**Douglas Hendrix 1, Jessica McKeon <sup>2</sup> and Kay Wille 3,\***


Received: 23 May 2019; Accepted: 13 June 2019; Published: 19 June 2019

**Abstract:** The dispersion quality of nanosilica (NS) is an essential parameter to influence and control the material characteristics of nanosilica-enhanced concrete. In this research, the dispersion quality of colloidal nanosilica in simulated concrete environments was investigated using dynamic light scattering. A concrete environment was simulated by creating a synthetic pore solution that mimicked the ionic concentration and pH value of ultrahigh-performance concrete in the fluid state. Four colloidal nanosilica samples were used, ranging in particle sizes from 5 to 75 nm, with differing solid contents and stabilizing ions. It was found that the sodium stabilized 20 nm NS sol remains dispersed at a solid concentration of 2 wt % through a variety of pH values with the inclusion of potassium ions. Calcium ions are a major contributor to the agglomeration of NS sols and only small concentrations of calcium ions can drastically affect the dispersion quality.

**Keywords:** ultrahigh-performance concrete; nanosilica; dynamic light scattering; zeta potential; pore solution

#### **1. Introduction**

The development of concrete research has accelerated in the areas of mechanical and durability properties due to potential benefits of using nanoparticles in concrete mixture design, including nanosilica (NS), nanosized TiO2, carbon nanotubes, and graphene oxide [1–6]. NS has been widely used due to its small particle size, spherical particle shape, and pozzolanic reactivity which are promising properties to further densify the microstructure of concrete [7–10]. In previous research, parameters of interest included the type of NS, the size of the particles or the surface area, the concentration of NS, addition or replacement of cement, and water to cement ratios [11–17]. One of the key challenges in densifying and, therefore, strengthening the microstructure lies in the dispersion quality of the nanoparticles during the mixing process. Figure 1 demonstrates this concept through three systems. System I is a well-dispersed cementitious system with cement and microsized silica, representative of ultrahigh-performance concrete, exhibiting a high particle-packing density and thus enhanced mechanical and durability properties in comparison to conventional concrete [18]. The addition of nanoparticles is a logical step to further increase the particle-packing density. System II highlights what commonly occurs with the inclusion of nanoparticles. These nanoparticles rapidly and irreversibly agglomerate, resulting in an undesired reduction of particle-packing density and thus leading to an inconsistency of mechanical properties. System III is the motivation for this research: Increasing particle-packing density through uniform dispersion of nanoparticles. The control of the dispersion

quality will control the concrete's microstructure and thus will have a direct effect on the material's mechanical and durability performance.

**Figure 1.** Illustrating the effects of poorly dispersed and well dispersed nanosilica (NS) in a cement system.

The use of NS in concrete has been proven to be a controversial topic due to significant variations of experimental results. Different researchers have reported that nanosilica increases, decreases, and has no effect on the mechanical properties of conventional concrete [19–21] and ultrahigh-performance concrete (UHPC) [22–24]. UHPC is a cementitious composite characterized by a compressive strength in excess of 150 MPa and enhanced ductility and durability properties as compared to conventional concrete. UHPC contains several constituents of varying sizes including sand, cement, silica powder, and silica fume, decreasing in size respectively. These different size gradations are essential to obtain dense particle-packing and favorable mechanical properties. [25–27]. With UHPC containing nanoparticles, the reason for discrepancies in mechanical properties is hypothesized to be a result of the variation in the dispersion quality of NS throughout the matrix. While high dispersion quality is expected to lead to enhanced material properties, poor dispersion quality as illustrated in Figure 1 (System II) compromises the improvement of mechanical properties [28,29].

NS particles are susceptible to form large agglomerates due to their very high specific surface area (50–750 m2/g). Colloidal NS, as opposed to a dry powder, provides NS particles in a well-dispersed state, free of significant agglomeration. These discrete particles have hydroxylated surfaces and achieve stability by possessing a net negative charge on the surface, strong enough to repulse the van der Waals attractive forces. If different ions are introduced to the sol, this surface charge can be altered and can lead to agglomeration and gelation of the NS [30–34]. Therefore, adding well-dispersed NS to the concrete mixture can lead to agglomeration and gelation resulting from the release of ions during cement hydration.

Understanding the dispersion mechanisms of NS is essential in the engineering and design of novel nanomaterial enhanced cementitious composites. Besides enhanced packing density, improved pore structure, and thus enhanced mechanical and durability properties, these new composites can exhibit specific functionality by inheriting properties of the added nanoparticles, such as air depolluting, elasticity control, and acceleration of hydration kinetics [35–38].

The goal of this research is to understand the mechanisms that cause NS destabilization and agglomeration in UHPC. It is hypothesized that even well-dispersed colloidal NS could result in agglomerates in the concrete environment of UHPC. The challenge is to define and investigate the dispersion quality of colloidal NS sols under concrete mixing conditions. In this paper, particle size distribution (PSD) through dynamic light scattering (DLS) and zeta potential (ZP) were measured and analyzed to evaluate the dispersion quality. NS sols were diluted, pH was altered, and added to a synthetic pore solution to further understand NS destabilization.

#### *1.1. Experimental Theory*

#### 1.1.1. Colloidal NS

Most commercially-available NS sols are stabilized at a pH range of 7–10. Silica particles are negatively charged as a result of the silanol groups (Si–O–H) on the surface. This provides a repulsive force between adjacent particles. Since the silica must be electrically neutral, counter ions such as potassium, sodium, and ammonia are inserted into the solution. These positively charged ions balance the charge by being attracted to the negatively charged surface of the silica particles [39]. NS sols can become destabilized for various reasons and once the van der Waals attractive forces are greater than the repulsive forces, the particles irreversibly agglomerate [40]. Silica can also be stabilized at low pH by removing all ions, resulting in slightly negatively charged particles, as this is the case with NS-20a used in this study.

The Derjaguin-Landau-Verwey-Overbeek (DLVO) theory predicts the stability of charged particles in a liquid medium by factoring in van der Waals attractive forces and electrostatic repulsion. DLVO theory predicts that if the electrostatic forces are greater than the van der Waals forces, the colloid will not agglomerate. DLVO provides a fundamental theory of the stability of particles in different pH and different ionic concentrations. However, in experimental practice with silica, DLVO breaks down at low pH. DLVO predicts at the isoelectric point at a pH of 2, the system should lack stability. However, at or near this point, colloidal silica has demonstrated stability [40]. This indicates that there are forces not considered by DLVO that significantly affect the stability, such as the steric hindrance effect. These properties of silica have made it difficult to predict stability, especially with the use of multivalent ions, varying particle sizes, and polymers. Therefore, experimental studies are necessary.

The pH of NS sols can affect the stability of the sol. At high pH (pH > 11), silica starts to dissolve, forming silicate ions, leading to destabilization [41]. At low pH, the repulsive forces are suppressed, resulting in a lower stability. Another potential source for NS destabilization is the ionic composition and concentration in the medium surrounding NS particles [42]. Monovalent and divalent ions, such as potassium (K+) and calcium (Ca2+) used in this study, alter the ionic charge of sols. This results in two mechanisms of agglomeration. First, the adsorption of potassium and calcium ions can lead to a reduction in the electrical double layer (further discussed in Section 1.1.3.). This reduces the electrostatic repulsive forces and attractive forces dominate. In addition to compressing the electrical double layer, these ions can form bridges between NS particles. A positively charged ion can neutralize a site on the silica surface. If two of these uncharged surfaces collide, the ion can coordinate with the oxygens of the silanol groups and the surface bonded water, forming a physical bridge between two silica particles [41].

For example, at a pH between 6 and 9, alkali metals at a high enough concentration can result in agglomeration. However, at a higher pH (pH > 10), it has been observed that potassium, rubidium, and cesium do not cause agglomeration at high concentrations. It is hypothesized that these ions form a complete double layer around the particle effectively shielding the silica surfaces from coming into contact [41]. Why this happens at a higher pH and not a lower pH is unknown, again supporting the observation that DLVO theory is not perfect.

These two effects of agglomeration are amplified with divalent ions such as calcium. A divalent ion neutralizes one negative surface charge and thus releases a hydrogen ion, while the originally divalent ion retains a positive charge. Since the ion still retains a positive charge, it can neutralize an additional site on an adjacent silica particle, forming a physical bridge between particles. The adsorption of calcium results in a mosaic of positive and negative charges on the surface which facilitates the agglomeration of particles as there are many sites for opposite charges to attract each other, as proposed by Goodman [43] and Iler [44]. When there are two different ions adsorbed onto the surface, such as potassium and calcium, coagulation can occur more rapidly. The addition of potassium reduces the overall repulsive forces by neutralizing some of the surface charge. Calcium then causes bridging and coagulation at lower concentrations compared to sols without salt [45,46].

#### 1.1.2. Dynamic Light Scattering

DLS was used to determine the PSD of colloidal NS particles or their agglomerates. DLS measures time-dependent fluctuations in the intensity of scattered light of particles in a suspension which move in a random Brownian motion [47,48]. These fluctuations are processed into an autocorrelation function and applied to fitting algorithms [49]. In this research, the cumulant method [50] has been used. It is the most common method to obtain a particle size distribution from the autocorrelation function.

Two measurements from DLS are primarily used in this research: Size (z-average) and polydispersity index (PDI). The z-average is the intensity-weighted mean hydrodynamic size of the particles, which is derived from the cumulants analysis. The PDI is a dimensionless number calculated from a fit to the cumulant data. It is a measure of how narrow or broad the PSD is. Smaller values indicate a narrow size distribution. Values greater than 0.5 indicate a very broad distribution and results become more qualitative than quantitative.

In this research, a lower z-average, closer to the manufacturers supplied value, indicates a better dispersion quality. A higher z-average indicates a lower dispersion quality, as agglomeration has occurred. In addition, lower PDI values are preferred, indicating a small range of particle sizes.

#### 1.1.3. Zeta Potential

Obtaining the zeta potential (ZP), or the electrokinetic potential, is a technique to quantify potential stability of suspended particles. By measuring the electrophoretic mobility of a particle, the ZP is calculated. A higher magnitude of ZP corresponds to a higher degree of electrostatic repulsion between the charged particles and, thus, a higher dispersion stability, or stronger agglomeration resistance. As a general rule of thumb, magnitudes greater than 30 mV indicate stability and magnitudes less than 30 mV indicate higher potential for agglomeration and coagulation [47,51].

The distribution of ions in a solution is affected by charged particles. These charged particles result in a higher concentration of ions close to the surface. In this region, known as the Stern layer, ions such as Na<sup>+</sup> and K<sup>+</sup> are strongly bound. An outer layer, known as the diffuse layer, contains ions that are less strongly bound. These two layers form an electrical double layer around each particle. These two layers extend a finite distance from a particle surface and are influenced by several factors including pH, ionic concentration and composition. The potential at the edge of electrical double layer with the surrounding environment is the ZP [40,47,52]. The thickness of the diffuse layer is the Debye length. The Debye length is a measure of how far the electrostatic forces persist from the particle surface. A larger Debye length keeps particles further apart which increases sol stability. In aqueous solutions of monovalent ions, the length is reciprocally proportional to the square root of the ionic concentration. In a 0.1 M solution (the maximum concentration of KOH in this work) the Debye length is 0.96 nm. In a 0.4 M solution of divalent ions, the Debye length is 0.28 nm [51]. This is a very short distance, which is one reason why NS can be difficult to stabilize in some environments.
