Steel slag’s unique physical properties result both from its chemical composition and its production method. Comparison to standard construction aggregates show a higher bulk density and specific gravity, higher hardness/toughness, coarser surface texture, and darker color.
3.1.1. Specific Gravity and Bulk Density. The bulk density and specific gravity
of steelmaking slag is higher than blast furnace slag and natural aggregates due to the increased content of iron and manganese. Specific gravity values for steel slag averages 3.1-3.7 for BOF and 3.2-3.8 for EAF, in comparison to 2.1-2.5 for air cooled blast furnace slag and 2.85 for dolomite.13,19,21,38 The resultant bulk density values are 1770- 2500 kg/m3 (156.1 lbs/ft3), in comparison to 1440-1600 kg/m3 (89.9-99.9 lbs/ft3) for dolomite.16,19 Beaver Valley Slag, a slag processing subcontractor, lists as-sold bulk densities of steel slag from 1410-1930 kg/m3 (88.0-120.5 lbs/ft3) depending upon the degree of compaction.17
3.1.2. Hardness/Toughness. Mineral hardness can be defined as the resistance to
breakage, and toughness is a relative factor indicating the ease of breakage. Of
importance for determining the processing and end use for slag is a quantitative factor to determine the amount of work required to break the particles. High toughness and hardness is beneficial for end use of the slag as it provides a more stable material. In construction applications it is desirable for the aggregate to maintain its shape, and resist breaking down into smaller particles. However the higher hardness and toughness, the more energy required to break the slag into desired particle sizes during initial processing (crushing). Estimated Moh’s hardness for steel slag is 6-7 compared to 3-4 for
dolomite.21 Crushing work index and grinding work index provide comparison values for crushing (breakage down to 1-2 cm) and grinding (breakage below 1-2 cm) operations respectively. The crushing work index for dolomite is 12.8, which is the same for slag.19 However, the grinding work index (ball mill) for dolomite is 13.9 versus 17.2 for slag.19 Work index has the units of kWh/ton, and a comparison shows it takes about 24% more energy to grind steel slag compared to dolomite. Steel slag has the double benefit of
requiring about the same energy for crushing as dolomite, but being more resistant to in- situ dust and fine generation. In using slag for asphaltic cement or road surfaces, the slag is more resistant to breakdown during application (mixing and lay down) and for the life of the constructed surface.
3.1.3. Particle Shape. The shape of crushed slag qualifies it as a premier
aggregate for road construction applications. The shape of crushed slag is described as cubical, in comparison to dolomite which is rounded or limestone which is flat and elongated.10 The cubical nature, along with rough surface texture, allows for greater aggregate shear resistance, compared to round or flat particles, which resists rutting of asphalt surfaces. As an estimate to quantify the shape for comparison, the angle of repose can be compared. The angle of repose for slag is 45-50°, while that of crushed dolomite is 37°, indicating it has higher particle-particle resistance when stacked.20,21
3.1.4. Surface Texture. The surface texture and hardness of slag provides
outstanding skid resistance, a highly sought after feature in pavement construction. The surface texture of slag is primarily a result of its cooling method. Most steel slags are air- cooled under ambient conditions, with final cooling accelerated with a water spray. The solidified material has a vesicular nature from gases entrained during the melt and generated upon cooling. The pores increase the surface roughness and area allowing for higher coefficient of friction and binder affinity. Slag resists polishing during the application life, as the new surfaces generated by impact or wear contain pores which renew the friction resistance. The two terms that quantify skid resistance are polished stone values (PSV, high values desirable) and aggregate abrasion values (AAV, low values desirable). The PSV for BOF slag is 54-57 and EAF slag is 58-63, compared to 53 for dolomite and 41-45 for limestone.13,38 Likewise the AAV or steel slag is 2-4
compared to 4 for sandy dolomite and 8 for limestone.13 Another quantification of abrasion resistance is the Los Angeles Abrasion Test, specified by ASTM C131-03 Standard Test Method for Resistance to Degradation of Small-Size Coarse Aggregate by Abrasion and Impact in the Los Angeles Machine.47 This test measures the degradation, as percent loss, due to abrasion in a rotating grinding drum. Steel slag has an exhibited value of 9-18% for BOF and 8-15% for EAF compared to 35-40% loss for air-cooled blast furnace slag and dolomite.21,38
3.1.5. Mechanical Swelling. The mechanical swelling of steel slag is a result of
hydration and carbonation of some of the compounds present in the matrix. Details of the chemical nature of hydration and carbonation are discussed in subsequent sections, while this section will focus on the mechanical effects from swelling.
Quantification of slag expansion was undertaken by Tsuchiya et al. in 1980 by studying the expansion characteristics of slag during the water immersion test.22 The method of their test was to measure the rate and total volume of expansion of four LD slags. Employing a constant-temperature-water-immersion test with molds used for measuring California Bearing ratio, data was obtained on expansion with time. Expansion ratios of 2-13% were achieved at one year, with the ultimate expansion reaching 2.0- 35.9% (time not given). A comparison between the final expansion ratio and the chemical composition suggests a proportional relationship with the amount of free lime (f-CaO). Slags A, D, B, and C had 3.5%, 6.1%, 8.9%, and 9.2% f-CaO respectively, and a
corresponding ultimate volume ratio of 2.0%, 2.6%, 35.9%, and 19.3%. A comparison of the other chemical constituent did not show a proportional correlation. It is assumed that primarily hydrates were formed (as opposed to carbonates), since the slags were
subjected to water immersion testing in a closed mold and not exposed to atmospheric carbon dioxide. No characterization was undertaken on the final aged slags.
A second round of tests was conducted on the highest expanding slag (B), where the slag was allowed to field age at zero, one, three, and six months before undergoing the immersion test. Field aging had the effect of significantly reducing the ultimate expansion ratio of the slag. At zero months field aging 35.9% expansion was exhibited, but this dropped to 18.2% expansion at one month field aging and 6.4% expansion at six months of field aging. Field aging and shortened the time needed to complete expansion, but lessened the ultimate expansion value. Qualitatively, these results show that the processes occurring in the field aging and water immersion are different. It is suggested that the free lime in field aging is subject to competing hydration and carbonation reactions from the H2O and CO2 in the air. Ultimately some carbonate forms, reducing
the available free lime for hydration in the immersion tests. The hypothesis for reducing the time to complete the reaction is that carbonate formation during field aging increases
crack formation and propagation, allowing greater surface area for hydration during the immersion tests.
Tsuchiya et al. found that most free lime in LD slag is present in unassimilated lumps, not distributed in solid solution.171 The free lime lumps are several mm in diameter at a density of one per cm3, and are observed as white to brown color in a blackish slag matrix.
In 1997 Kandhal and Hoffman published the results from a 1982 feasibility study of using cured steel slag fine aggregate in HMA mixtures.16 They noted that for raw slag, the unslaked lime will hydrate, causing large volume expansions in a few weeks, while the magnesium oxide hydrates more slowly causing volume changes that may occur after several years. Prior to their work, the Pennsylvania Department of Transportation
required that raw steel slag be moist cured under sprayed water conditions in a controlled stockpile for six months to alleviate expansion potential. Even with this requirement, several failures occurred where aged steel slag used as structure backfill expanded 3-6%. As a result, the Pennsylvania Testing Method (PTM) 130 was produced to better
characterize expansion characteristics of steel slag. This method was used to generate ASTM D4792-00 Standard Test Method for Potential Expansion of Aggregates from Hydration Reactions.48 Their research focused on the use of fine steel slag for hot-mix asphalt. Because the asphalt binder coats the aggregate particles and seals of the hydration route, this application serves as a better use for steel slag than unbound aggregate. Of use in their research is the evaluation of swelling potential from 10 slag sources. All samples were from Pennsylvania, with three from uncured stockpiles, and seven from stockpiles aged at six months or longer. Expansion tests were undertaken according to PTM-130, which uses a six-inch diameter mold filled with compacted slag and submerged in water at 71°C (160° F) for seven days. After seven days the sample is removed from the water bath, but kept saturated by adding water daily. Expansion measurements continue an additional seven days for a total of 14 days. A dial gauge measures vertical expansion from the initial height of 116.4 mm (4.584 in). Expansion values of 1.1-2.8% were observed for uncured slag, and 0.0-0.3% for cured slag. Further tests were undertaken to determine the particle size effect on expansion potential.
mm (0.185 in). The undersize (<4.8 mm) and oversize (4.8-50.8 mm) particles were then subjected to PTM-130 testing. Coarse particles showed 1.13-2.83% and 0.27-0.67% expansion for uncured and cured respectively. Fine particles showed 0.87-2.7% and 0.03- 0.23% expansion for uncured and cured respectively. These results indicate homogeneity in expansion potential versus particle size.
The testing time range for Tsuchiya et al. and Kandhal et al. was significantly different (>400 days vs. 14 days). However, a comparison of data from both sources at 14 days gives an average expansion on uncured slag of 0.7% for Tsuchiya and 1.9% for Kandhal. The difference lies in both the sample source (Tsuchiya was LD slag and Kandhal was unspecified), and that the Kandhal testing allowed the second half of the testing to have contact with air. The higher expansion rate may be a result of carbonation expansion taking place in addition to hydration expansion. Insufficient data was available from Tsuchiya to compare the cured slag expansion.
The literature sources reviewed have stated several times that the primary constituents in slag that lead to swelling are free lime (CaO) and free magnesia (MgO). However in the slag mechanical swelling studies reviewed16,22, no characterization work was undertaken to compare the phases before and after the swelling test. The volume increase contribution of free oxides alone versus other phases could not be determined. A comparison of the molar volumes of lime and magnesia to that of their corresponding hydroxides and carbonates provides an estimate of their individual expansion capacities. Lime has a molar volume of 16.76 cm3 (1.022 in3), and upon hydration this increases
97% to 33.08 cm3 (2.019 in3). The formation of calcium carbonate leads to 120% growth
(36.89 cm3 (2.251 in3)) from the oxide. Magnesia shows similar dramatic change, growing from 11.25 cm3 (0.6865 in3) to 24.63 cm3 (1.503 in3) upon hydration (119% increase), and to 28.03 cm3 (1.710 in3) upon carbonation (149% expansion) from the dry state. Other compounds in the slag (primarily silicates of the alkaline earth metals) also show a volume increase during hydrate and carbonate formation. A thermodynamic review of these minerals, including molar volume change, is covered in Section 3.4.