Structure Development of the Interphase between Drying Cellulose Materials Revealed by In Situ Grazing-Incidence Small-Angle X-ray Scattering

The nano- to microscale structures at the interface between materials can define the macroscopic material properties. These structures are extremely difficult to investigate for complex material systems, such as cellulose-rich materials. The development of new model cellulose materials and measuring techniques has opened new possibilities to resolve this problem. We present a straightforward approach combining micro-focusing grazing-incidence small-angle X-ray scattering and atomic force microscopy (AFM) to investigate the structural rearrangements of cellulose/cellulose interfaces in situ during drying. Based on the results, we propose that molecular interdiffusion and structural rearrangement play a major role in the development of the properties of the cellulose/cellulose interphase; this model is representative of the development of the properties of joint/contact points between macroscopic cellulose fibers.

. Schematic illustration of the preparation of water-swollen cellulose filament by precipitating a cellulose solution into ethanol, then washing with Milli-Q water for 7 days.

Sample system description of the µGISAXS measurements
As illustrated in Figure S4a, before the deposition of the water-swollen gel filament on the dried cellulose thin film, the sample stack can be described as a standard thin film multilayer. The observed peak 2 at = 0.0253 Å −1 in Figure 1g (red curve) can be assigned as the Yoneda peak of the dried cellulose thin film. As the corresponding critical angle for peak 2 ( = 0.0881 o , Figure   S5a After placing the water-swollen cellulose filament onto the thin film ( Figure S4b), the Scattering Length density (SLD) of the gel filament becomes the top subphase, as the x-rays penetrate the near-infinite thickness of the filament from its bottom, which makes it an effective subphase. The consequence is that the profile in Yoneda region changes significantly (Figure 1g), thus, the Yoneda peaks and critical angles need to be calculated relative to this subphase using the equation: As the cellulose content in the filament increases with drying time, it is anticipated that the SLD of the filament should be in between the SLD of water (9.4 × 10 -6 Å -2 ) and the dried thin film (12.5 × 10 -6 Å -2 , calculated from its critical angle 0.0881 o ). Before drying, the cellulose content in the filament is slightly higher than 1.5 wt%. Thus, the SLD of the gel filament should be larger than (9.4×0.985 + 12.5×0.015) 10 -6 Å -2 = 9.45 × 10 -6 Å -2 . Therefore, the SLD of wet thin film vs filament becomes smaller than 3.05 × 10 -6 Å -2 . The critical angle at = 0.04 o observed on Figure   S5b corresponds to SLD of 2.6× 10 -6 Å -2 . Indicating that this small peak is from the wet cellulose thin film. However, it is too weak and disappears during drying, thus, we do not track its change 5 with drying time. Moreover, the SLD of SiO2 layer vs filament coincides with the SLD of the dried thin film, that's the reason why the Yoneda peak 2 is still observed but shifts to lower . The newly appeared peak at = 0.0225 Å −1 (peak 1 in Figure 1g) is most probably due to the diffusion of the water molecules and cellulose molecular chains from the outermost layer of gel filament into the thin film ( Figure S4b), which will cause the dried thin film to swell. This is consistent with the sharp decrease of ∥ at the same drying time in Figure 1f.
After filament drying for 2 hours, a denser and rougher layer is formed at the interface, named the "interphase layer" ( Figure S4c). This is based on the observation that peak 2 (pink and purple curves in Figure 1g) become broader and the intensity-drop shifts to higher comparing to the dry thin film surface. This might be caused by the structural rearrangement of cellulose chains in the interphase layer in the later drying phase.    Figure S5.

µGISAXS data analysis
A Guinier-Porod empirical model 3 was used to fit both vertical and horizontal cuts. This model assumes a characteristic length scale in the system defined as , and a 'dimensionality' parameter ( ) to characterize the shape of the corresponding scattering object. It models the form factor for nonspherical objects with the following functional forms: where is the scattering vector, ( ) is the scattering intensity, is the radius of gyration, is During fitting, the Guinier form is used for ≤ 1 and the Porod form is used for ≥ 1 . The two forms are never used concurrently.
Representative vertical and horizontal cuts of µGISAXS data and the corresponding fitting curves were plotted in Figure S6. It is observed that this model is sufficient to fit both vertical and horizontal cuts of the µGISAXS data. Moreover, as discussed by Lenz et al., 4 fitting while neglecting the Distorted-Wave Born Approximation (DWBA) reflection terms only enables an extraction of an approximate value of . However, it is reliable enough to indicate how the structure changes normal to the surface.
In this Guinier-Porod model, 3,5 Porod exponents less than 3 are for 'mass fractals'. A Porod exponent = 5 3 ⁄ points to scattering from 'fully swollen' polymer chains (in a good solvent) and