Dear all,

I'm working on coarse graining individual CNTs so I can simulate an entire CNT array.

I'm using Build > Mesostructure > Coarse grain and then I simplify an entire CNT to a 1D structure with several beads conected by a line. To capture the mechanics, I'm assembling a forcefield that adds the following bead interactions:

Bond stretch

Angle bend

Torsion

Van der Waals interaction

The appraoch I'm using is a bit similar to the following paper:

S. Cranford, A single degree of freedom ‘lollipop’ model for carbon nanotube bundle formation, Journal of the Mechanics and Physics of Solids, 2010.

http://www.sciencedirect.com/science/article/pii/S0022509609001677

The difference is that I want to make the model even more coarse and randomize it. For that I wrote a script that generates a FF with random FF parameters per bead interaction. So far I have a working model (not callibrated yet) that shows the CG CNTs being attracted towards eachother over time by VdW forces, see the animation in the attachement (or the inserted image if it works). I've randomized the structure (random initial bond lengths and angles) so its closer the random morphology of a CNT in nature.

Furthermore, I intend to apply boundary conditions and prescribe displacements on the CG structure. Similar to the procedure I applied in the following paper: "Effects of single vacancy defect position on the stability of carbon nanotubes", Microelectronics reliability, 2012. I hope this will be also possible in the mesocite module or DPD with a script, otherwise I have a bigger problem.

The first question that I have is which module would you recommend for coarse graining up towards micrometer scales, DPD or Mesocite?

My current problem is that when I use continuum mechanics (CM) to predict the bending stiffness of a cylinder and compare it to the bending stiffness of 2 rods (between 3 beads) connected by a torsion spring the units are different.

For MD the Angle Bend energy is defined as:

E = K0/2*(theta-theta0)^2,

where the unit of bending stiffness is defined as:  K₀ : kcal/mol/rad²

When I convert this bending stiffness to standard units in CM I get the following:

%=== Bending conversion

Mol  = 6.02214179*1e23;          % Number of particles

kcal = 4184;                            % Joule = [N*m]

b_t = 1;                                   % kcal/mol/rad^2

b_t = b_t*kcal;                         % Joule/mol/rad^2 = N*m/mol/rad^2

b_t = b_t/Mol                           % N*m/rad^2 per particle

Every torsion spring I encounter has the units of Moment per radian: N*m/rad. So why is it N*m/rad^2 in every MD software while in CM the stiffness of a torsion spring is defined as N*m/rad?

There is probably a simple explanation, but I'm not seeing it.

Best regards,

René