Enabling water-based high-density nanoparticles assembly by using silk fibroin as an adsorbate

Abstract

Water-based fabrication holds promise for innovations in life sciences, electronics, and materials science at the biotic-abiotic interface by integrating living systems with high technologies. However, using water to create bio-nano interfaces is challenging, as it often requires surface pre-treatment and thermal processing, which are both harmful to living systems. Here, we propose silk fibroin (SF) as a natural adsorbate that enables the fabrication of high-density nanoparticle (NP) layers relying solely on water-based processing. We show that SF spontaneously adsorbs onto various NPs, enhancing intermolecular interactions and facilitating the wetting of otherwise hydrophobic substrates, resulting in densely packed NP layers. Several SF-adsorbed NP electronics are demonstrated with performance comparable to their conventional counterparts. This work offers significant utility by establishing a flexible and biocompatible approach for the fabrication of seamless, precisely controlled bio-nano interfaces.

Water-based fabrication is of potential use for a range of applications, but can be challenging. Here, the authors report the use of silk fibroin as an absorbate for the preparation of high density nanoparticle layers by water-based processing.

Introduction

The biotic-abiotic interface and the integration of living systems with conventional nanotechnologies are critical for advancing next-generation devices that seamlessly interface biological systems with engineered materials. This enables real-time sensing, adaptive functionality, and sustainable operation within living environments. This convergence is foundational for innovations in bioelectronics, data storage, personalized medicine, and environmentally responsive technologies^1–4. However, beyond addressing fundamental mismatches in mechanical properties and communication modalities between these domains^5–7, a critical challenge lies in preserving the intrinsic biological functions of labile and vulnerable biogenic substances when establishing such interfaces^8–10. Achieving this requires biocompatible fabrication approaches that not only maintain biochemical functionality but also meet the high standards of resolution and precision inherent in cutting-edge nanotechnology, ensuring mutual compatibility and functionality in both domains.

Water-based nanofabrication is a biocompatible approach for high-resolution processing that can bridge the challenging gap between biological systems and advanced technologies. However, water’s intrinsic physical properties, such as high surface tension and low volatility, pose significant challenges for surface wetting, which is necessary to form reliable interfaces and fabricate high-quality devices. Conventionally, these issues are typically addressed through surface pre-treatments (e.g., plasma, UV/Ozone, and acid/base treatments) to enhance interfacial wetting, followed by high-temperature annealing to remove organic residues and increase the crystallinity, thereby creating high-density and defect-free films^11,12. Unfortunately, both approaches are generally incompatible with most biological substances.

Extensive foundational studies have established the physicochemical principles governing protein adsorption (e.g., silk fibroin (SF)) under aqueous conditions, including its effects on wettability, secondary-structure transitions, and biological or material coating^13–18. While these works suggest that protein adsorption may offer a natural route for forming biocompatible interfaces, they focus primarily on fundamental interfacial behavior. Building upon this knowledge, recent efforts have begun translating such phenomena into aqueous nanoprocessing strategies^19,20. However, these approaches still rely on high-temperature annealing or solvent activation. A fully water-based route capable of producing high-density, electronically functional nanoparticles (NPs) layers without post-thermal processing remains unaddressed.

NPs can exhibit distinct functionalities when their size and composition are precisely controlled²¹. They can be directly dispersed in water through surface functionalization²², and their biocompatibility can be further enhanced by appropriate surface passivation²³. Therefore, improving the interfacial wettability of water-dispersible functional NPs to achieve uniform surface coverage while simultaneously forming high-density NPs layers without compromising their intrinsic properties can be a promising strategy to overcome existing technological challenges in water-based fabrication processes aimed at developing high-performance bio-nano interfaces.

We show here that SF can function as a natural amphiphilic adsorbate to enable water-based fabrication of high-density NP layers while preserving the original functionality of the NPs. When added to aqueous NP dispersions, SF spontaneously forms a nanoscopic adsorption layer around various NPs through multiple interaction routes (e.g., van der Waals, electrostatic, and hydrogen bonding interactions)^24,25. Analysis of adsorption behavior of SF and its role in enhancing intermolecular interactions between NPs during the coating phase points that the conformational rearrangement and network entanglement of adsorbed fibroin chains upon physical contact are key mechanisms enabling the self-assembly^26–28. These augmented interactions allow aqueous NP dispersions containing a small, precisely controlled amount of SF (~0.2 w/v%) to effectively wet hydrophobic substrates, such as polydimethylsiloxane (PDMS) and polytetrafluoroethylene (PTFE), resulting in the densely packed NP layers. Additionally, control over the polymorphism of the SF adsorption layer offers further benefits in the sequential stacking of water-based NP dispersions. Through this physico-chemical approach, high-density NP-based electronics can be formed on hard-to-wet surfaces via simple water processing with outcomes that retain the original functionalities of the NPs. These NP-based electronics exhibit electrical performances comparable to those of conventional inorganic solution-processed devices, with no observable degradation due to the presence of the adsorbed SF dielectric.

Results

Concentration-dependent adsorption of SF onto NPs

SF adsorbs spontaneously onto various NPs in a concentration-programmable manner, forming a controllable nanometric adsorption shell. This adsorption reshapes key NP surface characteristics, establishing the physico-chemical basis for the improved wetting and enhanced NP assembly (Fig. 1). This behavior arises from the amphiphilic structure of the SF chain, which allows it to function as a versatile natural surfactant^19,29. As illustrated in Fig. 1a, such adsorptions can occur through various interaction routes^24,25. This study explores the evolution of SF adsorption onto water-dispersed NPs with varying SF concentration (Fig. 1b). This study focuses on the adsorption-mediated NP-assembly behavior of SF within the optimal concentration regime required for electronic functionality, rather than benchmarking against conventional surfactants whose effects on wettability and rheology have been addressed previously¹⁹.

Fig. 1. Spontaneous adsorption of SF onto NPs.

a Amphiphilic nature of SF and its possible adsorption interactions with NPs. b A schematic illustration of the material system explored in this study. c, d Schematics and TEM images showing the evolution of SF adsorption onto NPs with increasing SF concentration in aqueous NP dispersion. Scale bars, 200 nm. e AFM topography images and corresponding AFM-IR absorption map at 1650 cm⁻¹ of the assembled NPs with varying SF concentrations. f Changes in hydrodynamic radius and zeta potential of PS NPs with varying SF concentrations. Data points and error bars indicate mean ± standard deviation (SD) (n = 3). g Adhesion force comparison between pristine PS NPs and those with different levels of SF adsorptions. Color bars and error bars indicate mean ± SD (n = 10).

The adsorption behavior of SF onto monodisperse, spherical polystyrene (PS) NPs (⌀ ~ 100 nm) is investigated via electron microscopy. Figure 1c illustrates the formation, growth, and saturation of SF adsorption layers on NP surfaces as the SF concentration is gradually increased (red arrows in Fig. 1d). Even at a low concentration of 0.02 w/v%, a thin SF adsorption layer (~3 nm) forms around the NPs. These observed features were neither visible in the pristine transmission electron microscopy (TEM) grids nor in the original NPs (see Supplementary Fig. 1). Additionally, capillary bridges (blue arrows in Fig. 1d) are observed between SF-adsorbed NPs as a result of intimate contact in the dispersed phase, followed by separation during the drying phase. As the amount of SF is increased in the dispersion, the adsorption layer thickens, reaching saturation. In dispersions with 0.2 w/v% SF, a thicker SF adsorption layer ( ~ 10 nm) with wider necks in the capillary bridges can be observed. With higher SF concentrations (i.e., 2.0 w/v%), the NP assembly appears fully encapsulated within the SF matrix. The adsorption behavior occurred regardless of the material composition, surface functionalization, shape, or size of the NPs (see Supplementary Figs. 2, 3). A high magnification scanning electron microscopy (SEM) image and corresponding energy dispersive X-ray spectroscopy (EDS) elemental mapping of nitrogen (N kα_1,2) confirmed that the thin membrane covering NPs and interparticle bridges, which are observed in TEM images, resulted from SF adsorption (see Supplementary Fig. 4). Please note that, unless otherwise specified, 0.02, 0.2, and 2.0 w/v% SF concentrations are denoted as SF002, SF020, and SF200, respectively, throughout the manuscript.

Chemical surface analyses also confirm that the thin membranes observed over the NP surfaces are SF adsorption layers. Atomic force microscopy-infrared spectroscopy (AFM-IR) analysis of SF-adsorbed NPs reveals a distinct IR spectrum, showing characteristic peaks of both pure SF films and pure PSNPs (see Supplementary Fig. 5). Specifically, IR absorption mapping at 1650 cm⁻¹, corresponding to the Amide I band (C = O stretching), visualizes the spatial distribution of adsorbed SF on NP surfaces (Fig. 1e). At lower SF concentration (i.e., SF002), weak IR absorption occurs at particle boundaries, indicating thin and sparsely distributed adsorption. In contrast, a substantial IR absorption is observed for NPs with a thicker and more uniform SF adsorption layer. For larger-scale particles ( ~ 20 µm), SEM-Raman correlative analysis is performed, clearly identifying the SF presence and providing additional conformational information about the adsorbed SF within these larger capillary bridges (see Supplementary Fig. 6).

The physical effect of SF adsorption on various NPs is confirmed via dynamic light scattering (DLS) analysis (Fig. 1f for PS NPs and Supplementary Fig. 7 for other NPs). As the SF content in aqueous PSNP dispersion (⌀ ~ 100 nm) increases, the hydrodynamic radius rises and plateaus by ~0.2 w/v%, and beyond this point, no significant differences are observed (p-value ~ 0.89).

The surface charge of the NPs changes in correspondence with the amount of SF in the dispersion. While the concentration at which hydrodynamic size or zeta potential saturation begins varies between different NP types, they all exhibited increases in size and changes in zeta potential with SF adsorption. As SF concentration increases, the zeta potential of the NPs approaches that of pure SF solution (~ −3.17 mV). Additionally, Supplementary Fig. 8 demonstrates that, at saturated SF adsorption level (i.e., SF200), the hydrodynamic radius distribution shifts upward relative to the original NPs.

The SF adsorption-induced change in NP surface energy (Fig. 1g, Supplementary Fig. 9) is confirmed by AFM force spectroscopy. The adhesion force between a diamond-like carbon colloidal probe and SF-adsorbed NPs exhibits a clear non-monotonic dependence on SF concentration: adhesion increases from the pristine ( ~ 14.4 nN) and SF002 conditions ( ~ 22.3 nN) to a pronounced maximum at SF020 ( ~ 32.3 nN) and then decreases again at SF200 ( ~ 21.8 nN). This result identifies that there is an optimal adsorption window maximizing the particle-to-particle interaction.

Collectively, these AFM results, together with electron microscopy, IR spectroscopy, and DLS, confirm both the concentration-dependent SF adsorption and its controllable mechanical influence on NPs (i.e., hydrodynamic radii, surface charge, and surface energy). Such tunable surface modification is expected to strongly influence intermolecular interactions among NPs during water-based assembly.

Enhanced intermolecular interactions by SF adsorbate

Significantly, SF adsorption can amplify both interparticle and interfacial adhesion through polymer-mediated interactions (Fig. 2). NPs in solution undergo Brownian motions that bring them into frequent diffusive contact with other particles, SF chains, and interfacial surfaces^30,31. However, to quantitatively analyze the adhesion dynamics at the nanoscale interface, a controlled and static approach is required, in which a fixed NP interacts with a hydrophobic solid surface under aqueous conditions at varying SF concentrations (Fig. 2a). Adhesion properties such as adhesion forces and energy dissipation are quantified by analyzing force-distance (FD) curves (Fig. 2b)³². Physical modeling of the approach and retract curves provides additional insights into the local elasticity changes in the interface region (i.e., 0 – 35 nm from the interface) and the molecular-level mechanics within the SF adsorption layer (see Supplementary Note 1, Supplementary Fig. 10).

Fig. 2. Enhanced particle adhesion by SF adsorption.

a Schematic of AFM setup used in this study to measure adhesion force and energy dissipation between NPs and a substrate in the presence of SF. b Example FD curve and analyzable information. c, d two-dimensional (2D) contour maps ( ~ 10 μm², 49 points) for adhesion force and energy dissipation between the colloidal probe and the substrate in the presence of SF with varying concentrations in the solution. Scale bars, 2 μm. Numbers in the lower left corner represent mean ± SD.

Figures 2c, d illustrate the SF-adsorption-dependent adhesion evolution across a 10 μm² area (N = 49 points) with increasing SF concentrations. SF adsorption leads to an initial rise in both adhesion force and energy dissipation, reaching 3.5 nN and 0.32 fJ at SF020, respectively, up from 1.7 nN and 0.12 fJ. However, at higher silk concentrations (i.e., SF200), both adhesion properties are compromised. The measured FD curves for each SF level are provided in Supplementary Fig. 11. In addition, contact time significantly affects the adhesion properties between two solids, as shown in Supplementary Fig. 12. Rapid increase in adhesion is observed when the contact time increases from 0.02 s to 10 s. Beyond 10 s, the adhesion properties begin to level off, indicating that the molecular dynamics within SF adsorption are approaching a steady state.

Concentration- and time-dependent adhesion behaviors can be attributed to the presence of the SF adsorption layer and its polymeric network evolution during physical contact between the NP and the solid surface (see Supplementary Note 2). Under low SF adsorption conditions (i.e., <0.002 w/v%), low adhesion and simple FD curves are observed (Supplementary Fig. 11a). When the SF adsorption is optimal and sufficient time for molecular dynamics within the adsorption layer is allowed, conformational rearrangements and network entanglements within the SF chains become more prominent, contributing effectively to stronger particle interactions (Supplementary Fig. 12). This phenomenon is reflected in increasing persistence and contour lengths in the SF adsorption layer as well as emerging bond rupture peaks (Supplementary Fig. 10d–f), a characteristic commonly observed in adhesion by polymers^{26–28,32,33}. However, when the adsorption layer becomes too thick and stiffer, its reduced ability to conform to surface asperities and the extra viscoelastic energy dissipated in the bulk can diminish the real contact area and weaken interfacial bonding, thereby potentially lowering the measured adhesion strength, as described in the supplementary text and observed in the previous study³⁴. These results highlight the need to precisely control SF concentration to induce the optimal level of adsorption, which can directly enhance the particle assembly during the aqueous coating procedure.

Wetting improvement and high-density NP coatings enabled by optimal SF adsorption

Significantly, the ability to control SF adsorption and fibroin-mediated intermolecular interaction improve the wetting of aqueous NP dispersions and enable high-density, uniform coating of NPs even on low-energy substrates (Fig. 3). Spin-coating of aqueous NP dispersions onto hydrophobic substrates like PDMS or PTFE reveals significant improvement in coating coverage depending on the SF adsorption level (Fig. 3a). Supplementary Figs. 13–15 show that less than SF002 adsorption is insufficient for optimal coating results. In contrast, at 0.02–1.0 w/v% of SF, surface coverage and packing density of NP assembly are gradually improved with the adsorption levels. When SF concentrations exceed 2.0 w/v%, SF becomes the continuous phase, embedding the NPs and disrupting ordered assembly. Regardless of NPs, these SEM results show that an SF020 adsorption level is the most effective range for high-density NP layer coatings while using minimal amounts of SF. When it comes to forming high-density and uniform NP layers, SF002 is insufficient, SF020 level gives optimal results for all studied NPs, and SF200 is excessive. Thus, hereafter, we refer to these concentrations as ‘insufficient’, ‘optimal’, and ‘excessive’, respectively. Figure 3b, c highlight the contrast between uniform, dense SiO₂ NP assembly on PDMS at optimal SF concentration (i.e., SF020) and the disrupted composite-like SF–SiO₂ layer with excessive SF (i.e., SF200). Excess free SF remaining in the dispersion can accumulate between NPs, leading to thicker coated films (Supplementary Fig. 16). This optimal adsorption level could vary upon the concentration and size of NPs. It should be noted that throughout this study, NP dispersions with low concentration ( < 10%) and nanoscale diameter (⌀ < 200 nm) were used.

Fig. 3. Water-based coating for high-density NP assembly enabled by optimal SF adsorption.

a A schematic illustrating the distinct coating outcomes of aqueous NP dispersions on hydrophobic substrates (e.g., PDMS, PTFE) at different SF adsorption levels. b, c Cross-sectional SEM images showing the SiO₂ NP layer coated on the PDMS layer with 0.2 w/v% and 2.0 w/v% SF adsorption, respectively. Scale bars, 500 nm. d Characteristic XPS substrate signals before and after NP coating on PTFE and PDMS with optimal SF adsorption. The color bars and error bars show the mean ± SD (n = 3 for original, n = 15 for NP-coated). e Schematic representation of a fully water-based sequential coating process for constructing multilayered NP layers. f SEM image of the sequentially coated NP layers and EDS mapping showing the elemental distribution in the multi NP layers. Scale bars, 300 nm.

Spin coating of various aqueous NP dispersions with optimal SF adsorption onto PTFE and PDMS surfaces results in high-density NP layers. X-ray photoelectron spectroscopy (XPS) analysis quantitatively confirms that these coatings fully obscure the underlying substrate, demonstrating effective surface coverage at the nanoscale. Figure 3d and Supplementary Fig. 17 illustrates the reduction in fluorine and silicon signals from PTFE and PDMS, respectively, after coating of NPs. XPS survey scans for each NP-coated substrate are shown in Supplementary Figs. 18, 19. These results quantitatively confirm that aqueous NP dispersions with optimal SF adsorption produced well-coated NP layers on hydrophobic substrates³⁵.

The SF-adsorbed NP assemblies formed on hydrophobic or elastic substrates remain well-adhered and electrically function even under moderate mechanical deformation (bending radius ≥ 20 mm), without noticeable delamination or loss of interconnectivity (Supplementary Fig. 20). Introducing a thin elastomeric encapsulation layer further enhances the stability and reliability of these assemblies, ensuring practical applicability in flexible or bio-integrated devices (Supplementary Fig. 21).

Furthermore, the structural conformation of the SF adsorption over NP layers can be manipulated via previously reported post-processing techniques like solvent annealing (e.g., water vapor or methanol vapor annealing)^36,37. Although solvent-induced crystallization may dehydrate the SF adsorption layer and alter its mechanical properties, its influence on interparticle adhesion at the optimal adsorption condition (i.e., SF020) is minimal (Supplementary Fig. 22). At this concentration, SF chains have already undergone substantial conformational rearrangement and network formation during adsorption, and further crystallization might mainly consolidate the existing adhesive interactions. Notably, crystallized SF adsorbates stabilize the interparticle packing against redispersion in water, allowing sequential coating of aqueous NP dispersions (Fig. 3e, f). Sequential spin-coating with crystallization of optimal SF-adsorbed SiO₂, zinc oxide (ZnO), and indium tin oxide (ITO) NPs enables the production of a multi-layered structure with clear and distinct boundaries. In contrast, sequential coating without crystallization results in partial intermixing of the previously deposited NP layers at the interfaces (Supplementary Fig. 23).

These results demonstrate that the optimal level of SF adsorption not only maximizes the fibroin-mediated intermolecular interactions but also enables high-density NP assembly on previously non-wettable substrates via water processing. In this aqueous process, the coating behavior is jointly governed by an interfacially adsorbed SF layer that defines NP assembly and by a minor residual free-SF fraction that modulates wetting^19,20. This interplay establishes an optimal adsorption window in which film coverage and uniformity are maximized. Furthermore, the polymorphism of SF offers additional benefits in processing tunability for building reliable multilayered NP heterostructures with distinct layer boundaries.

Preserved functionality of NP-based electronics at optimal SF adsorption

Optimal SF adsorption enables the water-based fabrication of high-density NP assembly, preserving the inherent electrical functionality of the original NPs^38,39. In this study, three types of NPs assemblies are formed via spin coating without thermal treatments, and their electrical performance is analyzed (Fig. 4). NP coatings with excessive SF adsorption, which produce composite-like layers, are excluded from this study, as they are beyond the scope of this study. All SF-adsorbed NP assemblies are treated with methanol to crystallize the SF adsorption layer, thereby improving the device stability against ambient humidity.

Fig. 4. High-density NP assembly electronics.

a A schematic of a high-density SiO₂ NP-based capacitor and its cross-sectional SEM image. b, c Current density-electric field and capacitance-frequency curves of SiO₂ NP-based capacitors with varying SF adsorption levels and silk film (n = 5). d A schematic of an Ag NW/ITO NP hybrid conductor and its cross-sectional SEM image. e Conductivity of ITO NP-based conductors as a function of SF adsorption levels (n = 5). f Conductivity and transmittance of Ag NW/ITO NP hybrid conductors with varying Ag NW concentrations (n = 6). g A schematic of a high-density ZnO NP-based transistor and its cross-sectional SEM image. h–j Transfer curves, subthreshold swing, and mobility of ZnO NP-based transistors with varying SF adsorption levels (n > 3). Scale bars in (a, d, g), 500 nm. The colored shade area in (c) and (f) indicates SD. The colored bars and error bars in (e, i, j) indicate mean ± SD.

Capacitors composed of SiO₂ NPs with a uniform and high-density NP layer are fabricated (Fig. 4a) and characterized to assess the influence of different SF adsorption levels on their insulation and capacitance characteristics (Figs. 4b, c). As shown in current density-electric field (J–E) curves (Fig. 4b), the pure SiO₂ NP capacitor and that with optimal SF adsorption exhibited comparable leakage currents and electrical breakdowns (10⁻⁷A/cm² and 5.2 MV/cm for pure NP layers, while 10⁻⁶ A/cm² and 4.9 MV/cm for NP with SF adsorption, respectively). On the other hand, the insufficient-SF adsorbed SiO₂ NP capacitor exhibits vertical cracks across the film (see Supplementary Fig. 24a), creating electrical paths between the electrodes and leading to poor insulating performance. J–E curves for devices with varying SF adsorption levels are provided in Supplementary Fig. 25. Figure 4c shows the frequency-dependent capacitance of the NP layers. The average dielectric constants for pure crystallized SF films (thickness ~274 nm), optimal-SF adsorbed SiO₂ NP capacitors (thickness ~321 nm), and pure SiO₂ NP capacitors (thickness ~ 346 nm) were 17.4 ± 0.9, 8.21 ± 0.5, and 5.97 ± 0.4 at 10 kHz, respectively, which are comparable to other solution-processed dielectrics^40,41. These results demonstrate that SF adsorption effectively facilitates the formation of high-density NP layers without compromising the intrinsic dielectric performance of the NP.

Figures 4d–f show a transparent NP conductor formed onto a PDMS strip. Optimal SF adsorption allows a continuous and closely packed ITO NPs layer on PDMS, causing the resultant film to exhibit more than double the conductivity of pure ITO NP layers and of the those formed with unoptimized/insufficient SF adsorption (Fig. 4e). These results suggest that, despite SF’s well-known dielectric nature, optimal usage of SF-adsorbate maintains physical connections between NPs, resulting in net conductivity enhancement. From a practical perspective, relying solely on zero-dimensional nanoparticles to form a strong percolation path is inefficient⁴². Hence, a hybrid percolation path using high-aspect-ratio (HAR) silver nanowires (Ag NWs) and ITO NPs is also explored. As shown in Fig. 4f, Supplementary Fig. 26a, introducing Ag NWs (HAR ~ 250) beneath the ITO NP layer increases conductivity by approximately three orders of magnitude, from 50 S/m to as high as 80,000 S/m. While the hybrid NP conductors’ conductivity is 100 times lower than sputtered ITO films ( ~ 10⁶S/m), it was still reasonable considering contact resistance and charge carrier scattering occurring at NP junctions^42,43. With appropriate encapsulation, the SF-adsorbed hybrid conductor formed on soft PDMS substrates can be mechanically stabilized, maintaining a consistent percolation network even under repeated tensile bending (up to 350 cycles at a bending radius of 20 mm; Supplementary Fig. 21b). Furthermore, compared to solution-processed conductive materials, the Ag NW/ITO NP hybrid conductor offers comparable or higher conductivity, underscoring the versatility of this approach^44–46. The optical transparency is compromised by increasing Ag NWs, as the NW meshes act as optical scattering points. Thus, the trade-off between conductivity and transmittance needs to be balanced depending on application requirements. Supplementary Fig. 26b shows the transparency of the Ag NW/ITO NP hybrid conductor at different Ag NW concentrations.

Analysis of the transfer characteristics of the semiconducting NPs layer with optimal SF adsorbates is illustrated in Figs. 4g–j, Supplementary Fig. 27. Pure NP transistors and those with optimal SF adsorption exhibit similar on/off current ratios of approximately 10⁵. On the other hand, ZnO NP layers with insufficient SF adsorbate exhibit poor surface coverage and defects (e.g., voids and cracks) (see Supplementary Fig. 24c), leading to a 70% failure rate, as shown in Supplementary Fig. 27. The subthreshold swings of both pure and SF-adsorbed ZnO NP transistors are measured to be around 250 mV/dec, showing no significant difference. In contrast, transistors with optimal-SF adsorption exhibit slightly higher mobility ( ~ 0.2 cm²/Vs) than the others (p-value ~ 0.018). This mobility improvement is attributed to the close NP packing and dielectric screening provided by the SF adsorption layer, which enhances charge carrier transport while reducing scattering from charged impurities in the NPs^42,47–49.

Discussion

Collectively, this study highlights the distinctive potential of SF as an enabler for water-based fabrication of high-density, functional NP layers. Experimental evidence shows that the amphiphilic nature of SF leads to its spontaneous adsorption on various NPs in aqueous environments, playing the role of an interparticle-binding surfactant. By examining NP interaction, an optimal SF concentration (i.e., SF020) is found for effective adsorption, which universally enhances both the wetting of aqueous NP dispersions and the close packing of NPs. The resulting high-density NP electronics, stabilized by strong interparticle bonds and low defect density, preserved the NPs’ original functionality. These NP electronics exhibit the electrical performance comparable to or greater than pure NP devices, thanks to higher packing densities and continuous charge carrier paths (as illustrated in Supplementary Fig. 28). This technique can be extended to plasmonic or photonic NP-based devices. For example, as shown in Supplementary Fig. 29, the Au NP and Au nanorod (NR) layers spin-coated on PDMS form uniformly and closely packed monolayers, and their localized surface plasmon resonance (LSPR) spectra remain essentially unchanged despite SF adsorption. Similarly, it is expected that SF can be utilized to form display layers composed of quantum dots (QDs), upconverting nanoparticles (UCNPs), or other materials, which can be potentially utilized in bio-photonic interfaces. This work offers a versatile strategy for the seamless integration of biological materials with nanotechnology, holding significant utility for bio-integrated systems and applications at the biotic-abiotic interface.

Methods

Materials

Water-based 10% carboxylate-modified polystyrene (PS, ⌀ ~ 100, 300, and 600 nm) nanosphere dispersions are purchased from Banglabs. Water-dispersed gold nanoparticles (Au NPs) and gold nanorods (Au NRs) are prepared with previously reported methods^50,51 and as described in this “Method” section. Water-based 20% zirconium oxide (ZrO₂, diameter ~ 50 nm) dispersion is purchased from USNANO. Gold (III) chloride trihydrate (HAuCl₄), sodium citrate, cetyltrimethylammonium bromide (CTAB), sodium borohydride (NaBH₄), silver nitrate (AgNO₃), L-ascorbic acid, water-based 50% silicon dioxide (SiO₂, Ludox TM-50, diameter ~ 22 nm) colloidal, 20% indium tin oxide (ITO, diameter ~ 18 nm), 20% zinc oxide (ZnO, diameter ~ 40 nm) dispersions, sodium carbonate (Na₂CO₃), and lithium bromide (LiBr) are purchased from Sigma Aldrich. Water-based 2% silver nanowires (Ag NWs, diameter ~ 120 nm, length ~ 20 μm) dispersion is purchased from MSE Supplies. Lacey carbon grid for transmission electron microscopy (TEM) measurement is purchased from Ted Pella. All nanoparticles (NPs), colloidal and dispersions are used as received without further purification and diluted accordingly with deionized water (DIW, 18.2 MΩ·cm), obtained from a Milli-Q Nanopure system.

Preparation of regenerated silk fibroin (SF)

SF is prepared with a process previously described³⁶. Briefly, Bombyx mori silk cocoons were cut and boiled in 0.02 M Na₂CO₃ (Sigma Aldrich) solution for 30 min to remove sericin. The dried and degummed silk was dissolved in 9.3 M LiBr (Sigma Aldrich) solution at 60 °C for 1 h. The dissolved SF solution was placed into a dialysis tube (Fisherbrand, MWCO 3.5 K), followed by dialysis against DIW for 3 days with at least 5 water changes. After dialysis, the solution was purified through centrifugation at 11,400 g for 20 min at 4 °C. The 30 min-boiled SF prepared in this manner exhibits a molecular weight distribution of ~219–258 kDa¹⁹. The resulting SF solution was diluted to 4.0 w/v% and served as a stock solution.

Preparation of Au NPs and Au NRs

AuNPs were synthesized by boiling 100 mL of 1.0 mM HAuCl₄ solution, followed by rapid injection of 10 mL of 38.8 mM sodium citrate under vigorous stirring. The reaction was maintained for 20 min until the solution turned deep red. AuNRs were synthesized using a seed-mediated growth method. The seed solution was prepared by mixing 0.4 mL of 50 mM HAuCl₄ with 40 mL of 0.2 M CTAB, followed by rapid addition of 4.8 mL of ice-cold 0.01 M NaBH₄ and stirring for 2 min at 25 °C. For the growth solution, 40 mL of 0.2 M CTAB was combined with 0.8 mL of 4 mM AgNO₃, 0.8 mL of 50 mM HAuCl₄, and 0.56 mL of 0.0788 M L-ascorbic acid. Subsequently, 0.1 mL of the seed solution was added to the growth solution, and the mixture was kept at 27 °C overnight to form NRs.

Preparation of SF-adsorbed NP suspensions with varying NP and SF concentration

The concentration of NPs in the water-based suspensions was appropriately adjusted according to the intended purpose, such as observing SF adsorptions in monolayer formation or coating of crack-free NP films. The concentrations of SF and NPs in the final suspensions were adjusted by mixing their stock solutions with DIW in appropriate volumetric ratios. Unless otherwise noted, all suspensions were prepared in unbuffered DIW with no added salts/ions and used at room temperature under near-neutral pH (typically pH 6.8–7.2 immediately after mixing). No wash/centrifugation step was applied to remove non-adsorbed free SF chains from the NP suspensions. Supplementary Table 1 shows an example of the mixing ratios of each stock solution and DIW used in this study.

Optical characterizations of SF-adsorbed NP suspensions and their coating results

A field emission TEM (JEOL F200, JEOL) was used to image NPs with or without SF adsorption at the acceleration voltage of 200 kV. NP suspensions were prepared according to Supplementary Table 1. For TEM, 5 μL of NP suspension was drop-cast onto lacey-carbon grids and air-dried overnight at ambient conditions, without any surface treatment or accelerated drying. Each grid was first surveyed at low magnification to locate regions where a single, close-packed particle layer spans a lacey opening; high-magnification images were then acquired to visualize the SF adsorption layer. The morphology and elemental composition of the samples were analyzed using scanning electron microscopy (SEM, EVO MA10, Zeiss). High-magnification images were acquired with a field emission SEM (FE-SEM, Sigma 300, Zeiss) equipped with an SE2 detector with a 60 µm aperture, operated at 7 kV with a working distance of 8.6 mm. To ensure electrical conductivity, the samples were mounted on aluminum stubs using conductive carbon tape and coated with approximately 13 nm of carbon using a sputter coater (Q150T ES Plus, Quorum). Elemental composition analysis was performed using an energy dispersive X-ray spectrometer (EDS, Ultimax 65, Oxford Instruments) detector with Aztec 6.0 software. The elemental maps were acquired using an energy range of 10 keV, with a pixel size of 0.00082 µm and accumulating three frames per map. Digital images are taken using a digital single-lens reflex (DSLR) camera (Rebel EOS-SL1, Canon).

Correlative AFM-IR imaging of SF-adsorbed NPs assembly (nanoscale)

The surface topography and selective chemical mapping of SF-adsorbed PSNPs (Ø ~ 500 nm) assemblies were obtained using an atomic force microscopy-infrared spectroscopy (AFM-IR, Dimension IconIR, Bruker). Prior to correlative imaging, IR spectroscopy was conducted on bare PSNPs and SF films within the 1800–800 cm⁻¹ range to determine their characteristic IR absorption bands. Correlative imaging of NPs assemblies with various SF concentrations (i.e., 0.02 and 0.2 w/v% SF) was performed in tapping mode using a gold-coated cantilever (PR-UM-TnIR-D, Bruker). In order to enhance the IR sensitivity and suppress the substrate signal, SF-adsorbed NPs assemblies were prepared over 70 nm gold-coated glass substrates. Neither the SF films nor the SF-adsorbed PSNPs assemblies were subjected to post-crystallization treatment. Height profiles were acquired by measuring cantilever deflections over a scan area of 500 nm², simultaneously recording IR absorbance at 1650 cm⁻¹ to map the spatial distribution of the SF-specific Amide I band.

Correlative Raman-SEM imaging of SF-adsorbed NPs assembly (microscale)

The morphology and conformation of the microscale SF-adsorbed PS particles (Ø ~ 20 µm) assemblies were investigated using the Fe-SEM (Sigma 300, Zeiss) with an in situ confocal Raman RISE setup. This setup enables confocal Raman imaging of the same sample positions from which the SEM images are taken⁵². Samples were mounted on aluminum stubs using conductive carbon tape and imaged as is with no additional coating. SEM images were acquired under high vacuum at low acceleration voltages EHT = 0.7 kV, with a secondary InLens detector, and working distances set to WD = 4-4.4 mm. Raman spectra were acquired using a Witec WRL-532-E-100-TP laser (λexcitation = 532 nm, P = 75 mW) equipped with a UHTS300VIS imaging spectrograph (600 g/mm grating), a PE cooled CCD, and a 3D piezoscan stage (250 μm x 250 μm x 250 μm). The laser was focused through a Zeiss LD EC Epiplan-Neofluar Vac Dic 100x objective (WD = 4 mm, NA = 0.75). Spectra were acquired at laser power P = 5 mW, integration time t = 1 s, and accumulations N = 30, while working at room temperature. Acquired Raman spectra were further processed using the Witec Control 6.2 software to perform cosmic ray removal, background subtraction, and smoothing using the Savitzky-Golay method. Corresponding optical micrographs were acquired using a DFK-33GX178 Imaging Source camera.

Hydrodynamic characterization of SF-adsorbed NP suspensions

Dynamic light scattering (DLS) and zeta potential measurements were performed using a Zetasizer Pro (Malvern) to determine the hydrodynamic radius of the particles and assess the stability of NP suspensions at varying SF concentrations (ranging from 0 w/v% to 2.0 w/v%). For reference, the zeta potential of a 5.0 w/v% aqueous silk solution was also measured from three different samples and found to be −3.17 ± 0.48 (s.d.) mV. To maintain a pH, 100 μL of 1X phosphate-buffered saline (PBS, pH 7.4) was added to each tested dispersion.

Adhesion characterization between SF-adsorbed colloidal probes and hydrophobic substrates

A bio-AFM (JPK Nanowizard, Bruker), along with a diamond-like carbon cantilever (Biosphere B300-FM, nanotools) or a colloidal cantilever (CP-PNPS-PS-A-5, sQube), was used to measure the interaction force between the NP layer and the cantilever. The cantilever approached the NP layer and was held for varying seconds from 0.02 s to 120 s after the contact, then retracted. The compressing force was set to 1 nN for characterizing time-dependent adhesion behavior; otherwise set to 10 nN. The adhesion force and energy dissipation between the NP layer and the cantilever were extracted from the force-distance (FD) curves using JPKSPM Data processing software. To characterize the changes in elastic modulus near the interface region, area-corrected fitting, tailored to the probe geometry, was performed on approach FD curves using the user-defined fitting function of OriginPro 2019 software. Similarly, worm-like chain (WLC) polymer extension model fitting was conducted on the observed elongation peaks in retracting FD curves in the same manner.

Optical and surface chemical characterization of SF-adsorbed NP coatings

Visual inspection for NP coatings (e.g., surface coverage and cracks, etc.) was performed using an optical microscope (OM) (Eclipse LV100, Nikon) with 20x and 100x objective lenses and top-view SEM observation. The thickness of NP coatings was measured using cross-sectional view SEM observation, cross-checked with a spectroscopic ellipsometer (RC2, J.A. Woollam). X-ray photoelectron spectroscopy (XPS) (K-Alpha plus, Thermo Fisher Scientific) was used to measure the atomic composition of the NP coating and the substrate.

Preparation of SF-adsorbed NP coatings for electronic devices

Water-based dispersions of 4% SiO₂, 10% ITO, 0.5 ~ 2% Ag NWs, and 10% ZnO with varying SF concentrations (0, 0.02, and 0.2 w/v%) were prepared for device fabrication. UV ozone-treated substrates were used for coating dispersions without silk, while pristine substrates were used for those with silk. For SiO₂ NP-based capacitors, a SiO₂ colloidal solution was spin-coated at 2000 rpm for 60 s onto ITO/glass substrates. Circular electrodes (⌀ ~ 300 μm) were thermally deposited on the SiO₂ NP layer coating. An 8 w/v% silk solution was spin-coated at 2000 rpm onto ITO/glass substrates for the comparison sample. For ITO/Ag NWs-based transparent conductors, Ag NWs dispersions were spin-coated at 4000 rpm for 10 s onto clean polydimethylsiloxane (PDMS) strips, followed by spin-coating ITO dispersions at 2000 rpm for 60 s over the Ag NWs layer. For bottom-gated ZnO NP-based field-effect transistors (FETs), a ZnO dispersion was spin-coated at 2000 rpm for 60 s onto a 250 nm-thick SiO₂/silicon substrate (As-doped, 0.001 ~ 0.005 Ohm-cm). Aluminum electrode pairs (gap ~100 μm, width ~800 μm) were thermally deposited on the ZnO NP layer. All NP layers were crystallized by spin-coating 50 μL of pure methanol at 2000 rpm for 60 s.

Characterization of SF-adsorbed NP electronic devices

The electrical properties of the aqueous NP-based devices were analyzed using a parameter analyzer (4200A-SCS, Keithley). The current density-electric field (J-E) curves for the SiO₂ NP layer and silk film (thickness ~ 700 nm) were obtained by linearly sweeping the voltage from −2 to 200 V across top-bottom aluminum electrodes. The measured current and voltage were converted to current density and electric field using the film thickness and electrode area. Capacitance per unit area was calculated by measuring the dielectric layer’s capacitance over 1 kHz to 1 MHz and dividing it by the electrode area. The dielectric constant was determined from the calculated capacitance per unit area, the vacuum permittivity, and the measured film thickness. The sheet resistance of ITO/Ag NWs-based transparent conductors was measured using a four-point probe (FPP-5000, Miller) and then converted to conductivity using the measured film thickness. The spectral transmittance of these conductors was measured using a custom-built microscope spectrometer setup (USB2000, Ocean Optics). The transfer characteristics of ZnO-based transistors were inspected via three-terminal measurements where drain electrodes were grounded while the source-gate voltage (VGS) and source-drain voltage (VDS) were manipulated. Electron mobility and subthreshold swing of the ZnO NPs layer were extracted from the forward sweep transfer curves.

Statistical analysis of the acquired data

At least three data points were collected for each measurement in this study. Mean and standard deviation were calculated from these data points and are presented as indicated in the figure captions. Analysis of variance (ANOVA) and paired t-tests with an alpha value of 0.05 were performed for mean comparisons using JMP Pro 17 software. No data were excluded from the analyses, and all experiments were conducted without randomization. The statistical significance is denoted as not significant (ns), * P < 0.05, ** P < 0.01, and *** P < 0.001.

Author contributions

Conceptualization: T.K. and F.G.O.; Methodology: T.K., C.K., N.G., N.O.S., and G.G.; Investigation: T.K., C.K., N.G., N.O.S., G.G., and S.B.; Visualization: T.K.; Funding acquisition: F.G.O.; Project administration: F.G.O.; Supervision: F.G.O.; Writing – original draft: T.K. and F.G.O.; Writing – review & editing: T.K., C.K., N.G., N.O.S., G.G. and F.G.O.

Peer review

Peer review information

Nature Communications thanks the anonymous reviewers for their contribution to the peer review of this work. A peer review file is available.

Data availability

The main data that support the findings of this study are available in this article and its Supplementary Information. The data generated in this study are provided in the Supplementary Information/Source Data file. Data is available from the corresponding author upon request. Source data are provided with this paper.

Competing interests

The authors declare no conflict of interest.

Supplementary information

The online version contains supplementary material available at 10.1038/s41467-026-68499-w.