UC Berkeley

Cellulosic biomass has the potential to be used as a sustainable feedstock for the production of liquid transportation fuels and other chemicals. However, this material also possesses an extreme resistance to structural and chemical degradation, known as recalcitrance, which currently prevents it from being used for this purpose. Therefore, an understanding of the origins of this recalcitrance and how to overcome it will be useful in the design of processes and technologies for the production of biofuels. To address these issues, we have performed all-atom molecular dynamics (MD) simulations of cellulose in various conditions and solvents to gain a molecular-level understanding of the forces and interactions that give rise to the macroscopic behaviors related to cellulose recalcitrance and how it is overcome.

Cellulose is a homopolymer of β-glucose connected by β-1,4 bonds found in the plant cell wall in the form of long, semicrystalline fibers called microfibrils. Within the two naturally occurring crystal structures of cellulose, I_β and I_α, glucose chains are arranged into flat sheets that are then stacked up upon each other, the polymerization axis of the chains aligned along the long axis of the microfibril. To better understand the forces that bind the chains of a microfibril together, we divided the interaction network of cellulose into three categories, which we then analyzed. These categories are:

1. Intrachain interactions: between neighboring glucoses within the same polymer chain.

2. Interchain interactions: between neighboring glucoses of different chains within the same sheet

3. Intersheet interactions: between neighboring glucoses of different chains in different sheets

The dominant intermolecular forces in the intrachain and interchain directions are OH--O hydrogen bonds (HBs), while in the intersheet direction, it is CH--O contacts (pseudo HBs) as well as van der Waals (vdW) interactions. We have examined the behavior of these three groups, and found that intersheet forces are most responsible for cellulose recalcitrance. The HBs of the intrachain and interchain interactions are severely disrupted at the microfibril surface by solvent exposure, seen by both HB geometries and HB occupancies. Conversely, by the same metrics, the intersheet CH--O HBs are robust and not weakened to any appreciable extent when at the microfibril surface. Counting up the amount of HBs that exist for each of the three interaction groups, we find that the number of intersheet HBs is far greater than the number of intrachain or interchain HBs. Also, energetically, sheet-sheet interactions are stronger than chain-chain interactions. In the interior of a microfibril, intersheet forces are ~1 kcal/mol-glucose stronger than interchain forces, and at the microfibril surface, the interchain forces weaken considerably, up to 50 %, whereas intersheet forces are only slightly affected.

Solvent water density behavior around the microfibrils is an important aspect of cellulose degradation, both for surface characterization, but more importantly, for its implications in the access of enzymes or chemical catalysts to cellulose surfaces. Cellulose imparts a specific, long-range structure into the surrounding water molecules that solvate it. Because of the amphiphilic nature of glucose, its water hydration layers exhibit the effects of both short and long-length scale hydrophobicity, the CH-presenting cellulose surfaces being the most hydrophobic. At short range, the HB network of water percolates around these CH groups while connecting with the glucose OH groups, while at long-range, the hydration shells of CH-presenting surfaces show the greatest amount of compressibility.

Ionic liquids (ILs), salts molten at or near room temperature, are one of the few solvents that are able to overcome the cellulose interaction network and dissolve the material. However, molecular and thermodynamic knowledge of how these liquids accomplish this is poor. By performing simulations of cellulose in a microfibril state and in a dissociated, dissolved state in a prototypical IL, 1-butyl-3-methylimidazolium (Bmim) chloride (Cl), and contrasting that behavior with the same in water, we are able to explain the solvent abilities of ILs with cellulose. Thermodynamically, the energy of dissolution is favorable in BmimCl but neutral to unfavorable in water, depending on temperature. From analysis of the three-dimensional solvent density distributions around the dissolved chains, chloride anions interact with the equatorial hydrophilic OH groups, forming strong HBs with them, while the cations preferentially solvate from the axial direction of the glucoses, where the hydrophobic CH groups are present, and have favorable contacts with the linker oxygen of the β-1,4 bond as well. Thus, BmimCl can satisfy the amphiphilic nature of glucose. This is in contrast to water, which cannot. In water, the energy of dissolution is not favorable because while water can form HBs with glucose's OH groups, interactions of water with its CH groups are ultimately undesirable.

Entropy also plays an important role in determining the thermodynamics of cellulose dissolution. Because of the long-range, highly ordered solvation structures that form around the sugar chains, solvent entropy favors the undissolved state of cellulose. Cellulose entropy, however, favors the dissolved state. The magnitudes of both of these preferences are less in BmimCl than in water because of the former's longer-range solvent-solvent forces and resulting liquid state structure. Summing the two entropic contributions together, the total entropy of dissolution is positive at all conditions and solvents tested except for in water at room temperature. Thus, in general, entropy changes favor cellulose dissolution.

Another solvent able to dissolve cellulose is lithium chloride (LiCl) in N,N-dimethylacetamide (DMA). Although the mechanism through which this is achieved is generally thought to be direct interaction of dissolved salt with cellulose, the specifics of how this is accomplished, and the relative contributions of the various components of the LiCl/DMA system to dissolving cellulose, are poorly understood. To address these issues, we performed MD simulations and free energy calculations related to the deconstruction of cellulose in LiCl/DMA, as well as in pure DMA, LiCl/water, and pure water. Calculation of the potential of mean force (PMF) of the deconstruction of a cellulose microfibril confirm that LiCl/DMA is a cellulose solvent, while DMA, LiCl/water, and water are not. Analysis of simulations of dissolved and undissolved cellulose in LiCl/DMA and LiCl/water reveals that solvent-mediated preferential interactions of dissolved ions with sugars are responsible for cellulose dissolution. By computing the three-dimensional density of ions around dissolved cellulose chains, we find that the localization of salt by the sugars is two orders of magnitude greater in DMA than in water. The ions near the sugars are able to disrupt cellulose's internal interaction network by forming O-Li⁺ and OH-Cl^- interactions, causing dissolution. Because DMA is a poor solvent for both LiCl and cellulose, these two species preferentially interact with each other to the exclusion of DMA, whereas since water is a good solvent for LiCl but a poor one for cellulose, no salt-sugar preferential interactions occur in it. We also find that opposite the case in water, in DMA, the high-density regions of Li⁺ cations are nearer to the cellulose atoms than those of the Cl^- anions. Quantification of the salt-sugar interactions via a coarse-graining force-matching analysis confirm that although they both are necessary, attractive sugar-Li⁺ interactions, and not sugar-Cl^- interactions, are the most important contributor to the disruption of cellulose's interaction network, and hence, to cellulose dissolution.