Insights into Supramolecular Sites Responsible for Complete Separation of Biomass-Derived Phenolics and Glucose in Metal-Organic Framework NU-1000.

The molecular origins of adsorption of lignin-derived phenolics to metal-organic framework NU-1000 are investigated from aqueous solution as well as in competitive mode with glucose present in the same aqueous mixture. A comparison of adsorption equilibrium constants (K) for phenolics functionalized with either carboxylic acid or aldehyde substituents demonstrated only a slight increase (less than a factor of 6) for the former according to both experiments and calculations. This small difference in K between aldehyde and carboxylic-acid substituted adsorbates is consistent with the pyrene unit of NU-1000 as the adsorption site, rather than the zirconia nodes, while at saturation coverage, the adsorption capacity suggests multiple guests per pyrene. Experimental standard free energies of adsorption directly correlated with the molecular size and electronic structure calculations confirmed this direct relationship, with the pyrene units as adsorption site. The underlying origins of this relationship are grounded in noncovalent π-π interactions as being responsible for adsorption, the same interactions present in the condensed phase of the phenolics, which to a large extent govern their heat of vaporization. Thus, NU-1000 acts as a preformed aromatic cavity for driving aromatic guest adsorption from aqueous solution and does so specifically without causing detectable glucose adsorption from aqueous solution, thereby achieving complete glucose-phenolics separations. The reusability of NU-1000 during an adsorption/desorption cycle was good, even with some of the phenolic compounds with greatest affinity not easiliy removed with water and ethanol washes at room temperature. A competitive adsorption experiment gave an upper bound for K for glucose of at most 0.18 M, which can be compared with K for the phenolics investigated here, which fell in the range of 443-42 639 M. The actual value of K for glucose may be much closer to zero given the lack of observed glucose uptake with NU-1000 as adsorbent.