Structure and Sulfur: Tuning the Viscoelastic and Surface Properties of Natural Keratin Fibers

Abstract

Natural keratin fibers, such as wool, possess a complex hierarchical structure that governs their mechanical properties and surface energy. However, the extent to which these characteristics are influenced by combined contributions of structural variations (e.g., fiber diameter, intermediate filament (IF) packing) and chemical composition (e.g., disulfide bond density) remains poorly understood. In this study, we investigate wool fibers from five sheep breeds (Merino, Polwarth, Cheviot, Eider, and Devon) to elucidate how these factors influence viscoelasticity and surface interactions. Using a multimodal approach integrating interfacial and bulk characterization methods, including inverse gas chromatography (IGC), atomic force microscopy-infrared spectroscopy (AFM-IR), X-ray photoelectron spectroscopy (XPS), uniaxial tensile testing, and synchrotron small-angle X-ray scattering (SAXS), we show that the nanometer-thick 18-methyleicosanoic acid (18-MEA) layer is consistently present across all wool types and plays a key role in governing hydrophobicity and surface heterogeneity. A controlled isothermal treatment at 200 °C, designed to cleave disulfide bonds, results in a nearly 40% reduction in specific surface area across all fiber types, accompanied by a significant decrease in tensile strength and 80% reduction in elongation at break for Merino and Devon wool, but limited influence on the mechanical properties of Eider fibers. Furthermore, rate-dependent tensile testing within the elastic regime reveals distinct viscoelastic responses among the fiber types, suggesting that the sulfur-rich protein matrix surrounding IFs and its structure contribute actively to stress partitioning. Altogether, when combined with conclusions from SAXS measurements of IF spacing, our work offers compelling insights into the role of the keratin-associated protein (KAP) matrix in shaping wool fiber mechanics. Differences in mechanical behavior among wool types, despite similar IF spacing or sulfur content, highlight the importance of matrix composition and cross-linking density, suggesting that the molecular architecture of the KAP network may be a dominant factor in determining fiber performance.

Introduction

Natural fibers are characterized by a hierarchical structure, originating from sugar-, mineral- or protein-based building blocks that assemble into complex nanocomposites. − These biologically derived materials often surpass synthetic counterparts in adaptability, sustainability, and multifunctionality, owing to their evolutionarily refined architectures. Although their properties are intricately linked to specific biological functions, this specialization can limit their direct applicability in engineered systems. As a result, while natural fibers play a vital role in daily life, from paper and packaging to construction materials and textiles, their properties must often be artificially tailored using engineering strategies such as surface treatments and synthetic polymeric coatings to expand their utility. This reliance on external sizing agents and surface finishes underscores the limitations in our current understanding of how intrinsic fiber architecture can be tuned directly to govern performance. It highlights the need for a deeper, mechanistic understanding of the hierarchical organization and intrinsic structure–property relationships in natural fibers, with the goal of enabling property tuning through molecular and structural design rather than through the addition of extraneous materials.

Such insight is particularly critical for α-keratin fibers; in terms of structural complexity and natural abundance, proteinaceous keratin fibers represent a promising class of biobased materials with significant untapped potential. Mammalian hair fibers such as human hair and sheep wool are among the most prominent examples. Their widespread availability and renewable nature make keratin-based fibers attractive candidates for sustainable material development. However, the structure of mammalian hair fibers is finely tuned to their biological role as a protective and thermally insulating outer layer and as a result, their mechanical properties can vary significantly across species, breeds, and individuals, and are not inherently optimized for industrial or load-bearing performance. Down feathers are a prime example of a keratinous material with exceptional thermal insulation properties. However, their low density and structural fragility limit the application of down to fillers and cushioning materials. This gap between evolutionary design and desired properties presents challenges in applications such as textiles, where functional demands often exceed the fiber’s native capabilities, thereby limiting their broader technological potential. A familiar example is the unwanted shrinkage of untreated wool textiles during washing. Wool fibers naturally tend to shrink and felt together during washing due to their natural crimp and scaly surface structure, which promotes irreversible fiber entanglement. To make wool machine-washable, manufacturers use shrink-proofing treatments that either degrade these surface scales or coat them with polymers. However, these treatments often significantly affect wool hydrophobicity, underscoring the importance of growing our understanding of the relationship between fiber structure and properties.

The structure of α-keratin fibers is highly complex and hierarchical and has been widely described as a nanocomposite consisting of an outermost layer, the cuticle, and an inner core, the cortex (Figure ). The cuticle (∼10 wt %) protects the inner part of the fiber against the environment and ensures structural integrity. It consists of thin, overlapping cuticle scales (Figure a) whose morphology correlates with fiber diameter and is affected by weathering and friction. Typically ∼ 500 nm thick (Figure b), the cuticle consists of multiple layers that differ in both thickness and chemical composition: the outermost 18-methyleicosanoic acid (18-MEA) layer, known as the F-layer, is covalently bound to the epicuticle via sulfoesters (Figure c); beneath it lies the disulfide-rich exocuticle, followed by the less cross-linked endocuticle. Rich in cystine, the cuticle contributes to chemical stability and mechanical strength through disulfide bond formation. Furthermore, because it is located on the exterior of keratinous fibers, the cuticle’s structure and composition can be readily modified by chemical and thermal treatments, and its surface chemistry is highly dependent on environmental and processing conditions. , The 18-MEA layer imparts hydrophobicity and, despite being only a few nanometers thick, plays a key role in surface interactions, in stark contrast to the hydrophilic cortex beneath. The cortex makes up the majority of a wool fiber (∼90 wt %) and is believed to be the primary contributor to its mechanical properties. It consists of cortical cells (ortho- (O), para- (P), and meso- (M) types), each with subtle chemical and structural differences. Within these cells, elongated macrofibrils are composed of intermediate filaments (IFs) with varying packing geometries, embedded in a matrix dominated by sulfur-rich keratin-associated proteins (KAPs). IFs form through the ordered assembly of α-helical keratin heterodimers and tetramers. The matrix, lacking long-range order, contains the highest concentration of disulfide bonds and swells significantly when hydrated. As IFs are mostly crystalline, moisture interactions are largely governed by the matrix, which has higher free volume. In the hydrated state, the sulfur-rich matrix behaves like a cross-linked polymer gel earning it a description as an amorphous region.

1.

(a) Representative SEM images of the cuticle surface of various wool fibers, alongside optical micrographs of typical fiber cross-sections for Merino, Eider, and Devon (scale bars = 20 μm). (b) TEM image of a Devon fiber cross-section highlighting the cuticle and cortex. (c) Schematic illustration of the wool fiber structure, emphasizing the nanometer-thick hydrophobic 18-MEA layer with its chemical structure, and a simplified representation of the fiber as a skin–core system, where the cortex consists of an intermediate filament-reinforced matrix.

This structural organization has led to the widely accepted model of keratin fibers as fiber-reinforced composites, where IFs serve as the reinforcing phase embedded within a gel-like KAP matrix characterized by a high density of disulfide bonds. In this two-component framework, the α-helical IFs are aligned along the fiber axis, maintaining structural integrity, while the matrix acts as the amorphous, load-transferring phase. This model has been instrumental in explaining the mechanical behavior of keratin fibers, including extension, bending, torsion, and swelling. To probe these properties, various experimental techniques have been used, − with uniaxial tensile testing being the most common due to its simplicity. Keratin fibers typically exhibit high extensibility, with a breaking strain exceeding 40%, which increases further in the wet state. Feughelman used uniaxial tensile testing with the fiber-reinforced model to attribute the ″Hookean″ region of the stress–strain curve to the stretching of IF α-helices and hydrogen bonding in the matrix. Further extension induces a reversible α-to-β transformation in IFs. Reported results of the Young’s modulus of human hair range from 2 to 7.5 GPa, while recent measurements across 16 wool types yielded similar values (3–7 GPa) at 50% relative humidity and 23 °C. Keratin fibers also display pronounced viscoelasticity, with strong strain-rate sensitivity observed in creep, relaxation, and rate-dependent tests. ,− Such fiber-reinforced model has also been used to explain the plasticizing effect of water, which weakens the matrix while leaving the crystalline IFs largely unaffected throughout the load–extension curve. ,,,

However, a fiber-reinforced composite model does not account for interfacial behavior, as it overlooks the cuticle, nor does it capture key structural variations across hierarchical length scales. Fiber diameter, for instance, significantly influences bending and torsion, suggesting a more prominent role for the cuticle in bulk mechanical properties. Additionally, it does not capture variations in cortical structure and IF alignment inherent to various wool fibers. Transmission electron microscopy (TEM) studies of Merino, Polwarth, Cheviot, Eider, and Devon fibers revealed breed-specific nanoscale architectures. Fine fibers like Merino and Polwarth contain closely packed, aligned IFs within a high proportion of O-cortex cells (∼500 nm), resulting in a bilateral cortical arrangement, which stabilizes high crimp. In contrast, coarser fibers from Cheviot, Eider, and Devon exhibit reduced O-cortex content, smaller cortical diameters (∼260–280 nm), and lobulated structures with wavy lamellar IF arrangements and lower crimp. Despite these differences, IF center-to-center spacing remains consistent across breeds (approximately 10 nm in M- and O-cortex, and 11 nm in P-cortex). Moreover, the assumption of a fully amorphous matrix has been challenged. As early as the 1960s, Crewther proposed a structured matrix composed of ∼ 2 nm globules based on mechanical studies of reduced and alkylated fibers. More recent TEM analyses support this view, revealing a grain-like matrix organization with domains ranging from 2–4 nm.

In this work, we investigate how IF organization and disulfide cross-linking govern viscoelastic behavior and surface interactions in keratin fibers. We provide new insights into their hierarchical mechanics and interfacial properties beyond the conventional fiber-reinforced composite model. Utilizing wool fibers, we aim to deepen the understanding of surface and mechanical characteristics independently and in their interactions with each other. Our central hypothesis is that the viscoelastic behavior of wool fibers is primarily influenced by the distribution and interaction of the KAP matrix with IFs, and that surface properties (particularly the presence and integrity of the 18-MEA layer) modulate interfacial interactions, indirectly affecting mechanical performance. We expect that thermal treatment targeted at disulfide bond reorganization will alter both the surface chemistry and the internal matrix structure, and that these changes will manifest in measurable differences in viscoelastic and interfacial properties.

To test these hypotheses, we selected five sheep wool types (Merino, Polwarth, Cheviot, Eider, and Devon – sorted by increasing fiber diameter) chosen for their well-characterized morphological and structural differences, while maintaining a consistent elemental composition and sulfur content. A summary of these characteristics is provided in Table S1, Section S1 of the Supporting Information (SI). We employ a combination of surface-sensitive techniques – atomic force microscopy (AFM), inverse gas chromatography (IGC), X-ray photoelectron spectroscopy (XPS), and AFM-based infrared spectroscopy (AFM-IR) – to characterize the surface morphology, chemistry, and heterogeneity of the wool fibers. This includes investigations on the persistence and thermal response of the 18-MEA layer. Bulk mechanical properties are evaluated through tensile testing before and after thermal treatment. By varying the strain rate within the elastic regime, we directly assess viscoelastic behavior and attempt to correlate it with matrix content and organization. To probe the bulk structure, we use synchrotron small-angle X-ray scattering (SAXS), analyzed using an approach based on Porod’s law , as well as a form-free Monte Carlo approach, − to assess IF packing and infer matrix distribution within the cortex. These structural insights are complemented by bulk compositional characterization methods, including Fourier-transform infrared spectroscopy (FT-IR) and thermogravimetric analysis (TGA). Our integrated approach allows exploration of how hierarchical structure (from molecular cross-linking to mesoscale IF packing) affects macroscopic mechanical response, and enables us to establish relationships between surface (cuticle) and bulk (cortex) properties.

Results and Discussion

Surface Morphology and Chemistry of the Wool Cuticle

The keratin fiber surface is dominated by overlapping scales arranged in a characteristic roof-shingle pattern (Figure a). The surface properties of the fiber are strongly influenced by the nanometer-thick layer of 18-MEA (Figure c), a covalently bound fatty acid responsible for the hydrophobic properties of the outer cuticle. We first studied the impact of topography and composition on surface energy for the five representative wool types – Merino, Polwarth, Cheviot, Eider, and Devon – by combining IGC, XPS, and AFM-IR (Figure ).

2.

(a) Octane isotherms (n = 2) of the five wool types measured at 30 °C and 0% relative humidity (RH). (b) Specific surface areas (SSAs) calculated from BET analysis of the octane isotherms. (c) Representative dispersive surface energy (γ _D ) profiles plotted as a function of surface coverage, with corresponding exponential decay fits. (d) γ _D distribution profiles illustrating surface heterogeneity across the wool types. (e) XPS C 1s spectra of the fiber surfaces for Merino, Eider, and Devon. (f) AFM-IR spectra collected from three distinct surface locations on a Devon fiber, showing the presence of the 18-MEA layer covalently bound via a sulfoester linkage. (g) AFM topography images of the Devon fiber surface at multiple positions, highlighting variations in scale structure and surface roughness.

IGC was first used to assess the surface adsorption behavior of the fibers (Figure a-d). In line with previous studies, octane was used to obtain Brunauer–Emmett–Teller (BET) isotherms (Figure a), all of which exhibited linear behavior. BET analysis enables the calculation of the specific surface area (SSA) for the different fiber types (Figure b), with values ranging from ∼0.4 m²·g^–1 to ∼0.8 m²·g^–1. These values are in agreement with other reports. , Although some studies have reported significantly higher SSA for Merino (2.24 m²·g^–1), such discrepancies may arise from differences in fiber sourcing, processing, or measurement conditions. Here, repeated measurements yielded highly consistent SSA values across all samples, supporting the robustness of our methodology. Finer fibers, such as Merino and Polwarth, exhibit higher SSA values due to the greater number of individual fiber surfaces exposed per unit mass. Overall, SSA decreases with increasing fiber diameter; however, notable deviations from the expected trend based on a smooth, solid fiber model suggest a more complex influence of surface morphology and potential internal porosity. For fibers with a solid circular cross-section, SSA is theoretically given by the relationship:

$$SS{A}_f=\frac{4}{\rho D_f}$$ 1

where ρ is the density of the polymer and D _f is the fiber diameter. According to this model and considering no significant change in ρ, Devon fibers, whose diameters are nearly five times larger than those of Merino, should exhibit SSA values one-fifth as large. Instead, they show only a 2-fold reduction, indicating that additional structural features may contribute to the accessible surface area. Devon fibers display a high linear density of scales and contain a central medulla (Figure a), both of which may increase SSA. Eider fibers also deviate from the expected trend, exhibiting higher SSA values than Cheviot and Devon despite lacking a medulla, suggesting that surface roughness, scale morphology, or density variations may also play a role.

These deviations from the theoretical SSA model suggest that surface chemistry and structural complexity may also play a role in fiber–adsorbate interactions. Solid surfaces often exhibit structural and chemical organization distinct from that of the bulk material. The degree of geometrical disorder and variation in chemical composition at the surface plays a critical role in governing adsorption behavior. As the surface becomes increasingly differentiated into discrete adsorption sites (defined as local minima of the adsorption potential), the distribution of adsorption energies broadens. This variation is referred to as surface heterogeneity, and reflects the complexity of interactions occurring at the fiber interface. , To investigate these surface characteristics, IGC was also employed to determine the dispersive surface energy (γ _D ) profiles of the wool fibers (Figure c,d). Specifically, γ _D was plotted as a function of surface coverage (n/n ₀), allowing estimation of γ _D at 0% and 100% surface coverage (corresponding to infinite dilution and monolayer coverage, respectively). At infinite dilution, only small amounts of probe molecules are introduced into the chromatographic column containing the solid to be investigated, minimizing intermolecular interactions and enabling the assessment of the most energetically active adsorption sites. ,

As shown in Figure c, initial γ _D values vary slightly among the wool types, with Cheviot exhibiting the highest surface energy. With increasing surface coverage, γ _D decreases exponentially for all samples. This trend reflects the preferential occupation of high-energy adsorption sites at low coverage, followed by the gradual filling of lower-energy sites. The exponential decay observed is characteristic of heterogeneous surfaces, where the plateau region corresponds to the average (bulk) surface energy. At low coverage, Merino, Eider, and Devon display overlapping γ _D profiles, while Cheviot maintains higher values. Overall, γ _D values decrease from 50 to 70 mJ·m^–2 at low coverage to ∼40 mJ·m^–2 at higher coverage. Additional γ _D values across different coverages are presented in Figure S1 (Section S2, SI). While the extrapolated γ _D at full coverage (n/n ₀ = 100%) should theoretically align with surface energy values obtained via tensiometry (typically ranging from 17 to 26 mJ·m^–2), some deviation remains, likely due to methodological differences. In general, IGC yields higher surface energy values compared to contact angle-based methods. Unlike tensiometry, where the fiber surface interacts with a liquid and may swell, IGC measurements are conducted under dry conditions. Prior to analysis, samples were purged by nitrogen (N₂), which reduces moisture content and likely affects surface energy. Thermogravimetric analysis (TGA) confirmed a mass loss of ∼5% after 2 h of N₂ purging (Figure S2, Table S2, Section S3, SI), indicating significant dehydration prior to surface energy measurement by IGC.

Beyond average surface energy values, IGC also provides insight into surface heterogeneity through γ _D distribution profiles (Figure d), which reflect the range of adsorption energies across different surface sites. A broader distribution indicates a greater number of adsorption sites, reflecting increased dispersive physisorption capacity, and, consequently, higher surface heterogeneity. Among the wool types, Cheviot clearly exhibits the broadest γ _D distribution, suggesting a more heterogeneous surface compared to the others. Additionally, the higher distribution maximum implies that a larger portion of the Cheviot fiber surface possesses elevated γ _D values. This trend is consistent with previous tensiometry results. Such differences may arise from variations in surface morphology and chemistry, as well as structural features. Notably, the composition, amount, and coverage of the 18-MEA layer have been shown to depend on fiber diameter. In general, both the absolute amount and chemical makeup of the bound fatty acids vary with breed and fiber diameter. As fiber diameter increases, the total quantity of bound fatty acids decreases linearly, and for fine fibers, it has been reported that more 18-MEA is covalently bound than required to form a complete monolayer, potentially contributing to enhanced hydrophobicity. ,

In the surface energy profiles, we focused solely on the γ _D , as the polar gas probes exhibited minimal interaction with the wool fibers. This suggests that the polar contribution to the overall surface energy is negligible and could not be reliably quantified. This observation indicates that a substantial portion of the 18-MEA layer remains on all the fibers’ surfaces, effectively shielding underlying amino acids from polar interactions. FT-IR analysis further supports this interpretation, revealing residual lipid content on the fiber surface (Figure S5, Section S4, SI). To gain deeper insight into the local surface chemistry, we employed surface-sensitive techniques, including XPS (Figure e) and AFM-IR (Figure f). We measured the XPS C 1s spectra of three wool types (Merino, Eider, Devon) selected for their differences in fiber diameter. The spectral profiles are characteristic of untreated wool, with C–C/C–H peaks corresponding to the hydrocarbon backbones of fatty acids and amino acid side chains. The NH–C=O signal corresponds to the amide linkage characteristic of proteins while peaks attributed to C–O, C–N, and C–S species are characteristic of both keratin and keratin-associated proteins residues. The surface composition is dominated by carbon species (>70 at%), further confirming a high concentration of hydrocarbon-rich material on the wool fiber surfaces. Closer examination of the FT-IR spectra indicates distinct regions associated with 18-MEA (Figure S5, Section S4, SI). A sharp band at 1745 cm^–1 (>C=O) corresponds to covalent bonding of 18-MEA to the cuticle via a sulfoester linkage. This band is known to persist after wool carding and bleaching, and typically only disappears following more intensive treatments, such as plasma etching. However, this feature is often difficult to resolve using conventional FT-IR due to low intensity and overlap with the intense amide I band characteristic of polypeptides. To overcome this limitation, we employed nanoscale-resolved AFM-IR spectroscopy (Figure f). In this surface-sensitive context, a band is visible around 1745 cm^–1 in spectra collected from three distinct locations on a Devon fiber. While the intensity varies across these points, this variability is likely influenced by the high surface roughness, which complicates spectral acquisition at arbitrary locations. The AFM topography images of the Devon fiber (Figure g) reveal a rough surface morphology, shaped by the natural curvature of the fiber and the characteristic scale structure. These scales often exhibit steep slopes or lateral cracks, and within individual scales, the surface may appear wrinkled. Altogether, these morphological features contribute to pronounced surface heterogeneities in topography and composition, corroborating the SSA and γ _D trends obtained through IGC.

Effects of Sulfur on Fiber Mechanics and Surface Chemistry

Disulfide bonds play a critical role in defining both the interfacial and mechanical properties of wool fibers, and are known to undergo cleavage at elevated temperatures. To gain mechanistic insights on the role of sulfur in shaping the physicochemical characteristics of wool fibers, we subjected three of the wool types (Merino, Eider, and Devon) to two heat treatments: 1 h at 120 and 200 °C, respectively (Figure ). Cheviot was excluded from this analysis due to its outlying γ _D distribution profile, while Polwarth was omitted because its fiber diameter, SSA, and γ _D distribution closely resembled those of Merino. The 120 °C treatment was designed to assess potential surface rearrangements following the evaporation of residual moisture, while the 200 °C was anticipated to result in disulfide bond cleavage.

3.

(a) Thermogravimetric analysis (TGA) curves showing mass loss during 1 h isothermal treatments at 120 and 200 °C for Merino, Eider, and Devon wool fibers. (b) Peak mass loss observed during the heat ramp phase of TGA measurements. Insets illustrate progressive yellowing of wool fibers with increasing temperature (D – Devon, M – Merino, E – Eider). (c) Tensile strength and (d) elongation at break as a function of thermal treatment temperature. (e) Specific surface areas (SSAs) derived from octane BET isotherms at room temperature, after thermal treatment. (f) Dispersive surface energy (γ _D ) distributions (n = 2) following 1 h treatment at 200 °C for all three wool types. (g) X-ray photoelectron spectroscopy (XPS) S 2p spectra of Merino wool reveal a reduction in C–S bonds and an increase in oxidized sulfur species (S–O and S=O). (h) Bulk Fourier-transform infrared (FT-IR) spectra confirm the formation of cysteic acid (S=O) with increasing heat exposure.

TGA isotherms recorded at 120 and 200 °C in air confirmed these expected compositional changes (Figure a). During the initial heating ramp, all fibers exhibited a mass loss of approximately 9–10 wt%, primarily attributed to water evaporation (Figure S3, Section S3, SI). Notably, the 120 °C isotherms showed no further mass loss, whereas the 200 °C treatment resulted in an additional ∼ 2 wt% reduction likely due to the release of volatile sulfur-containing compounds such as hydrogen sulfide, carbonyl sulfide, and carbon disulfide. While fibers with larger diameters exhibited greater mass loss, the temperature at which water loss peaked remained consistent across wool types (Figure b), occurring near 60 °C with minimal variation (1–2 °C). In addition to compositional changes, heat treatment also induced visible alterations in fiber coloration. Samples transitioned from white or creamy hues at room temperature and 120 °C to a yellow-orange tone at 200 °C (Figure b), suggesting chemical modifications that may occur either uniformly throughout the fiber or be localized near the surface.

To determine whether these changes occur in the bulk of the wool, we performed single-fiber uniaxial tensile tests, which measure the average mechanical properties of the fiber. The mechanical consequences of thermal exposure were captured in the tensile strength and elongation at break values derived from stress–strain curves (Figure c,d and Figure S7, Section S5, SI). Across all fiber types, mechanical performance declined with increasing temperature, with the most pronounced reduction observed at 200 °C. An exception was noted for Eider fibers, which showed a slight increase in tensile strength at 120 °C, consistent with their previously reported superior mechanical properties. For Merino and Devon, tensile strength at 200 °C was approximately half that of untreated fibers. Moreover, Merino exhibited a 10-fold reduction in elongation at break, with the stress–strain curve lacking any plateau-like deformation behavior (Figure S7, Section S5, SI). While thermal effects on keratin fiber composition and mechanics have been studied, most prior work focused on wet conditions, complicating direct comparison. We assume that fibers in this work are dry and that the IFs are unaffected by the heat treatment since α-helix denaturation for dry fibers is reported to start at about 230 °C. This agrees with our TGA results, in which decomposition onset temperature is above 200 °C in both nitrogen and oxidative atmosphere (Figure S4, Section S3, SI). Earlier studies report that fiber plasticity in water decreases with declining sulfur and cystine content. This observation aligns with the trends reported here, where disulfide bond cleavage and the volatilization of sulfur-containing byproducts at elevated temperatures result in a marked decline in mechanical performance. Overall, the mechanical data suggest that thermally induced disulfide bond reduction alters the internal cortical structure and composition, leading to measurable changes in bulk mechanical characteristics across fiber diameters and types.

To further explore the impact of thermally induced disulfide bond reduction on fiber surfaces, we extended our analysis to include IGC measurements on the heat-treated samples following the same protocol as for the untreated fibers (Figure e,f). The SSA values for the three wool types consistently decrease with increasing thermal treatment temperature, with Merino and Eider exhibiting the most substantial reductions (30–40%) following disulfide-bond cleavage at 200 °C (Figure e). Despite this clear trend, no corresponding topographic modifications were detected using SEM and AFM (Section S6, SI). The analysis of surface heterogeneity, as assessed by the γ _D distribution profiles (Figure f), reveals more nuanced effects. While Merino retains a distribution similar to that of the untreated fiber even after heating to 200 °C, both Eider and Devon show more significant changes. Eider exhibits a broadened distribution, indicative of increased surface heterogeneity, whereas Devon shows a marked shift in the distribution maximum toward higher γ _D values, prompting to significant compositional changes affecting the energy of adsorption sites. Interestingly, Merino’s surface heterogeneity appears largely unaffected by heat treatment, despite the observed decrease in SSA, suggesting that different structural and chemical factors may be at play. To further probe the chemical nature of these surface changes, we examined the sulfur environment using XPS (Figure g). The S 2p spectra for Merino reveal a clear decrease in the peak associated with disulfide bonds, accompanied by the emergence of a smaller peak near 169 eV, most prominent at 200 °C. This new feature corresponds to S–O bonds in sulfonate groups, specifically cysteic acid, formed either through further oxidation of cysteine residues generated by disulfide bond cleavage, or the decomposition of the 18-MEA layer. Additionally, the relative atomic percentage of sulfur on the fiber surface decreased significantly (from ∼2–3 at% down to below 0.5 at%). These findings indicate that the 18-MEA layer is partially removed and are corroborated by bulk FT-IR analysis (Figure h for Devon and Figure S6, Section S4, SI for the other wools), which show a distinct absorption band near 1080 cm^–1, characteristic of the symmetric S=O stretching vibrations in sulfonic acids, further confirming the formation of cysteic acid both within the cortex and the cuticle. Altogether, these surface characterization techniques revealed a complex interplay between structural and chemical transformations induced by heat treatment, extending beyond simple disulfide bond cleavage or 18-MEA debonding. The observed reduction in sulfur content, along with bond reorganization (i.e., cysteine oxidation), appears to influence surface heterogeneity through distinct mechanisms, variably affecting SSA and surface energy across fiber types and diameters, despite their initially similar elemental compositions.

Viscoelastic Response Governed by Cortical Matrix Structure

Because tensile strength and elongation at break are influenced by both surface and bulk compositional and structural changes, we sought to isolate the contribution of the fiber cortex to the bulk viscoelasticity of wool and examine how this varies across different fiber types. Most of the sulfur content in wool fibers is concentrated in the KAP matrix within the cortex. Unlike IFs, which exhibit long-range crystallinity, the KAP matrix lacks ordered structure and is therefore expected to be the primary contributor to the viscoelastic response of wool fibers. This hypothesis motivated our use of single-fiber tensile testing at varying strain rates, complemented by synchrotron SAXS, to extend and contextualize the previously described thermal and surface analyses (Figure ).

4.

(a) Rate-dependent stress–strain curves for representative fibers tested at four strain rates: 0.01%·s^–1, 0.1%·s^–1, 1%·s^–1, and 10%·s^–1. Data are shown for three wool fiber types: Polwarth, Eider, and Devon. (b) Semilogarithmic plot illustrating the strain-rate dependence of the normalized rate-dependent modulus E_r for all wool types. Modulus values are normalized to the lowest strain rate (0.01%·s^–1) to highlight the increase with strain rate. Dashed lines connecting data points serve as visual guides and do not represent fitted trends. For clarity, data points at each strain rate are slightly offset. (c) Scattering plot covering a q-range from 0.06 nm^–1 to 45 nm^–1 for all wool types. The shaded region indicates the q-range corresponding to the lattice parameter D. Assuming hexagonal packing, the structural parameter derived from SAXS corresponds to D, which is directly related to the mean IF distance a.

Previous studies have explored the viscoelastic behavior of keratin fibers using creep measurements and strain-rate dependent tensile tests, where strain-rate sensitivity served as a mechanical parameter to characterize viscoelasticity. , However, these tests typically extended fibers beyond their elastic region (2–4% strain), limiting each strain-rate experiment to a single fiber and obscuring true rate-dependent behavior due to high natural variability, largely resulting from a broad variation in fiber cross-sectional area and the occurrence of a medulla. For example, delicate Merino fibers with high crimp and thin cross sections failed at forces around 30 mN, while bulkier Devon fibers with medullae endured forces in the hundreds of mN (Figure a). To address this variability, we performed rate-dependent tensile tests within the elastic region (up to 1.5% strain), enabling each fiber to be tested at four strain rates (0.01, 0.1, 1, and 10%·s^–1) spanning 4 orders of magnitude (Figure a,b). This approach, adapted from cellulosic fibers studies, allowed for practical test durations from several minutes to under a second. To ensure the mechanical response captured was truly viscoelastic, cyclic tests confirmed that all applied strains remained within the elastic regime (Figure S8, Section S5, SI). Additionally, strain rates were applied in both ascending and descending order to verify consistency and rule out sequence-related artifacts.

Representative stress–strain curves for Polwarth, Eider, and Devon fibers show a consistent increase in slope with rising strain rate (Figure a). Eider fibers, however, display a distinctly nonlinear elastic response with a lower initial slope, which does not appear to result from fiber crimp. Moreover, the spacing between curves is not uniform; for instance, the 0.1%·s^–1 and 1%·s^–1 responses are nearly undistinguishable, particularly in the case of Eider wool. This behavior becomes more apparent when examining the relative increase in modulus (defined as the slope of the stress–strain curve in the elastic regime) as a function of strain rate (Figure b). To improve comparability of rate effects and minimize variability within each wool type, especially in coarser fibers, all modulus values (E) were normalized to the value at the lowest strain rate (0.01%·s^–1), denoted as E _0.01. The resulting normalized modulus is expressed as E _r = E/E _0.01. Notably, a modulus plateau is visible between 0.1%·s^–1 and 1%·s^–1, with varying prominence across fiber types. Devon fibers, the largest among those tested, typically contain a medulla and have the smallest cortical cells. They also exhibit a higher proportion of P-cortical cells, less ordered IF packing, and lower IF alignment along the fiber axis. In stark contrast, Merino fibers are much thinner, lack a medulla entirely, and are dominated by O-cortical cells. These fibers feature highly periodic IF packing and strong alignment, representing a structurally more ordered system. Despite these pronounced structural differences, both Devon and Merino fibers show a similar increase in modulus with strain rate by approximately 20–25% between 0.01%·s^–1 and 10%·s^–1. Cheviot fibers fall between these two structural extremes while exhibiting a comparable mechanical response. In contrast, Eider and Polwarth fibers display markedly higher strain rate dependence, with ∼60% and ∼80% increases in modulus across the range of tested strain rates, respectively.

Crystallinity values of the fibers range between 10% and 18% with Devon showing the highest degree of crystallinity and Merino and Eider the lowest. Polwarth and Cheviot exhibit intermediate values around 12%. These findings suggest that IF long-range order and cortical cell type distribution alone cannot account for the observed differences in viscoelastic behavior. Instead, the sulfur-rich KAP matrix, with its structural and composition variations across wool types, may play a more critical role in rate-dependent mechanical properties. Since all applied strains were below the yield point, keratin in the IFs likely remained in its α-helical conformation, and elongation was primarily accommodated by bond deformation within the matrix. Although often described as an amorphous network cross-linked via disulfide bonds, prior studies suggest the matrix may contain a grain-like substructure with feature sizes of 2–4 nm. , Our results support the idea that local gradients in matrix composition, network topology, and microstructure can significantly influence viscoelastic properties, thereby challenging the conventional assumption of a fully homogeneous matrix.

To further investigate the interaction between IFs and the surrounding KAP matrix, we performed synchrotron SAXS on nonoriented wool fiber samples. The resulting scattering profiles (Figure c) include the wide-angle region, where characteristic peaks of α-keratin (q ∼ 6 nm^–1) and β-keratin (q ∼ 13 nm^–1) are observed. , The SAXS data was evaluated using an approach utilizing Porod’s law , (Figure S11, Table S3, Section S7, SI) and applying a Monte Carlo-based fitting approach with a hard-sphere model (Figure S12, Table S4, Section S7, SI). While there are slight variations in the scattering values among the different wool types, the same key features are consistently observed across all samples (Figure c). Region (ii), centered around q = 1.33 nm^–1 (∼4.5 nm spacing), aligns with literature reports on lipid lamellae in human hair, where second and third-order peaks are observed at q = 2.66 nm^–1 (2.36 nm) and at q = 3.99 nm^–1 (1.57 nm), respectively. , These lamellar structures are thought to arise from bilayer arrangements in the cell membrane complex (CMC), and while more prominent in human hair, our results further confirm that lipids remain present in both the surface and cortex of processed wool fibers. Here, we observe in most samples a peak at q = 2.66 nm^–1. Region (i), with a peak at q = 2.4 nm^–1 (∼2.6 nm), may also correspond to lipid domains or filament subunits. A sharp peak at q = 0.93 nm^–1 (indicated by an arrow) corresponds to the 6.7 nm axial stagger, a structural feature resulting from the packing of keratin coils into intermediate filaments. Of particular relevance to IF–matrix interactions is region (iii), where a broad feature between q = 0.6–0.8 nm^–1 reflects the scattering from the IF and their center-to-center spacing. This spacing D can provide indirect insight into matrix volume and organization. The scattering feature in this region arises from a convolution of form and structure factor contributions, the form factor describing the scattering from the individual IFs, and the structure factor capturing their partial ordering. The Monte Carlo fits reveal a broad size distribution (Figure S11, Section S7, SI), which is attributed to the filament’s cross-sectional dimensions. While SAXS offers an averaged signal across the fiber cross-section, TEM data for Merino fibers reveal that IF spacing increases from approximately 10 nm in the O-cortex to about 12 nm in the P-cortex. This trend suggests that matrix area expands with increasing disorder in IF packing, reinforcing the idea that matrix structure plays a critical role in defining load transfer within the fiber.

Both SAXS evaluation methods revealed consistent trends, with closely matching results for Devon (Section S7, SI). However, the form-free Monte Carlo fitting approach yielded values for Merino, Polwarth, Cheviot, and Eider that align more closely with TEM measurements. Using a hexagonal lattice model and the lattice parameter D obtained from region (iii) in the SAXS data, we calculated the average IF center-to-center spacing a with eq 2 for each wool type (Figure S12, Section S7, SI). These values reveal notable differences across wool types that do not follow expected trends based on the assumption of increased IF disorder with increasing fiber diameter: Devon exhibits the smallest a at 9 nm, Cheviot at 10.2 nm, while Merino, Polwarth, and Eider show the largest values at approximately 11.5 nm. Given the progression in cortical arrangementfrom small-diameter fibers with high O-cortex content and regular IF packing (e.g., Merino, Polwarth) to larger-diameter fibers with predominant P-cortex and more disordered IF organization (e.g., Eider, Devon)it is likely that the idealized hexagonal lattice model becomes increasingly distorted. − Nevertheless, the SAXS-derived values for IF center-to-center spacing remain in good agreement with TEM data, despite sample preparation differences (i.e., staining and embedding for TEM). , Importantly, the observed trend in a cannot be explained solely by cortical cell type distribution. For instance, although O-cortex content typically decreases with increasing fiber diameter, this does not align with the measured evolution of IF spacing. Assuming a constant IF radius, the larger a values observed in Merino, Polwarth, and Eider suggest a greater volume of matrix material between filaments (inset, Figure c). This increased matrix volume may help explain the more pronounced viscoelastic responses observed in Merino, Polwarth, and Eider compared to Cheviot and Devon. However, Merino’s relatively moderate modulus increase with strain rate, especially when contrasted with the stronger rate sensitivity of Polwarth and Eider, suggests that additional structural factors intrinsic to the KAP matrix may also play a critical role.

Altogether, when combined with conclusions drawn from thermally induced sulfur bond reorganization, the viscoelastic and SAXS measurements offer compelling insights into the role of the KAP matrix in shaping wool bulk mechanical and viscoelastic properties. Merino and Devon fibers exhibit similar decreases in mechanical performance following thermal treatment and show comparable, moderate increases in modulus under varying strain rates. Yet, their maximum IF center-to-center spacing a differs significantly, indicating that neither matrix volume nor IF packing geometry alone can account for their mechanical behavior. Instead, differences in matrix composition and internal structure, such as disulfide bond density and degree of cross-linking, appear to be more influential. This interpretation is reinforced by the behavior of Eider fibers, which have an a comparable to Merino but respond very differently to both thermal treatment and strain-rate variation. Despite similar sulfur content and matrix volume, Eider fibers are less affected by high-temperature treatments, suggesting that a smaller fraction of sulfur may be involved in forming a continuous cross-linked KAP network. This lower cross-linking density may contribute to the observed nonlinearity in Eider’s mechanical response at low strain, and its prominent strain-rate sensitivity. These compositional and network topology differences may also influence the grain-like substructure of the matrix, which likely varies across wool types. These insights underscore the need for further investigation into the molecular architecture and cross-linking behavior of the KAP matrix, which may ultimately govern the mechanical performance of wool fibers more than previously recognized.

Relating Structure and Mechanical Properties

We utilized IGC and SAXS to investigate structural parameters at both the surface and cortex levels of wool fibers (Figure a). At the surface, which is dominated by the cuticle layer, we observed that SSA decreases with increasing hair diameter d and with thermal treatment. This trend suggests that coarser fibers and heat-treated samples have smoother or less porous surfaces. In the case of treated samples, this is likely a result of heat-induced bond reorganization. Within the cortex, SAXS was used to indirectly probe the KAP matrix by measuring the average spacing a between IFs. We found that a decreases with increasing fiber diameter, which we interpret as a reduction in matrix volume or density in the case of thicker fibers (Figure a). When correlating these structural parameters with the rate-dependent tensile modulus, only weak trends emerged. The lack of correlation strength indicates that mechanical behavior arises from a complex interplay of surface and bulk features, and cannot be attributed to a single structural parameter.

5.

(a) Schematic illustration of the relationship between fiber diameter d or heat treatment temperature T with specific surface area (SSA) and cortex structure (captured in the IF–IF center distance a obtained by SAXS). Line of fit equation for the dependence of a on d is y = −0.053x + 12.490 with R² = 0.50 and p-value for a two-sided test of no correlation is 0.01. (b) Relationship between Young’s modulus at 1%·s^–1 and SSA across all wool fiber types. Line of fit equation is y = −1.649x + 4.193 with R² = 0.56 and p-value for a two-sided test of no correlation is less than 0.001. (c) Relationship between tensile strength and SSA for Merino, Eider, and Devon wool fibers treated at three temperatures (22 °C, 120 °C, 200 °C). Line of fit equation is y = 464.7x – 183.9 with R² = 0.99 and p-value for a two-sided test of no correlation is 0.002, y = 281.7x + 135.6 with R² = 0.27, p-value: 0.002, and y = 642.4x – 127.6 with R² = 0.85, p-value: 0.01 for Merino, Eider, and Devon, respectively.

To further explore the influence of surface structure on mechanical performance, we examined the relationship between SSA and modulus (Figure b). Fibers with higher SSA (indicative of a more porous or textured surface) tend to exhibit lower modulus values. This observation aligns with the trend across fiber diameters: smaller-diameter fibers, which have higher SSA, appear to be more sensitive to surface effects. Thermal treatment impacts both surface and bulk properties (Figure c). Degradation likely initiates at the fiber surface, reducing SSA, while disulfide bond reorganization within the cortex contributes to a loss in mechanical strength. Overall, these findings highlight the need to consider both cuticle and cortex contributions when interpreting the mechanical response of keratin fibers.

Conclusions

Together, these findings challenge the traditional model of wool fibers as simple fiber-reinforced composites dominated by IF alignment and packing. Instead, they suggest that structural and compositional variations within the KAP matrix and the cuticle play a central role in determining both surface properties and bulk mechanical performance. Using a multimodal approach, we confirmed that the nanometer-thick 18-MEA layer is consistently present across all wool types and plays a critical role in maintaining hydrophobicity and modulating surface energy. IGC and AFM-IR measurements revealed that surface heterogeneities in structure and composition vary across fiber types, with Cheviot wool exhibiting the broadest distribution of dispersive surface energies. Thermal treatments at 200 °C led to a consistent reduction in specific surface area (up to 40%) and a significant loss in mechanical performance, including an 80% drop in elongation at break for wool types exhibiting similar viscoelastic behavior, such as Merino and Devon, despite substantial differences in fiber diameter and IF organization. XPS and FT-IR analyses confirmed the cleavage of disulfide bonds and the formation of cysteic acid, underscoring the essential role of sulfur cross-linking in preserving both surface and bulk integrity. Importantly, rate-dependent tensile testing within the elastic regime revealed distinct viscoelastic responses among wool types that were not captured by traditional tensile testing. Polwarth and Eider fibers exhibited the strongest rate dependence, with modulus increases of up to 80% between 0.01%·s^–1 and 10%·s^–1. Complementary SAXS measurements revealed variations in matrix content and IF spacing, supporting the observed differences in viscoelastic response and reinforcing the crucial role of the matrix in modulating nanoscale stress transfer. The similar mechanical behavior observed in Merino and Devon fibers (despite their contrasting structural features) alongside the distinct response of Eider fibers, which exhibit reduced sensitivity to sulfur removal and pronounced rate-dependent behavior, highlights the importance of matrix composition, cross-linking density, and grain-like substructure in determining fiber performance. Further investigation is essential to better understand the properties of the KAP matrix and how coupling between surface and bulk structures affect the mechanical response of wool fibers. The fiber surface, dominated by the cuticle and the 18-MEA layer, is a gateway to the fibers’ cortex, which largely governs mechanical behavior. Our results show that fibers with high SSA and smaller diameters both tend to exhibit lower modulus values. Heat treatment affects the surface and the cortex of wool fibers, leading to chemical changes such as disulfide bond cleavage and reorganization. These alterations result in consistent decreases in tensile strength, elongation at break and SSA. Altogether, this work lays a critical foundation for developing a more complete, multiscale mechanical model of keratin fibers by establishing direct links between hierarchical structure in keratin IFs and KAP matrix, sulfur-based chemistry, and functional properties. A stronger focus on in-situ mechanical and spectroscopic characterization of the matrix, , ideally in combination with targeted chemical modifications, will be essential to resolve these remaining questions. To gain a holistic view on the interplay of the cortex’s structures, it will be crucial to perform shear testing providing information on the interactions between IFs and KAPs. Such efforts will ultimately enable the rational design of keratin-based materials with tunable mechanical and interfacial performance.

Materials and Methods

Materials and Heat Treatment

All five wool Roving used in this study were sourced from Paradise Fibers, located in Spokane, Washington (USA). The fleeces were prewashed and carded by the manufacturer to remove excess lanolin and debris. The Roving samples analyzed were wool tops (longer fibers remaining after scouring and combing) and contained only minimal residual grease (1–2%), as specified by IWTO-10-2003.

The wool fibers were heat treated in a Carbolite tube furnace (TZF 15/610, Germany) under constant nitrogen flow (ca. 0.3 L·min^–1) from room temperature to 120 or 200 °C at a heating rate of 5 °C·min^–1 and a holding time of 1 h for IGC and FT-IR measurements. For the reported XPS and tensile testing experiments, a glass oven (Buchi B-585, USA) was utilized performing the same heating protocols with a vacuum pump (V-100) at 20 mbar.

Scanning Electron Microscopy (SEM)

Scanning electron microscopy (SEM) was utilized to examine the surface morphology of the wool fibers, providing high-resolution images of the scale-like fiber surface. Six samples were examined: untreated and heat-treated fibers from Devon, Eider, and Ultrafine Merino breeds. Three strands of wool were selected for each sample, and all of the fibers were coated with a 9 nm layer of osmium tetroxide (OsO₄) for electrical conductivity (Filgen OPC60A Osmium Coater). Approximately 10 images were collected per sample, ensuring that all three fibers were imaged, using the EPIC SEM Hitachi S-3400 at 5 kV. The highest-quality image was selected for each sample.

Inverse Gas Chromatography (IGC)

Specific surface areas and dispersive surface energies of the wools were determined using an Inverse Gas Chromatograph Surface Energy Analyzer (IGC SEA, Surface Measurement Systems, UK). Three mm internal diameter silane-treated glass columns were filled with ca. 300 mg fibers. Nitrogen was used as carrier gas and methane was used as reference probe for the correction of the dead volume. The specific surface areas of the wools were determined using the Brunauer–Emmett–Teller (BET) method, with octane serving as a nonpolar probe. , Each sample was tested minimum in duplicate, and the average value was calculated under conditions where the BET equation exhibited good linearity in the p/p₀ range from 0.1 to 0.35 (R² ≥ 0.983) (see Figure a). Additionally, four nonpolar n-alkane probes, octane (C8), nonane (C9), decane (C10), and undecane (C11), were injected to evaluate the dispersive surface energy profiles of the wool fibers over a defined surface coverage range (n/nm = 0.002–0.1). All solvents were HPLC grade (>99.9%). Due to insufficient interaction with the monopolar acidic and basic probes (dichloromethane and ethyl acetate, respectively), the specific surface energies and acid–base characteristics of the wool fibers could not be determined. All measurements were conducted at 30 °C and 0% relative humidity at a flow rate of 10 mL·min^–1. Prior to testing, samples were preconditioned at the measurement temperature (30 °C) for 2 h. Each column was measured twice. The data were analyzed using SEA analysis software, incorporating the PeakCom and Dorris–Gray methods. ,, The dispersive surface energy (γ _D ) distributions were obtained by integrating the cumulative exponential decay functions of the γ _D profiles over the whole coverage range.

Thermogravimetric Analysis (TGA)

TGA was performed using a Netzsch TG 209 F3 Taurus to assess the mass loss in wool fibers under conditions simulating IGC. Wool fiber bundles were purged with nitrogen at a flow rate of 10 mL·min^–1 for 2 h at 25 °C, with two samples analyzed per fiber types. Additionally, isothermal measurements at 120 and 200 °C were conducted on two samples per wool type. These tests began at 25 °C and employed a heating rate of 20 °C·min^–1 followed by an isothermal step at 120 or 200 °C, using an air purge at 50 mL·min^–1 to monitor mass loss during thermal treatment.

Fourier Transform-Infrared Spectroscopy (FT-IR)

FT-IR spectra were acquired with a Cary 630 FT-IR Spectrometer with a diamond attenuated total reflectance module (Agilent Technologies, USA). Spectra were obtained over 64 scans at a resolution of 2 cm^–1 in a scan range between 4000 cm^–1 and 850 cm^–1. Data was background corrected, smoothed and normalized using the rubberband method and Fourier transformation, using python 3.9.12 with the following packages: numpy, pandas, scipy, matplotlib, and spectrochempy.

X-ray Photoelectron Spectroscopy (XPS)

XPS measurements were performed on untreated and temperature treated wool fiber bundles using a NEXSA G2 system (ThermoFisher Scientific, USA). An Al Kα X-ray source with a 50 μm spot size was employed, and a field emission gun was used for charge compensation. Survey scans (10 scans, 0.1 eV step size) and high-resolution elemental scans of the C 1s and S 2p regions (6 scans, 0.1 eV step size) were acquired.

Atomic Force Microscopy (AFM)

8 × 8 μm² and 5 × 5 μm² tapping mode topography images were recorded with a Bruker FastScan AFM with triangular SiN-based FastScan Bruker AFM probes (nominal resonance frequency of 1400 kHz and nominal spring constant of 18 N·m^-1 with a nominal tip radius of 5 nm) at ambient conditions. AFM-IR measurements were obtained with a Bruker Dimension IconIR enabling acquisition of three local IR spectra for wavenumbers between 800 cm^–1 and 1800 cm^–1. AFM probes of the Bruker TNIR-D-10 series were utilized in tapping mode with a gold coating on the tip, a nominal spring constant of 40 N·m^–1 and a nominal resonance frequency of 300 kHz. During AFM-IR measurements, the sample chamber was purged with air, providing relative humidity levels below 10% on the sample surface. For all measurements, single fibers were prepared by gluing them to a glass slide with several small droplets of nail polish.

Mechanical Testing

Strain-rate dependency of single fibers was investigated by tensile testing with displacement-controlled ramps until 1.5% strain with a fiber span/gauge length of 10 mm. The displacement rates were increased from 0.06 mm·min^–1 to 60 mm·min^–1 corresponding to strain-rates of 0.01%·s^–1, 0.1%·s^–1, 1%·s^–1, and 10%·s^–1. A preload of 1 mN was applied and all experiments were performed in a dynamic mechanical analyzer (TA Instruments DMA 850, USA) at 30–60% relative humidity and temperature of 22 °C. Overall, for obtaining the rate-dependent results, 9 to 16 individual fibers per wool type were successfully tested and the evaluation procedure was analogous to the protocol presented in M. Zizek et al. (fibers evaluated: Merino (n = 16), Polwarth (n = 14), Cheviot (n = 9), Eider (n = 16), and Devon (n = 16)). The rate-dependent modulus was obtained by determining the slope until 1% strain. To obtain an average value of the cross-sectional area of the wool fibers, several fibers of each wool type – number was depending on the heterogeneity of the fiber diameter – were embedded in a glycol methacrylate resin and the resin was left 24 h to cure. The embedded sample was placed in a microtome and cut at a minimum of two different positions along the length of the fiber. Since Eider and Devon were previously showing a high degree of variation in cross-section, each individual fiber of these two wool types tested for the rate-dependent results were embedded and cross sections at two different positions of the fiber. For each of those cuts, an optical image of the cross section was recorded. The outlines of the cross-sectional area of the fiber were traced manually in an image analysis software (Paint.NET). Then, the image was binarized and the cross-sectional area of the fiber was calculated on the binarized image. , In Table S1 in the SI, the cross-sectional areas (average ± standard deviation) of the five sheep wool types are summarized.

To investigate heat treatments, single fiber tensile tests of Merino (n_22 °C = 11, n_120 °C = 5, n_200 °C = 9), Eider (n_22 °C = 8, n_120 °C = 10, n_200 °C = 12), and Devon (n_22 °C = 11, n_120 °C = 10, n_200 °C = 10) fibers were performed until failure with 1 mN preload, a span/gauge length of 10 mm and a displacement rate of 0.6 mm·min^–1 on the same device. The stress–strain curves per fiber type and temperature level were evaluated for tensile strength and elongation at break. The average values of the cross-sectional areas were used to calculate the stress from the experimental force data.

Synchrotron Small-Angle X-ray Scattering (SAXS)

SAXS measurements were performed at beamline 5ID in the Dupont-Northwestern-Dow (DND-CAT) sector at the Advanced Photon Source in Argonne National Lab. Two configurations were used to collect data from sample distances 200.12 and 1011.4 mm to access a Q-range of 0.0036 Å^–1 to 4.642 Å^–1 where Q = 4π/λ sin­(θ/2). λ is the wavelength of the X-ray beam, which for these experiments is 0.7293 Å and θ is the deflection angle. For each sample, data was collected using three Rayonix detectors: LX170HS (hs102), LX170HS (hs103), and a MarMosaic225 (mm019). The sample-to-detector distances were calibrated using LaB6 SRM660a (hs102), Ag Behenate (hs103), and Spun Si 7200 line/mm (mm019) standards, yielding distances of 200 mm (hs102), 1013 mm (hs103), and 8524 mm (mm019), respectively. Repeated measurements were taken with the data from each detector grouped and averaged to obtain curves with propagated uncertainties. Background scans were processed identically and subtracted from the sample data. Background-subtracted curves from all three detectors were merged using intensity-weighted overlapping regions to produce a continuous scattering curve for each sample.

Powdered fiber samples were placed between two pieces of 3 M “Magic Tape” with fiber covering the entire interior surface of the tape. These samples were then attached to the sample holder with additional “Magic Tape”.

For SAXS evaluation, two approaches were applied. First, a Porod-based evaluation , was applied by determining the Porod’s region and after processing, fitting the features of interest related to the IFs and matrix (Figure S11, Section S7, SI). [McSAS3GUI] was utilized using a hard sphere model from [sasmodels] to fit the data. McSAS uses Monte Carlo methods to fit scattering data and produces form-free size distributions (Figure S12, Section 7, SI). The approach is commonly applied to porous, solid systems , and has been utilized to study self-assembly.

For the calculation of center-to-center distances for IFs a hexagonal geometry was utilized to convert from D to a (Inset in Figure c). ,

$$a=\frac{2D}{\surd 3}$$ 2

Glossary

ABBREVIATIONS

18-MEA

18-methyleicosanoic acid

AFM

atomic force microscopy

AFM-IR

atomic force microscopy combined with infrared spectroscopy

BET

Brunauer–Emmett–Teller

C8

octane

C9

nonane

C10

decane

C11

undecane

CMC

cell membrane complex

FT-IR

Fourier transform infrared spectroscopy

γ _D

dispersive surface energy

IF

intermediate filament

IGC

inverse gas chromatography

KAP

keratin-associated protein

M-cortex

meso-cortex

N₂

nitrogen

O-cortex

ortho-cortex

P-cortex

para-cortex

SAXS

small-angle X-ray scattering

SEM

scanning electron microscopy

SSA

specific surface area

TEM

transmission electron microscopy

TGA

thermogravimetric analysis

WAXS

wide-angle X-ray scattering

XPS

X-ray photoelectron spectroscopy

All the experimental data presented and evaluated in this work are available at Czibula, C., & Chazot, C. A. C. (2025). Structure and Sulfur: Tuning the Viscoelastic and Surface Properties of Natural Keratin Fibers [Data set]. Zenodo. https://doi.org/10.5281/zenodo.16513009.

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acsmaterialsau.5c00130.

• Further experimental results including sample description, IGC, TGA, SAXS, tensile testing and imaging (AFM, SEM) data (PDF)

C.A.C.C. and C.C. jointly conceptualized the research. C.A.C.C. supervised the project and secured funding. C.C. coordinated collaborative efforts for data acquisition and analysis and performed experiments involving AFM, AFM-IR, FT-IR, TGA, XPS, and uniaxial tensile testing. J.B.S. led the acquisition and analysis of IGC data. J.B.C., N.C.F., and K.T.G. conducted the synchrotron SAXS experiments, while C.C., G.J.S., K.T.G., and J.J.R. carried out SAXS data analysis. D.S.S. performed SEM imaging of wool fibers. M.T. assisted with fiber microtome cutting and optical cross-sectional imaging. C.C. drafted the initial manuscript. J.B.S. contributed to the first draft and reviewed changes. C.A.C.C. edited the manuscript and oversaw its development. All authors reviewed and provided feedback on the final manuscript. CRediT: Caterina Czibula conceptualization, data curation, formal analysis, investigation, methodology, visualization, writing - original draft, writing - review & editing; Jana Bianca Schaubeder data curation, formal analysis, investigation, methodology, visualization, writing - original draft, writing - review & editing; Glen Jacob Smales formal analysis, methodology, software, writing - review & editing; Julia B Chatterjee data curation, investigation, methodology, writing - review & editing; Natalie C Fisher data curation, investigation, methodology, writing - review & editing; Deborah S Silverstein investigation, methodology, writing - review & editing; Michael Thoman investigation, methodology, writing - review & editing; Kayla T Ghezzi formal analysis, investigation, methodology, writing - review & editing; Jeffrey J Richards formal analysis, funding acquisition, supervision, writing - review & editing; Cécile A. C. Chazot conceptualization, data curation, formal analysis, funding acquisition, investigation, methodology, supervision, visualization, writing - review & editing.

The authors declare no competing financial interest.

This paper was published ASAP on October 30, 2025, with an older version of Figure 1. The corrected version was reposted on October 31, 2025.

Published as part of ACS Materials Au special issue “2025 Rising Stars in Materials Science”.