*2.3. Experiment Samples*

The trehalose, LiOH, KOH and urea used in the experiment were purchased from Shanghai Aladdin Reagent Company. The purity of trehalose was over 99%, and the purity of alkali and urea was over 99.9% for both. Both deuterated urea and heavy water were purchased from Sigma-Aldrich China. The deuteration rate of heavy water was over 99.9%, and the deuteration rate of urea was over 98%.

In this study, we used SANDALS and MPI to observe 4 kinds of experimental sample with different chemical components. They were aqueous solutions of trehalose/LiOH/urea, trehalose/KOH/urea, trehalose/LiOH and trehalose/KOH. The molar ratio of each component in the sample was consistent with that in Prof. Zhang's experiment [2]. Each kind of solution included three different deuterated ratio test samples, i.e., full hydrogen, full deuterium and half deuterium. All samples were measured at room temperature (25 ◦C) and −10 ◦C. However, the trehalose LiOH/urea aqueous solution froze at −10 ◦C in the neutron scattering experiment, so we raised the temperature of this sample to −5 ◦C. Each sample was measured in a flat sample cell of titanium–zirconium alloy with a capacity of 1.3 mL for approximately 6 h. Empty sample cells were measured for approximately 4 h for background subtraction. The sample composition and symbols are shown in Table 1.

**Table 1.** Sample labels, chemical components, deuterium ratios and molar ratios of the samples.


For the EPSR simulation and the following discussion of the results, the symbols of the atoms in trehalose were as shown in Figure 1. Oxygen and hydrogen atoms of H2O were labeled as OW and HW, respectively; alkali cations were labeled as Li, Na and K, respectively; the oxygen and hydrogen atom of hydroxide anion were labeled as OOH and HOH, respectively; and the carbon, oxygen, nitrogen and hydrogen atoms in the urea molecule were labeled as CU, OU, NU and HU, respectively. The label of each atom and the parameters of reference potential used in EPSR simulation are listed in Table 2(1,2). Among them, epsilon and sigma are the parameters of the Lennard-Jones reference potential (L-J), and q is the charge of the atoms [13–17]. The information for the EPSR simulation box is shown in Table 3. The data about NaOH aqueous solutions can be found in our previous study [13].

**Figure 1.** Structure of a trehalose molecule and its atomic labels used in EPSR simulation. All carbon atoms are labeled as C; all hydrogen atoms connected to the oxygen atom are labeled as H; the remaining hydrogen atoms connected to the carbon atom are labeled as M; the oxygen atom linking the two glucose rings is labeled as O1; the oxygen atoms on the glucose ring is labeled as O2; the oxygen atoms on the hydroxyl group connected to the glucose ring are labeled as O3; and the oxygen atom on the methyl group is labeled as O4. In order to describe the distribution of atoms around trehalose with spatial angles, we set up a coordinate system with O1 as the coordinate origin. The blue arrow connecting two C atoms at the glucose rings represents the direction of the *X*-axis; the red arrow pointing vertically to the blue arrow from O1 is the direction of the *Y*-axis; the *Z*-axis is the cross product of the *X*-axis and the *Y*-axis, which is approximately perpendicular to the paper.



**Table 3.** Cubic simulation box atomic number density and number of molecules used in EPSR.


#### **3. Results**

The neutron scattering structure factor and EPSR simulation results of different samples are shown in Figure 2. As shown in the figure, the EPSR simulation results and the neutron scattering profiles were in good agreement. Therefore, the structure reconstructed by EPSR represented the most probable atomic structure of the solutions.

**Figure 2.** Experimental neutron total scattering profiles (dashed black lines), the EPSR fitted neutron structure factors (red lines) and their differences (blue dots). (**a**) Trehalose in LiOH/urea aqueous solution, (**b**) trehalose in LiOH aqueous solution, (**c**) trehalose in KOH/urea aqueous solution, (**d**) trehalose in KOH aqueous solution. Data have been offset for clarity.

EPSR gave us the most probable, all-atom structures in trehalose alkali/urea aqueous solution. The simulation, as shown in Figure 3, qualitatively showed the distribution diagram of Li+, Na+, K+ ions and urea molecules around trehalose. In order to improve the statistics, the ions and atoms in the figure were the number of atoms gathered in the 100-frame EPSR simulation conformation. Three things could be seen directly. The first thing was that those ions and urea molecules were anisotropically distributed around

trehalose because of steric repulsion, so we built a coordinate system around its O1 atom (Figure 1). Thus, RDF could be calculated at different space angles. The second thing was that ions concentrated around the hydrophilic oxygen atoms of trehalose, and smaller ions were much more easily penetrated into the glucose ring, from K<sup>+</sup> to Na<sup>+</sup> to Li<sup>+</sup> (Figure 3a). The third thing was that larger urea molecules could not directly interact with trehalose, and they were further away from trehalose than the ions (Figure 3b).

**Figure 3.** The schematic diagram of the accumulation of alkali metal ions and urea molecules around trehalose. (**a**) Li+ (blue dots), Na+ (dark blue dots), K+ (red dots) ions are distributed around a trehalose molecule. (**b**) Distribution of Li<sup>+</sup> and the oxygen atoms of urea (red dots) around a trehalose molecule.

Then, we quantitatively analyzed the distribution of alkali, urea and water, as shown in Figure 3. We paid specific attention to the hydration shell around the oxygen atoms (we call them Os hereafter) of the glucose rings. They included all of the hydrophilic atoms on the glucose rings.

Because Os are electronegative, the first hydration shell had to be electropositive. There are three kinds of electropositive atom in an alkaline system, i.e., hydrogen atoms of urea and water and cations of alkali. Thus, there were 3 × 4 = 12 kinds of g(r) between

them and the Os. There was only one g(1)(r) larger than 1, i.e., it was between cation and the Os. From the previous study, we knew that cations play a role in breaking the intra- and inter-molecular hydrogen bonds of cellulose. So, we first examined the RDF distribution between the cations and oxygen atoms of trehalose. Figure 4 compares the RDFs of Li+/Na+/K+ ions and O1~O4. Here, we separated Os into two groups. O1 and O2 were in one group, while O3 and O4 were in the other group. We discussed their RDFs with cations, respectively. In group one, both O1 and O2 were inside the glucose rings. g(1)O1-K(r)~5, centered at r1~2.6 Å. This means that the first K<sup>+</sup> shell around O1 was centered about 2.6 Å, where the K+ concentration was five times larger than the average K<sup>+</sup> concentration in the solution. g(1)O1-Na(r)~10, located at r1~2.4 Å, and g(1)O1-Li(r)~20 was at r1~2.0 Å. Thus, smaller cations were more close to O1, and their concentrations were 10 and 20 times their averaged concentration in solution. Similar things happened around O2. In group two, both O3 and O4 were in the periphery of the glucose rings. The spaces around them were, thereby, relatively open. g(1)O-K(r)~1.8, centered at r1~2.8 Å; g(1)O-Na(r)~1.9, centered at r1~2.45 Å; and g(1)O3-Li(r)~1.4, centered at r1~2.1 Å. Larger ions more easily concentrated around them.

**Figure 4.** RDF distributions between the alkali metal ions and the solute oxygen atoms. g(r)s between Li+, Na+, K+ and O1 (**a**), O2 (**b**), O3 (**c**), O4 (**d**) of trehalose.

From these comparisons, it was found that the Li+ ions were closest to the two oxygen atoms inside the glucose ring, i.e., O1 and O2, while larger ions, such as Na+ and K+, were more likely to be enriched near O4. Because the LiOH urea aqueous solution had the most powerful solubility, these comparisons confirmed that the cations in the alkaline solution directly interact with the glucose ring. An interesting result was that K+ ions with a relatively larger ionic radius were more concentrated around oxygen atoms away from the center of the glucose ring. This may explain why the KOH/urea aqueous solution had a better dissolving effect on chitin [18]. We suppose that, in the dissolution process of chitin, breaking the association between the side groups may be crucial, while, in the dissolution process of cellulose, breaking the interaction between the glucose backbones is a prerequisite.

Then, we looked for the components in the second layer of the hydration shell. The first hydration shell was electropositive, so the second shell had to be electronegative. There were four electronegative solvent atoms in the system, i.e., OW, NU, OU and OOH. They themselves or their complexations composed the second hydration layer. We listed all of their g(r) with the Os and compared their structures with gO1-Li(r) in LiOH/urea aqueous solution. Figure 5 assumes that they were isotropically distributed. We added two dash lines to indicate their possible boundary. It is easy to see that all of those g(r), except g(1)OW-O3 /O4(r), were smaller than 1. Therefore, three conclusions can be made, i.e., the second hydration layer almost did not exist; it was anisotropically distributed, or it only existed around O3 and O4 on the periphery of the glucose ring. We checked them one by one.

**Figure 5.** RDF distributions of some solvent atoms, i.e., Li, OW, OU, NU and OOH, around Os atoms in trehalose LiOH urea aqueous solution. (**a**–**d**) is the RDF distribution around O1 to O4, respectively. The dashed lines are used to indicate the boundary of the second hydration shells.

As shown in Figure 3, the distributions of cations and urea around O1 and O2 were very asymmetric. To quantitatively study the asymmetry of cationic and electronegative atoms around O1, we calculated their g(r) as a function of spatial angle. In the calculation, the setting of the coordinate axis was as shown in Figure 1. We first set a cone with an apex angle of 60◦ and the Y coordinate axis as the axis of symmetry. Then, we placed the apex of this cone at O1. Finally, we let the cone take O1 as the center of rotation and rotated it around the *X*-axis and calculated the g(r) changes with the rotation angle. The calculation results showed that g(1)O1-Li(r) reached its maximum value between 90◦ and 180◦, which coincided with the *Z*-axis. Figure 6 shows gO1-Li(r), gO1-OW(r), gO1-OU(r), gO1-NU(r) and gO1-OOH(r) along the *Z*-axis. g(1)O1-OW(r), g(1)O1-OOH(r) and g(1)O1-NU(r) were larger than 1.0. Therefore, electronegative OW, OOH, NU or their complexation composed the second hydration layer.

**Figure 6.** RDF distributions of Li, OW, OU, NU and OOH around O1 atom along *Z*-axis. The dashed lines are used to indicate the boundary of the second hydration layer.

The existence of urea is crucial for dissolving cellulose in green solvent. Previous NMR observations claimed that the amino group of urea forms an electronegative complexation with OH− through hydrogen bonding. Such complexations wrap around cations and eventually form cellulose–NaOH–urea–H2O inclusion complexes (ICs) [5,19]. To clarify this, we listed all of the possible complexations between OH− and urea (Figure 7a). There were four possibilities, i.e., HU as the proton donor to form a hydrogen bond with OW (HB1) and OOH (HB2), NU as the proton acceptor (HB3) and OU as the proton acceptor (HB4). Figure 7b gives the RDFs between OW and NU, OW and HU and HW and NU. We used 2.9 ± 0.3Å as the donor–acceptor distance constraint and linear bond (±20◦) as the angle constraint to define the hydrogen bond. The possibility of forming a hydrogen bond for HB1 and HB3 was 78.5% and 10.9%, respectively. Thus, HU prefers forming a hydrogen bond with OW. Figure 7c shows the RDFs between OOH and NU, OOH and HU and HOH and NU. If we used 2.5 ± 0.4 Å as the donor–acceptor distance constraint and (±20◦) as the angle constrain to define the hydrogen bond, the possibility of forming a hydrogen bond for HB2 was 51.9%. Thus, HU can also form a hydrogen bond with OOH. Finally, Figure 7d is the RDF between OW and OU and HW and OU. They could form a hydrogen bond (HB4), but they were away from the second hydration layer, as shown in Figure 6. Therefore, as a proton donor, the amino group of urea can hydrogen bond with hydroxy group to form complexations. The electronegative complexation forms the second hydration layer.

Temperature effect on dissolution was also investigated. Figure 8 shows the neutron scattering profiles of trehalose in LiOH urea aqueous solutions at 25 ◦C and −5 ◦C and their differences. It can be seen that there was no obvious difference between the two different temperatures. Therefore, low temperature does not change the atomic structure of the solutions. It only increases the stability of the solution, making it difficult for the dissolved cellulose to re-aggregate.

**Figure 7.** Schematic representation of the four important hydrogen bonds that urea forms with water and OH− (**a**). The colors of the four atoms C, H, O and N are cyan, white, red and blue, respectively. RDF distributions between the atoms of urea and H2O/OH− in trehalose LiOH urea aqueous solution (**b**–**d**).

**Figure 8.** Neutron total scattering profiles of trehalose in LiOH/urea aqueous solution at 25 ◦C and −5 ◦C and the difference between them (green lines).
