Article pubs.acs.org/JPCB
Translational Entropy and Dispersion Energy Jointly Drive the Adsorption of Urea to Cellulose Pan Chen,†,‡ Yoshiharu Nishiyama,*,§,∥ Jakob Wohlert,‡ Ang Lu,*,⊥ Karim Mazeau,§,∥ and Ahmed E. Ismail*,# †
Aachener Verfahrenstechnik, RWTH Aachen University, Turmstrasse 46, D-52064 Aachen, Germany Wallenberg Wood Science Center, and the Department of Fibre and Polymer Technology, KTH Royal Institute of Technology, SE-10044 Stockholm, Sweden § CERMAV, Univ. Grenoble Alpes, F-38000 Grenoble, France ∥ CERMAV, CNRS, F-38000 Grenoble, France ⊥ College of Chemistry and Molecular Science, Wuhan University, Wuhan 430072, China # Department of Chemical and Biomedical Engineering, West Virginia University, Morgantown, West Virginia 26505, United States ‡
S Supporting Information *
ABSTRACT: The adsorption of urea on cellulose at room temperature has been studied using adsorption isotherm experiments and molecular dynamics (MD) simulations. The immersion of cotton cellulose into bulk urea solutions with concentrations between 0.01 and 0.30 g/mL led to a decrease in urea concentration in all solutions, allowing the adsorption of urea on the cellulose surface to be measured quantitatively. MD simulations suggest that urea molecules form sorption layers on both hydrophobic and hydrophilic surfaces. Although electrostatic interactions accounted for the majority of the calculated interaction energy between urea and cellulose, dispersion interactions were revealed to be the key driving force for the accumulation of urea around cellulose. The preferred orientation of urea and water molecules in the first solvation shell varied depending on the nature of the cellulose surface, but urea molecules were systematically oriented parallel to the hydrophobic plane of cellulose. The translational entropies of urea and water molecules, calculated from the velocity spectrum of the trajectory, are lower near the cellulose surface than in bulk. As urea molecules adsorb on cellulose and expel surface water into the bulk, the increase in the translational entropy of the water compensated for the decrease in the entropy of urea, resulting in a total entropy gain of the solvent system. Therefore, the cellulose−urea dispersion energy and the translational entropy gain of water are the main factors that drive the adsorption of urea on cellulose.
1. INTRODUCTION The promise of a scalable, environmentally sustainable process for the dissolution of biomass has been a major driver in green chemistry research. Finding a viable method, which would allow its use as a widescale replacement for petroleum-based feedstock, has led to the renewed investigation of solvent systems. As an example, while urea (CO(NH2)2) has been used as a common reagent for over a century,1 only recently has the addition of urea to precooled aqueous alkaline solutions been shown to facilitate the dissolution of crystalline polysaccharides, such as cellulose,2 chitin,3 and chitosan.4 Since then, while the role of urea in the dissolution process has been investigated both experimentally and theoretically, no clear consensus has yet been reached. One key question has been whether urea interacts directly with the polysaccharides5−8 or if the effect is indirectly mediated by the solvent.9−13 The controversy on the role of urea in cellulose dissolution actually parallels its role as a denaturing agent for proteins and as a hydrotrope in the increased solubility of hydrophobic drug © 2017 American Chemical Society
molecules. It has been long unclear whether the molecular mechanism consists of direct interactions between urea and the protein (or solute) or if urea indirectly causes denaturation by altering the local water structure. In the last decade, both simulations14−22 and small-angle X-ray and neutron scatterings have shown an accumulation of urea molecules around the solutes or proteins.23 Statistical thermodynamics also identifies two major driving forces for solubilization: accumulation of urea as a hydrotrope around the solute and reduced activity of water in the bulk.24 The accumulation of urea around the solute seems to be relatively well established. On the other hand, the molecular interactions driving the accumulation of urea is still not clear. Some molecular dynamics (MD) simulation studies highlight the role of hydrogen bonding,25 whereas others consider van Received: November 26, 2016 Revised: February 20, 2017 Published: February 21, 2017 2244
DOI: 10.1021/acs.jpcb.6b11914 J. Phys. Chem. B 2017, 121, 2244−2251
Article
The Journal of Physical Chemistry B
Figure 1. (a) Surface distribution function of urea around the cross section of the cellulose fibril as a function of urea concentration. (b) Experimental and simulated urea−cellulose adsorption isotherms. The amount of urea considered to have adsorbed on the cellulose surface was measured at cutoffs of 0.53, 0.93, and 1.2 nm, corresponding to the distance between the center of mass (COM) of urea and the closest nonhydrogen atom of cellulose. The amount of urea adsorbed in the MD simulations was scaled by 1/6. (c) Snapshot of the cellulose fibril coated by urea in 0.30 g urea/mL solution.
der Waals (nonpolar) interactions to be more important.26 Urea is a polar molecule that forms relatively strong hydrogen bonds (HBs), as can be expected from its heat of sublimation (nearly 100 kJ/mol),27,28 but its dissolution in water is endothermic.29 Thus, the presence of hydrogen bonding does not necessary mean that such polar interactions are the driving force of urea solvation. Thawing in the quaternary urea/water/alkali hydroxide/ cellulose system has been previously studied by differential scanning calorimetry.9,11 Here, the cellulose−alkali, hydroxide− water, and urea−water domains appear to behave independently, suggesting that urea and cellulose do not have strong mutual interactions. The same conclusion is derived from 13C and 1H NMR experiments of the cellobiose/urea/aqueous NaOH system, where the cellobiose−aqueous NaOH chemical shifts and relaxation behavior are not influenced by the presence of urea.5 On the other hand, the accumulation of urea molecules in the proximity of the solid cellulose surface is predicted by MD simulations and corroborated by the enhanced spin relaxation of the surface cellulose chains measured by solid-state NMR spectroscopy.26,30 Again, the driving force for the putative accumulation of urea around cellulose remains an open question. In this study, we used adsorption isotherm experiments to quantify the accumulation of urea molecules on cellulose surfaces. MD simulations were then performed to compute the various thermodynamic contributions associated with this. These calculations established that Lennard-Jones dispersion interactions, as well as the gain in the translational entropy of water expelled from the cellulose surface, are the primary molecular mechanisms driving the urea solvation process.
of urea on cellulose surfaces at room temperature is shown in Figure 1b. For urea concentrations between 0.01 and 0.30 g/ mL, about 5% of the dissolved urea is adsorbed on the cellulose surface, independent of concentration. Enrichment of urea on cellulose surfaces was also observed in the MD simulations. Although the 18-chain bundle represents the lower bound on the smallest possible microfibril unit,31 we find that the effective specific surface area of microcrystalline cellulose is much larger in our simulations than that observed in experiments. For example, using solid-state NMR spectroscopy, the estimated crystallinity of the microfibril is only 60% as large as the crystallinity of a simulated 9 × 10 chain fibril, if all “noncrystalline” chains were on the surface. Similarly, the experimental crystallinity of cellulose is only one-third of that measured in our simulations. Furthermore, multiple crystallites can form water-inert aggregates with typical widths of about 20 nm,32 further decreasing the specific surface area by about a factor of three. Altogether, the number of simulated urea molecules in the proximity of cellulose was scaled down by a factor of 6 to compare with the experimental isotherm. The experimental data and simulated accumulation results agreed well with each other when either one or two solvation layers was considered. Even when the second layer was considered, the apparent adsorption still showed a Langmuir-type curve. 2.2. Urea and Water Interaction with Cellulose. In the first adsorption layer, urea formed more HBs with cellulose than water; it acted mostly as a hydrogen donor, which is logical as it contains four hydrogen atoms capable of donating to all cellulose O atoms (Figure S4 and Table S3). A striking feature is that urea donates to the acetal oxygen atoms, O1 and O5, of cellulose, whereas water barely interacts with these “buried” atoms. A similar result is also obtained by Wernersson et al. in MD simulations using another force field.26 Urea donated systematically about 20% more to the O6 hydroxyl group of cellulose than to the O2 group, independent of concentration. Urea donated the same amount of HBs to the O2 and O3 hydroxyl groups of cellulose but accepted more HBs from O2 than O3. The hydrogen atom of the HO3 hydroxyl group was significantly engaged in intrachain HBs with the ring oxygen O5 atom of the next residue. The dihedral angle distribution, p(cos θ), between the urea molecule plane and cellulose crystal surfaces is shown in the right-hand side of Figure 2. Urea tended to orient parallel to the
2. RESULTS AND DISCUSSION 2.1. Urea Adsorption: Experiments and Simulation. Figure 1a shows the distribution function of urea relative to the cellulose surface as a function of urea concentration. An intense peak is observed, corresponding to the first solvation layer, particularly at lower urea concentrations; only small bumps can be seen for the second and third solvation layers. To estimate the number of adsorbed molecules on the surface, we defined urea layers by the minima of the surface distribution function: 0.53, 0.93, and 1.2 nm. The experimental adsorption isotherm 2245
DOI: 10.1021/acs.jpcb.6b11914 J. Phys. Chem. B 2017, 121, 2244−2251
Article
The Journal of Physical Chemistry B
Figure 2. (Left) Density profiles of urea normal to the interfaces for various cellulose surfaces along with superimposed snapshots. (Right) The orientation parameter distribution, p(cos θ), between the plane of bulk and adsorbed urea molecules and different cellulose surfaces or the hydrophobic 200 plane.
hydrophobic 200 surface (|cos θ| > 0.87), normal to the 010 surface (|cos θ| < 0.5), and oblique to the other two surfaces (0.5 < |cos θ| < 0.87). When the pyranose (200) plane of cellulose was used as the reference plane, urea co-aligned with cellulose (Figure 2, right). Water molecules also tended to align on the cellulose crystal surfaces (Figures 3 and S5), although they had less orientational freedom compared to bulk water. On the hydrophobic 200 surface and 11̅0 surface, water tended to co-align with the surface, whereas co-alignment was largely avoided on the hydrophilic 010 and 110 surfaces (Figure S5). As the urea concentration increased, the water orientation was more disordered and became closer to that of bulk water for urea concentrations of ≥0.30 g/mL. 2.3. Residence Time and Diffusion of Urea on Cellulose Surface. The adsorption of urea on cellulose is a
dynamic process. Urea molecules, even in the first adsorption layer, were still mobile enough to diffuse back into the bulk phase after coming into close contact with cellulose. On average, in the individual surface simulations, every urea molecule spent about one-quarter of the total simulation time within 0.53 nm of a cellulose surface. The residence time of urea molecules near the cellulose surface ranged from 20 ps to 10 ns (Figure 4a), with most molecules spending less than 200 ps near the surface, indicating the very fast exchange between bulk and adsorbed urea. Figure 4b shows the MSDs for urea molecules with residence times longer than 3 ns, either in bulk or absorbed on the surface. The diffusion constant of urea in the bulk phase varied from 2.15 × 10−5 to 2.78 × 10−5 cm2 s−1, compared to the experimental values of 1.14 × 10−5−1.37 × 10−5 cm2 s−1 in the same concentration range,33 similar to the calculated diffusion 2246
DOI: 10.1021/acs.jpcb.6b11914 J. Phys. Chem. B 2017, 121, 2244−2251
Article
The Journal of Physical Chemistry B D bulk × 105 = −2.16c + 2.80
(1)
Dsurf × 106 = −5.07c + 4.58
(2)
The diffusion constant of urea in the proximity of cellulose surfaces is nearly an order of magnitude smaller than that in the bulk. This confinement leads to a loss in entropy, as will be discussed in the next section. 2.4. Driving Force of Urea Adsorption on Cellulose. 2.4.1. London Dispersion and Electrostatic Interactions. Electrostatic interactions accounted for nearly two-thirds of the total urea−cellulose interaction energy (Figure S6). However, urea−cellulose HBs, the main contributor to the urea−cellulose electrostatic interactions, are weaker than the urea−water HBs, as the partial charges of the hydroxyl groups of cellulose are smaller compared to those of water. Thus, the driving force of urea accumulation near the cellulose surface appears to be the result of London dispersion interactions rather than electrostatic interactions. The carbonyl O and N atoms in urea molecules have LJ ϵ values higher than those of the water oxygen in all three force fields (Tables S1 and S2), leading to a greater nonpolar stabilization. The occurrence of urea−cellulose HBs was directly related to the accumulation of urea on the cellulose surface. We followed the change in the amount of urea adsorbed as a function of the amplitude of the particle atomic charges of urea. Three MD simulations performed using the CHARMM force field on cellulose immersed in 0.30 g/mL of urea solution were compared, and the charge of urea was scaled by 2/3, 3/3, and 4/3, respectively. The number of urea molecules adsorbed on the cellulose surface as a function of simulation time at different cutoffs is shown in Figure S7. If electrostatic interactions were the main driving force, then the increased partial charge would lead to a higher amount of urea adsorbed. However, more urea molecules were adsorbed when their charges were reduced. When the charges were multiplied by 4/3, the amount of urea adsorbed in the first adsorption layer was almost unchanged but that in the second and third layers slightly increased. Therefore, the London dispersion stabilization between urea and cellulose is more important than the electrostatic interaction in driving urea adsorption. 2.4.2. Entropy. The solvent entropy was quantified using the two-phase thermodynamic (2PT) method35,36 (details in Supporting Information). Adding either cellulose or urea to pure water led to a decrease in water entropy. The decreased entropies per water molecule caused by urea, cellulose, and both urea and cellulose are denoted, respectively, by
Figure 3. (Left) Orientation parameter (cos θ) between the plane of either bulk or adsorbed water molecules and various cellulose surfaces. (Right) Snapshot of the model exposing the hydrophobic 200 surface in 0.15 g urea/mL. The analyzed water molecules are displayed as green spheres for adsorbed molecules and blue spheres for bulk molecules.
ΔSwurea/water = Swurea/water − Swwater < 0
(3)
ΔSwcellulose/water = Swcellulose/water − Swwater > 0
(4)
ΔSwurea/cellulose/water = Swurea/cellulose/water − Swwater < 0
(5)
Decomposing the total entropy into translational, rotational, and vibrational entropies showed that the variation of the total entropy of water was proportional to the translational entropy, but that of urea was due to the variation in both rotational and translational entropies, of which the translational entropy was more important (see Figure S8). The translational entropy increased monotonically as a function of the distance from cellulose surface (Figure 5a,b). Compared to that in bulk urea, the entropy of the urea in the first solvation shell was reduced by about 14%, which can be intuitively explained by the
Figure 4. (a) Residence time distribution of urea on cellulose. (b) Mean-squared displacements (MSDs) of bulk and adsorbed urea versus simulation time for different urea concentrations.
constant of pure urea for the OPLS model parameterized into the GROMOS force field.34 The bulk and surface diffusion constants decreased linearly with increasing urea concentration (c) according to 2247
DOI: 10.1021/acs.jpcb.6b11914 J. Phys. Chem. B 2017, 121, 2244−2251
Article
The Journal of Physical Chemistry B
Figure 5. (a) Variation of translational entropy of each water molecule in the cellulose−urea−water system as a function of distance from cellulose surface. (b) Variation of translational entropy of urea in the cellulose−urea−water system as a function of distance from the cellulose surface. Only the 0.3 g/mL system is shown. (c) The number of adsorbed urea and expelled water molecules in the first solvation shell (≤0.53 nm) of cellulose surface as a function of urea concentration. (d) The excess molar entropy of water (ΔSw) in urea/cellulose/water, cellulose/water and urea/water systems and that of urea (ΔSu) and solvent (ΔSsolvent) due to the presence of cellulose.
entropy of the expelled water. On average, one urea molecule expels between 2.2 and 2.5 water molecules from the first adsorption shell. The transfer of urea from the bulk to the cellulose surface lowered the concentration of bulk urea by approximately 5%, leading to an increase in the bulk water entropy of 0.4 × 10−3 to 1.4 × 10−3 J mol−1 K−1 per mole urea, as calculated from Table S4. The decrease in the entropy of urea adsorbed on the cellulose surface (Figure 5b) is always lower than the increase in the entropy of water. Thus, the net solvent entropy change is always positive (see Figure 5d).
translational freedom being restricted to a single dimension. The loss of the rotational entropy in urea at the surface is related to the confined orientation when absorbed, as observed previously. When urea dissolves in water and forms HBs with water molecules, the newly formed urea−water HBs are weaker than the broken water−water HBs. The dissolution of urea in water is endothermic, so the process is entropy-driven. Although water loses entropy when urea is added, urea gains more entropy in comparison to its solid state more than compensating the loss in water entropy leading to the negative dissolution free energy. The same holds for the dissolution of crystalline cellobiose,37 but not cellulose, because the translational and rotational entropies are simply proportional to the number of molecules, whereas the molecular surface is roughly proportional to the total mass. In a urea−cellulose−water system, the decrease in the entropy of water should be the sum of the cellulose and urea contributions, if urea and cellulose do not interact. However, the calculated total entropy loss of water in the ternary system was less than the sum of the contributions from the cellulose and urea surfaces ΔSwurea/cellulose/water > ΔSwcellulose/water + ΔSwurea/water
ΔSw = ΔSwurea/cellulose/water − (ΔSwcellulose/water + ΔSwurea/water) > 0
(7)
ΔSu = ΔSuurea/cellulose/water − ΔSuurea/water > 0
(8)
ΔSsolvent = ΔSw + ΔSu > 0
(9)
We therefore conclude that the entropy gain of water, especially the translational entropy, also drives the adsorption of urea on cellulose. The dependence of water entropy on urea concentration would be instructive to adjust the amount of urea toward exploring the best dissolution condition of cellulose together with alkaline ions. Urea is also found to prevent the selfaggregation of hydrophobic chains by acting as a “glue” intermediate.18 In this study, we focused on crystalline cellulose surfaces, but the same mechanism probably holds for cellulose
(6)
This is because urea adsorbs on cellulose and correspondingly expels partially confined water directly from the cellulose surface to the bulk (Figure 5c), leading to an increase in the 2248
DOI: 10.1021/acs.jpcb.6b11914 J. Phys. Chem. B 2017, 121, 2244−2251
Article
The Journal of Physical Chemistry B
built with exposed 110, 110̅ , 010, and 200 surfaces. The fibril was centered in a periodic simulation box with approximate dimensions of 9 × 9 × 4.15 nm3. The chains in the fibril were covalently bonded to their own periodic images in the direction of the fibril axis to simulate infinite chains. The fibril was surrounded by various ratios of water and urea molecules corresponding to the experimental concentrations. A detailed description of the simulation boxes and compositions is displayed in Figure S2. Another series of models, each with only one surface exposed to solvent, were generated to compare the adsorption properties of urea on different surfaces, as shown in Figure 2a. The chains have DP = 8 and were chemically linked to their periodic images. The surface exposed to a 0.15 g urea/mL solution, either 200, 110, 11̅0, or 010, also directly interacted nonchemically with their images in either the x or y direction, as in an ideally perfect crystal, to represent an infinite surface. All simulations were performed with GROMACS version 4.6.4.39 Energy minimizations were performed using the steepest descent and conjugate gradient methods. MD simulations were carried out by first heating the models from 0 to 300 K in 6 ns, followed by 100 ns production runs. Bond lengths were constrained using the LINCS algorithm.40 The equations of motion were solved using the standard leapfrog algorithm41 with a time step of 2 fs. The pressure was regulated at 1 bar using a semi-isotropic Berendsen barostat42 with compressibilities of 4.5 × 10−5 and 7.9 × 10−7.43 The temperature was controlled by the velocity-rescaling algorithm of Bussi et al.44 MD frames were saved every 20 ps. 5.1. Force Fields. We primarily use the GROMOS 56Acarbo force field,45 in which one LJ parameter was optimized46 as well as the corresponding urea34 and SPC water47 models. Two additional carbohydrate force fields, CHARMM C3648 and GLYCAM06,49 were also used for the 0.30 g urea/mL system. For these two force fields, TIP3P water50 and their respective urea models were used.51,52 We note briefly that while many additional urea force fields have been developed for the study of urea−protein interaction in the last two decades, because of different development strategies, the resulting parameters (especially the atomic partial charges)53 vary significantly between models. Although some of them are popularly used, their quality have not been carefully assessed. 5.2. HBs. Urea−cellulose and water−cellulose HBs were identified according to the geometric criteria of having a donor−acceptor−hydrogen angle of
