Tapan G. Desai

Molecular dynamics simulations of polymer transport in nanocomposites

Molecular dynamics simulations on the Kremer-Grest bead-spring model of polymer melts are used to study the effect of spherical nanoparticles on chain diffusion. We find that chain diffusivity is enhanced relative to its bulk value when polymer-particle interactions are repulsive and is reduced when polymer-particle interactions are strongly attractive. In both cases chain diffusivity assumes its bulk value when the chain center of mass is about one radius of gyration R(g) away from the particle surface. This behavior echoes the behavior of polymer melts confined between two flat surfaces, except in the limit of severe confinement where the surface influence on polymer mobility is more pronounced for flat surfaces. A particularly interesting fact is that, even though chain motion is strongly speeded up in the presence of repulsive boundaries, this effect can be reversed by pinning one isolated monomer onto the surface. This result strongly stresses the importance of properly specifying boundary conditions when the near surface dynamics of chains are studied.

Structure, surface excess and effective interactions in polymer nanocomposite melts and concentrated solutions.

The Polymer Reference Interaction Site Model (PRISM) theory is employed to investigate structure, effective forces, and thermodynamics in dense polymer-particle mixtures in the one and two particle limit. The influence of particle size, degree of polymerization, and polymer reduced density is established. In the athermal limit, the surface excess is negative implying an entropic dewetting interface. Polymer induced depletion interactions are quantified via the particle-particle pair correlation function and potential of mean force. A transition from (nearly) monotonic decaying, attractive depletion interactions to much stronger repulsive-attractive oscillatory depletion forces occurs at roughly the semidilute-concentrated solution boundary. Under melt conditions, the depletion force is extremely large and attractive at contact, but is proceeded by a high repulsive barrier. For particle diameters larger than roughly five monomer diameters, division of the force by the particle radius results in a nearly universal collapse of the depletion force for all interparticle separations. Molecular dynamics simulations have been employed to determine the depletion force for nanoparticles of a diameter five times the monomer size over a wide range of polymer densities spanning the semidilute, concentrated, and melt regimes. PRISM calculations based on the spatially nonlocal hypernetted chain closure for particle-particle direct correlations capture all the rich features found in the simulations, with quantitative errors for the amplitude of the depletion forces at the level of a factor of 2 or less. The consequences of monomer-particle attractions are briefly explored. Modification of the polymer-particle pair correlations is relatively small, but much larger effects are found for the surface excess including an energetic driven transition to a wetting polymer-particle interface. The particle-particle potential of mean force exhibits multiple qualitatively different behaviors (contact aggregation, steric stabilization, local bridging attraction) depending on the strength and spatial range of the polymer-particle attraction.

Thermal transport in graphene-based nanocomposite

Using molecular dynamics simulations, we study model graphene nanoplatelets and carbon nanotubes in an organic matrix. We demonstrate that, despite relatively high interfacial thermal resistance between the filler and the matrix, the thermal conductivity enhancement of the nanocomposite can be very significant. Our results suggest that agglomeration and low aspect ratio of the conductive nanofiller additive are primarily responsible for the limited conductivity enhancement reported to date. Mapping of the simulation results on the homogenization model, accounting for interfacial resistance, allows us to predict the full potential of the nanocarbon filler addition for thermal conductivity enhancement.

Comparison of ReaxFF, DFTB, and DFT for phenolic pyrolysis. 1. Molecular dynamics simulations.

A systematic comparison of atomistic modeling methods including density functional theory (DFT), the self-consistent charge density-functional tight-binding (SCC-DFTB), and ReaxFF is presented for simulating the initial stages of phenolic polymer pyrolysis. A phenolic polymer system is simulated for several hundred picoseconds within a temperature range of 2500 to 3500 K. The time evolution of major pyrolysis products including small-molecule species and char is examined. Two temperature zones are observed which demark cross-linking versus fragmentation. The dominant chemical products for all methods are similar, but the yields for each product differ. At 3500 K, DFTB overestimates CO production (300-400%) and underestimates free H (~30%) and small C(m)H(n)O molecules (~70%) compared with DFT. At 3500 K, ReaxFF underestimates free H (~60%) and fused carbon rings (~70%) relative to DFT. Heterocyclic oxygen-containing five- and six-membered carbon rings are observed at 2500 K. Formation mechanisms for H2O, CO, and char are discussed. Additional calculations using a semiclassical method for incorporating quantum nuclear energies of molecules were also performed. These results suggest that chemical equilibrium can be affected by quantum nuclear effects at temperatures of 2500 K and below. Pyrolysis reaction mechanisms and energetics are examined in detail in a companion manuscript.

Effect of crosslink formation on heat conduction in amorphous polymers

We performed molecular dynamics (MD) simulations on amorphous polyethylene (PE) and polystyrene (PS) in order to elucidate the effect of crosslinks between polymer chains on heat conduction. In each polymer system, thermal conductivities were measured for a range of crosslink concentration by using nonequilibrium MD techniques. PE comprised of 50 carbon atom long chains exhibited slightly higher conductivity than that of 250 carbon atom long chains at the standard state. In both cases for PE, crosslinking significantly increased conductivity and the increase was more or less proportional to the crosslink density. On the other hand, in the PS case, although the thermal conductivity increased with the crosslinking, the magnitude of change in thermal conductivity was relatively small. We attribute this difference to highly heterogeneous PS based network including phenyl side groups. In order to elucidate the mechanism for the increase of thermal conductivity with the crosslink concentration, we decomposed en...

Improvement of heat transfer efficiency at solid-gas interfaces by self-assembled monolayers

Using molecular dynamics simulations, we demonstrate that the efficiency of heat exchange between a solid and a gas can be maximized by functionalizing solid surface with organic self-assembled monolayers (SAMs). We observe that for bare metal surfaces, the thermal accommodation coefficient (TAC) strongly depends on the solid-gas interaction strength. For metal surfaces modified with organic SAMs, the TAC is close to its theoretical maximum and is essentially independent from the SAM-gas interaction strength. The analysis of the simulation results indicates that softer and lighter SAMs, compared to the bare metal surfaces, are responsible for the greatly enhanced TAC.

Thermal transport in nanoclusters

Nonequilibrium and equilibrium molecular dynamics simulations are employed to study the thermal transport in sintered silicon nanoclusters made of 15 nm diameter nanoparticles arranged on a simple cubic lattice. Both simulation techniques indicate a reduction in the thermal conductivity from ∼120 W/m K (bulk) to 1.5 W/m K (nanoclusters) at 500 K. This dramatic reduction is attributed to the reduced thermal conductivity of nanoparticle (15 W/m K) and most prominently to the nanosized constriction resistance due to necking between the two nanoparticles. Comparison with the existing models, radial distribution function and vibrational analysis show that the phonon transport in the nanosized neck region is ballistic rather than diffusive.

Comparison of ReaxFF, DFTB, and DFT for phenolic pyrolysis. 2. Elementary reaction paths.

Reaction paths for the loss of CO, H2, and H2O from atomistic models of phenolic resin are determined using the hybrid B3LYP approach. B3LYP energetics are confirmed using CCSD(T). The energetics along the B3LYP paths are also evaluated using the PW91 generalized gradient approximation (GGA), the more approximate self-consistent charge density functional tight binding (SCC-DFTB), and the reactive force field (ReaxFF). Compared with the CCSD(T)/cc-pVTZ level for bond and reaction energies and barrier heights, the B3LYP, PW91, DFTB(mio), DFTB(pbc), and ReaxFF have average absolute errors of 3.8, 5.1, 17.4, 13.2, and 19.6 kcal/mol, respectively. The PW91 is only slightly less accurate than the B3LYP approach, while the more approximate approaches yield somewhat larger errors. The SCC-DFTB paths are in better agreement with B3LYP than are those obtained with ReaxFF.

Molecular-dynamics simulations of the transport properties of a single polymer chain in two dimensions

Molecular-dynamics simulations are conducted to elucidate the critical factors affecting the transport properties of isolated polymer chains in strictly two dimensions. The relevance of surface inhomogeneity is critically examined. We unequivocally find that surface inhomogeneity is critical in obtaining transport behavior consistent with the recent measurements of surface diffusion for polymers adsorbed at the solid-liquid interface. For a systematic investigation of this point, heterogeneity was introduced by decorating the surface with impenetrable elements and we find that chain diffusivity crossed over from Rouse-type behavior to reptationlike with increasing surface coverage of obstacles. This transition in behavior occurred when the mean distance between obstacles is approximately equal to the end-to-end distance, Re, of the two-dimensional chain. Our results underscore the importance of surface disorder (not only literal obstacles but by reasonable extension also to other types of disorder) in determining the transport behavior of chains adsorbed to solids.

Density fluctuation correlation length in polymer fluids

We employ molecular dynamics simulations and integral equation theory to study the real space collective density correlations in chain polymer fluids over a wide range of concentrations. For a degree of polymerization of 80 and low polymer concentrations, simulation and theory both find the typical behavior that the density–density correlation function decays monotonically with an associated correlation length that decreases with polymer concentration. In contrast, at high polymer concentrations the correlation function decays in an oscillatory manner with a correlation length that grows with fluid density. The correlation length is found to be smallest at a reduced polymer density corresponding to the onset of monomer scale oscillations in the density correlation function. Based on the theory, qualitatively identical behavior is found for chains of 2000 monomers. The observed density dependence of the correlation length is analogous to the dependence of the charge screening length on ionic strength in el...