(339a) Determining Spinnability of Carbon Nanotube Dispersions in Wet Spinning Via Shear Viscosity of Dispersions

Unprecedent properties of nanomaterials can be utilized by manufacturing them into macroscopic structures. Assemblies of nanomaterials have been widely manufactured using the liquid phase, including both a molten state and in form of a solution. Particularly, the complex phase and flow behaviors of solutions facilitate various approaches to understand and control the manufacturing process.

Wet spinning, a representative solution processing, is the fiber spinning process where the spinning dope passes through a narrow spinning line and solidifies into fiber. Especially, when the spinning dope flows through a narrow spinning line, the flow occurs under the high shear-rate regime. The dynamic behavior of materials in this regime affects the structure of the as-spun fiber which determines the success of the fiber formation, becoming a matter of interest in the field of wet spinning.

In the wet spinning of polymer fibers, the spinnability of dopes, the ability of dopes to spin fibers continuously, has been determined using their shear viscosity. For the quantitative evaluation of the flow behavior from the shear viscosity, the power law viscosity model has been employed due to its simplicity. The power law viscosity model, known as the Ostwald-de-Waele relationship, describes the shear viscosity η [Pa∙s] as a function of the shear rate γ [s^-1] by the following equation:

η = mγ^n-1

where is the consistency index m [Pa∙sⁿ], and n is the power law index [dimensionless]. We can quantify the flow behavior of fluids through . When n=1, the fluid exhibits the Newtonian flow behavior which has a constant shear viscosity. For n>1, the fluid exhibits the shear-thickening behavior, while the fluid with exhibits the shear-thinning behavior. As n approaches to 0, the shear-thinning behavior becomes stronger. The evaluation of the spinnability through the quantified flow behavior of dopes enables to discover appropriate dope conditions, advancing the wet spinning.

Among promising materials applied in the wet spinning, carbon nanotubes (CNTs) have been frequently processed in macroscopic-fiber form. The wet spinning of CNT fibers greatly advanced by the solution processing for homogeneous CNT dispersions based on the superacid and surfactant system. However, unlike the wet spinning of polymer dopes, the flow behavior of CNT dispersions within the spinning line has not been investigated. The rod-like structure and large aspect ratio of CNTs similar to the long-chain characteristics of polymers enabled us to apply the methods for polymer dopes into CNT dispersions. For instance, the phase behavior of CNT dispersions and characteristics of CNTs can be analyzed using the Onsager theory which is based on polymer thermodynamics. The similarity between polymers and CNTs inspired us to determine the spinnability of CNT dispersion in wet spinning using the power law viscosity model as used in polymers. Herein, we presented the correlation between the spinnability and the flow behavior of CNT dispersions through the power law viscosity model.

We investigated the flow behavior of CNT dispersions at various concentrations using three commonly used single-walled CNT products (eDips, Tuball, SG101) in the wet spinning of CNT fibers. The surfactant-based aqueous CNT dispersions were obtained after the ultrasonication. Prior to evaluate the shear-thinning behavior between the dispersions through the power law viscosity model, we investigated the shear-rate range which can represent the flow behavior of the dispersion within during the wet spinning, enabling proper fitting of the model. The shear rate γ in the spinning line was estimated using the following equation, assuming laminar flow:

γ=4Q/Ï€r³

where Q is the volume rate of the spinning rate, and r is the radius of the spinning needle. In our wet spinning, the shear rate within the spinning line was estimated as 787.33 s^-1. The shear-rate range from 100 to 400 s^-1 and from 200 to 800 s^-1 which are close or contains the estimated spinning shear rate were investigated for the power law viscosity model. We quantitatively compared the suitability of the fitting ranges using the normalized difference between the expected and experimental viscosity values at the estimated shear rate of spinning. Despite similar R² values of the fittings, the normalized differences where the fitted range was from 100 to 400 s^-1 were greater than that were fitted within the range of 200 to 800 s^-1 for all CNT dispersions. This suggested that the fitting range which includes the spinning shear rate more accurately predicts the shear viscosity at the spinning shear rate than other. In this work, we determined the spinnability of the dispersion using the power law index from the power law viscosity model fitted within the shear rate range of 200 to 800 s^-1.

We defined the spinnability as the ability to spin fibers continuously with measurable properties. The spinnability was classified into the ‘good’ spinnability when the fiber spinning is continuous and the fiber properties are measurable, and the ‘bad’ spinnability wherein the fiber spinning is discontinuous and the fiber properties are unmeasurable. eDips, Tuball, and SG101 had the good spinnability when was lower than 0.418 (0.16 wt.%), 0.438 (0.35 wt.%), and 0.738 (0.48 wt.%), respectively. Although CNT products exhibited different ranges of the spinnable , the spinnability and fiber properties were increased as approached to 0. In other words, the spinnability of the dispersion was proportional to the shear-thinning behavior of the CNT dispersion.

When the CNT dispersion flows, CNTs encounter each other, resulting in the friction forces between them. The friction forces induce the disentanglement and reorientation of CNTs towards to the shear direction, facilitating the flow of the dispersion in the shear direction. As a result, the shear viscosity of the dispersion decreases, which known as shear-thinning behavior. The weak shear-thinning behavior was observed at the dispersion with low CNT concentration which had large free space within the dispersion. Based on polarized optical microscopic images and the magnitude of shear viscosity, we confirmed that the dispersion at low concentration had large free space. When the free space within the dispersion is large, the amount of contact between CNTs during the flow will be minimized, decreasing the possibility of collision between them. The friction forces will not be sufficient to induce the disentanglement and reorientation of CNTs, exhibiting the weak shear-thinning behavior. In the wet spinning, as CNTs within the dispersion maintain an isotropic conformation even at the spinning shear rate, the resulting CNT fiber will exhibit a low contact area between CNTs within the fiber. Consequently, the ‘bad’ spinnability was observed in this dispersion.

On the other hand, the strong shear-thinning behavior was observed at the dispersion with high CNT concentration which had small free space within the dispersion. The increase of the birefringence and shear viscosity at high CNT concentration indicates the small free space within the high CNT concentration dispersion. The amount of contact between CNTs will large enough to induce the friction forces resulting in the disentanglement and reorientation of CNTs. The CNT fiber will have a large contact area between CNTs within the fiber, exhibiting the ‘good’ spinnability.

The correlation between the spinnability and the shear-thinning behavior of CNT dispersion arises from the fiber integrity which is derived from the strong van der Waals forces between CNTs. If CNTs are aligned along the fiber axis, the total amount of van der Waals forces between CNTs within the fiber will be maximized as the contact area between CNTs increases. We hypothesized that CNT fibers from CNT dispersions which has the ‘good’ spinnability have the large contact area between CNTs within the fiber. In contrast, the dispersion exhibiting the weak shear-thinning behavior are not able to form fibers due to the small contact area between CNTs which derives from the isotropic conformation of CNTs within the fiber.

We can extend our hypothesis to other materials with similar fiber integrity as CNTs. When the fiber integrity of the material is governed by the strong van der Waals force of the graphitic carbon structure, the spinnability will be maximized as the dispersion close to the ideal shear-thinning behavior. Similar to CNT, graphene oxide (GO) has the strong van der Waals force on its graphitic plane. GO fibers maintain their structure through the strong van der Waals forces between adjacent layers. The spinnability of GO dispersions in the wet spinning was found to be proportional to the shear-thinning behavior of the dispersions. Therefore, in the wet spinning of the carbon materials with graphitic structure, we suggest that the spinnability of the dispersion can be determined through the degree of the shear-thinning behavior of the dispersion.

In summary, we quantitatively investigated the correlation between the shear-thinning behavior and the spinnability of CNT dispersion through the power law index which approaches to 0 if the dispersion has stronger shear-thinning behavior. The dispersions were more spinnable as approached to 0, suggests that the spinnability of CNT dispersions is proportional to the shear-thinning behavior of the dispersion. Our work presents a simple method to determine the spinnability of CNT dispersion using the shear viscosity of CNT dispersion.