Structure-Dynamics-Property Relationships in Silk Fibroin Films Probed via Solid-State NMR

Abstract

Silk fibroin films offer tunable mechanical properties, yet the structural origins of this tunability are incompletely understood. Here, we show how solvent choice and post-treatment govern the mechanics of silk fibroin films by combining solid-state NMR relaxation measurements (¹³C T ₁/T _1ρ; ¹H T ₁/T _1ρ), X-ray diffraction, and TG-DTA analyses. Formic acid-derived films exhibited a 6-fold higher Silk II content and greater backbone constraint (longer ¹³C T _1ρ) than water-derived films, resulting in higher initial tensile strength. Methanol treatment had opposite effects on mechanical strength in formic acid-derived and water-derived films. In the former, it relaxed backbone constraints (shorter ¹³C T _1ρ) without changing the Silk II content, reducing strength. In the latter, it created nanoscale Silk II domains (<5 nm by ¹H T _1ρ), increasing strength. These findings demonstrate that the mechanical response of silk fibroin films is governed not solely by Silk II content but also by the interplay among domain size and chain dynamics.

Introduction

Silk fibroin (SF) is the main structural protein of silk fibers and is a natural polymer material that possesses excellent biocompatibility, mechanical strength, and processability. − The Bombyx mori fibroin consists of a 391 kDa heavy chain (H-chain), a 26 kDa light chain (L-chain), and a glycoprotein known as P25. The primary structure of the H-chain comprises a large repetitive sequence region flanked by nonrepetitive N- and C-terminal domains. This repetitive region is organized into 11 domains, each separated by approximately 43-residue linker sequences containing charged amino acids, and it is built on repeating Gly-X units (where X = Ala, Ser, Tyr, Val, or Thr). − These regular repetitive sequences form β-sheet crystals. Through such molecular design, SF forms a hierarchical structure composed of crystalline and amorphous regions, which imparts the high tensile strength and toughness of natural silk fibers. Regenerated silk materials can be processed into various forms such as films, fibers, and gels, and their crystalline structures and molecular arrangements vary significantly depending on the regeneration conditions. ,− In particular, the choice of solvent used to dissolve silk proteins and the post-treatments, such as annealing with alcohol, water vapor, or heat, are key factors that govern the secondary structure and crystallinity, ultimately determining the physicochemical properties of the regenerated materials. ,−

SF has two distinct structures: the Silk I structure observed in the silk gland and the Silk II structure formed after fiber formation. The Silk I structure consists of repeated type II β-turns, in which intra- and intermolecular hydrogen bonds alternate to form a relatively stable conformation. − In contrast, the Silk II structure is a highly ordered form composed of antiparallel β-sheets and is the stable structure formed in silk fibers. ,− The proportions of these conformations and amorphous regions with random coil conformation have been shown to change depending on the processing conditions, which in turn affect the properties of the resulting material. ,,

The crystalline structure of SF varies with the solvent and processing conditions. Among these solvents, formic acid has attracted considerable attention as an important solvent in silk-based biomaterials research. Compared with films prepared from other solvents, films cast from formic acid solutions exhibit several notable advantages: (1) they become water-insoluble without additional chemical or physical post-treatments, (2) they show high tensile strength after drying, and (3) they readily form highly crystalline Silk II structures. − These features make formic acid-derived films (FA films) particularly attractive for biomedical applications that require strong mechanical performance and long-term stability in aqueous environments. The fact that formic acid promotes β-sheet crystallization and induces the reorganization of intermolecular hydrogen bonds contrasts with conventional regeneration solvents such as water and hexafluoro-2-propanol (HFIP), which usually require post-treatments with methanol or water vapor to achieve comparable insolubility and mechanical strength. , In addition, several studies have reported that simple post-treatments, such as water vapor treatment or alcohol treatment, can further modify and control the mechanical properties of FA films that are already water-insoluble. The ability to tune the mechanical behavior using such simple post-treatments is another practical advantage of using FA-processed SF. However, previous studies indicate that the mechanical properties of FA films are too complex to be explained solely from the content of β-sheet crystals. Rajkhowa et al. reported that the ethanol treatment of FA films increased the Silk II content, as evidenced by infrared (IR) absorption spectra, and enhanced both stress and strain. In contrast, Beena et al. demonstrated that water vapor and methanol treatment caused little change in the secondary structure according to IR spectra, although the stress was increased. Moreover, Um et al. reported that methanol treatment of FA films induced almost no formation of the Silk II structure. Taken together, these findings show that changes in mechanical properties do not necessarily exhibit a simple correlation with changes in the Silk II content. This suggests that, in addition to the amount of β-sheet crystals, factors such as the crystal size and distribution, molecular mobility, and hydration state may play important roles in governing the mechanical properties of FA fibroin films. Thus, although formic acid is frequently used in silk processing, the molecular-level origin of the characteristic mechanical properties of FA films remains insufficiently understood. In particular, there are few systematic studies that compare formic acid-derived and water-derived films under identical post-treatment conditions and evaluate not only the secondary structure but also chain dynamics, domain size, and amount of bound water. In this study, we aimed to clarify the correlation between the structure and mechanical properties of fibroin films prepared from a fibroin solution dissolved in formic acid through molecular-level structural analyses, mainly using solid-state nuclear magnetic resonance (NMR). SF films were prepared from a formic acid solution (FA) and an aqueous solution (W) and further subjected to water vapor treatment (WT) or methanol treatment (MeT) to obtain six types of films. For these samples, quantitative secondary structure analysis was performed using ¹³C DP (direct polarization)-MAS NMR, and the molecular mobility and domain size were evaluated from ¹³C and ¹H relaxation time measurements. The crystallinity and hydration state were analyzed by using X-ray diffraction (XRD) and thermal analysis. Based on the obtained results, the correlations among crystal size, molecular mobility, bound water state, and mechanical properties were examined to understand the structure-property relationship of FA films at the molecular level. This study shows that the mechanical properties of SF films are governed not only by the Silk II content but also by the combined contributions of crystal size, molecular mobility, and bound water state, providing new insight into the structure-property relationship of silk materials.

Materials and Methods

Preparation of SF Films

Cocoons were obtained from B. mori silkworms (Ehime Sansyu, Japan). Raw silk was degummed in a 0.05% (w/v) sodium carbonate solution at 90 °C for 60 min to remove sericin, followed by thorough rinsing with distilled water. The degummed fibers were dissolved in 9 M LiBr at 37 °C with shaking at 200 rpm for 1 h. The resulting solution was dialyzed against distilled water at 10 °C for 3 days using a cellulose dialysis membrane (molecular weight cutoff = 12,000–14,000 Da) to remove the salts, yielding a regenerated aqueous SF solution. The concentration of the prepared SF solution was 2.7–4.0%. The SF aqueous solution was then lyophilized at −30 °C and 30 Pa to obtain an SF sponge. The sponge was dissolved in FA (Wako) to prepare a 3% (w/v) SF-FA solution, which was left to stand for 24 h. Then, 13 mL of this solution was cast into acid-resistant polymethylpentene Petri dishes with a diameter of 7 cm (SUNPLATEC, Japan) and slowly dried at 20 °C and atmospheric pressure under a draft to prepare SF-formic acid films (FA). The FA films were then immersed in 80% (v/v) aqueous methanol for 24 h and then dried under a draft to prepare methanol-treated films (FA-MeT). Water vapor-treated films (FA-WT) were obtained by placing FA films in a closed container with water at the bottom for 3 days under room temperature and atmospheric pressure to provide a high-humidity environment. Water-based SF films (W) were prepared by casting 13 mL of the SF aqueous solution into 7 cm polystyrene dishes (AS ONE, Japan) and slowly drying at room temperature and atmospheric pressure. The treatment processes were identical with those used for FA films. All films were air-dried and then stored in a vacuum desiccator until use. The names, solvents, and treatment conditions of each film are summarized in Table . For ¹³C NMR measurements, biologically ¹³C-labeled cocoons were prepared by feeding silkworm larvae with an artificial diet (Silkmate 2S, NOSAN, Japan) containing uniformly ¹³C-labeled glucose (CIL, US). Each larva was fed 100 mg/day ¹³C glucose mixed with 2 g of the diet from day 3 to day 5 of the fifth instar. Films were prepared from the obtained cocoons following the same procedures as described above.

1. Sample Identification and Preparation Conditions of the SF Films.

sample name	solvent used	post-treatment
FA	formic acid	-
FA-WT	formic acid	water vapor
FA-MeT	formic acid	80% methanol
W	water	-
W-WT	water	water vapor
W-MeT	water	80% methanol

Solid-State NMR Measurements

The secondary structure and molecular dynamics of each film were evaluated by using solid-state NMR spectroscopy. All measurements were performed on an ECA600II spectrometer (JEOL, Japan). Each film sample (∼40 mg) was packed into a 3.2 mm zirconia rotor and measured at an MAS rate of 10 kHz. ¹³C CP (cross-polarization)-MAS NMR spectra were recorded with a recycle delay of 5 s, a ¹H 90° pulse width of 2.6 μs, a contact time of 2 ms, and 1000–10000 scans. During CP, the ¹H and ¹³C RF field strengths were 67 and 46 kHz, respectively. ¹³C DP-MAS NMR spectra were obtained with a recycle delay of 30 s, a ¹³C 90° pulse width of 3.3 μs, and 1500–10000 scans. This recycle delay exceeds five times the longest ¹³C T ₁ value of Ala Cβ among all films (2.81 s), ensuring the quantitative reliability of the spectra. The ¹³C T ₁ measurements were carried out using the Torchia method with a recycle delay of 5 s, a ¹H 90° pulse width of 2.6 μs, a contact time of 2 ms, and delay intervals of 0.01, 0.1, 0.5, 1, 3, 5, 7, 10, and 15 s, with 64 scans. The ¹³C T _1ρ measurements were conducted using the spin-lock CP-MAS method with a recycle delay of 5 s, a ¹H 90° pulse width of 2.6 μs, a contact time of 2 ms, and spin-lock durations of 0.05, 0.1, 0.3, 0.5, 1, 5, 10, and 20 ms with 64 scans. The ¹H T ₁ measurements were performed using the inversion–recovery CP-MAS method with a recycle delay of 5–7.5 s, a ¹H 90° pulse width of 2.6 μs, a contact time of 2 ms, and several delay intervals with 64 scans. The ¹H T _1ρ measurements were obtained using the spin-lock CP-MAS method with a recycle delay of 5–7.5 s, a ¹H 90° pulse width of 2.6 μs (ν₁ = 96 kHz), a contact time of 0.1 ms, and spin-lock durations of 0.01, 0.05, 0.1, 1.0, 5.0, 10, 15, 20, 30, and 50 ms, with 256 scans.

The observed magnetization intensity M(τ) as a function of delay or spin-locking times τ was analyzed using the following Kohlrausch–Williams–Watts (KWW) equation to estimate the relaxation time (T ₁ and T _1ρ):

$$M(\tau )=M(0)\times e^{-{\left[\frac{\tau }{T_1}\right]}^{\beta }}(0<\beta \le 1)$$

Tensile Testing

Tensile tests were conducted using a tabletop testing machine (TCLS-1L, Sanko, Japan). Briefly, each film was cut into strips measuring 30 × 10 mm, and the gauge length was set to 15 mm. The tests were performed at a cross-head speed of 1 mm/min using a 100 N load cell. The tensile breaking stress, tensile breaking strain, and Young’s modulus were calculated from the stress–strain curves obtained. Each sample was measured four times, and average values were used.

Thermogravimetric Analysis

The thermal properties were evaluated using a thermogravimetry/differential thermal analysis instrument (TG/DTA7200, Hitachi High-Tech Analysis Corp., Japan). α-Alumina was used as the reference material, and an alumina crucible was employed as the sample container. The sample mass was set to 10–15 mg, and the measurements were conducted under an O₂ atmosphere using oxygen with a purity higher than 99.5%. The heating rate was set to 5 °C/min from 30 to 300 °C and then to 20 °C/min from 300 to 1000 °C.

XRD Measurements

The crystalline states of the samples were characterized by using an X-ray diffractometer (D8 ADVANCE, Bruker, Germany). Cu Kα radiation (λ = 0.154 nm) was used as the X-ray source. A silicon sample holder, which prevents X-ray reflection, was used to mount the specimens. During the measurement, the samples were rotated at 20 rpm to minimize the effects of orientation and thickness inhomogeneity. The diffraction patterns were recorded over a 2θ range of 4°–36°, with a step size of 0.0196° per step and a counting time of 3 s per step.

Results and Discussion

Mechanical Properties of the SF Films

The tensile breaking stress, tensile breaking strain, and Young’s modulus of the SF films are summarized in Figure a, and the corresponding numerical values are provided in Table S1. The stress–strain curves are shown in Figure b. The maximum tensile breaking stress was 67.8 MPa for the FA film and 53.2 MPa for the W film, indicating that the FA film exhibited about 1.3 times higher strength. Films prepared using formic acid as the solvent are considered to exhibit a higher initial strength because β-sheet crystallization proceeds during solvent removal. When examining the effects of treatment on FA films, the water vapor-treated film (FA-WT) retained about 90% of the maximum tensile breaking stress of the untreated film (61.3 MPa) while exhibiting a 1.4-fold increase in tensile breaking strain. In contrast, the methanol-treated film (FA-MeT) showed a decrease in tensile breaking stress to 44.4 MPa (about 35% lower) and a large reduction in Young’s modulus from 41.6 to 10.9 MPa. These results indicate that water vapor treatment enhances the flexibility of FA films, whereas methanol treatment causes a decrease in stiffness and strength.

1.

Mechanical properties of SF films prepared using formic acid (FA) and water (W) and their post-treated films (FA-WT, FA-MeT, W-WT, and W-MeT). (a) Tensile breaking stress, tensile breaking strain, and Young’s modulus were obtained from the stress–strain curves. In each film, different letters indicate significant differences (Tukey–Kramer test, p < 0.05). (b) Stress–strain curves for each film.

In contrast to the FA films, both water vapor and methanol treatments increased the tensile breaking stress of the W films. The maximum tensile breaking stresses of the W-WT and W-MeT films were 60.9 and 67.0 MPa, respectively, both higher than that of the untreated W film (53.2 MPa). In particular, the W-MeT film exhibited about 1.3 times higher tensile breaking stress than the untreated W film, suggesting that the structure was reinforced by the treatment. The tensile breaking strain also increased after both water vapor and methanol treatments, with the W-WT film showing approximately 1.5 times greater elongation.

Focusing on Young’s modulus, both FA and W films showed a significant decrease after treatment. In the FA films, Young’s modulus decreased from 41.6 MPa for the untreated sample to 22.7 MPa after water vapor treatment and to 10.9 MPa after methanol treatment. Similarly, in the W films, Young’s modulus decreased from 37.7 to 17.1 MPa after water vapor treatment and to 10.5 MPa after methanol treatment. These results indicate that in both systems the treatments made the materials more flexible. This change was considered to result from molecular rearrangement and relaxation at the crystal–amorphous interfaces induced by the treatments.

Molecular Conformation of the SF Films

To clarify the structural characteristics of the SF films, we quantitatively evaluated the conformations of each sample using ¹³C DP-MAS NMR measurements. Figure shows the ¹³C DP-MAS spectra of the films. Peaks corresponding to the Silk II and Silk I/random coil (rc) structures are indicated for Ala Cβ, Ala Cα, and Ser Cβ. In the ¹³C DP-MAS method, direct excitation of ¹³C nuclei without CP provides spectra with a high quantitative reliability. The fractions of the Silk II and Silk I/rc structures were determined by deconvoluting the Ala Cβ peak (Figure ), which sensitively reflects the secondary structure and thus provides reliable quantitative evaluation. The results showed that the Silk II content of the FA film was 64%, about six times higher than that of the W film (11%), indicating that formic acid strongly promotes the crystallization of SF. Regarding the effect of the treatments, the Silk II content in the FA films remained almost constant at (64% for FA, 66% for FA-WT, and 62% for FA-MeT, indicating that high crystallinity was maintained. In contrast, in the W films, the Silk II content remained low in the W (11%) and W-WT (14%) samples but increased to 37% in the W-MeT sample. These results show that secondary structural changes induced by treatment are minor in the FA films, whereas the crystallization behavior of the W films varies significantly depending on the treatment conditions.

2.

¹³C DP-MAS NMR spectra of the SF films. Peaks corresponding to Silk II and Silk I/rc structures are indicated. For Ala Cβ, Silk II and Silk I/rc signals appear at 20.5 and 17.2 ppm, respectively. For Ala Cα, Silk II and Silk I/rc signals appear at 49.6 and 51.2 ppm, respectively. For Ser Cβ, Silk II and Silk I/rc signals appear at 64.2 and 61.1 ppm, respectively.

3.

Deconvolution of the Ala Cβ region in the ¹³C DP-MAS spectra used to quantify the fractions of Silk II and Silk I/rc structures in the SF films. The Silk II component is shown in blue and the Silk I/rc component in orange, together with the percentage fractions of each component.

The same tendency was also confirmed by the XRD results (Figure ). Peak assignments were made based on previous studies, where the diffraction at 2θ = 20° was assigned to Silk I and those at 2θ = 19° and 21° to Silk II. , In the FA film, distinct peaks corresponding to Silk II (2θ = 19° and 21°) were observed, indicating that Silk II crystals were predominantly formed. In the FA-WT film, the intensity of the Silk I peak at 20° increased in addition to these Silk II peaks, suggesting that Silk I crystals were newly formed or increased by water vapor treatment. In contrast, the FA-MeT film exhibited very weak peaks, and clear crystallinity was not detected by XRD, although ¹³C NMR showed Silk II content comparable to that of the other FA films. These results suggest that the FA-MeT film contains microcrystals too small to be detected by XRD and that solid-state NMR can detect microcrystals on a nanometer level. ,

4.

X-ray diffraction patterns of the SF films, with peaks at 2θ ≈ 20° assigned to Silk I and peaks at 2θ ≈ 19° and 21° assigned to Silk II.

Collectively, these results indicate that in the FA films more of the Silk II structure is initially formed due to the solvent effect, and subsequent structural changes are governed by the formation of Silk I crystals induced by water vapor treatment and by the microcrystallization of Silk II induced by methanol treatment. In contrast, in the W films, crystallization is promoted by the treatments, but the resulting crystal structures differed depending on the treatment method: the ¹³C DP-MAS results suggest that Silk I is mainly formed via water vapor treatment, whereas Silk II is predominantly formed via methanol treatment.

Molecular Motion of the SF Films

The structural differences among the samples partially explain the variations in mechanical properties as shown in Figure ; however, changes in Silk II content alone cannot account for all of the observed differences. For example, although the FA and FA-MeT films exhibited nearly the same Silk II content, their crystallinity differed significantly according to the XRD data, and the stress decreased by 35%. In contrast, the FA and FA-WT films showed almost no change in Silk II content, yet the strain increased by approximately 1.4 times. These results indicate that the changes in mechanical properties are influenced not only by the quantitative aspects of the crystalline structure but also by dynamic factors such as molecular chain mobility and domain size. To evaluate the molecular dynamics of each sample, we measured the ¹³C and ¹H relaxation times. The ¹³C T ₁ is sensitive to very fast motions on the MHz time scale and reflects local fluctuations of molecular chains, whereas the ¹³C T _1ρ corresponds to motions around 96 kHz and reflects segmental motion of the polymer backbone. In addition, the ¹H relaxation times (T ₁ and T _1ρ) can be used to estimate the domain size through spin diffusion. By combining these measurements, the treatment-induced changes in molecular dynamics were analyzed to clarify the relationship between the structures and mechanical properties of the SF samples.

Evaluation of Local Molecular Motion by ¹³C T₁

The experimental and fitted T ₁C plots of the Silk II and Silk I/rc components at Ala Cβ for each film are shown in Figure S1, and the T ₁C values of Ala Cβ are summarized in Figure . We experimentally confirmed that the T ₁C values of Ala Cβ increased with the temperature, indicating that higher T ₁C values correspond to higher molecular mobilities for Ala Cβ in the samples used in this study. Because Silk II has a stable structure comprising stacked antiparallel β-sheets and Silk I includes metastable turn conformations, the molecular motion of fibroin chains is expected to be more restricted in Silk II than in Silk I/rc.

5.

T ₁C of the Ala Cβ for Silk II and Silk I/rc components in the SF films. Higher T ₁C values correspond to faster local molecular motion. Error bars (standard deviation σ) are smaller than the symbols.

The T ₁C of Silk I/rc was larger than that of Silk II in both the FA and FA-MeT films, indicating higher local molecular motion for the Silk I/rc component. In contrast, in the FA-WT film, the two components showed nearly identical T ₁C values; compared with the FA film, the T ₁C of Silk I/rc decreased while that of Silk II increased. These results suggest that water vapor treatment decreased the mobility of the Silk I/rc component and increased that of Silk II. Previous studies reported the Silk I crystals form upon water vapor treatment. Therefore, the decrease in the molecular mobility of the Silk I/rc component can be attributed to the formation of Silk I and the resulting alignment or orientation of molecular chains, which reduced the local flexibility. This interpretation is consistent with the XRD results, where the intensity of the Silk I peak increased after treatment. The increased local mobility of Silk II is likely due to a higher amount of bound water around the Silk II domains after the water vapor treatment. Bound water is known to act as a plasticizer in SF and other polymers by weakening intermolecular hydrogen bonds and increasing chain flexibility. , In the FA-WT film, the larger low-temperature weight loss in the TG-DTA data indicates an increased amount of bound water. Therefore, it is reasonable to assume that the additional bound water partially replaces intermolecular hydrogen bonds with hydrogen bonds to adsorbed water, thereby enhancing local molecular motion. In contrast, the T ₁C values of the W films decreased after treatment, consistent with the increase in Silk II content, indicating that the treatments promoted crystallization and suppressed fast molecular motions.

Evaluation of the Molecular Dynamics of the Fibroin Chain Backbone Using ¹³C T_1ρ

Next, the ¹³C T _1ρ values of Ala Cα and the CO of the Ala and Gly residues were evaluated for each film. T _1ρC is sensitive to molecular motions near the spin-lock frequency (56 kHz region) and reflects the segmental motion of the polymer backbone on this time scale. Figure S2 shows the decay curves of Ala Cα, Cβ, and CO for the FA-WT film, and Figure summarizes the T _1ρC values of all films. No significant differences were observed between the conformations (Silk I/rc and Silk II) in any of the films, but clear differences were found among the films. For instance, the T _1ρC values of Ala Cα and CO showed the following trend: FA-WT > FA > FA-MeT ≈ W ≈ W – WT ≈ W-MeT. According to previous studies of the relationship between T _1ρC and molecular mobility, these results indicate that the backbone mobility was lowest in FA-WT and relatively higher in FA-MeT and W films.

6.

T _1ρC relaxation times of the Ala Cα and CO for the SF films. T _1ρC provides information on the segmental dynamics of the fibroin backbone. Error bars (standard deviation σ) are smaller than the symbols.

The relationship between T _1ρC and molecular mobility of polymer chains has been reported in several previous studies. − Shefer et al. compared various glassy polymers and found that shorter T _1ρC values correspond to higher average mobilities. They also reported that the annealing of polymers reduces segmental motion by increasing local chain packing density and decreasing free volume. Yamaguchi et al. investigated the temperature dependence of T _1ρC for each carbon atom in poly­(γ-benzyl l -glutamate), which forms an α-helical structure. They found that side-chain carbons showed a minimum T _1ρC value, whereas the backbone Cα did not exhibit a minimum even at high temperatures but rather showed a continuous decrease in T _1ρC. This result indicated that the correlation time (τ_c) of the backbone Cα is higher than 1/ω₁C, meaning that backbone motion is slower than side-chain motion. Kishore et al. measured the relaxation times for two types of spider dragline silks and cocoons. They reported that the T _1ρC of the carbonyl carbon in Argiope keyserlingi dragline silk was shorter than that in Nephila edulis dragline silk, suggesting that A. keyserlingi silk exhibits higher molecular mobility on slower time scales and longer length scales. From these studies, it can be concluded that at room temperature, the molecular motions observed through the T _1ρC values of the polymer backbones correspond to a motional regime where the τ_c is longer than 1/ω₁C and shorter T _1ρC values indicate higher molecular mobility. Given that Silk II comprises highly ordered antiparallel β-sheets with restricted chain mobility, the backbone carbons in silk fibroin are expected to reside in the slow-motion regime (τ_c > 1/ω₁). Furthermore, the T _1ρC values of the Silk I/rc components showed trends consistent with those of Silk II across all films (Figure ), suggesting that the backbone carbons of all conformational components reside in the same motional regime, where shorter T _1ρC values are interpreted as indicating higher segmental mobility of the backbone in the slow-motion regime (τ_c > 1/ω₁).

Comparing the investigated films, it was found that the backbone mobility was lowest in the FA-WT film, followed by the FA film, while the FA-MeT and W films exhibited higher mobility. The decrease in mobility from FA to FA-WT can be attributed to the formation of Silk I crystals in the amorphous regions, where increased crystallinity suppresses the segmental motion of the backbone. In contrast, in the FA-MeT film, the T _1ρC became shorter despite the Silk II content remaining almost unchanged, suggesting that the molecular constraints were relaxed and the backbone motion was activated due to the refinement of Silk II crystals.

In the W films, the change in T _1ρC caused by the treatments was small and no significant difference in backbone mobility was observed. Previous studies using dynamic light scattering measurements have reported that SF molecules in formic acid exist in a compact conformation, whereas those in an aqueous solution are more extended. This suggests that the fibroin chains are more entangled in W films than in FA films, making it difficult for molecular reorientation or packing rearrangement to occur during treatment. As a result, little change in backbone mobility was induced by the treatments in the W films.

Estimation of Domain Size through the ¹H Relaxation Time

The domain size in each film was estimated from solid-state ¹H relaxation time measurements. The ¹H T ₁ values of the Silk I and II phases are usually different. In the solid state, however, ¹H spin diffusion occurs, and relaxation is averaged among protons within the diffusion distance. When the domain size of coexisting Silk I and II becomes smaller than several tens of nanometers, each T ₁H value of two phases becomes identical. A similar criterion holds for T _1ρH on the several-nanometer scale. The spin diffusion distance depends on the ¹H T ₁ values, the ¹H spin diffusion coefficient, and the assumed phase model. The maximum diffusive path length is reported to be approximately 20–50 nm for the spin–lattice relaxation time (T ₁H) and 2–5 nm for the spin–lattice relaxation time in the rotating frame (T _1ρH) assuming a typical ¹H spin diffusion coefficient of 500–1000 nm²/s and a three-dimensional diffusion model; however, these values should be regarded as only rough estimates. When domains with sizes of approximately 5–20 nm exist, T ₁H becomes averaged, whereas T _1ρH does not. −

Figure shows the T ₁H and T _1ρH values of the Ala Cβ. For all films, the T ₁H values of Ala Cβ were nearly identical between the Silk II and Silk I/rc components, indicating that relaxation is averaged within the spin diffusion distance (20–50 nm). The same trend was observed for other carbon sites, suggesting that Silk II and Silk I/rc coexisted within this spatial scale in all films. In contrast, distinct differences were observed among the films in the T _1ρH values. In five of the films (except for W-MeT), the T _1ρH values differed between the Silk II and Silk I/rc components, indicating that the domain sizes are within the range of 5–20 nm. In the W-MeT film, however, the T _1ρH values of the two components were nearly identical, indicating efficient spin diffusion that leads to averaged relaxation behavior on the T _1ρH time scale.

7.

T ₁H and T _1ρH relaxation times of the Ala Cβ for the SF films. Differences between the T ₁H and T _1ρH values reflect domain sizes probed using ¹H spin diffusion. Error bars (standard deviation σ) are smaller than the symbols.

Therefore, the domain size in W-MeT is smaller than that on this length scale. These results indicate that in most of the FA and W films Silk II and Silk I/rc exist as separate domains on the 5–20 nm scale, whereas in the W-MeT film, the two structures coexist at the nanometer scale, forming small crystalline domains. It should be noted that these domain size estimations are based solely on ¹H spin diffusion arguments and should be regarded as approximate.

Interaction with Water Molecules

The molecular dynamics of fibroin chains are considered to be affected by both the amount and properties of solvent molecules remaining in the films. Therefore, TG-DTA measurements were performed to evaluate the weight loss associated with residual solvent in each film (Figure ). In all films, a decrease in mass was observed in the TG curves, and an endothermic peak appeared in the DTA curves at around 50–80 °C, indicating desorption of residual solvent molecules. The weight loss was approximately 7% in all cases.

8.

(a) TG curves in the low-temperature region (room temperature to 300 °C). (b) TG curves over the full temperature range (room temperature to 700 °C). (c) DTA curves in the low-temperature region (room temperature to 300 °C). (d) DTA curves over the full temperature range (room temperature to 700 °C).

Since the W film was prepared using only water as the solvent, the desorbed molecules in this temperature range can be attributed to water. The FA film also exhibited weight loss in the same temperature range, suggesting that the desorbed molecules were likewise water. Comparison of the desorption temperatures (Table ) showed that both FA and W films released water at approximately 52 °C, whereas the temperatures for FA-WT and FA-MeT shifted to higher values, at 82.6 and 68.6 °C, respectively. These results indicate that the amount of strongly retained water differs among the silk fibroin samples; the order of the desorption temperatures was FA-WT > FA-MeT > FA = W (Figure a and Table ), although the total amount of adsorbed and retained water is similar for all samples. Water merely adsorbed on the surface would be easily removed at temperatures much lower than 100 °C, whereas more strongly retained water requires a higher temperature to evaporate. Accordingly, FA-WT contains more strongly retained water than FA-MeT, while a larger amount of water merely adsorbed on the surface of silk fibroin is present in the FA and W samples.

2. Desorption Temperatures of Water Molecules Released from the SF Films Obtained Using TG-DTA Analysis.

	temperature[°C]
FA	51.7
FA-WT	82.6
FA-MeT	68.6
W	51.7

Correlation between Mechanical Properties, Conformation, and Molecular Dynamics

The relationships among the structure, molecular dynamics, and mechanical properties of the films were summarized and are schematically illustrated in Figure . Comparing the untreated FA and W films revealed clear differences in the crystalline structure, backbone dynamics, and mechanical properties. First, the Silk II content of the FA film (64%) was approximately six times higher than that of the W film (11%), reflecting the strong crystallization-promoting effect of formic acid during solvent evaporation. Second, this higher crystallinity was accompanied by greater backbone constraint, as evidenced by the longer T _1ρC values of the FA film (∼4 s) compared with those of the W film (∼2 s). Third, these structural and dynamic differences collectively account for the higher tensile breaking stress of the FA film (67.8 MPa) relative to that of the W film (53.2 MPa). These results indicate that solvent choice directly governs not only the crystalline structure but also the backbone dynamics of the resulting films, which both contribute to the mechanical performance.

9.

Schematic illustration of the effects of solvent choice (formic acid or water) and post-treatment (water vapor or methanol) on the Silk I/II content, crystal size, and domain distribution in silk fibroin films.

Next, water vapor treatment (FA-WT) significantly increased the tensile breaking strain while maintaining the tensile breaking stress on the FA film, although the Silk II fraction remained nearly constant. This can be explained by the formation of Silk I, as indicated by the XRD and T ₁C measurements, and by the increase in bound water, as revealed by the TG-DTA analysis. Water vapor treatment induced the rearrangement of part of the amorphous molecular chains into Silk I crystals and the incorporation of bound water between molecular chains. These contributed to the increased flexibility of the film without disrupting the Silk II crystalline network. Zhang et al. compared the properties of silk films stretched under humid conditions with those treated with methanol and reported that bound water plays a crucial role in enhancing flexibility, supporting the present interpretation.

In contrast, in the FA-MeT film, the tensile breaking stress decreased by approximately 35%, from 67.8 to 44.4 MPa, despite little change in the Silk II content. This result suggests that methanol treatment caused refinement of Silk II crystals. The absence of clear XRD peaks in the FA-MeT film suggests that the crystallites are too small to produce detectable diffraction, indicating refinement to the nanometer scale. Such local and heterogeneous refinement likely caused inefficient stress transfer between crystals, resulting in a reduction in the overall stress of the film.

Meanwhile, both water vapor and methanol treatments increased both the stress and strain in the W films. Although the Silk II content slightly decreased in these films after water vapor treatment, the formation of Silk I promoted molecular chain alignment and the resulting increase in crystallinity contributed to higher strength. In contrast, methanol treatment markedly increased the Silk II content, which led to higher stress. Furthermore, the identical T _1ρH values for Silk II and Silk I/rc in the W-MeT film indicated that the two structures coexist within regions smaller than the characteristic ¹H T _1ρ spin diffusion length scale. This nanoscale structure, in which fine crystals are homogeneously dispersed within the matrix, is considered to facilitate efficient stress transfer and thereby enhance the tensile strength. This tendency is consistent with previous reports showing that smaller crystal sizes in silk materials lead to higher stress. This size range is comparable to the dimensions of β-sheet nanocrystals reported in B. mori silk fibroin (∼2–4 nm in width and thickness), supporting the interpretation that the domains detected by ¹H T ₁ and T _1ρ are on the scale of β-sheet crystals.

Collectively, these results indicate that in formic acid-derived films, the degree of Silk II crystallization primarily governs the tensile strength, while Silk I formation and crystal refinement induced by the treatments control fibroin flexibility and stiffness. In contrast, in water-derived films, recrystallization and further domain refinement during treatment contribute to the improvement of the mechanical properties of SF. Thus, the interplay among the Silk I/II structures, crystal size, bound water, and molecular chain dynamics determines the mechanical response of SF films.

Conclusion

This study elucidated the correlation between the structure and mechanical properties of SF films prepared from formic acid and aqueous solutions through molecular-level analyses mainly using solid-state NMR, together with XRD and thermal analysis. The results revealed that the mechanical response of the films cannot be explained solely by the Silk II content but is determined by the interplay among Silk I content, crystal size, molecular dynamics of fibroin chains, and bound water.

Formic acid-derived films showed a high Silk II fraction upon formation and therefore high tensile strength. Water vapor treatment introduced bound water between the fibroin chains and increased Silk I formation; in both solvent systems, this led to greater extensibility, while largely maintaining the Silk II network. Methanol treatment promoted Silk II formation, but its effect was strongly dependent on the initial solvent system. In the FA films, where Silk II was already abundant, methanol did not further increase the Silk II content but instead produced heterogeneous crystal refinement at the nanometer scale. This microstructural change relaxed constraints on local regions and caused inefficient stress transfer between crystallites, lowering the overall stiffness and strength of the material. Meanwhile, in the W films, methanol produced small Silk II crystals more uniformly dispersed within the matrix, thereby increasing the strength.

Solid-state NMR relaxation provided sensitive probes of backbone dynamics (T _1ρ) and domain size (via ¹H spin diffusion), allowing us to connect molecular mobility and nanoscale organization with macroscopic mechanics. Overall, our results suggest that the mechanical behavior of regenerated silk is governed not only by the amounts of Silk I/II crystals but also by crystal size, the uniformity of domain distribution, chain dynamics, and interactions with water. These insights obtained from formic acid- and water-derived silk fibroin films provide a practical basis for understanding how solvent choice and post-treatment such as water vapor and methanol treatments govern the flexibility and strength of regenerated silk films.

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acsomega.6c01268.

• Tensile properties and¹³C T₁ and T_1ρ relaxation curves (PDF)

The authors declare no competing financial interest.