Articulo 12
Articulo 12
Articulo 12
a r t i c l e i n f o a b s t r a c t
Article history: The micro-capsules used for drug delivery are fabricated using polylactic acid (PLA), which
Received 28 March 2017 is a biomedical material approved by the FDA. A coarse-grained model of long-chain PLA
Revised 10 October 2017 was built, and molecular dynamics (MD) simulations of the model were performed using a
Accepted 13 October 2017 MARTINI force field. Based on the nonlocal theory, the formula for the initial elastic modulus
Available online 25 October 2017 of polymers considering the nonlocal effect was derived, and the scaling law of internal
characteristic length of polymers was proposed, which was used to adjust the cut-off radius
Keywords: in the MD simulations of PLA. The results show that the elastic modulus should be computed
Polylactic acid using nonlinear regression. The nonlocal effect has a certain influence on the simulation
Molecular dynamics simulation results of PLA. According to the scaling law, the cut-off radius was determined and applied
Nonlocal theory to the MD simulations, the results of which reflect the influence of the molecular weight
Scaling law change on the elastic moduli of PLA, and are in agreement with the experimental outcome.
Cut-off radius © 2017 Published by Elsevier Ltd on behalf of Chinese Society of Theoretical and Applied
Mechanics.
R
Project supported by the National Natural Science Foundation of China (no. 11272360), the Natural Science Foundation of Guangdong
Province (no. 2014A030313793), the Special Program for Applied Research on Super Computation of the NSFC-Guangdong Joint Fund (the
second phase), and National Supercomputer Center in Guangzhou.
∗
Corresponding author.
E-mail address: [email protected] (T. Xiao).
https://doi.org/10.1016/j.camss.2017.10.003
0894-9166/© 2017 Published by Elsevier Ltd on behalf of Chinese Society of Theoretical and Applied Mechanics.
acta mechanica solida sinica 30 (2017) 630–637 631
carrier drugs. Therefore, the complexity has increased con- system. A standard MARTINI force field would include four
siderably in terms of the structural design and fabrication of types of basic particles: polar (P), nonpolar (N), apolar (C), and
the carrier capsules; wherein if the potential designs are fab- charged (Q). These four types of particles are further divided
ricated one by one for testing, there would be an impractical into eighteen basic particles of P1–P5, C1–C5, Na, Nd, Nda, N0,
amount of time and fiscal investment required to select a fi- and so on. Normally all the diameters of the particles are fixed
nal design. In comparison, computational simulation meth- at 0.47 nm. The movement of particles is controlled by the
ods have the advantages of low cost, high computation speed, strength of interaction of 10 energy levels.
and the ease of manipulating structure and material param- In molecular simulations, the nonlocal effect caused by
eters, which makes it a widely adapted means in biological long-range forces should be taken into consideration. By ap-
membrane research [3]. Computational molecular simulation plying the cut-off radius, the computation time can be re-
is capable of being introduced as a tool for preliminary re- duced on the premise of certain computational accuracy be-
search in the process of identifying a feasible material and ing assured for simulations of rather large molecular systems.
its available structures for the micro-capsule design. It is also Based on nonlocal theory, an attempt has been made to deter-
worth noticing that the elastic moduli of micro-capsule ma- mine the value of cut-off radius required for simulations of the
terials usually need to be obtained after comparing the data same polymer at different molecular weights, and a formula
from the theoretical model with those from the multi-body of elastic modulus was derived for the nonlinear analysis of
experiments [4, 5], because they are difficult to be measured simulation results. The nonlocal theory [17], which was de-
directly through indentation tests on an individual capsule. veloped by Eringen from the early nonlocal integration theory
Fortunately, the elastic moduli as well as phase transition in 1980s, is a set of stress gradient methods simply containing
temperatures of micro-capsule materials can also be evalu- one scaling parameter. The essence of this theory is that at the
ated using computational simulations. microscale, the stress at a point in material will be affected by
To fabricate the micro-capsules, the commonly selected the stress state of other neighboring points. In 1990s, Aifan-
material was initially albumin, but was later replaced by li- tis rewrote the scaling parameter in the linear elastic form
posome which is more bio-compatible. However, the limited of Laplacian operator with a strain gradient [18], and then in
stability of liposome makes the capsules prone to bursting, so 2010 Askes and Gitman exhibited the possibility of combining
an increasing number of bio-compatible polymers are applied the two above-mentioned methods [19]. The historical devel-
to the fabrication of micro-capsules. Among these polymers, opment of these methods were reviewed in reference [20]. As
polylactic acid (PLA) is non-toxic, degradable with products a continuum theory concerning scaling effects, the nonlocal
that have no adverse effects on the human body, and is a fre- theory has been widely used to study problems such as wave
quently used material for drug carriers and in tissue engineer- propagation, dislocation and crack singularity. It has been in-
ing [6]. It is possible to modify PLA by attaching PEG and chain troduced into micro and nano structural mechanics in 2003
segments of the targeting molecule, or other biological macro- and used to solve the problems of beams [21], bars [22], plates
molecules in various proportions to it, which would adjust the [23], cylindrical shells (for example, carbon nanotubes) [24],
physical properties of the material [7, 8] for the fabrication of and spherical shells [25] at microscale.
medical scaffold or drug carriers. There exists considerable In this work, two macromolecular systems of PLA at the
experimental research on the physical properties of PLA [9], molecular weights of 50 k and 100 k were first built using the
and concurrently, some full-atomic molecular dynamics (MD) coarse-grained method. The PLA system was almost fully an-
simulations have been performed for a qualitative analysis on nealed and equilibrated through cooling simulation and then
the physical properties of small molecule PLA-PEG [10]. Since the curves of self-diffusion coefficient and density of material
polymers with greater molecular weight tend to exhibit higher versus temperature were acquired. After that, the two PLA sys-
strength, greater elastic modulus and better stability, the poly- tems at different molecular weights were separately stretched
mer materials of larger molecular weight are often adopted by MD simulations, and the resulting stress–strain curves were
for the design and fabrication of the carrier capsules. How- acquired and linearly regressed. Meanwhile, the scaling law
ever, the computational cost is too immense when full-atomic for describing the change of the internal characteristic length
MD simulation is performed on polymers with large molecular and the formula for computing the polymer elastic modulus
weight, thus a coarse-grained MD simulation method should were derived based on the nonlocal theory. Finally, the cut-off
be used to simplify the computation. radius was adjusted according to the change of internal char-
The MARTINI force field [11], proposed in 2004 by Marrink acteristic length and the simulations were rerun. The stress–
et al., is a coarse-grained force field initially used to study the strain curves were nonlinearly fitted and compared with the
system behavior of surface-active agents. Given the advantage experimental results.
of having open source modules, many researchers were able to
use and gradually improve upon this force field based on their
specific research topics, which greatly expanded the applica- 2. Force field parameterization and
tion of the force field to biomolecules such as proteins and car- pretreatment
bohydrates [12–15], and furthermore to polyester, polyolefin
and other artificially synthesized polymers [16]. The funda- An isotactic PLA molecule of 10 repetitive units was built in
mental concept of the MARTINI force field is to treat four or Materials Studio (MS). The single PLA molecule was simu-
more non-hydrogen atoms as a coarse-grained particle, and lated through molecular mechanics, and then the centroid
then redefine the non-bonding parameters for the interac- distance between two PLA units, 0.434 nm, was adopted as
tion between the particles to simplify the calculation of the the bond length of coarse-grained PLA particles. The elastic
632 acta mechanica solida sinica 30 (2017) 630–637
Fig. 3 – The change of PLA (Mw = 50 k) density with Fig. 5 – Stress–strain curves of PLA (Mw = 100 k) in tensile
temperature and the glass transition temperature. simulations when cut-off radius r = 1.2 nm.
Group C1 C3 Groups κ E
−5
A 1.28 1556 A+B 4.96 × 10 1.20
B 1.12 −1661 A+C 4.67 × 10−5 1.21
C 1.21 44 A+D 1.15 × 10−4 1.10
D 1.39 2567 A+E 7.56 × 10−5 1.16
E 1.14 −264 B+C 5.22 × 10−5 1.21
The average value of E is B+D 6.53 × 10−5 1.23
1.18 GPa with the standard B+E 1.57 × 10−5 1.14
deviation of 0.04 C+D 7.42 × 10−5 1.20
Fig. 6 – Micro unit cell is stretched along the z-axis to the C+E 2.18 × 10−4 1.20
range of cut-off radius r. D+E 8.98 × 10−5 1.16
When the unit cell system is subjected to uniaxial tension There are five sets of data, from group A to E, in Fig. 4. Each set
in the z direction, as shown in Fig. 6, Eq. (3) can be simplified was nonlinearly fitted by a cubic polynomial of Eq. (6) to calcu-
after substituting Hooke’s law σ = Eε: late coefficients C1 and C3 . Five sets of C1 and C3 were acquired,
from which each two sets were selected to form a system of
2 d2 equations as per Eq. (7), for the solution of elastic modulus E
1 − ( e0 a ) T = Eε (4)
d z2 and coefficient κ. Ten results were obtained and tabulated in
Table 1. The average value of the elastic modulus of PLA at
The physical essence of nonlocal theory is taking the in- molecular weight of 50 k is 1.18 GPa with standard deviation
fluence from the neighboring material into account, which is of 0.04, which matches well with the experimental result of
in correspondence with the concept of cut-off radius in sim- 1.20 GPa reported in reference [9]. It shows a good reproducibil-
ulations. For a simulated system, the internal characteristic ity of the five simulations. The initial elastic modulus of PLA
length should rest within the range of cut-off radius r adopted is mainly determined by the coefficient C1 and modified by C3 ,
for simulation computation, beyond which the noncovalent which describes the influence of nonlocal effect.
forces acting on molecules will be isolated and the nonlo- Similarly, the three sets of data in Fig. 5 were nonlinearly
cal effect no longer exists. As shown in Fig. 6, a micro unit fitted and calculated. The results were tabulated in Table 2,
cell of length z0 is stretched to the cut-off radius r after the which are different from the experimental results (with the
acta mechanica solida sinica 30 (2017) 630–637 635
Table 2 – Simulation results of PLA (Mw = 100 k) when Table 3 – Simulation results of PLA (Mw = 100 k) when
r = 1.2 nm. r = 1.7 nm.
Group C1 C3 Groups κ E
Group C1 C3 Groups κ E
−5
A 1.67 2363 A+B 9.76 × 10 1.44
B 1.35 −981 B+C 5.79 × 10−4 1.92 A 2.28 1319 A+B −5
8.35 × 10 2.17
C 0.14 −3061 C+A 2.82 × 10−4 1.01 B 2.05 −1417 A+C 2.64 × 10−4 1.93
C 2.08 553 A+D 5.16 × 10−4 1.60
Note: The units of C1 , C3 , and E are GPa, and κ is a dimensionless D 3.01 2746 A+E 1.64 × 10−4 2.06
constant. E 2.05 −54 B+C 1.33 × 10−4 2.07
The average value of E is B+D 2.32 × 10−4 2.38
2.02 GPa with the standard B+E 2.49 × 10−6 2.05
deviation of 0.21 C+D 4.28 × 10−4 1.84
C+E 3.76 × 10−5 2.05
D+E 3.43 × 10−4 2.07
Group C1 C3 Groups κ E
−4
A 2.158 −1464 A+B 5.93 × 10 3.03
B 1.639 −2339 B+C 2.12 × 10−4 2.13
C 2.647 2421 C+A 1.26 × 10−4 2.34
√
a ∝ < R2 >, was proposed based on the concept that force
Table 5 – Elastic moduli E versus self-diffusion coeffi-
cients D at temperature 300 K. transfers with entanglement points, and a method for esti-
mating the value of cut-off radius for molecular simulations
PLA (Mw = 50 k, r = 1.2 nm) PLA (Mw = 100 k, r = 1.7 nm) is given accordingly. The research results suggest that: (1) the
Group E D Group E D elastic modulus should be acquired through nonlinear regres-
sion. (2) The nonlocal effect has a certain influence on the
A 1.20 1.48 A 2.17 1.9
simulation results of PLA. (3) According to the scaling law, the
B 1.21 1.47 B 2.07 2.01
C 1.21 1.47 C 1.93 1.98 cut-off radius is determined and applied to the MD sim-
ulations, of which the results reflect the influence of the
Note: The unit of E is GPa, and the unit of D is × 10 − 11 m2 /s. molecular weight change on the elastic moduli of PLA with
considerable accuracy, and are in good agreement with the ex-
perimental outcome.
impedes the relative slippage between polymer molecules
during the process of tensile simulations and raises the value
of elastic modulus. Conversely, the decrease of chain entan- references
glement with the decrease of molecular weight weakens the
intermolecular force, and the elastic modulus declines corre-
spondingly. When the molecular weight decreases to a certain
[1] S. Tinkov, R. Bekeredjian, G. Winter, C. Coester, Microbubbles
critical value, the polymer no longer has tensile strength. as ultrasound triggered drug carriers, J. Pharm. Sci. 98 (2009)
The elastic moduli and related self-diffusion coefficients 1935–1961.
at 300 K of PLA for two different molecular weights were tab- [2] S.P. Qin, C.F. Caskey, K.W. Ferrara, Ultrasound contrast
ulated in Table 5, respectively. It shows that elastic moduli as microbubbles in imaging and therapy: physical principles
well as self-diffusion coefficients have a tendency to increase and engineering, Phys. Med. Biol. 54 (2009) R27–R57.
[3] M. Stepniewski, A. Bunker, M. Pasenkiewicz-Gierula,
when PLA molecular weight grows. Larger self-diffusion co-
M. Karttunen, T. Rog, Effects of the lipid bilayer phase state
efficient implies an increase in mobility and a decrease in
on the water membrane interface, J. Phys. Chem. B 114 (2010)
frictional interaction between PLA chain segments, which is 11784–11792.
unfavorable to load transfer. But as mentioned earlier, the [4] L. Hoff, Oscillations of polymeric microbubbles: effect of the
elastic performance of polymers is in fact predominately de- encapsulating shell, J. Acoust. Soc. Am. 107 (2000) 2272–2280.
termined by the chain entanglement of molecules. PLA with [5] A.N. Damir, B. Khismatullin, Radial oscillations of
larger molecular weight has longer chains and a greater num- encapsulated microbubbles in viscoelastic liquids, Phys.
Fluids 14 (2002) 3534–3557.
ber of entanglement points, and thus greater elastic modulus.
[6] D. Garlotta, A literature review of poly(lactic acid), J. Polym.
The fluctuation of curves in Fig. 7 is quite different from Environ. 9 (2001) 63–84.
that in Fig. 4 because the intermolecular entanglement and [7] A.P. Mathew, K. Oksman, M. Sain, Mechanical properties of
the conformation of molecules are more stable with the in- biodegradable composites from poly lactic acid (PLA) and
creases of molecular weight and chain length, leading to microcrystalline cellulose (MCC), J. Appl. Polym. Sci. 97 (2005)
lower probability of slippage, plastic deformation, and sudden 2014–2025.
change of polymer conformation. In actual polymer systems, [8] C.C. Chen, J.Y. Chueh, H. Tseng, H.M. Huang, S.Y. Lee,
Preparation and characterization of biodegradable PLA
the degree of crystallization increases with larger molecular
polymeric blends, Biomaterials 24 (2003) 1167–1173.
weight and better stability of conformation, so more elasticity [9] I. Engelberg, J. Kohn, Physico-mechanical properties of
and brittleness is observed. On the contrary, more plasticity is degradable polymers used in medical applications: a
observed in the polymer system while the degree of crystal- comparative study, Biomaterials 12 (1991) 292–304.
lization decreases. [10] H.Z. Xiang, W.H. Yang, C.B. Hu, Y.H. Mu, J.F. Li, Study of the
mechanical and thermal properties of poly(lactic acid) and
poly(ethylene glycol) block copolymer with molecular
dynamics, J. Polym. Anal. Charact. 15 (2010) 235–244.
6. Conclusions [11] S.J. Marrink, A.H. de Vries, A.E. Mark, Coarse grained model
for semiquantitative lipid simulations, J. Phys. Chem. B 108
MD simulation is widely used in the study of polymer prop- (2004) 750–760.
erties, through which the intrinsic characteristics of certain [12] S.J. Marrink, H.J. Risselada, S. Yefimov, D.P. Tieleman, A.H. de
polymers in its stable conformation can be roughly deter- Vries, The MARTINI force field: coarse grained model for
biomolecular simulations, J. Phys. Chem. B 111 (2007)
mined. In this work, the coarse-grained models of long-chain
7812–7824.
PLA were built and simulated through the MD method, us- [13] L. Monticelli, S.K. Kandasamy, X. Periole, R.G. Larson,
ing the MARTINI force field. The change of self-diffusion D.P. Tieleman, S. Marrink, The MARTINI coarse-grained force
coefficient and density with temperature, and the glass transi- field: extension to proteins, J. Chem. Theory Comput. 4 (2008)
tion temperature were acquired from the simulations. Using 819–834.
a third-order polynomial to approximate the nonlocal stress [14] C.A. Lopez, A.J. Rzepiela, A.H. de Vries, L. Dijkhuizen,
P.H. Hunenberger, S.J. Marrink, Martini coarse-grained force
of polymer under the condition of comparatively small strain,
field: extension to carbohydrates, J. Chem. Theory Comput. 5
the formula for computing the initial elastic modulus consid-
(2009) 3195–3210.
ering the nonlocal effect based on the nonlocal theory was de- [15] J.J. Uusitalo, H.I. Ingolfsson, P. Akhshi, D.P. Tieleman,
rived. By combing polymer physics with the nonlocal theory, S.J. Marrink, Martini coarse-grained force field: extension to
the scaling law of internal characteristic length of polymers, DNA, J. Chem. Theory Comput. 11 (2015) 3932–3945.
acta mechanica solida sinica 30 (2017) 630–637 637
[16] G. Rossi, I. Giannakopoulos, L. Monticelli, N.K.J. Rostedt, [24] S.A. Fazelzadeh, E. Ghavanloo, Nonlocal anisotropic elastic
S.R. Puisto, C. Lowe, A.C. Taylor, I. Vattulainen, T. Ala-Nissila, shell model for vibrations of single-walled carbon nanotubes
A MARTINI coarse-grained model of a thermoset polyester with arbitrary chirality, Compos. Struct. 94 (2012) 1016–1022.
coating, Macromolecules 44 (2011) 6198–6208. [25] E. Ghavanloo, S.A. Fazelzadeh, Nonlocal elasticity theory for
[17] A.C. Eringen, On differential equations of nonlocal elasticity radial vibration of nanoscale spherical shell, Eur. J. Mech. A
and solutions of screw dislocation and surface waves, J. Appl. Solid 41 (2013) 37–42.
Phys. 54 (1983) 4703–4710. [26] L. Monticelli, S. Kandasamy, X. Periole, R. Larson,
[18] E.C. Aifantis, On the role of gradients in the localization of D.P. Tieleman, S.J. Marrink, The MARTINI coarse grained force
deformation and fracture, Int. J. Eng. Sci. 30 (1992) 1279– field: extension to proteins, J. Chem. Theory Comput. 4 (2008)
1299. 819–834.
[19] H. Askes, I.M. Gitman, Review and critique of the stress [27] S.J. Marrink, H.J. Risselada, S. Yefimov, D.P. Tieleman, A.H. de
gradient elasticity theories of Eringen and Aifantis, Adv. Vries, The MARTINI force field: coarse grained model for
Mech. Math. 21 (2010) 203–210. biomolecular simulations, J. Phys. Chem. B 111 (2007)
[20] H. Askes, E.C. Aifantis, Gradient elasticity in statics and 7812–7824.
dynamics: an overview of formulations, length scale [28] C. Wei, Thermal expansion and diffusion coefficients of
identification procedures, finite element implementations carbon nanotube-polymer composites, Nano Lett. 2 (2002)
and new results, Int. J. Solids Struct. 48 (2011) 1962–1990. 647–650.
[21] J. Fernandez-Saez, R. Zaera, J.A. Loya, J.N. Reddy, Bending of [29] P.-G. de Gennes, Scaling Concepts in Polymer Physics, Cornell
Euler–Bernoulli beams using Eringen’s integral formulation: University Press, New York, 1985.
a paradox resolved, Int. J. Eng. Sci. 99 (2016) 107–116. [30] P.J. Flory, Molecular theory of rubber elasticity, Polym. J. 17
[22] W. Sumelka, R. Zaera, J. Fernandez-Saez, A theoretical (1985) 1–12.
analysis of the free axial vibration of non-local rods with [31] S.H. Wu, Chain structure and Entanglement, J. Polym. Sci.
fractional continuum mechanics, Meccanica 50 (2015) Part B Polym. Phys. 27 (1989) 723–741.
2309–2323. [32] W.W. Graessley, S.F. Edwards, Entanglement interactions in
[23] P. Lu, P.Q. Zhang, H.P. Lee, C.M. Wang, J.N. Reddy, Non-local polymers and the chain contour concentration, Polymer 22
elastic plate theories, Proc. R. Soc. A 463 (2007) 3225–3240. (1981) 1329–1334.