Rheological behavior of particle-filled polymer suspensions and its influence on surface structure of the coated electrodes

2.1 Materials

PVA (M _w ∼ 7.56 × 10⁴ g·mol⁻¹, 98–99% alcoholysis degree) and PEO (M _w ∼ 1 × 10⁵ g·mol⁻¹) were purchased from Aladdin Industrial Corporation in Shanghai, China. Ludox TM-50 colloidal silica (average diameter of 22 nm, 50 wt%) was obtained from Sigma-Aldrich in the USA. All reagents were used as received without further purification, and deionized water was used in all experiments.

2.2 Suspensions preparation

To prepare the suspensions, PVA or PEO and silica nanoparticles were used following a specific preparation protocol. First, 6 g of PVA powder was added to 94 g deionized water and continuously mechanically stirred at 95℃ for 48 h (≤100 rad·min⁻¹) to prepare a homogeneous aqueous polymer solution with a mass concentration of 6%w/w. A similar operation was performed for PEO blank solution, but at room temperature. Next, colloidal silica was added to the polymer solution in batches to achieve the desired concentrations; the amount of single addition was about 1 g with an interval of 1 h. The mixture was then continuously mixed for at least 24 h (≤100 rad·min⁻¹, room temperature) to break silica agglomerates and to ensure a homogeneous suspension. The concentration of polymer was 6%w/w, while the concentrations of silica nanoparticles were 20, 25, 30, and 35%w/w. The samples were left to rest for more than 1 day before use. All the sample codes used are shown in Table 1.

2.3 Rheological measurements

Rheological measurements were performed using an Anton Paar MCR102 rheometer (Graz, Austria) with parallel plates (25 mm diameter, 316L stainless steel). The parallel plates are spaced 1.0 mm apart. Viscosity measurements were conducted within the range of 1–100 s⁻¹, taking into account the possible coating velocity. Dynamic viscoelastic modulus measurements included frequency sweep (strain amplitude of 0.1%, frequencies ranging from 0.158 to 100 rad·s⁻¹) and amplitude sweep (angular frequency of 6.28 s⁻¹, strain amplitudes ranging from 0.01% to 100%). The three-interval Thixotropy Test (3ITT) consisted of pre-shear and three intervals. Pre-shear for 20 s (shear rate of 0.25 s⁻¹) was performed to eliminate the time-dependent changes of suspensions. The three intervals were in order of rest interval for 50 s (constant small shear rate of 0.5 s⁻¹), load interval for 50 s (high shear rate of 100 s⁻¹), and a recovery interval for 300 s (shear rate returned to 0.5 s⁻¹). The hysteresis loop measurements were obtained by increasing the shear rate from 0 to 100 s⁻¹ for 300 s and decreasing it from 100 to 0 s⁻¹ for 300 s. The temperature was maintained at 25°C for all runs.

3.1 Steady shear of suspensions

Figure 1 displays the shear rate dependence of suspensions with varying particle concentrations. The PVA or PEO blank media without mixed particles are mostly Newtonian with a low viscosity. It was obvious that the addition of silica nanoparticles leads to an increase in viscosity across the whole range of shear rates regardless of the polymer used. As viscosity reflects the friction between adjacent layers in liquid, the addition of particles increased the friction between polymer molecules, due to the inhibiting role in motion of polymer chain segments (27). As a result, a thickening effect to polymer suspension was observed. According to Eisenberg’s model, the interaction of polymer segments with silica reduces the mobility of polymer and results in the formation of immobilized and restricted mobility regions around the filler particles (28).

Figure 1

Flow curve of PVA/SiO₂ and PEO/SiO₂ suspensions: (a) viscosity vs shear rate and (b) shear stress vs shear rate.

There were differences between PVA and PEO. For PVA, as the silica increases, the suspension changes from Newtonian fluid to shear-thinning fluid. While for PEO, the suspension was still Newtonian fluid even though the silica increased as much as 35%w/w. Since the molecular weights of PVA and PEO used were similar, the different response of viscosity to particle concentration was attributed to function group and structure. Figure 2 shows the molecular structures of the two polymer binders.

Figure 2

Molecular structure of polymer binders: (a) rigid PVA and (b) flexible PEO.

Both PVA and PEO are linear polymers, whose conformation in solution can be described through the random walk concept to be a flexible coil. In suspensions, when the polymer coil approaches the surface of the particle, the polymer segment replaces the solvent on the particle surface and adsorbed with conformation transition (29). Relevant studies proposed that the adsorption mechanism of PEO to silica particles is the hydrogen bonding between the ether oxygen of PEO and isolated silanol groups on the silica surface (30–32). Likewise, hydroxyl groups on PVA can also provide hydrogen bonds for adsorption (33,34). In addition, carbon-based particles commonly used in desalination also exhibit weak adsorption due to van der Waals forces, even without strong hydrogen bonding. Despite the similar driving force for adsorption, the difference in molecular flexibility leads to a disparity in the adsorption state on particles.

Compared with the C–C single bond of PVA, the C–O single bond on PEO main chain corresponds to a lower potential barrier of internal rotation, resulting in a smaller rigid factor, making PEO more flexible than PVA. Owing to the better flexibility, it was easier for PEO to transform from random coils to an adsorbed conformation through internal rotation. The adsorbed conformation can be described as segments attached to the particle surface, loops of segments that extend into the solution between the chains, and tails of dangling ends of the chains (35). PVA with poor flexibility preferred extended loops and dangling tails to generate polymer bridging, through which two or more particles are linked together, thereby producing effective flocculation and forming a crosslinked microstructure (36,37). When sheared, shear-thinning was observed due to the destruction of microstructure.

However, PEO tends to completely cover the particle surfaces due to the flexible chain. Repulsion between adsorbed layers made the suspensions to be thermodynamically stable and prevented flocculation (38–41). Therefore, the suspension remained Newtonian fluid since there is no microstructure destruction. Under the conditions where the particle surfaces are highly covered with adsorbed polymer, the colloidal interactions between particles are very weak. Reflected in the flow curve, the viscosity of PEO samples with the same filler addition has always been lower than that of PVA.

With the addition of silica less than 25%w/w, the shear rate dependence of viscosity for PVA/SiO₂ was nearly a Newtonian fluid. As the concentration of silica reached 25%w/w, the viscosity decreases as the shear rate increases, and the suspension begins to show a slight shear-thinning behavior. Moreover, the shear-thinning tendency is progressively pronounced with the increasing particle concentration. Shear-thinning can be attributed to the shear-induced breakdown of flocculated structures (42,43).

For typical flocculated suspensions, polymer bridges the particles together, forming a partial flocculated structure (Figure 3a), and a whole flocculation network forms when the number of bridges reaches a critical level (Figure 3b). It is generally believed that bridging flocculation occurs for long polymer chains and low surface coverage of particles (44,45). This requires enough uncovered particles in the range of bridging distance to form effective bridging flocculation. Above a critical concentration (∼25%w/w), increasing the number of particles within the bridging distance, analogous to introducing more cross-linking agent, requires greater shear to destroy the network, resulting in large shear stress as shown in Figure 1(b). When the particle concentration is below a critical concentration, full coverage of polymer on the particle surface is achieved, and the suspensions can be sterically stabilized by polymer chains. Although flocculation occurs locally, the number of bare particles is not enough to form a cross-linked network via bridging. Therefore, the suspension is Newtonian fluid or slightly shear thinning caused by local microstructure destruction.

Figure 3

Schematic representation of the two roles played by particle concentration in a polymeric suspension: (a) steric stabilization at low particle concentration and (b) bridging flocculation at high particle concentration.

3.2 Dynamic rheological behavior of suspensions

Dynamic test focuses on the mechanical properties of suspensions. To further detect the change in microstructure under shear, frequency sweep and amplitude sweep were used on PVA/SiO₂ and PEO/SiO₂ suspensions. Figure 4(a) and (b) shows the storage modulus (G′) and loss modulus (G″) measured by frequency sweep with an amplitude of 0.1%. It can be observed that for PVA/SiO₂, G′ is greater than G″ in the whole range and shows little dependence on frequency, indicating a solid-like structure and the formation of a cross-linked network in the suspension. On the contrary, for PEO/SiO₂, G″ is greater than G′ at low frequency and increases as the frequency increases, indicating a viscous fluid and a wrapping structure in the suspension. Meanwhile, these reasons also lead to the instability of G′ data in the low-frequency region (these data have been omitted for clarity).

Figure 4

Storage modulus (G′) represented by solid symbols and the loss modulus (G″) represented by open symbols, as a function of angular frequency and shear strain for PVA/SiO₂ and PEO/SiO₂ suspensions with varying particle concentrations: (a) moduli vs angular frequency for 30%w/w, (b) moduli vs angular frequency for 35%w/w, (c) moduli vs shear strain for 30%w/w, and (d) moduli vs shear strain for 35%w/w.

Figure 4(c) and (d) shows the G′ and G″ measured by amplitude sweep with an angular frequency of 6.28 s⁻¹. For PVA/SiO₂, there existed a critical value in strain, critical strain (γ*), at which G′ was equal to G″, intersection of the two curves. Below γ*, G′ was greater than G″ and barely changed as the strain increased, indicating the presence of cross-linked network in PVA/SiO₂. However, as strain increased near γ*, G′ decreased rapidly, down to lower than G″. The critical strain point reflects the transformation from solid-like to liquid. This transformation indicates the breakdown of the cross-linked flocculated structures as it cannot withstand tensile stress. However, a similar transformation was not observed in PEO/SiO₂, and G′ was lower than G″ in the whole range, indicating the absence of cross-linked structures in the suspension. These findings are consistent with the flow curve results presented earlier.

3.3 Thixotropic behavior of suspensions

Thixotropy, which is helpful for maintaining the shape and thickness after coating, is regarded as the evidence of cross-linking. As shown in Figure 5(a), 3ITT under the Rot-Rot-Rot mode was performed to investigate the viscosity recovery on time, wherein continuous shear measurements at alternating small and large shear rate were carried out. From the flow curve above, PVA/SiO₂ was shear-thinning fluid, so it can be seen that the viscosity decreased noticeably from the first low shear stage to the high shear stage. Then as the shear rate decreased to 0.5 s⁻¹, the viscosity began to recover, returning to its initial value in approximately 30 s. Since viscosity rooted in the cross-linked structure, fast recovery of viscosity meant fast rebuilding of cross-linking. On the other hand, for PEO/SiO₂, the viscosity remained approximately the same in all three stages, indicating the absence of cross-linking in the suspension.

Figure 5

Thixotropic behavior of PVA/SiO₂ and PEO/SiO₂ suspensions: (a) 3ITT and (b) hysteresis loops.

From the hysteresis loop measurement in Figure 5(b), for PVA/SiO₂, yielding was obvious in the shear-up curve, stress did not increase linearly, and the rate of increasing decreased (the rate is equivalent to viscosity). In contrast, in the shear-down curve, stress decreased linearly in the high shear rate range. Then the viscosity increased as the shear rate decreased due to the rebuilding of the cross-linking. The recovery of viscosity came in sight during the shear going down rather than shear completely removed, illustrating that the destruction and rebuilding of the network were dynamically competing, which existed all through the shear process. This may be the reason for the fast recovery of viscosity. While, for PEO/SiO₂, the shear-down curve almost superimposed on the shear-up curve, showing a linear increase or decrease. This typical rheological behavior of Newtonian fluid draws a conclusion that the microstructure barely changed under shear. Flexible polymers such as PEO are adsorbed and coated on particles at multiple points, making it challenging to pull away from all sites at the same time, especially in cases of multi-point adsorption with high coverage (46). Therefore, the enveloped structure was difficult to be destroyed under short time and low shear, resulting in no shear-thinning observed.

3.4 Influence of rheological behavior on coating

When particle-filled polymeric suspensions are used in a coating process, the viscosity needs to be within a suitable range. We define this range as “processing viscosity window.” High viscosity (∼10⁴ mPa·s) was too thick to flow smoothly, leaving “brushmark” on the electrode surface (Figure 6a, PVA-6/Si-35), which will deteriorate not only the appearance but also the uniformity and performance in desalination. Low viscosity suspension (∼10² mPa·s) spread easily, but causes problems in controllability of size and thickness, and the effective particles are prone to precipitation, which also increases the drying time of the electrode. To obtain a smooth and uniform electrode, low viscosity in coating and high viscosity in drying was profitable. Thus, appropriate shear-thinning is beneficial for molding through the coating method.

Figure 6

Digital photos of coatings prepared from suspensions with different rheologies: (a) high viscosity leaving “brushmark” from PVA-6/Si-35 suspension, (b) suitable viscosities such as PVA-6/Si-25 with Tyndall effect in suspension, and (c) encapsulated structure of PEO-6/Si-30 is prone to cracking.

Appropriate shear thinning results from the formation and destruction of the internal network structure in the suspension. In the PVA/SiO₂ suspension, particles are linked together by polymers forming a strong network structure. Thixotropic test results also indicate that fast recovery of viscosity and fast rebuilding of cross-linked structure during subsequent drying process guaranteed the formation of network before most of the solvent evaporated. This was beneficial for coating method. The remaining high-strength network was enough to override the capillary force during evaporation, obtaining a robust coating on the current collector (Figure 6b, PVA-6/Si-25).

However, the viscosity of PEO/SiO₂ is slightly lower than that of PVA at the same amount of particle addition. For example, the viscosity of PEO-6/Si-30 is close to that of PVA-6/Si-25. Although most PEO samples are also within the “processing viscosity window,” the flexible chain of PEO is adsorbed and covered on filler particle surfaces. The suspension can be sterically stabilized and uniformly distributed by steric effects of polymer chains. In this case, although partial flocculation may occur, the wrapping structure is not enough to form an overall cross-linked network through bridging flocculation, resulting in a weak mechanical strength of the coating. At the same time, the strength of the stacked structure formed by encapsulated particles was not sufficient to withstand the capillary force, leading to cracking of the coating after drying (Figure 6c, PEO-6/Si-30). Obviously, a coating with crack morphology cannot be applied for CDI.