Calculation of single chain cellulose elasticity using fully atomistic modeling

Cellulose nanocrystals, a potential base material for green nanocomposites, are ordered bundles of cellulose chains. The properties of these chains have been studied for many years using atomic-scale modeling. However, model predictions are difficult to interpret because of the significant dependence of predicted properties on model details. The goal of this study is to begin to understand these dependencies. We focus on the investigation on model cellulose chains with different lengths and having both periodic and nonperiodic boundary conditions, and predict elasticity in the axial (chain) direction with three commonly used calculation methods. We find that chain length, boundary conditions, and calculation method affect the magnitude of the predicted axial modulus and the uncertainty associated with that value. Further, the axial modulus is affected by the degree to which the molecule is strained. This result is interpreted in terms of the bonded and nonbonded contributions to potential energy, with a focus on the breaking of hydrogen bonds during deformation. transition their use from an investigative method to a predictive tool useful for cellulose-based application design. Application: This study lays the groundwork for understanding the predictions of atomistic models and to help tomistic modeling of cellulose has been used to A complement experimental measurements of cellulose nanocrystals by providing the atomic level detail necessary to predict structural, energetic, and mechanical characteristics, and to gain a fundamental understanding of the atomic-scale origins of these characteristics. Such models frequently are used to predict elastic properties because the simulation methods are relatively simple and results can be compared to experimental data. Predictions of elasticity in the axial (chain) direction for the la or Iß polymorphs of crystalline cellulose have been reported from atomistic model-based studies for more than 20 years [l]. Unfortunately, quantitative comparison of elasticity results from model-to-model or to experimental measurements has been difficult because of the significant effect of variations in the simulation methods, atomic interaction models, and configuration of the modeled structures. For example, the axial modulus predicted using a model of 1x1x10 Iß unit cells was ~11% smaller than that predicted using a model of 4x4x10 Iß unit cells [2]. In another study, a difference in the initial equilibrium unit cell length of 0.1% caused a 22% variation in the axial modulus prediction [3]. The goal of this paper is to clearly illustrate some of these effects such that the meaning of predictions made using atomistic models can be understood more fully. We report results of a molecular dynamicsand molecular mechanicsbased study focusing on the most basic structure, a single cellulose molecule, to understand the effects of some variables on model predictions. Many chains of different lengths coexist in a bulk polymer, and so the variation of structure, energetic, and elastic prop erties with chain length is insignificant. However, at the nanoscale, properties are likely to be influenced by so-called end effects, where the molecule is short enough that the contribution the ends of the chains make is comparable to that of the main body of the chain. To characterize this effect, we focused this study on model cellulose chains with different lengths and having both periodic and nonperiodic boundary conditions. These models are used to predict elasticity in the axial direction with three commonty used calculation methods. Results are discussed in terms of the effects of chain length, boundary conditions, calculation method, and strain on the model-predicted axial modulus, and the uncertainly associated with those predictions.