular dynamics for the prediction ... - rc.library.uta.edu
TRANSCRIPT
Molecular Dynamics for the Prediction
of the Interfacial Shear Stress and
Interface Dielectric Properties of
Carbon Fiber Epoxy Composites
RAJNI CHAHAL, ASHFAQ ADNAN, KENNETH REIFSNIDER,
RASSEL RAIHAN, YUAN TING WU, VAMSEE VADLAMUDI
and MUTHU RAM PRABHU ELENCHEZHIAN
Proceedings of the American Society for Composites—Thirty-third Technical Conference
ABSTRACT
The thermoset epoxy resin Diglycidyl ether of Bisphenol F (EPON 862),
crosslinked with the Diethylene Toluene Diamine (DETDA) hardening agent, are
utilized as the polymer matrix component in many graphite (carbon fiber)
composites. Since it is difficult to experimentally characterize the interfacial region,
computational molecular modeling is a necessary tool for understanding the
influence of the interfacial molecular structure on bulk-level material properties. The purpose of this research is to evaluate and compare the interfacial shear stress
and dipole moment for the pristine carbon fiber composite and the one with the moisture
content at the interface. Molecular models are established for Carbon fiber reinforced
EPON 862-DETDA polymer with and without the moisture content at the interface.
Interatomic interactions are defined by Reactive Force Field (ReaxFF). Material
characteristics such as polymer mass-density and dipole moment are investigated near the
polymer/fiber interface. It is determined that a region exists near the carbon fiber surface in
which the polymer mass density and dipole moment are different than that of the bulk
values. It can further be seen that material having larger values of dipole moment in
interface region have comparatively lesser values of interfacial shear stress.
_____________
1Rajni Chahal, PhD Student, Department of Mechanical & Aerospace Engineering, The University of Texas at Arlington, TX- 76019, U.S.A. 2Ashfaq Adnan, Associate Professor, The University of Texas at Arlington, TX- 76019, U.S.A. 2Kenneth Reifsnider, Director of the Institute for Predictive Performance Methodologies at University of Texas at Arlington Research Institute, TX – 76118, U.S.A. 2Rassel Raihan, Research Engineer at University of Texas at Arlington Research Institute, TX – 76118, U.S.A. 3Yuan Ting Wu Post-Doctoral Student, Department of Mechanical & Aerospace Engineering, The University of Texas at Arlington, TX- 76019, U.S.A. 4Vamsee Vadlamudi, PhD Candidate, Department of Mechanical & Aerospace Engineering, The University of Texas at Arlington, TX- 76019, U.S.A. 4Muthu Ram Prabhu Elenchezhian, PhD Student, Department of Mechanical & Aerospace
Engineering, The University of Texas at Arlington, TX- 76019, U.S.A.
INTRODUCTION
Composites are heterogeneous material systems which are generally made up of
multiple phases (i.e., fiber, defects, matrix, and voids in structural composites)
which have different conductivities. Since, they are substantially light in weight and
deliver other desirable properties, i.e., corrosion and wear resistance, as compared
to metals, ceramics or other materials, they are being used in a variety of different
sectors, especially the aerospace industry. Thermoset epoxy/carbon fiber composites have long been one of the primary
material choices for modern aerospace applications. These materials exhibit an
excellent strength to weight ratio and can be easily manufactured to have specific
performance characteristics.
The aerospace industry certification approach follows a strict fail-safe
philosophy which sometimes restricts the use of many technological breakthroughs.
Reifsnider et al. [1] showed the evidence that dielectric properties can extract
material state information and can be utilized to predict the performance of these
complex material systems. In the current research, it is aimed to observe and
establish similar relation on molecular level. Dielectric investigation of material can get information at different level, i.e.,
polarization mechanisms such as ionic (molecular), dipolar (orientational), electronic,
interfacial (Maxwell–Wagner–Sillars) polarization and hopping charge polarization [2].
The following Figure 1 shows the different types of polarization and their effect on the
dielectric response and its corresponding effective frequency range.
Figure 1. Dielectric response of material constituents at
broad band frequency ranges.
Figure 2. Through thickness measurement of dielectric response for off-
axis loading of woven glass epoxy coupons.
Using Broadband Dielectric Spectroscopy (BbDS), Raihan et al. studied the
dielectric response of woven cross ply glass reinforced epoxy composite measured
through the thickness of the coupons and those results were compared with all other
variables. [3] An example of those data is shown in Figure 2, along with a general
interpretation of the relationship of the micro-details with the dielectric
measurements. The changes in dielectric response are not monotonic; they show
unique changes in magnitude and direction depending on the changes in the
mechanisms of damage. It was observed over a wide range of composite materials
and applied conditions (mechanical, thermal, electrical, chemical) that the through
thickness dielectric response is directly and uniquely related to the internal
microstructure and changes in local morphology Nandini et al. studied the dielectric responses of degraded and non-degraded
composite prepreg materials. The partial crosslinking was observed which showed
the result of decrease in the dielectric parameters. [4] Also, from the static tensile
test it was proved that the mechanical strength was lesser for the degraded sample
and the dielectric relaxation strength was higher as shown in Figure 3. Dielectric
relaxation Strength (DRS) is the difference of the real part of the dielectric property
between lower and high frequency.
Figure 3. Dielectric Response vs. Mechanical Strength of
degraded and non-degraded composites.
Polar molecules are contained inside of heterogeneous materials and may have
interfaces with different electrical properties. In the presence of an applied electric
field it will polarize the material by orienting the dipole moments of polar
molecules and charge accumulation at the interfaces of dissimilar materials. Molecular Dynamics is a computational tool for understanding the interfacial
behavior of heterogeneous materials such as CFRP composites. Chunyu Li et al. [5]
used COMPASS force-field to observe the interface of crosslinked epoxy in the
presence of carbon fiber and calculated it’s mechanical. Hadden [6] used OPLS
United Atom force-field developed by Jorgensen and co-workers to study the
tensile and shear properties of Carbon Fiber reinforced epoxy composites.
Whereas traditional force fields are unable to model chemical reactions because
of the requirement of breaking and forming bonds (a force field's functional form
depends on having all bonds defined explicitly), Reax Force Field (ReaxFF)
eschews explicit bonds in favor of bond orders, which allows for continuous bond
formation/breaking. In the ReaxFF, the potential energy is defined as a function of
bond order with energy penalties for nonequilibrium configurations. The ReaxFF
was initially developed to model bond dissociation and formation in carbon-based
materials [7]. Odegard et al. [8] proved ReaxFF to be working for bulk epoxy
system using parameters obtained by Liu et al. [9].
An important part of ReaxFF is the charge equilibration procedure. Charge
equilibration (QEq) procedure [10] approximates the partial charges on atoms by
minimizing the electrostatic energy of the system. Charge equilibration is
mathematically formulated as the solution of a large sparse linear system of
equations. This solve needs to be performed accurately at each time-step because it
significantly impacts forces and total energy of the system. Since partial charges on
atoms are fixed in conventional MD, this is not a consideration for conventional
methods. Chakrabarty et al. [11] studied the piezoelectric properties of carbon
nanotube polymer nanocomposites using Dreiding Force Field. In this paper, we have used ReaxFF force-field to study the interfacial and
dielectric properties of Carbon Fiber/EPON 862-DETDA nanocomposite system.
Molecular dynamics study is performed to evaluate interfacial shear stress (ISS) and
dipole moment for carbon fiber composites. Mechanical and electrical properties,
such as interfacial shear strength and dipole moment have been calculated and
compared for both pristine carbon fiber composite systems and systems having
moisture content at their interfaces.
MOLECULAR STRUCTURE OF CARBON FIBER/EPOXY
SYSTEM Carbon Fiber
At the molecular level carbon fibers consist of stacked layers of graphene. This
is a two-dimensional layer with carbon-carbon sp2 hybridized covalent bonds that
result in outstanding mechanical properties [15].
Figure 4. Predicted internal structure for carbon fibers, surface atom structures, and
aromatic carbon ring structure. Image courtesy of NASA Report 4084 © L.T.
Drzal [15].
For the creation of the graphite surface, a program was written to create sheets of
carbon based on the aromatic pattern for which graphite surfaces are well known. The
simulated graphite surface was constructed from 3 sheets of stacked graphene, each
sheet containing 448 carbon atoms for a total of 1344 atoms. The graphene sheets were
oriented along the x-y plane, with periodic boundary conditions in the x and y-
direction, and had an interlayer spacing of 3.35Å. The graphite structures were relaxed
using a series of MM and MD simulations in LAMMPS [16]. While equilibrating the
graphite, the z-direction box coordinate was chosen to implement interlayer spacing for
periodic boundary conditions. Thus, the top surface was influenced by the bottom
surface and visa-versa, representing many layers of bulk graphite. Initial graphite
structures are shown in Figure 4 as visualized in Ovito [17].
Figure 5. Carbon Fiber.
After equilibration, Graphite density is 2.2 g/cc, while the reported density =
2.09– 2.23 g/cm³.
Epoxy
The modeled epoxy system was composed of 64 EPON 862 (Diglycidyl ether of
Bisphenol F) crosslinked with 32 DETDA (Diethylene Toluene Diamine) molecules [18]. Figure 4 shows the molecular structure of EPON 862 monomer & DETDA
before crosslinking. During the crosslink process, each amine group in DETDA can
react with two epoxide groups in EPON 862. Therefore, a molar ratio of 2:1 of
EPON 862 to DETDA molecules is necessary for a stoichiometric mixture.
Figure 6. Molecular Structures of EPON 862 and DETDA monomers.
64 EPON-862 & 32 DETDA molecules are put in a larger box and is minimized
and equilibrated using ReaxFF with the parameterizations of Liu et al. [9]. The initial
density of the system was 0.3 g/cc. The model was further allowed to shrink under NPT
ensemble for nearly 150 ps at 300 K, using 0.1 fs of timestep. After this step, the final
density of epoxy system was 1.12 g/cc. To release the residual stresses in the system,
equilibration is performed in NVT ensemble for 50 ps at the same timestep. Average
residual stress after this step is close to zero. Figure 7 shows typical molecular structure
of equilibrated and crosslinked EPON-DETDA molecules.
Figure 7. Molecular structure of equilibrated and densified Epoxy with ReaxFF. Coloring scheme: CPK.
CARBON FIBER/EPOXY SYSTEM:
Without Water Content
The nano-scale material system in this study is composed of Carbon fiber and
EPON 862-DETDA matrix. Specifically, a Carbon fiber is embedded in polymer
matrix. Previously equilibrated and compressed small epoxy blocks are put above and
below the Carbon fiber. The size of Carbon fiber was previously chosen such that its
length and width was approximately equal to the size of two epoxy blocks put together
in x and y direction. Therefore, final model is having 8 epoxy blocks each above and
below the fiber. This model contains 8784 atoms including 1344-C-atom Carbon fiber,
128 EPON 862 & 64 DETDA molecules. The dimensions of the system are ~ 3.5 x 3.5
x 8 nm. Average density of the system is ~ 1.18 g/cc after equilibration.
Figure 8. Molecular model of Carbon fiber inside epoxy matrix. Coloring scheme: CPK.
In the molecular dynamics, the entire material system is represented with the
ReaxFF potential. The system is replicated across periodic boundaries in each of the
3-dimensions, thereby making Carbon fiber & polymer chains infinitely long.
Before equilibration, a minimization is performed to find configuration that will
hopefully be in local potential energy minimum, and the new atom positions to be
computed. This model is further equilibrated in NVT ensemble for 70 ps at a
timestep of 0.1 fs. Figure 8 shows the final CF/epoxy model after minimization and
equilibration, which is ready for pull-out simulations.
With Water Content
Epoxy has been believed to absorb moisture from the atmosphere by diffusion
[14, 19, 20]. The absorbed moisture eventually finds its way to the interface and is
available for hydration of the interface. In this study, 10 water molecules (0.531% w/w) are included at both the interface
regions of Carbon fiber and epoxy. The final model contains 8844 atoms including
1344-C-atom Carbon fiber, 128 EPON 862 & 64 DETDA molecules. The dimensions
of the system are ~ 3.5 x 3.5 x 8 nm. Average density of the system is ~ 1.181 g/cc.
Figure 9. Molecular model of Carbon fiber inside contaminated epoxy matrix. Coloring scheme: CPK.
This model is further minimized and equilibrated for 100 ps at a timestep of 0.1 fs. Figure 9 shows the final CNT/epoxy model after minimization and equilibration.
RESULTS AND COMPARISONS
Interfacial Shear Stress Calculation
PULLOUT SIMULATIONS OF PRISTINE & CONTAMINATED CARBON
FIBER REINFORCED COMPOSITE
The simulation cell with periodic boundary condition in y-z plane as shown in
Figure 8(a) is composed of a fragment of Carbon fiber totally embedded inside
EPON-DETDA matrix. To capture the complete pull-out of Carbon fiber from
epoxy matrix, periodic condition has been removed in the pullout-direction (i.e. x
direction). Size of simulation cell in pullout-direction of Carbon fiber is chosen to
be slightly larger than the Carbon fiber length. This prevents Carbon fiber from
going out of the simulation cell after it has been completely pulled out of epoxy
matrix. Box dimensions are 8 x 3.5 x 8 nm.
Figure 10. Snapshots from the MD Simulation of Carbon Fiber Pullout from Pristine CFRP.
The pull-out simulations of Carbon fiber from epoxy matrix are carried out by
applying a constant velocity of 0.01 Å/fs to approximately 3 Å layer of carbon atoms at
the pullout side of Carbon fiber. Nearly 3 Å right and left part of epoxy matrix is held
fixed during the pullout. The simulation is run for ~ 40,000 MD time steps of 0.1 fs.
Figure 11. Snapshots from the MD Simulation of Carbon Fiber Pullout from Contaminated CFRP.
Interfacial Shear Stress Calculation
In the carbon fiber reinforced composites, the bonding strength between the
Carbon fiber and epoxy resin can be evaluated by interfacial bonding energy. The
nature of interfacial bonding energy primarily comes from the electrostatic and van
der Waals forces in the molecular system. The initial and final total energy of the pullout simulation for pristine CFRP and
contaminated CFRP are listed in Table I. The potential energy of both the CFRP system
increased as the Carbon fiber was pulled out of the epoxy. In the pullout simulation, the
entire Carbon fiber and epoxy resin were not held fixed. The potential energy of the Carbon
fiber and epoxy resin increased due to the changes of their configurations during the
pullout. The deformation of the Carbon fiber and epoxy resin during the pullout has
influence on the pullout energy [21]. After the pullout, the difference in final and
initial potential energies was noted for both the cases. Gou et al. [21] defined the pullout energy as:
(1)
In the same reference, Gou et al. defined the relation between pullout energy
and interfacial shear strength for the case when fiber has circular cross-section. This
can be modified to relate pullout energy and interfacial shear strength for
rectangular fiber cross-section:
2
(2)
(3)
Where:
L = Length of the Carbon Fiber in pullout-
direction b = Width of the Carbon Fiber
TABLE I. EFFECT OF MOISTURE CONTENT ON ISS VALUE OF CFRP.
Model EInitial
Interfacial Shear
EFinal EPullout Stress, ISS
(kcal/mol) (kcal/mol) (kcal/mol) (Mpa)
CFRP 1118328.3 1119250.5 922.2 188
(Pristine) CFRP
(Moisture Content = 1124884.9 1125453.6 568.7 116
0.53 % w/w)
From equation (1) & (2), the interfacial shear strength between the carbon fiber
and epoxy was calculated to be about 188 MPa & 116 MPa for pristine &
contaminated CFRP, respectively.
Dipole Moment Calculation
In the electrically heterogeneous material, dispersion occurs in the bulk properties
from the charging of the interfaces within the material. This phenomenon does not arise
from dielectric relaxation in the bulk phases of the material, but it is a consequence of
the boundary conditions on the field at the interfaces between phases. The dielectric behavior is associated with the non-uniform distribution of free
electronic charges across the interface between the dissimilar dielectric materials
under the influence of electric field. Carbon fiber and epoxy have their own free
charge carries concentration and associated charge carries mobility. To achieve
current continuity through materials 1 and 2 there will have to be a charge carrier
concentration discontinuity across the interface. This interfacial charge build-up or
polarization, as function of frequency gives rise to the dielectric dispersion
exhibited by inhomogeneous system [22].
In physics, the electric dipole moment is a measure of the separation of positive
and negative electrical charges within a system, that is, a measure of the system's
overall polarity [23].
Figure 12 shows pristine and contaminated CFRP under the influence of
Positive electric field of magnitude 0.2 eV.
(a) (b) Figure 12. (a) Pristine CFRP Under Positive Electric Field, (b)
Contaminated CFRP Under Positive Electric Field.
After the application of electric field, compute dipole/chunk command has been
used to calculate the dipole moment for the specified chunk thickness along z-direction.
This compute calculates the x, y, z coordinates of the dipole vector and the total dipole
moment for each chunk, which includes all effects due to atoms passing through periodic
boundaries. For chunks with a net charge the resulting dipole is made position independent
by subtracting the position vector of the center of mass or geometric center times the net
charge from the computed dipole vector. Figure 13 & 14 show the variation of average total
dipole moment calculated over a slice (in x-y plane) of thickness = 0.5
Å for the pristine & contaminated composite system, respectively.
Figure 13. Average Dipole Moment of Pristine CFRP along simulation box Z-axis.
Figure 14. Average Dipole Moment of Contaminated CFRP
along simulation box Z-axis.
Further, to calculate average interfacial dipole moment, dipole moment values
have been averaged for the interface regions in Figure 13 & 14. The calculated values for the average interfacial dipole moment are reported in
Table II below.
TABLE II. EFFECT OF DIRECTION OF ELECTRIC FIELD (EZ = ± 0.2 V/Å) ON
AVERAGE DIPOLE MOMENT OF PRISTINE AND CONTAMINATED CFRP.
Average Dipole Moment Average Dipole Moment
Model (charge*Å) (charge*Å)
Ez = 0.2 V/ Å Ez = -0.2 V/ Å
CFRP 5.99 6.07
(Pristine)
CFRP 10.8 10.9 (Moisture Content = 0.531 % (80.3 % increase) (79.6 % increase)
w/w)
Table II shows that percentage increase in the average values for dipole moment
for pristine CFRP and the moisture absorbed CFRP also seems to be independent of
the direction of electric field (Ez = ± 0.2 V/Å). It is nearly 80 % in both the cases.
TABLE III. COMPARISON OF ISS AND AVERAGE DIPOLE MOMENT FOR
PRISTINE AND CONTAMINATED CFRP.
ISS Average Dipole Moment
Model (MPa) (charge*Å)
@Ez = 0.2 V/ Å
CFRP 188 5.99
(Pristine)
CFRP 116 10.8 (Moisture Content = 0.531 % (38.30 % decrease) (80.3 % increase)
w/w)
We know that Polarization Density is directly proportional to Dipole Moment.
Thus, Dipole Moment, hence polarization, increases if there are external impurities
(here water) are present at the interface.
CONCLUSIONS
Reactive force field is used to evaluate the mechanical (Interfacial Shear Stress)
and electrical (Dipole Moment) properties of Pristine and Contaminated carbon
fiber reinforced epoxy composite. As calculated from pullout simulations, the ISS
value for Pristine CFRP is found to be 188 MPa, while the ISS value is lowered to
116 MPa when 0.531 % w/w moisture content was present in the system. Further,
average dipole moment is evaluated at the interface region of Pristine and
Contaminated CFRP is 5.99 charge x Å and 10.8 charge x Å, respectively.
It can be observed that for pristine composite system, value of dipole moment is
lower than that for composite with water content at the interfaces, while it has better
Interfacial Shear Stress than latter. It seems to be consistent with the experimental
results obtained by Reifsnider et al. [1]. Therefore, lesser the value for interfacial
dipole moment, hence interfacial polarization, better are the mechanical properties
(interfacial shear stress) of the composite material. As predicted by Tan et al. [11] in certain cases, the effect of Efield could not be
completely observed due to the charge equilibration (QEq) [10, 12] method used in
the simulations. With the further development of this method, such as the
polarizable charge equilibration method (PQEq), the polarization can be better
described to reveal more details of the effects of Efields in molecular simulations.
REFERNCES
1. Reifsnider, K. L., Fazzino, P., Majumdar, P. K., & Xing, L. 2009. “Material state changes
as a basis for prognosis in aeronautical structures,” Aeronautical Journal, 113(1150), 789-
798 (Silver Award, 2010, Royal Aeronautical Society). 2. Kremer F., Schoenhals A., Broadband Dielectric Spectroscopy, in Springer-Verlag, 2002.
3. Raihan, R., Adkins, J. M., Baker, J., Rabbi, F., & Reifsnider, K. "Relationship of dielectric
property change to composite material state degradation." Composites Science and
Technology 105 (2014): 160-165.
4. Nandini, A., Shute, N., Elenchezhian, M.R.P., Vadlamudi, V., Raihan, R., Reifsnider, K.,
“Dielectric property investigation of degraded pre-preg and performance prediction of the
final composite part” Proceedings of SAMPE 2018, Long Beach CA May 21-24 2018.
5. Li, C., et al. 2012. “Atomistic simulations on multilayer graphene reinforced epoxy
composites,” Composites Part A: Applied Science and Manufacturing. 43(8): p. 1293-1300. 6. Hadden, C. 2015. "Molecular Modeling of Epon 862-Detda / Carbon Composites,"
Dissertation, Michigan Technological University. 7. van Duin, A.C.T., Dasgupta, S., Lorant, F., and Goddard, W.A. 2001. “ReaxFF: A reactive force
field for hydrocarbons,” Journal of Physical Chemistry A, 105(41): 9396-9409.
8. Odegard, G.M., Jensen, B.D. 2014. “Predicting mechanical response of crosslinked epoxy
using ReaxFF,” Chemical Physics Letters, Volume 591, Pages 175-178. 9. Liu, L.C., Liu, Y., Zybin, S.V., Sun, H., and Goddard, W. A. 2011. “ReaxFF-/g: Correction of
the ReaxFF Reactive Force Field for London Dispersion, with Applications to the Equations of
State for Energetic Materials,” Journal of Physical Chemistry A. 115(40): 11016-11022. 10. Rappé, A. K. & Goddard, W. A. 1991. “Charge Equilibration for Molecular Dynamics
Simulations,” J. Phys. Chem. 95, 3358–3363. 11. Tan, S., Xia, T., Shi, Y., Pfaendtner, J., Zhao, S., & He, Y. 2017. “Enhancing the Oxidation
of Toluene with External Electric Fields: a Reactive Molecular Dynamics Study,”
www.nature.com/scientificreports. 12. Aktulga, H. M., Fogarty, J. C., Pandit, S. A. & Grama, A. Y. 2012. “Parallel reactive molecular
dynamics: Numerical methods and algorithmic techniques,” Parallel Comput. 38, 245–259.
13. Chakrabarty, A. 2008. “Carbon Nanotube Polymer Nanocomposites for Electromechanical
System Applications,” Doctoral dissertation, Texas A&M University. 14. Frankland, S.J.V. & Hinkley, J.A. 2008. “Molecular Dynamics Simulations of Adhesion at
Epoxy Interfaces,” NASA Technical Report. 15. Drzal, L. T., Bascom, W. D. 1987. “The surface properties of carbon fibers and their
adhesion to organic polymers,” Report/Patent Number: NASA-CR-4084, NAS 1.26:4084. 16. Plimpton, S. 1995. “Fast Parallel Algorithms for Short-Range Molecular Dynamics,” J
Comp Phys, 117, 1-19.
17. Stukowski, A. 2010. “Modelling Simul. Mater. Sci. Eng.,” 18, 015012.
18. Komuves, F., Kelkar, D. A., Mohan, R., and Kelkar, A. V. 2011. "Prediction of Mechanical
Properties of EPON 862 (DGEBF) cross-linked with Curing Agent W (DETDA) and
SWCNT using MD Simulations," in 52nd AIAA/ASME/ASCE/AHS/ASC Structures,
Structural Dynamics and Materials Conference, Denver. 19. El-Sa'ad, L., Darby, M. I. & Yates, B. 1990. “Moisture absorption by epoxy resins: the
reverse thermal effect,” Journal of Material Science 25, 3577-3582. 20. De'Nève, B. & Shanahan, M.E.R. 1993. “Water absorption by an epoxy resin and its effect on
the mechanical properties and infrared spectra,” Polymer, Volume 34, Issue 24, Pages 175-178.
21. Gou, J., Minaie, B., Wang, B., Liang Z., and Zhang, C. 2005. "Computational and
experimental study of interfacial bonding of single-walled nanotube reinforced
composites," Computational Materials Science, vol. 31, pp. 225-236.
22. Raju, G.G. 2003. “Dielectrics in Electric Field,” New York: Marcel Dekker.