Microribbons composed of directionally self-assembled nanoflakes as highly stretchable ionic neural electrodes

Significance

Direct-ink writing has been attracting significant attention as a three-dimensional printing method while symmetric structures can be mainly printed when a single ink is used. Here, we present the formation of asymmetric microstructures during printing and ink drying, induced by a symmetry-breaking self-assembly process of aligned two-dimensional nanoflakes in the viscous polymer matrix. The obtained structural variation in the printed microribbons can be controlled in a reproducible and standardizable manner by tuning the rheological properties of the printing ink. We further demonstrated the spontaneous transformation of as-printed microribbons to ultrastretchable springs upon stimulation that show superior nanoscale ionic transport behavior as soft and highly stretchable neural electrodes.

Abstract

Many natural materials possess built-in structural variation, endowing them with superior performance. However, it is challenging to realize programmable structural variation in self-assembled synthetic materials since self-assembly processes usually generate uniform and ordered structures. Here, we report the formation of asymmetric microribbons composed of directionally self-assembled two-dimensional nanoflakes in a polymeric matrix during three-dimensional direct-ink printing. The printed ribbons with embedded structural variations show site-specific variance in their mechanical properties. Remarkably, the ribbons can spontaneously transform into ultrastretchable springs with controllable helical architecture upon stimulation. Such springs also exhibit superior nanoscale transport behavior as nanofluidic ionic conductors under even ultralarge tensile strains (>1,000%). Furthermore, to show possible real-world uses of such materials, we demonstrate in vivo neural recording and stimulation using such springs in a bullfrog animal model. Thus, such springs can be used as neural electrodes compatible with soft and dynamic biological tissues.

Evolution endows biological materials with complex structural variation by guiding the assembly of building blocks across multiple length scales, contributing to their superior mechanical properties and diverse functions (1, 2). Examples of such materials include hierarchically arranged proteins in animal silk and hair (3, 4) and site-specific orientation of stiff fibrils in animal bone, shell, and tooth (5–7), rendering these biomaterials both flaw-tolerant and mechanically strong. In particular, locally anisotropic arrangement of cellulose in pinecones, seedpods, and awns (8–10) enables them to change their morphology to adapt to various environments for their seed dispersal. To mimic these ingenious strategies for smart materials, researchers have been proposing various assembly techniques to incorporate structural anisotropy and diversity into synthetic materials (11, 12).

Self-assembly, as one of the few practical assembly methods for constructing nanomaterials, is a process of spontaneously organizing building blocks into ordered structures without human intervention (13). Studies on the automatic construction of the building blocks enabled by internal interaction forces (e.g., covalent and noncovalent interactions) or external stimuli (e.g., electric field, magnetic field, and light) help to understand the complex biological behavior and contribute to the development of nanofabrication techniques as well (14). However, it is challenging to realize programmable structural variations in self-assembled materials since highly symmetric and uniform structures usually dominate such assemblies as a result of uniform equilibrium interactions (15, 16).

On the other hand, ionic transport widely exists in plants and animals, and it helps to maintain normal physiological activities (17, 18). Interestingly, ionic transport through a confined nanoscale fluidic channel is significantly different from that in bulk (19, 20). The ionic current in a nanofluidic channel with a charged surface is dominated by monopolar counterions at low salt concentrations, and it can largely enhance ionic conductance up to several orders of magnitude compared with that of bulk solution (21). Therefore, incorporating structural variations into nanofluidic ionic assemblies may enable new properties and broaden their applications. Considering the same ionic-conducting nature of nanofluidic ionics and biological tissues, nanofluidic ionic assemblies may have promising applications in bioelectronic devices (22). In particular, soft and stretchable nanofluidic ionic conductors are promising for replacing conventional rigid and dry metal or inorganic oxide electrodes (23).

Here, we show the formation of an ionic conducting microribbon composed of directionally self-assembled two-dimensional (2D) nanoflakes in a polymeric matrix during three-dimensional (3D) printing, which can spontaneously transform into an ultrastretchable spring upon stimulation and serve as a soft and highly stretchable neural electrode. The 2D nanoflakes in the polymer matrix are prealigned when extruded from the printing nozzle then follow a process of solvent-evaporation-assisted asymmetric self-assembly on the substrate, leading to the formation of a ribbon with site-specifically varied mechanical properties. Upon stimulation, the ribbons with graded structures morph their shape into ultrastretchable helical springs, which can be applied as nanofluidic ionic conductors with high conductivity and large elasticity. We further demonstrate their application as soft and biocompatible neural electrodes for reliable neural recording and stimulation.

Results and Discussion

Formation of Directionally Self-Assembled Microribbons.

A hybrid solution, composed of 2D graphene oxide (GO) dispersed in sodium alginate (SA) viscous solution, was used as the 3D printing ink. GO, as a typical monolayer nanosheet (SI Appendix, Fig. S1) with masses of oxygen-containing functional groups in the plane, is proved to be a superior reinforcement nanofiller. The extremely high aspect ratio of GO nanosheets endows them with remarkable anisotropy (24) that provides the basis for the construction of structural variation observed in the printed GO/SA ribbons. We observed that 2D nanofillers in a viscous polymeric matrix attempt to directionally self-assemble on a substrate and form site-specific structural variations.

Fig. 1A shows the formation of a directionally self-assembled ribbon with a graded structure. Fig. 1 A, i schematically shows that the printing ink is composed of randomly distributed GO flakes in SA polymeric matrix, which is evidenced by the scanning electron microscope (SEM) image (Fig. 1 B, i). A diffraction ring of (002) plane in the 2D wide-angle X-ray diffraction (WAXD) pattern of the GO/SA ink shows nearly uniform intensity at all azimuthal angles and a low Herman’s orientation factor (f) of 0.006 (Fig. 1 C, i), indicating the isotropic nature of the printing ink. During printing, these GO flakes are then aligned with their planes parallel to the flow direction (or with their director n perpendicular to the flow direction) when extruded from a nozzle (Fig. 1 A, ii and B, ii), which can be ascribed to the large shear force in the nozzle and thus increases the GO orientation (f = 0.253; Fig. 1 C, ii). Notably, the circular cross-section of the extruded filament shows that the parallelly aligned GO flakes are radially distributed (Fig. 1 A, ii, B, ii, and SI Appendix, Fig. S2), which can be ascribed to the radial flow of the ink induced by swell of the extrudate and thus the generated extensional rate leads to radial alignment of the GO flakes (25, 26). Z = 1/Oh (Oh is the Ohnesorge number; see details in Materials and Methods) is used to characterize and analyze the printability of the ink (27). Z of our printing ink is on the order of 10⁻¹, indicating continuous and smooth extrusion can be realized during the direct-ink-writing printing. In contrast, Z >2 is usually required to achieve single-drop ejection in inkjet printing for drop-on-demand printing (27).

Fig. 1.

Formation of directionally self-assembled ribbon with graded structure. (A) Schematic illustration showing the evolution of the orientation of GO during the extrusion and drying process (i–v). (B) SEM images showing the evolution of GO orientation during printing and deposition process (i–v). (Scale bars, 50 μm.) (C) The 2D WAXD patterns and their corresponding Herman’s orientation factors (f) showing the change of orientation degree (i–v).

When the extruded filament is deposited on a substrate, the radially distributed GO flakes continue to reorient until the water fully evaporates (Fig. 1 A, ii–v). Note that the motion range of GO flakes in different portions of the cross-section is different. With the confinement of substrate, GO flakes in the bottom portion tend to lie flat when contacting with the substrate, forming a planar alignment (i.e., a typical bricks-and-mortar structure). In contrast, as the top portion is exposed to the ambient environment the water evaporates and the viscosity of the left ink thus increases, which causes an enhanced energy barrier for the movement of GO flakes in the top portion and greatly restricts their motion, resulting in only a small deflection from their original location and forming a homeotropic alignment (Fig. 1 B, iii–v). As a result, a sharp contrast of GO orientation in the bottom and top portions of the obtained ribbons was observed. During the process, the reorientation of GO flakes contributed to the increased orientation degree of GO flakes, and it was proved by the stepwise narrowed azimuthal-integrated intensity distribution curves (SI Appendix, Fig. S3) and increased f from 0.253 to 0.782 (Fig. 1 C, ii–v).

Site-Specific Mechanical Properties of the Graded Structure.

The directional self-assembly of GO flakes leads to a significant structural variation of the printed microribbon, thus resulting in different mechanical properties along the direction of its thickness. Fig. 2A shows distinctly different morphologies between the top and bottom portions of the ribbons. As shown in Fig. 2 A–C, loosely and vertically distributed GO flakes can be seen in the top portion (Fig. 2B), while an orderly and planarly stacked microstructure is observed in the bottom one (Fig. 2C). As shown in Fig. 2D and SI Appendix, Fig. S4, the top surface shows a rough surface with a root-mean-square roughness value of 470.5 nm, while that of the bottom one is only 39.0 nm with a smaller friction coefficient.

Fig. 2.

Difference in the morphology and mechanical properties between the top and bottom portion of the printed ribbons. (A) SEM image of the cross-section of a ribbon showing the different morphology in the top and bottom portion. (B) SEM image of the cross-section showing the top portion. (C) SEM image of the cross-section showing the bottom portion. (D) Friction coefficients of the top and bottom surfaces. (Insets) AFM images showing the topography of the top and bottom surfaces. (E) Nanoindentation measurement of partial loading–unloading curves of the top and bottom surfaces. (F) Comparison of the modulus and hardness as a function of the distance to the top and bottom surfaces. (G) Scratch test of the top and bottom surfaces.

As revealed by the partial loading–unloading nanoindentation curves in Fig. 2E, the bottom portion shows a sharper force response and a smaller maximum contact depth than that of the top one under an applied normal force of 3 mN, indicating Young’s modulus (E_y) and hardness (H) of the bottom portion are higher than the top one. As shown in Fig. 2F, E_y and H of the bottom portion remain constantly high as the indentation depth increases, while the values of the top portion show an increasing trend until reaching values similar to the bottom one. In addition, the scratch test is shown in Fig. 2G, further indicating their structural variation. The applied sliding results in a typical scratch under a constant normal force of 3 mN (SI Appendix, Fig. S5). The scratch depth (normal displacement of the nanoindentation tip) of the bottom surface is larger than that of top one, while the resistant forces (lateral forces) of both surfaces are similar, which can be ascribed to the loosely packed GO flakes in the top areas.

Controllable Structural Variation.

The size of the printing nozzle largely influences the alignment of GO flakes and thus the resultant structural variation degree. As revealed by the computational fluid dynamics simulations (Fig. 3A), small diameter of the nozzle outlet renders a large shear stress to the flowing ink, leading to a higher alignment of GO flakes with their planes parallel to the flowing direction. In addition, the speed of the ink flow is accelerated when flowing through a tapered nozzle, and the generated elongation rate further enhances the orientation degree of GO flakes in the extruded filament (SI Appendix, Fig. S6). The shear-force-induced prealignment of GO flakes strongly influences the resultant orientation degree in the printed ribbons. From the 2D WAXD results shown in Fig. 3B, a higher orientation degree (evidenced by f) of the printed ribbons can be obtained using printing nozzles with a narrower nozzle outlet.

Fig. 3.

The effect of nozzle diameter, rheological property of ink, and temperature of substrate on the resultant structural variation of the printed ribbons. (A) Distribution of calculated shear stress in a tapered printing nozzle with different outlet diameters. (B) The 2D WAXD patterns of the printed ribbons using nozzle outlets with different diameters. (C) Apparent viscosity of printing inks with different water content as a function of shear rate. (D) POM images of the extruded filament ink with high water content (94.3%) before and after drying. The insets in D, iii and iv are their corresponding conoscopic figures. (Scale bars, 300 μm.) (E) POM images of extruded filament ink with low water content (87.0%) before and after drying. The insets in E, iii and iv are their corresponding conoscopic figures. (Scale bars, 300 μm.) (F) Distribution curves of azimuthal-integrated intensity of the printed ribbons using printing inks with different water content. (G) Distribution curves of azimuthal-integrated intensity of the printed ribbons deposited on substrates with different temperature.

The rheological properties of the printing ink significantly affect the motion of the GO flakes on a substrate during the drying process, and thus eventually determine the resultant structural variation of the obtained ribbons. As shown in Fig. 3C, the printing ink shows a typical shear-thinning behavior. By controlling the water content in the printing ink, the viscosity can be adjusted. A low water content leads to high viscosity of the printing ink that greatly restricts the motion of the aligned GO flakes during the drying process, and thus causes a large structural variation to the printed ribbons. Notably, the decrease of water content also enhances the storage modulus (G′) of the printing ink (SI Appendix, Fig. S7 A–E), which helps shape maintenance of the extruded filaments on the substrate and makes the ink components hard to move; with water content less than 89.3%, printing inks show larger G′ than loss moduli (G′′) (SI Appendix, Fig. S7 A–E), indicating that they possess more solid-like behavior when resting on the substrate. In contrast, increased loss tangents (tan δ = G′/G′′) are observed when enhancing the water content (SI Appendix, Fig. S7F), proving these inks have more fluid-like behaviors and the GO flakes are thus realigned more easily after being extruded. In addition, the capillary number (Ca) describes the relative effect of viscous force versus surface tension of the printing ink (28, 29). The Ca of our ink approaches or even exceeds 1 as the water content decreases (see details in Materials and Methods), indicating the viscous effect becomes dominant, and the movement of GO is thus restricted. Alternatively, if the viscosity is rather low, the capillary effect would dominate and the commonly seen coffee-ring effect would appear (30).

The effect of rheological properties of the printing ink on the resultant GO alignment can be also evidenced by polarizing optical microscopy (POM) (Fig. 3 D and E). The GO flakes of the printing ink with a high water content (94.3%) can move easily and form a planar (homogeneous) alignment after the extruded filament becomes dry, which can be proved by the completely darkening POM image (Fig. 3 D, i and iii) and its conoscopic figure showing a Maltese cross in the center (Fig. 3 D, iii, Inset). Rotation of the dried ribbon about the microscope axis does not alter the dark image and the cross (Fig. 3 D, iv), further proving the formation of the planar alignment of GO flakes after the printing ink dries on the substrate. In contrast, a lower water content (87.0%) of the printing ink causes a higher energy barrier for these GO flakes to move. After being aligned when extruding from the printing nozzle, the GO flakes in the top portion retain their homeotropic alignment after drying. A sharp contrast in the intensity of transmitted light can be found in Fig. 3 E, i and its 45° rotation in Fig. 3 E, ii, revealing a higher viscosity of printing ink tends to retain the alignment of GO flakes after being extruded onto the substrate. When the axis of the dried ribbon is along the direction of the polarizer, it shows a completely dark POM image, while the ribbon reappears after a 45° rotation. Their corresponding conoscopic figures change from the dark cross (Fig. 3 E, iii) into two arc-shaped isogyre patterns (Fig. 3 E, iv), which indicates the existence of biaxial phase and retaining homeotropic alignment in the dried ribbon.

Thus, the structural variation of the printed ribbons can be controlled by adjusting the rheological properties of the printing inks. Fig. 3F shows the distribution curves of azimuthal-integrated intensity of the printed ribbons by using inks with different viscosities; a lower water content of the printing ink leads to a more broadening peak distribution, indicating its large structural variation, while a higher water content of the printing ink reduces its viscosity and contributes to an easy motion of GO flakes, thus leading to a more uniform structure and thus a smaller structural variation in the printed ribbons, which is evidenced by the increased f (SI Appendix, Fig. S8A) when water content increases.

A more practical way to control the structural variation of the printed ribbons is to adjust the temperature of the deposition substrate. It is noted that the saturated vapor pressure of water will increase as the temperature rises (the values are 814, 2,339, 12,344, 31,176, and 101,325 Pa at temperatures of 4, 20, 50, 70, and 100 °C, respectively), which will affect the evaporation rate of water. High temperatures tend to drive the solvent out fast and increase the viscosity of the ink quickly. As a result of increased viscosity, the movement of ink components is further restricted. As shown in Fig. 3G, higher temperature of the substrate induces a larger structural variation of the printed ribbons, which is proved by the decreased f as the temperature increases (SI Appendix, Fig. S8B);

Spontaneous Transformation to Ultrastretchable Springs.

The structural variation of the printed ribbons enabled by the directionally self-assembled 2D nanoflakes can be employed to realize complex 3D architectures that are beyond the capability of conventional 2D fabrication techniques. As schematically illustrated in Fig. 4A, a printed ribbon can spontaneously transform into an ultrastretchable spring in water. First, the ribbon was cross-linked in Ca²⁺ aqueous solution, and the microstructure of GO alignment was fixed to prevent its dissolution in the following water treatment. Then the cross-linked ribbon underwent an anisotropic swelling in water and finally transformed into an ultrastretchable spring.

Fig. 4.

Structural variation enabled the spontaneous transformation to ultrastretchable springs. (A) Schematic illustration of the morphing of a printed ribbon to an ultrastretchable spring. (B) Schematic illustration of the shape-morphing mechanism. (C) Snapshots showing the process of the transformation of a printed ribbon in water. (D) Simulation results showing the process of a shape-morphing ribbon. (E) The radius and pitch of the resultant springs as a function of the temperature of substrate during the ink drying process. (F) Deflection angles of the resultant springs as a function of the temperature of substrate during the ink drying process. (G) Photographs of an ultrastretchable spring under different tensile strains and their corresponding calculated induced strain distribution (along the tensile direction).

The mechanism for the spontaneous transformation of the printed ribbons is based on the formation of the structural variations. The SA polymeric matrix was confined by the aligned GO flakes and the swelling of the matrix in water was thus limited and guided (Fig. 4B). Notably, the dominant swelling direction was along the director n of the aligned GO flakes, since it was easier to expand these layered structures than others. To be more specific, the GO flakes in the bottom portion form a horizontally located bricks-and-mortar structure, so the dominant swelling direction (with a swelling ratio of a₁) is along the thickness of the ribbon (director n). In contrast, the GO flakes in the top area are vertically aligned with a small in-plane deflection angle to the longitudinal axis of the ribbon. Thus, the dominant swelling direction (with a swelling ratio of a₁) of the top area is along the deflection angle in the top surface (director n). As suggested by Timoshenko’s thermal expansion bilayer model (31), a large in-plane swelling of the top portion is restricted by the less deformed bottom one, whereas the bottom portion is in turn forced to deform along with the top one. The generated bending moment thus renders the ribbon to bend along the dominant swelling direction of the top area. Since its swelling direction is not along the ribbon, a spring architecture thus forms. As shown in Fig. 4C, the printed ribbon can suddenly snap into a helical spring after swelling to a certain extent in water (Movie S1). It is noted that the transformation of the printed ribbon to helix is enabled by the asymmetric microstructures, which is different from the reported deformations induced by the variation in materials (32).

The mechanism of structural variation enabled shape morphing is further proved by the finite-element analysis results (Fig. 4D). As guided by the simulations, different geometric dimensions of the springs can be obtained by controlling their structural variation. Therefore, the printing nozzle, rheological property of the ink, and temperature of the substrate can be rationally adjusted to meet the desired architectures of the spring. For example, the temperature of the substrate significantly affects the structural variation of the printed ribbon and thus the resultant architecture of the spring. As shown in Fig. 4 E and F, the radius, pitch, and deflection angle of the obtained springs increase when raising the temperature of the substrate. These trends can be explained by the enlarged structural disorder of the ribbon after drying on hot substrates.

The obtained spring shows an ultrahigh stretchability. As shown in Fig. 4G, it can be stretched to more than 1,000% of its initial length. As revealed by the finite-element analysis result, when a strain of 1,050% is applied to the spring, the maximum induced strain (MIS) distributed in the stretched spring is less than 0.1%. More specifically, the outside surface of the stretched spring is in compression, while its internal surface is in tension. Both tensile and compressive MISs rise along with the increase of the applied strain (SI Appendix, Fig. S9). In addition, the temperature of the substrate can influence the stretchability of the obtained springs. A lower temperature of the substrate during the ink drying favors a higher stretchability of the obtained spring (SI Appendix, Fig. S10A). For example, springs with large ultimate stretchability of >2,000% can be obtained from the ribbons when the ink is dried on a 4 °C substrate, while that of only about 300% is reached if the ink is dried on a 70 °C substrate. In addition, the MIS distributed in fully stretched springs with a low drying temperature is also smaller than those with a high drying temperature (SI Appendix, Fig. S10B). Nevertheless, the largest MIS in the stretched spring is nearly two orders of magnitude lower than the fracture limit (∼18%) of the cross-linked ribbon (SI Appendix, Fig. S11), making it possible as an ultrastretchable ionic conductor.

Highly Conductive and Stretchable Nanofluidic Ionic Conductors.

The structurally stretchable springs can be served as high-performance nanofluidic ionic conductors. Fig. 5A shows the test setup for measuring the ionic conductivity of such springs. Typical current–voltage curves at various salt concentrations show linear current response to the applied voltage (Fig. 5B). Fig. 5C compares the ionic conductivities of the obtained springs with the printed SA ribbons and the salt bulk electrolyte under different electrolyte concentrations. The ionic transport characteristics of the spring exhibit two apparent features. At high electrolyte concentrations the ionic conductivity of the spring linearly rises as the increase of the electrolyte concentration, which is similar to the behavior of the salt bulk electrolyte, whereas at low electrolyte concentrations its ionic conductivity becomes independent of the electrolyte concentration and remains several orders of magnitude larger than that of bulk electrolyte, which is due to its surface-charge-governed ionic transportation at this concentration regime.

Fig. 5.

Highly conductive and stretchable nanofluidic ionic conductors. (A) Schematics showing the test setup for measuring the ionic conductivity of the spring. (B) Representative current–voltage curves of the spring. (C) Ionic conductivity of the obtained spring, SA ribbon, and bulk solution as a function of electrolyte concentration. (D) Schematics showing the molecular structure of the ink components and ion transportation in subnanoscale channel of the spring. (E) X-ray diffraction patterns of the printed ribbon and SA ribbon. a.u., arbitrary units. (F) Comparison of the resistance change of the obtained spring with reported electronics conductors and hydrogel conductors under stretching.

The ionic conductivity of the spring is higher than that of the printed SA ribbon and can also be proved by the electrochemical impedance spectroscopy (SI Appendix, Fig. S12), where an impedance value of ∼525.5 Ω at 1 kHz for the spring and ∼1,844.0 Ω for the printed SA filament was measured. The high ionic conductivity of the spring can be ascribed to two main reasons: its chemical composition and highly aligned subnanoscale ionic transportation channel (Fig. 5D). The ionic conductivity of an ionic conductor at low electrolyte heavily depends on the charge density of the conductor. The composed GO and SA of the printed ribbon, both of which contain masses of carboxyl group (–COO⁻), contribute to a high density of negative charge and a high ionic conductivity (21). Additionally, as shown in Fig. 5E, there are well-aligned 2D subnanoscale channels with an interspace of 8.6 Å in the printed ribbon, which can efficiently boost the smooth transportation of positive ions (33). In contrast, the lower ionic conductivity of the printed SA ribbon can be ascribed to a lower charge density and the lack of well-aligned channels.

Notably, the high ionic conductivity of the spring remains nearly invariable even under a stretch of >11 times of its initial length (Fig. 5F). In comparison, hydrogel conductors, as the most typical type of ionic conductor, follow a theoretic resistance change of λ² along with the stretch of λ due to the deformation of their intrinsic polymer network (34). Even worse, electronic conductors show a more dramatic change in resistance when stretched due to the breakdown of their rigid or brittle materials (35, 36). Our spring exhibits excellent performance in high ionic conductivity as well as high stability in conductivity under ultralarge deformations.

Application of the Spring as a Soft and Elastic Neural Electrode.

Conventional neural electrodes are usually made of dry, rigid, electron-conducting, and static materials (metal or metal oxide electrodes), which are quite foreign and incompatible with biological tissues with a wet, soft, ion-conducting, and dynamic nature (37). The large mechanical mismatch between these hard electrodes (e.g., the elastic modulus of ca. 100 GPa for metal electrodes) and soft neural tissues (e.g., elastic modulus of ca. 100 kPa) may incur small contact areas or even cause wound or glial scar to the tissue, resulting in large resistances against a good signal-to-noise ratio (SNR). Moreover, electrochemical reaction at the tissue–electrode interface may take place when injecting electric currents to the tissue from these electron-conducting electrodes, and thus the incurred environment change at the neural interface, such as production of hazardous chemicals, pH change, or generation of local heat, poses serious issues (such as acute inflammation and chronic neural disorder) to the tissue (22). Surprisingly, the soft and ion-conducting nature of the helical spring with high ionic conductivity and highly stable conductivity under large deformations shows great potential for the neural interface applications.

As a proof of concept, we implanted and integrated the obtained springs on the sciatic nerve of a bullfrog (Fig. 6A) as neural electrodes for in vivo compound action potential (CAP) recording and neural stimulation. Due to its aqueous nature and ultrastretchable helical architecture, the spring with helical architecture could conformably contact with the highly deformable and dynamic nerve fiber (Fig. 6B). As the experimental setup shown in Fig. 6C, two Pt wires were used as the stimulating electrodes at one end of the sciatic nerve, and the two spring electrodes at another end recorded the evoked CAPs. Upon stimulated by a series of monophasic rectangle waves that were applied to the Pt wires, the evoked CAPs were thus transmitted along the nerve and recorded by the spring electrodes (Fig. 6D).

Fig. 6.

Application of the obtained springs as neural electrodes for action potential recording and neural stimulation. (A) Schematic of a bullfrog. (B) Photograph of a spring integrated on the sciatic nerve of a bullfrog. (C) Schematic showing the experimental setup for recording the CAPs of the sciatic nerve. (D) A recorded CAP using the spring electrodes. (E) Recorded CAPs using the spring electrodes under different voltages of stimulation with a frequency of 15 Hz. (F) Comparison of recorded CAPs using the spring electrodes and Pt wires. (G) Twitch forces evoked in the gastrocnemius muscle of the bullfrog using the springs as the stimulation electrode under various applied voltages with a frequency of 0.5 Hz.

Different stimulating voltages with a constant frequency of 15 Hz and a width of 1 ms led to a series of CAPs with different amplitudes (Fig. 6E). The recorded CAPs show a high SNR due to the high ionic conductivity of the spring electrodes and their compatible contact with the nerve fiber. In addition, the peak of the evoked CAPs increases as the applied stimulating voltage enhances and reaches a plateau after a voltage of 0.5 V, which can be ascribed to the saturated transmembrane potential of the nerve cell. Notably, the evoked CAPs recorded by the spring electrodes show higher amplitudes compared with that recorded by the Pt wires (Fig. 6F), which show great advantages over the conventionally rigid, dry, and geometrical mismatched metal electrodes for the neural interface. In addition, the spring electrodes were used as the stimulating electrodes for muscle electrical stimulation. The evoked action potentials by the spring electrodes led to the contraction of the gastrocnemius muscle of the bullfrog. The twitch force of the gastrocnemius muscle rose linearly along with the increase of voltages (0.2 V to 0.7 V) applied to the spring electrodes (Fig. 6G and SI Appendix, Fig. S13). Note that higher stimulating voltage will not induce hazardous electrochemical reaction at the interface due to the ion-conducting nature of the spring electrodes, so the workable voltage of stimulation is not limited to the restriction of harmful water electrolysis. These results prove that the fabricated springs can be used as a highly compatible and efficient electrode for neural recording and stimulation. In addition, the materials (GO and SA) of the spring electrodes are generally pretty safe according to the toxicity evaluation in the recent studies (38, 39). It should be noted that SA is usually used as bioresorbable material, while it is hard for GO to be absorbed by biological tissues. However, we believe the reported asymmetrical self-assembly strategy may be applied to other nanoflakes that may bring new properties such as biodegradability and bioresorbability (40, 41) or other functionalities into the system.

Conclusion

In summary, we report the spontaneous formation of ribbons composed of directionally self-assembled 2D nanoflakes in a polymeric matrix during printing. Controllable manipulation of asymmetric self-assembly of building blocks was realized by using a composite ink containing GO flakes and SA polymeric matrix dispersed in water. The printing ink formed a graded structure through a process containing two key steps. First, randomly distributed GO flakes were aligned along the extruded filament driven by the large shear force when extruded from a printing nozzle. Once the ink was deposited on the substrate, the aligned GO flakes then reoriented discrepantly at different thicknesses of the extruded filament, and finally formed a ribbon with structural variation. The asymmetric self-assembly of GO flakes enabled site-specific mechanical properties of the printed ribbons at different thicknesses. It was demonstrated that the structural variation could be further enlarged by swelling in water after cross-linking in Ca²⁺ ions, leading to an enhanced mismatch along the thickness. As a result, the printed ribbons could be spontaneously transformed into springs with controllable helical architectures in water. The obtained spring can work as an elastic nanofluidic ionic conductor with high and stable ionic conductivity under high tensile strains (>1,000%). Further, we demonstrated the application of the nanofluidic ionic spring as neural electrodes for action potential recording and neural stimulation, showing its high potential for neural interface applications.

Materials and Methods

Preparation of the GO/SA Printing Ink.

GO was prepared according to the typical Hummers’ methods (42). To obtain GO/SA aqueous ink with different viscosities, GO and SA with a weight ratio of 200 mg:1,000 mg were mixed in a series of water with different volumes ranging from 8 mL to 20 mL. Then, these components were mixed in a speed mixer at 3,500 rpm for 5 min, following by mechanical mixing at 100 rpm for 5 min, then underwent another mixing in the speed mixer at 3,500 rpm for 5 min. Unless otherwise stated, the GO/SA ink used elsewhere in this work is at the weight ratio of GO:SA:H₂O = 1:5:60.

Fabrication Process of the Printed Ribbon and the Spring.

Multiple parallel filaments were designed using a commercial software (3ds Max) and the corresponding G-codes were generated by another commercial software (Ultimaker Cura). The ink was then loaded in an injection syringe and connected to a nozzle with different outlet diameters that was fixed on a 3D printer (Anycubic I3 MEGA). The ink was extruded by a syringe pump (Lead Fluid TSD01) at a feeding rate of 200 μL/min to draw filaments on a polyethylene terephthalate film fixed on a bed with controllable temperature and left to dry for further use. The obtained ribbons were then immersed into 0.5 M CaCl₂ aqueous solution for 5 s and then transferred into deionized water to swell for shape morphing into springs.

Characterization of the GO/SA Printing Ink, the Printed Ribbon, and the Spring.

The rheological properties including viscosity and modulus of the printing ink were measured by a rheometer (MCR 301; Anton Paar). The storage and loss modulus of the printing ink was recorded within a shear sweep from 10⁻¹ to 10⁴ Pa at a constant frequency of 1 Hz. Reynolds (Re = αρu/η), Weber (We = αρu²/γ), and Ohnesorge [Oh = (We)^0.5/Re = η/(γρα)^0.5] numbers are used to analyze the printability of our ink, where α is the nozzle diameter (micrometers), ρ is the ink density (grams per cubic centimeter), u is the ink velocity at the nozzle (meters per second), η is the ink viscosity (millipascal second), and γ is the surface tension (millijoules per square meter). Z = 1/Oh is usually used to analyze and evaluate the printability of ink. For our ink system, the nozzle diameter is 260 μm, the ink density is 1.11 g/cm³, the ink viscosity is on the order of 10³ mPa⋅s at shear rates between 10³ to 10⁴ s⁻¹, and the surface tension is on the order of 10² mJ/m², and thus Z is on the order of 10⁻¹.

In addition, capillary number (Ca, Ca = ηu/γ) describes the relative effect of viscous forces versus surface tension. For water drying on a substrate (with a low viscosity on the order of 10⁻³ Pa⋅s), Ca is the on the order of 10⁻⁷. For our highly viscous ink, the viscosity (10²∼10⁴ Pa⋅s at low shear rates) is several orders of magnitude higher than that of water, where the ink velocity (u = R/t_dry, where R is the diameter of extruded filament and t_dry is the drying time) is on the order of 10⁻⁵ m/s and surface tension is on the order of 10⁻¹ J/m², and thus Ca is on the order of 1.

The thickness of GO flake and surface topography of the printed ribbon were measured using atomic force microscope (AFM) (Oxford Instruments Asylum Research, Inc.). The printing ink and extruded filament were freeze-dried and their morphologies were measured by an SEM (Zeiss, Merlin). The strain–stress curves of the cross-linked ribbon with a gauge length of 10 mm were measured at a load rate of 1 mm/min using a Shimadzu AGS-X universal testing machine. Friction coefficients were measured using a multifunctional surface mechanics tester (MFT-5000; RETC-Instruments, Inc.) equipped with a sliding diamond ball (R = 6.33 mm). Nanoindentation measurements were done using a triboindenter (TI 950; Hyitron) equipped with a Berkovich diamond tip (R = 100 nm), and the scratch test was also conducted in the triboindenter with a conical diamond tip (R < 1μm). X-ray diffraction spectra of the printed ribbon and SA ribbon were measured using a laser Al Kα radiation (Escalab 250Xi; Thermo Scientific). The 2D WAXD patterns of the printed ribbons were recorded by a single crystal diffractometer (Bruker D8). Multiple printed ribbons were piled together and mounted with their thickness toward the incident X-ray beam. Azimuthal angle intensity distributions [I(ϕ)] of (002) plane at 2θ ≈ 10.3° were analyzed by GADDS software, and the Herman’s factors (f) was calculated to evaluate the GO orientation according to the following equations:

$$\left({\text{cos}}^2\phi \right)=\frac{{\displaystyle \int }I\left(\varphi \right)\text{sin}\varphi \text{cos}{\varphi }^{2}d\varphi }{{\displaystyle \int }I\left(\varphi \right)\text{sin}\varphi d\varphi }$$ [1]

$$f=1-3\left({\text{cos}}^2\varphi \right).$$ [2]

The ionic conductivities of the obtained spring, SA ribbon, and bulk solution at different concentrations of KCl were calculated by their corresponding current–voltage curves measured by an electrochemical workstation (CHI760E; CH Instruments, Inc.).

In Vivo Animal Experiments.

The in vivo bullfrog tests were approved under contract SYXK (Jing) 2009-0022 by the Ethics Committee of Tsinghua University, Beijing, China. The gastrocnemius muscle with sciatic nerve from the bullfrog (weight: 348 g) was extracted with the double pithing method. Two of the spring electrodes were immersed into Ringer’s solution and then intertwined onto the sciatic nerve and connected to a physiological signal processing device (RM-6240BD; Chengdu Instrument Factory). For recording the signal of CAPs by the spring electrodes, two Pt wire electrodes were placed on the bottom surface of the sciatic nerve. A series of monophasic rectangle waves with different voltages (15 Hz, a width of 0.2 ms) were applied on the Pt wire electrodes, and the sampling rate of the spring electrodes was 20 kHz. As a comparison, another two Pt wire electrodes replaced the spring electrodes and recorded the CAPs. For stimulating the gastrocnemius muscle by the spring electrodes, a series of monophasic rectangle waves with different voltages (0.5 Hz, a width of 1 ms) were applied on the spring electrodes, and the twitch force of the gastrocnemius muscle was recorded by a force sensor.

Finite-Element Analysis.

Computational fluid dynamics simulations were conducted to understand the influence of shear stress on the resultant GO orientation. The GO/SA printing ink is considered to be incompressible, and the printing ink follows the typical non-Newtonian flow behavior. By fitting the data from Fig. 3C by the power law viscosity model$\eta =K{\dot{\gamma }}^{n-1}$, we obtain the corresponding parameters K = 771.24 kg/(m·s) and n = 0.223 and apply them to the simulation. The Navier–Stokes equation shown below guides the dynamics of the velocity filed V of the printing ink:

$$\mathrm{▽}·V=0$$ [3]

$$\mathrm{▽}·[-p\mathbf{I}+\eta (\mathrm{▽}V+\mathrm{▽}{V}^T)]=\rho (V·\mathrm{▽})V+\rho {\partial }_{t}V,$$ [4]

where ρ is the density of the fluid and measured to be 1.11 g/cm³ and I is the unity matrix. The equation is solved in COMSOL Multiphysics software with corresponding geometry of printing nozzles, and the shear stress distribution is then obtained from the results.

The induced strain of a spring under tensile strain is simulated using COMSOL Multiphysics software. The mechanical properties of the ribbon are obtained according to the strain–stress curve of the ribbon in the wet state (SI Appendix, Fig. S8) and applied into simulation of the stretched spring. Different tensile strains are loaded to the spring, and the corresponding induced strain and stress distribution in the spring are then calculated. The transformation of the printed ribbon to the spring is simulated using ABAQUS software. The shape-morphing mechanism is ascribed to the different swelling along the thickness of the printed ribbon. Thus, swelling ratio and direction can be easily tuned in each layer of top and bottom and applied into the simulation, which determines the final helical architectures of the obtained springs.

Data Availability.

All data, materials, and associated protocols are included in the paper and SI Appendix.