FULL PAPER WWW.C-CHEM.
ORG
Implicit Solvent Coarse-Grained Model of Polyamidoamine
Dendrimers: Role of Generation and pH
Leebyn Chong, Fikret Aydin, and Meenakshi Dutt*
Highly branched polymers such as polyamidoamine (PAMAM)
dendrimers are promising macromolecules in the realm of
nanobiotechnology due to their high surface coverage of tun-
able functional groups. Modeling efforts of PAMAM can pro-
vide structural and morphological properties, but the inclusion
of solvents and the exponential growth of atoms with genera-
tions make atomistic simulations computationally expensive.
We apply an implicit solvent coarse-grained model, called the
Dry Martini force field, to PAMAM dendrimers. The reduced
number of particles and the absence of a solvent allow the
capture of longer spatiotemporal scales. This study character-
izes PAMAM dendrimers of generations one through seven in
Introduction
Dendrimers are an unique group of highly branched polymers
that comprise of an multifunctional core with a radial shell of
relatively short chains of monomers that split into additional
short chains with each increasing generation or shell size.[1]
The terminal ends on the exterior shell are sites suitable for
groups that can provide unique functionalities to the den-
drimer.[2] Maximized terminal groups and highly symmetric
structures can be achieved with starburst topologies, products
of divergent and convergent synthesis.[1,3] This architecture
has been considered a model system for ultrasoft colloids with
characteristics of polymers and hard spheres.[4]
Polyamidoamine (PAMAM) and its derivatives have been
some of the most investigated dendrimers, experimentally and
computationally.[5–13] Its biocompatibility opens it up to a sig-
nificant number of biomedical applications which include
drug delivery,[14] imaging micelles,[15] and protein mimicry.[16]
The importance of higher-generation dendrimers lies in the
exponential increase of terminal groups capable of being func-
tionalized and coordinated. Studies of higher-generation
dendrimers show promise as a cell transfection agent in
disrupting large vesicles as well as delivering genetic mate-
rial.[17,18] In the realm of inorganic nanomaterials, its uses
include nanoparticle catalysis[19] and decontamination.[9–11]
Many of the above applications require an understanding of
PAMAM structures in a variety of conditions and solvents. Scat-
tering experiments have attempted to fit measurements such
as radius of gyration for a number of generations in different
pH environments.[20] Computational studies have investigated
various characterizations of PAMAM using atomistic molecular
dynamics (MD),[6]–[8,21] coarse-grained (CG) MD,[13,22] and atom-
istic Monte-Carlo (MC) methods.[21] All-atom models are com-
920 Journal of Computational Chemistry 2016, 37, 920–926
acidic, neutral, and basic pH environments. Comparison with
existing literature, both experimental and theoretical, is done
using measurements of the radius of gyration, moment of
inertia, radial distributions, and scaling exponents. Additionally,
ion coordination distributions are studied to provide insight
into the effects of interior and exterior protonation on counter
ions. This model serves as a starting point for future designs
of larger functionalized dendrimers. V C 2015 Wiley Periodicals,
Inc.
DOI: 10.1002/jcc.24277
putationally expensive to simulate due to the exponentially
increasing number of atoms and bonds of higher-generation
dendrimers. Future research will likely necessitate simulations
of systems beyond the nanoscale with multiple dendrimers,
such as self-assembled dendrimersomes.[23] The high compu-
tational cost of current models is one of the limitations for
large-scale simulations. One possible solution involves model-
ing the dendrimer in an implicit solvent, which would
decrease the number of particle interaction calculations and
thus decrease the computational cost. Lee et al.[22] used the
Martini CG force field developed by Marrink et al.[24] to model
PAMAM conjugated with polyethylene glycol. This CG method
groups approximately four nonhydrogen atoms including
water as one bead. Arnarez et al.[25] have recently simulated
40 nm lipid vesicles by developing another version of the
Martini CG force field called Dry Martini, an implicit solvent
model.
In this contribution, we develop and use a Dry Martini
model of PAMAM dendrimers to study its static properties in
aqueous solutions. Our study spans seven generations (G1 to
G7) in acidic, neutral, and basic pH environments. We repara-
metrize the force field such that the radius of gyration is in
agreement with corresponding results from experiments and
all-atom simulations. We make comparisons with regards to
the ratio of the principal moments of inertia tensor and the
radial density distributions. Finally, we characterize the ion
coordination statistics of the dendrimers. The significance of
this contribution comes from the application of a new
L. Chong, F. Aydin, Meenakshi Dutt
Department of Chemical and Biochemical Engineering, Rutgers the State
University of New Jersey, Piscataway, New Jersey 08854
E-mail: meenakshi.dutt@rutgers.edu
C 2015 Wiley Periodicals, Inc.
V
WWW.CHEMISTRYVIEWS.COM
 WWW.C-CHEM.ORG
Figure 1. Martini CG representation of G1 PAMAM at neutral pH as half of
a starburst dendrimer or a dendron.
implicit solvent CG model to PAMAM dendrimers with charac-
terizations for controllable variables including generations
and pH.
FULL PAPER
Methodology
All simulations are conducted using the Large-scale Atomic/
Molecular Massively Parallel Simulator (LAMMPS) package.[26]
Non-bonded interaction parameters originate from the Dry
Martini force field of Arnarez et al.[25] The PAMAM Martini
structure are derived from Lee et al.[22] and includes bond and
angle parameters. A temperature of 298 K is maintained with
a Langevin[27] thermostat in a canonical ensemble, with a
damping parameter set to 4.0 ps along with a time step of 10
fs. The cubic simulation box has a dimension of 13.5 nm.
Upon equilibration of the energy and radius of gyration, each
system is simulated for 10 ns and sampled every 0.1 ns. To
ensure reproducibility of results, the simulations are repeated
for four random seeds.
The Martini CG-scheme involves mapping approximately
four nonhydrogen atoms to 0.47 nm diameter beads of vary-
ing levels of polarity.[24] For PAMAM at a neutral pH, the CG
dendrimer is composed of three groups[22]: internal tertiary
amine junctions (56 g/mol), amide branches (70 g/mol), and
terminal protonated primary amines (31 g/mol). In acidic envi-
ronments (pH < 5), high concentrations of hydronium ions
lead to the protonation of all amines via the nitrogen’s lone
pair of electrons. However, only protonation of terminal
amines is assumed to occur in neutral pH environments due
to water accessibility and stable solvation of primary amines
over tertiary amines. Basic pH conditions (pH > 10) allows for
deprotonation of all amines including the terminal groups. Fig-
ure 1 illustrates this dependency of the CG moieties on pH
and the interaction representations. Internal amines are con-
sidered relatively nonpolar “N0” with respect to the other
groups. The more electronegative acyl group on the branch
amides allows it to be designated as moderately polar “P3.”
Protonated amines (11e) are classified as charged species
“Qd” capable of being hydrogen bond donors. Since neutral
and low pH levels result in net positive PAMAM dendrimers,
CG representations of the hydrated chloride (21e) counter
ions (107.453 g/mol) are added to those systems and are also
considered charged species “Qa” but with the ability to be
hydrogen bond acceptors. Such groupings primarily affect the
van der Waals interactions modeled by the Lennard-Jones (LJ)
potential:
 12  6 !
rij rij
ULJ 5 4eij 2 (1)
rij rij
The parameters rij is the effective bead diameter (0.47 nm)
and eij is the energy parameter, which is correlated with the
interaction strength. As part of the Dry Martini force field, a
matrix of all possible pairs of bead identities details the inter-
action energy based on the polar/nonpolar relation and
charge.[25] A matrix of energy parameters is provided in Sup-
porting Information Table SI1.
Nonbonded interactions also include long-range electro-
statics, which is vital for modeling the electrostatic screening.
Dry Martini electrostatics follows the same scheme as wet Mar-
tini electrostatics, where the relative permittivity is set to 15
Journal of Computational Chemistry 2016, 37, 920–926 921
 FULL PAPER WWW.C-CHEM.ORG

angle neutron scattering (SANS) experiments[20] and atomistic

MD simulations.[6]

Results and Discussion

To validate the Dry Martini force field for PAMAM dendrimers,
comparisons of the radius of gyration, the ratio of the eigen-
values of the moment of inertia tensor, and the radial density
distributions are made with respect to other simulation and
SANS results. The radius of gyration provides an approximate
measure of the dimensions of the polymeric structures. The
following equation is used to calculate radius of gyration (Rg):
PN
mi ri2
R2g 5 Pi51
N
(4)
i51 mi

where N is the number of CG beads encompassing the

PAMAM dendrimer and mi and ri are the mass and positions of
the ith bead, respectively. Supporting Information Figure SI1

Figure 2. Scaling analysis of the radius of gyration (Rg) with respect to a)

number of CG groups and b) molecular weight.

and the Coulombic potential cutoff is given by 1.2 nm.[24,25]

The neighbor list cutoff is extended to 1.4 nm and is updated
at every iteration. Particle-particle particle-mesh (PPPM) solvers
are in place to handle the long-range electrostatics.
Functions and parameters for the bonds and angles are
close to the values determined by Lee et al.[22] Bonds and
angles are treated as harmonic and cosine squared potentials,
respectively:

Ub 5Kb ðr2r0 Þ2 (2)

2
Ua 5Ka ðcos ðuÞ2cos ðu0 ÞÞ (3)

For angles with amides as vertices, the angle constant Ka is

given by 35 kJ/mol with an equilibrium angle h0 given by
1608. Angles with tertiary amines as vertices have Ka 5 10 kJ/
mol and h0 5 1208. All bonds in the PAMAM dendrimer are
considered equal with a bond constant Kb given by 2000
kJ mol21 nm22. Angle parameters are summarized in Support-
ing Information Table SI2. Deviating from Lee et al.[22] with
regards to the equilibrium bond length r0, we increase r0 from Figure 3. Images of the equilibrated conformations of G1 – G7 PAMAM
dendrimers at low, neutral, and high pH values. Tertiary amines, amides
0.42 to 0.5 nm to accommodate for the swelling of the den- and terminal amines are colored in yellow, red, and blue, respectively.
drimer. Our parameterization is guided by mapping the radius [Color figure can be viewed in the online issue, which is available at
of gyration of G7 PAMAM to published values from small- wileyonlinelibrary.com.]

922 Journal of Computational Chemistry 2016, 37, 920–926 WWW.CHEMISTRYVIEWS.COM

 WWW.C-CHEM.ORG FULL PAPER

ues obtained from experiments as the model parameters have

been developed with respect to G7.
The scaling relation between the molecular weight (M) or N
as a function of the radius of gyration is provided in Figure 2.
We observe that scaling exponent at neutral pH (measured to
be N 5 R3:160:1
g ) is a little bit less than scaling exponents at low
and high pH (given by N 5 R3:360:1 g and N  Rg3.26 6 0.09,
respectively). Previous all-atom simulation and Small-Angle X-
ray Scattering studies[21,28] on PAMAM dendrimers at various
solvent conditions as well as studies using MD and MC
approaches[29,30] of theoretical dendrimers in ideal or theta sol-
vents demonstrate the scaling relation of N 5 R3g . In addition,
CG simulations of PAMAM dendrimers[31] give the scaling rela-
tion of N 5 R3:1
g . This indicates that our results, especially at
neutral pH environment, are in good agreement with previous
theoretical, computational, and experimental studies.
Figure 3 shows equilibrated conformations of the seven
PAMAM dendrimers in low, neutral, and high pH. We observe
a majority of the PAMAM dendrimers to exhibit a globular
shape which is in qualitative agreement with TEM images.[32]
The shape of the dendrimers can be characterized by calculat-
ing the ratios, Iz/Ix and Iz/Iy, of the principal moments of iner-
tia. These values represent the average of the three principal
moments of inertia in decreasing order Iz > Iy > Ix where the

Figure 4. Ratios of the eigenvalues of the moment of inertia tensor for

PAMAM at a) low, b) neutral, and c) high pH. Comparisons with Han
et al.[7] is shown for high pH. [Color figure can be viewed in the online
issue, which is available at wileyonlinelibrary.com.]

compares the measurement of the radius of gyration from

atomistic MD,[7,21] wet Martini MD simulations,[22] and SANS
experiments.[20] For all generations, the radius of gyration is
observed to increase with acidity. Underestimation of experi-
mental values of the radius of gyration appears to be the
greatest for G4 and G5, with a maximum of 24.7% deviation Figure 5. a) Asphericity and b) relative shape anisotropy calculated from
for G4 at high pH. As expected, the radius of gyration for G7 the gyration tensor. [Color figure can be viewed in the online issue, which
at neutral and low pH accurately matches corresponding val- is available at wileyonlinelibrary.com.]

Journal of Computational Chemistry 2016, 37, 920–926 923

 FULL PAPER WWW.C-CHEM.ORG

tion seven for neutral pH. Due to the unequal composition

and charge between the dendrimer core and shell, the electro-
static interaction of the terminal groups with each other and
the counter-ions may be inhibiting the spherical shape of
PAMAM. Figure 5b depicts the relative shape anisotropy
decreasing with higher generations. A value of 1 signifies an
ideal linear chain while 0 represents symmetric conforma-
tions.[33] Comparisons with atomistic MD simulations from
Maiti et al. confirm similar trends in the relative shape anisot-
ropy.[6] Also, measurements of the aspect ratios using atomis-
tic MD simulations of the dendrimers only covers high pH
values and demonstrate a consistent decrease in the ratio with
increasing generation.[6,7] Although there are large variations
for the ratios corresponding to G1, the Dry Martini and the
atomistic models are in relatively good agreement on the
PAMAM dendrimer approaching a spherical shape for higher
generations.
We investigate the effect of pH on the radial distribution of
monomers for G1 to G7 PAMAM dendrimers by binning the
monomers based on their radial distances from the center-of-
mass of the dendrimer into spherical shells of thickness
Dr 5 0.1 nm. The radial distribution of tertiary amine, amide,

Figure 6. Number of monomers as a function of radial distance from

center-of-mass of a) G1, b) G4, and c) G7 PAMAM dendrimers at low, neu-
tral, and high pH. The production simulations have been run for a total
time of 10 ns and each data point has been averaged over time and four
simulations using different random seeds. [Color figure can be viewed in
the online issue, which is available at wileyonlinelibrary.com.]

axes rotate along with the dendrimer. The ratios are plotted
against dendrimer generation in Figure 4. With a decrease in
aspect ratio of the minor-major axes for all pH levels, PAMAM’s
ellipsoid shape becomes less eccentric as the number of gen- Figure 7. a) Average fractions of chloride ions that are uncoordinated and
b) average number of CG groups coordinating with each chloride (taking
erations increases. In addition, asphericity and relative shape
into account of only ions that have at least one coordination) as a function
anisotropy are calculated from the gyration tensors.[33] Figure of generation. [Color figure can be viewed in the online issue, which is
5a shows indications that the asphericity increases at genera- available at wileyonlinelibrary.com.]

924 Journal of Computational Chemistry 2016, 37, 920–926 WWW.CHEMISTRYVIEWS.COM

 WWW.C-CHEM.ORG
Figure 8. Probability distribution breakdown of CG groups coordinating
chlorides at a) neutral and b) low pH. [Color figure can be viewed in the
online issue, which is available at wileyonlinelibrary.com.]
and terminal amine for G1, G4, and G7 dendrimers at varying
pH conditions are shown in Figures 6a–6c, respectively. The
radial distribution plots for the remaining generations are
given in Supporting Information Figure SI2. Our results demon-
strate that the distribution of tertiary amines shifts to the right
as pH decreases, which indicates that the monomers of terti-
ary amine prefer to locate near the periphery of the dendrimer
due to the protonated tertiary amines at the core of the den-
drimer. There is also a minor shift toward the periphery when
pH decreases from high to neutral but this is observed only
for the G6 and G7 dendrimer. For G1 dendrimers, we do not
observe significant change in the distribution of the mono-
mers with pH. Similarly, amides and terminal amines prefer to
locate away from the core of the dendrimers when pH
decreases from high to low due to the protonated tertiary
amines, which keep the dendrimer branches away from each
other, resulting in swelling and their location closer to the
periphery. The amount of shifting becomes more prominent
with the number of protonated monomers increasing from
neutral to low pH. However, the difference in the shifting
between neutral and low pH decreases at small generations.
This could arise from a more effective electrostatic repulsion
between terminal amines at neutral pH when the size of the
FULL PAPER
dendrimer is small since there are fewer number of uncharged
monomers that cause screening between the charged mono-
mers. Our results are consistent with earlier experimental[20]
and computational studies,[22,34] which demonstrate an
increase in the monomer density towards the periphery of the
dendrimers at low pH.
An earlier work demonstrated and characterized the adsorp-
tion behavior of surface-grafted PAMAM dendrons with car-
boxylate groups on its terminal ends by calculating
distributions and averages of the coordination with ions.[35] In
this current study, the maximum cutoff for defining coordina-
tion is 0.5811 nm or 0.47(2(21/621)11). This threshold encom-
passes the LJ potential energy minimum at 21/6r with an
additional (21/621)r skin width. The fraction of uncoordinated
chloride ions is shown in Figure 7a. There is an expected trend
of increasing coordination with the surface area of the PAMAM
dendrimer. Additionally, the greater net positive charge of
PAMAM at low pH promotes coordination with chloride ions.
For the chlorides that are coordinated, Figure 6b illustrates the
average coordination number per ion to increase with genera-
tion and acidity. Decomposition of the specific CG moieties
coordinating with the chloride ions are depicted in Figure 8 as
probability distributions. The distributions are constructed by
normalizing the averaged counts by the number of possible
coordination for each CG moiety. This is to prevent the bias
arising from the higher concentration of internal amines and
amides with respect to terminal amines and chlorides. Signifi-
cant differences in distributions are observed for each pH level.
At neutral pH, the electrostatic potentials dominate the inter-
action and cause adsorption of the chloride ions to the exte-
rior of the dendrimer. Low pH conditions are characterized by
the enhanced diffusion of chloride ions to the highly positive
interior of the PAMAM dendrimer. Ion coordination analysis
cannot be applied to high pH systems because of the lack of
ions.
Conclusions
We have developed and used a Dry Martini force field of
PAMAM dendrimers to study the role of dendrimer generation
and pH on its size, structure, shape and ion adsorption statis-
tics. The size, shape, and structural predictions of the PAMAM
dendrimers using the new model are in agreement with earlier
investigations using atomistic MD simulations and experi-
ments.[7,20]–[22] Results on the radius of gyration, scaling expo-
nents, ratios of the principal moments of inertia, and the radial
density distributions follow predictable trends as reported in
literature.[6,7,20]–[22,32,34] Coordination numbers and distribu-
tions characterizing the adsorption statistics of ions onto the
PAMAM dendrimer demonstrate electrostatic effects of chlo-
ride ions on the surface of PAMAM at neutral pH and ion diffu-
sion into the dendrimer at low pH. The CG model can be used
to investigate PAMAM dendrimers with larger generations or
systems with multiple dendrimers. The methodology and tools
presented can be used to design novel dendrimers with
desired ion adsorption profiles.
Journal of Computational Chemistry 2016, 37, 920–926 925
 FULL PAPER WWW.C-CHEM.ORG
Acknowledgments
The authors acknowledge the use of computational resources at the
Rutgers Discovery Informatics Institute (RDI2) and the Texas
Advanced Computing Center (TACC) through Extreme Science &
Engineering Discovery Environment (XSEDE) allocations TG-
DMR140125 and TG-MCB090174 for the completion of the work.
Keywords: polymers  dendrimers  molecular dynamics  coarse-
graining
How to cite this article: L. Chong, F. Aydin, M. Dutt. J. Comput.
Chem. 2016, 37, 920–926. DOI: 10.1002/jcc.24277
] Additional Supporting Information may be found in the
online version of this article.
[1] D. Tomalia, Prog. Polym. Sci. 2005, 30, 294.
[2] P. G. Z. Benini, B. R. McGarvey, D. W. Franco, Biol. Chem. 2008, 19, 245.
[3] D. Tomalia, H. Baker, J. Dewald, M. Hall, G. Kallos, S. Martin, J. Roeck, J.
Ryder, P. Smith, Polym. J. 1985, 17, 117.
[4] M. Ballauff, C. N. Likos, Angew. Chem. Int. Ed. 2004, 43, 2998.
[5] J. Mijović, S. Ristić, J. Kenny, Macromolecules 2007, 40, 5212.
[6] P. K. Maiti, T. Çag
ın, G. Wang, W. Goddard, Macromolecules 2004, 37, 6236.
[7] M. Han, P. Chen, X. Yang, Polymer (Guildf). 2005, 46, 3481.
[8] P. Carbone, F. M€ uller-Plathe, Soft Matter 2009, 2638.
[9] M. R. Mankbadi, M. Barakat, M. H. Ramadan, H. L. Woodcock, J. N.
Kuhn, J. Phys. Chem. B 2011, 115, 13534.
[10] R. Qu, Y. Niu, C. Sun, C. Ji, C. Wang, G. Cheng, Microporous Mesoporous
Mater. 2006, 97, 58.
[11] Y. Niu, R. Qu, C. Sun, C. Wang, H. Chen, C. Ji, Y. Zhang, X. Shao, F. Bu,
J. Hazard. Mater. 2013, 244–245, 276.
[12] K. Kono, R. Ikeda, K. Tsukamoto, E. Yuba, C. Kojima, A. Harada, Biocon-
jug. Chem. 2012, 23, 871.
[13] H. Lee, R. G. Larson, J. Phys. Chem. B 2009, 113, 13202.
[14] S. Sadekar, H. Ghandehari, Adv. Drug Deliv. Rev. 2012, 64, 571.
[15] J. Chen, J. Guo, B. Chang, W. Yang, Colloid Polym. Sci. 2012, 290, 517.
926 Journal of Computational Chemistry 2016, 37, 920–926
[16] J. Yang, S. Cao, J. Li, J. Xin, X. Chen, W. Wu, F. Xu, J. Li, Soft Matter
2013, 9, 7553.
[17] J. F. Kukowska-Latallo, A. U. Bielinska, J. Johnson, R. Spindler, D. A.
Tomalia, J. R. Baker, Proc. Natl. Acad. Sci. USA 1996, 93, 4897.
[18] Z. Y. Zhang, B. D. Smith, Bioconjugate Chem. 2000, 11, 805.
[19] J. Zheng, S. Lin, X. Zhu, B. Jiang, Z. Yang, Z. Pan, Chem. Commun.
2012, 48, 6235.
[20] Y. Liu, C.-Y. Chen, H.-L. Chen, K. Hong, C.-Y. Shew, X. Li, L. Liu, Y. B.
Melnichenko, G. S. Smith, K. W. Herwig, L. Porcar, W.-R. Chen, J. Phys.
Chem. Lett. 2010, 1, 2020.
[21] P. K. Maiti, T. Çagın, S.-T. Lin, W. A. Goddard, Macromolecules 2005, 38,
979.
[22] H. Lee, R. G. Larson, Macromolecules 2011, 44, 2291.
[23] V. Percec, D. A. Wilson, P. Leowanawat, C. J. Wilson, A. D. Hughes, M.
S. Kaucher, D. A. Hammer, D. H. Levine, A. J. Kim, F. S. Bates, K. P.
Davis, T. P. Lodge, M. L. Klein, R. H. DeVane, E. Aqad, B. M. Rosen, A. O.
Argintaru, M. J. Sienkowska, K. Rissanen, S. Nummelin, J. Ropponen,
Science 2010, 328, 1009.
[24] S. J. Marrink, H. J. Risselada, S. Yefimov, D. P. Tieleman, A. H. de Vries,
J. Phys. Chem. B 2007, 111, 7812.
[25] C. Arnarez, J. J. Uusitalo, M. F. Masman, H. I. Ing olfsson, D. H. de Jong,
M. N. Melo, X. Periole, A. H. de Vries, S. J. Marrink, J. Chem. Theory
Comput. 2015, 11, 260.
[26] S. Plimpton, J. Comput. Phys. 1995, 117, 1.
[27] T. Schneider, E. Stoll, Phys. Rev. B 1978, 17, 1302.
[28] S. Rathgeber, M. Monkenbusch, M. Kreitschmann, V. Urban, A. Brulet,
J. Chem. Phys. 2002, 117, 4047.
[29] G. Giupponi, D. M. A. Buzza, J. Chem. Phys. 2004, 120, 10290.
[30] M. Murat, G. S. Grest, Macromolecules 1996, 29, 1278.
[31] P. K. Maiti, Y. Li, T. Cagin, W. A. Goddard, J. Chem. Phys. 2009, 130,
144902.
[32] C. L. Jackson, H. D. Chanzy, F. P. Booy, B. J. Drake, D. A. Tomalia, B. J.
Bauer, E. J. Amis, Macromolecules 1998, 31, 6259.
[33] H. Arkın, W. Janke, J. Chem. Phys. 2013, 138, 054904.
[34] Y. Liu, V. S. Bryantsev, M. S. Diallo, W. Goddard, J. Am. Chem. Soc.
2009, 131, 2798.
[35] L. Chong, M. Dutt, Phys. Chem. Chem. Phys. 2015, 17, 10615.
Received: 21 August 2015
Revised: 24 October 2015
Accepted: 8 November 2015
Published online on 17 December 2015
WWW.CHEMISTRYVIEWS.COM