The silica fume is sourced from a local ferrosilicon alloy factory. The silica fume is originally formed as dust created from burning ferrosilicon raw material, coke and silica in an electric arc furnace, which is then collected in a gas cleaning plant. The dust is filtered and contained in special plastic bags as silica fume byproduct. Fig. (2a) a shows a photograph of the dry silica fume on a glass plate. It consists of extremely fine solid particles, light grey in color and odourless. In general, the color of silica fume can range from light to dark grey, depending on the production process [15]. The moisture content of the dry silica fume sample (from the bag) is 0.82%.

Fig. (2b) shows a picture from Scanning Electron Microscope (SEM) of the silica fume sample. The particles are shown as spherical in shape with the diameter mostly less than 0.2μm. Therefore, in a dry form, the silica fume is respirable. Since the particles are extremely fine, the specific surface area reaches between 13,000 m²/kg to 30,000 m²/kg, which is much higher compared to Ordinary Portland Cement (OPS), i.e. approximately 150-430 m²/kg [16, 17]. Larger size particles (lumps) are the results of binding together with smaller ones due to moisture. Therefore, sieve analysis of silica fume from the bag might result in a larger particle size distribution if lumps are not removed.

The chemical composition of the silica fume is predominantly SiO₂ that comprises 94% by weight. The other 6% consists of other chemical components, namely: Fe₂O₃, K₂O, MgO, CaO, Na₂O, Al₂O₃, and P₂O₅. The specific gravity of the solids (G_s) of the silica fume is 2.2, at which the unit weight (γ) ranges between 200–350 kg/m³, 500–650 kg/m³, 700–1000 kg/m³ and 1400 kg/m³, when it is prepared in undensified, densified, micropelletized, and slurry conditions, respectively [18]. The G_s and the γ makes the silica fume to be much lighter than normal silica sands (G_s = 2.65). The liquid limit of the silica fume is 70%. The liquid limit is an important parameter for the flowability during compaction grouting and thus determines the preparation of samples for rheology and strength testing in this study. The high liquid limit value is associated with very large specific surface area when compared to cohesive soils such as silt and clay. The larger the specific surface area, the larger amount of water is needed to make the material behave like liquid. Table 1 summarizes the physical properties of the silica fume used in this study.

The Cemented Silica Fume Paste (CSFP) is designed by mixing the silica fume with cement binder and tap water at appropriate proportion. The cement binder used is an Ordinary Portland Cement (OPC) sourced from Cahya Mata Sarawak that conforms with the standard requirement specified in Malaysian Standard MS EN 197-1 [19]. The CSFP is soluble in the alkali solution. The amount of water determines the consistency of the CSFP in the solution.

Table 2 lists the design mixture that was used for the rheological properties test and UCS test in the experiments. There are three variations in terms of water-to-solid (W/S) ratio by weight, i.e. 0.8, 0.9 and 1.0. Note that the term W/S is synonymous with water content. We chose the W/S ratio to avoid confusion since the mixture has two types of solids, i.e. cement and silica fume. Also, the W/S ratio is a more practical term compared to water content. The three variations of W/S ratio are all slightly above the liquid limit, to ensure that the CFSP flows in paste-like consistency. Each W/S ratio is varied with different proportions of cement content ranging from 0% to 30% with a 5% increase. Note that the cement content is calculated in the percentage (%) by weight. Since the total amount of solid is fixed, the amount of cement replaces the amount of silica fume. The variation of cement content is investigated to determine its effects on rheological properties and shear strength of cemented silica fume paste at an early age. There are a total of twenty-one (21) specimens used in the rheological properties tests and forty-two (42) specimens used to determine the shear strength by performing the Unconfined Compressive Strength (UCS) tests. For the UCS tests, each mix has two specimens. The actual W/S ratio was recalculated after each test, since the water might dissipate during the UCS specimen preparations and tests. The prepared specimens of the design mixture included the control sample with a 0% proportion of cement.

Fig. (3) shows the workflow of the experiments. Prior to rheology and strength tests, the silica fume material was physically and chemically characterized, including scanning electron microscopy. Based on the Atterberg limit values, a trial mix was conducted to determine the range of Water to Solid ratio (W/S), at which the CFSP has a good workability (flowable) consistency but is not too liquid that might cause the segregation of cement and reduction in strength.

Rheological properties test was performed to determine the rheological parameters such as shear stress and viscosity of cemented silica fume paste when some of the silica fume is replaced with the cement binder (OPC). The test was conducted using a rotational viscometer instrument, HAAKE Viscotester 7 plus, in accordance with ASTM D2196 Standard Test Methods for Rheological Properties of Non-Newtonian Materials by Rotational Viscometer [20]. All test conditions and settings prescribed in the standard method were followed precisely. The tests were all performed at a controlled room temperature of 25^oC. The sample of cemented silica fume paste was prepared by manually mixing the silica fume, the cement binder (OPC) and the tap water according to the design mixture (Table 2). The CSFP material components were thoroughly mixed in a glass beaker using a spatula for about 5 minutes.

The test specimen was made by pouring the sample into a 0.5 litre can to reach 25 mm from top. The specimen was shaken vigorously on the shaker for about 10 minutes. After 10 minutes, the specimen was removed from the shaker and was allowed to stand undisturbed for 60 minutes at 25^oC prior to testing. The rheology test was started no later than 65 minutes after removing the can from the shaker. The instrument was placed on the adjustable stand and the viscometer was lowered to a level that would immerse the spindle to a proper depth. The spirit level was attached to level the instrument to align the specimen horizontally. The selected spindle was tilted and inserted into one side of the center of the surface of the materials, followed by attaching the spindle to the instrument. The viscometer was then lowered until the groove (immersion mark) on the shaft just touched the materials. The container was slouched in a horizontal plane until the spindle was located in approximately the center of the container so that the test would run in a region that was undisturbed by lowering the spindle. The digital viscometer was turned on and adjusted to the selected speed (unit in rotation per minute, rpm) for the materials under the test. The data was registered in the viscometer instrument at the following standard shear rates (1/s): 0, 1, 2, 4, 6, 10, 12, 20, 30, 50, 60, 100, and 200. The viscometer was allowed to run until the digital reading had stabilized and the digital viscometer gave a direct reading in Pascal and centipoises for shear stress and viscosity, respectively. The same procedure was repeated for every other design mixture.

Unconfined Compressive Strength (UCS) test was selected to determine the strength of the CSFP samples at an early age when subjected to axial stress. The axial stress was applied to a cylindrical CSFP specimen without lateral constraint (i.e., no confining pressure) while observing the axial strains corresponding to various stress levels. The axial stress at which failure in the soil specimen occurs is referred to as the UCS shear strength (Pa). The failure was defined when the axial stress dropped from the peak, followed by a physical appearance in the sample in the forms of cracks, disintegration of the material or excessive deformation. The test was conducted in accordance with ASTM D2166 “Standard Test Method for Unconfined Compressive Strength of Cohesive Soil” [21].

The test specimen was prepared in a small cylindrical PVC mold measuring 38 mm × 76 mm. The specimen was made soon after the rheology test was done. The molding and specimen extraction took about 10 minutes. After extraction, the diameter and length of the specimen was measured twice to get the average values. The specimen was placed centrally between the two loading plates of the UCS machine and the base loading plate was moved upward until the top of the specimen reached the top loading plate. The specimen was compressed vertically at a displacement rate of 1mm/min, and the readings were taken at the vertical displacement intervals, ΔL of 0.3 mm. The specimen was compressed until the specimen failed. The failed specimen was collected, weighed and placed in the oven for 24 hours for the W/S ratio calculation (i.e., moisture content).

Fig. (3). Methodology flow chart.

Figs. (4-6) show the observable shear stress and viscosity measurements against shear rate from the viscometer instrument reading for samples with W/S of 0.8, 0.9 and 1.0, respectively, with all variations of cement contents. The measurements are presented in log-log scale. The shear stress and viscosity start to be observable (i.e., > 0) at a shear rate of 10/s and above. The shear stress and viscosity data showed to be dependent on the shear rate. The shear stress increases as the shear rate increases, while the viscosity decreases as the shear rate increase (opposite). The same trend applies to all three samples with different W/S and cement contents.

Fig. (4). Shear stress and viscosity of samples with W/S 0.8: (a) Shear stress vs shear rate, (b) Viscosity vs shear rate.

Fig. (5). Shear stress and viscosity of samples with W/S 0.9: (a) Shear stress vs shear rate, (b) Viscosity vs shear rate.

Fig. (6). Shear stress and viscosity of samples with W/S 1.0: (a) Shear stress vs shear rate, (b) Viscosity vs shear rate.

The UCS measures the ability of the specimen to sustain the axial stress without lateral confinements. The UCS shear strength relies on the cohesivity of the material. The UCS shear strength, s_u was calculated as: (Eq 1).

(1)

where σ_max is the maximum axial stress (at failure). Fig. (7a) shows the relationship between UCS shear strength and the cement content. A good linear correlation between shear strength and cement content was found, where the shear strength increases as the cement content increases. The relationship signifies that cement binder has a significant effect on the cohesivity of the material at an early age. Fig. (7b) shows the W/S (actual) versus the cement content. It shows a fair correlation between cement content and the W/S; whereas the cement content increases, the W/S decreases. Note that the actual W/S was the W/S conducted after the UCS test (via oven drying). The actual values of W/S differed slightly from the design W/S (Table 2), i.e. the ones used during preparation.

Fig. (7). UCS shear strength and W/S versus cement content: (a) UCS shear strength vs cement content, (b) W/S vs cement content