Dear LAMMPS experts,
I’m interested in studying the welding of polymers subjected to shear. My box consists of two equilibrated melts that are entangled with within themselves, but not between one another.
The polymer chains are color coded for clarity, but they use the same potential (Kremer-Grest). What I’m trying to study is if the deformation contributes to that diffusion of chains.
I’ve been running several simulations using a nonperiodic system where I apply a vibration using fix move wiggle on piston atoms to create compressional waves using NVE in the region of interest, and a Langevin Thermostat at the bottom to scatter coherent waves as best I can. But to achieve higher levels of extension / shear, I’m think I’m going to need to use a nonequilibrium thermostat like fix nvt/sllod in conjunction with fix deform wiggle because lower periods and higher amplitudes are causing the chains to heat up unphysically.Here is my conundrum. Fix nvt/sllod does not allow for nonperiodic boundary conditions in the direction of deformation for fairly intuitive physical reasons, and since I am interested in measuring the strength of the middle interface itself, a homogeneous extension protocol using uniform deformation wouldn’t really be measuring the strength of the interface which is what I’m after. So I’m using fix addforce on a set of grip atoms like so.
-------------------- time & grip sizes --------------------
variable t equal time
variable Nu equal count(upper)
variable Nl equal count(lower)variable Nu0 equal {Nu} variable Nl0 equal {Nl}
-------------------- SMD-style spring forces along z --------------------
Fu = -K [ z_com_u(t) - z_com_u(0) - v0 * t ]
Fl = -K [ z_com_l(t) - z_com_l(0) ]
variable Fuz_tot equal “-v_Kspring*(c_comU[3] - v_zU0_hold - v_v0*time)”
variable Flz_tot equal “-v_Kspring*(c_comL[3]-v_zL0_hold)”
variable Flz_tot equal “-v_Kspring*(c_comL[3] - v_zL0_hold )”variable fzU_atom equal “v_Fuz_tot / v_Nu0”
variable fzL_atom equal “v_Flz_tot / v_Nl0”Apply only z forces (x,y = 0)
fix addU upper addforce 0.0 0.0 v_fzU_atom
fix addL lower addforce 0.0 0.0 v_fzL_atom-------------------- gauge region & diagnostics --------------------
Gauge = slab between grips (excludes the grips themselves)
variable lower_grip_zmax equal v_zlo+v_gthick
variable upper_grip_zmin equal v_zhi-v_gthickregion gauge block INF INF INF INF v_lower_grip_zmax v_upper_grip_zmin units box
group gauge region gaugecompute sGauge gauge stress/atom NULL
compute sumGauge gauge reduce sum c_sGauge[1] c_sGauge[2] c_sGauge[3]
variable Ax equal lx*ly
variable volGauge equal v_Ax*(v_upper_grip_zmin-v_lower_grip_zmax)variable pGauge equal “- (c_sumGauge[1]+c_sumGauge[2]+c_sumGauge[3]) / (3.0*v_volGauge)”
variable pxxGauge equal “- c_sumGauge[1] / v_volGauge”
variable pyyGauge equal “- c_sumGauge[2] / v_volGauge”
variable pzzGauge equal “- c_sumGauge[3] / v_volGauge”
When I switch to nvt/sllod, I know I’m going to need to live with the periodic boundary conditions in the Z dimension, but if I want to measure the “real” interface strength in the middle of the box between the two melts, I’m going to have some atoms spanning the periodic boundary condition in Z that will be part of both sets of grips. This, to me, seems like a recipe for a crash.
So, here are the ideas I’m contemplating and I welcome any and all feedback.
-
After the welding simulation is complete, I find those atoms that span the Z periodic boundary and cut the bonds of those atoms with a custom script or pick so I can proceed with the boundary driven uniaxial extension test. Yes it will be unphysical bond breaking, but my logic is that they will be part of the grip atoms and aren’t part of the real region of interest which is the “real” middle interface.
-
I unwrap in the Z dimension only, then quench the welded sample as normal in the NPT ensemble like I have been. Then proceed as normal. My hesitation here is I’m having difficulty picturing what’s going to happen to the polymers that span the Z dimension and even more difficulty thinking about how that’s going to change the stress transfer in my melt.
-
Make copies of all Z spanning polymers so that the local connectivity and chain entanglement structure within the box are kept consistent, then I expand the box to compensate for these duplicated chains, then quench, then do boundary driven extension.
Please let me know if any of these ideas are workable. Or if there’s another option outside of what I’m thinking.
Thanks in advance,
Erik