THE LOAD DEPENDENCE OF THE MICRO-HARDNESS OF THE BLAST FURNACE SLAG

Deposits of old blast-furnace slag are an environmental problem. The slag’s hardness is an important for calculation of the energy cost for crushing and grinding process. Due to its porosity, the measurement of (macro) hardness must be adapted so that the indentation dimensions are suited to the character of the slag. Thus, it is necessary to apply loads in the range of micro-hardness. The purpose of this paper is to evaluate the influence of load on the micro-hardness the Indentation Size Effect (ISE) using Meyer’s, Hays-Kendall and PSR methods. ISE for all samples is “normal”, the slag’s basicity affects micro-hardness and ISE.


Introduction
Since the 18 th century, large plants with several blast furnaces produced enormous amounts of slag. According to data published by Ling et al. [1], the annual production of granulated blast furnace slag in China is around 15 million tons. About 17 million tons of iron and steel slags (1:1) were sold in 2011 in the USA [2]. The worldwide pig iron and steel slag production, as estimated by Kalyoncu [3], was approximately 200 million tons in 2000.
Because the dumping of the slag has a potentially undesirable impact on the environment, it is advisable to recover metals from the slags and utilize them in other ways. As demonstrated Veselská & Majzlan [4], the layers of old iron slag are generally not dangerous with regard to possible toxicity. However, they are dangerous due to the unstable subsoil. It was found that the radioactivity of iron slag in Slovakia [5,6] and in Romania [7] does not exceed the limit permitted by the legislation.

Materials and Methods
The experimental material is the slag of iron smelting in charcoal blast furnaces. Samples were found in the territory of Slovakia. The "Slovak furnace" (Windische Ofen) the bloomery operated by water power was the dominant equipment until the middle of the 19 th century. The defensive war against the Osman Empire (1541-1685), repeated rebellions against the absolutist rule of the Hapsburg dynasty (1605 -1711), and the conservatism of the enterprisers caused the lag in the blast furnace process; the first blast furnace was built only in 1692. The iron metallurgy in Slovakia reached its peak in the middle of the 1870s (59 ironworks with blast furnaces); only two blast furnaces are in operation today. A list of analysed slags is in Table 1. Their microstructure generally consists of fayalite (Fe2SiO4) with globular iron particles, pores, blows, and cracks. Typical microstructures of analyzed slags are in Fig. 1; for example, the slag No. 2 (Remetské Hámre) with diopside is in cell 2). The samples are numbered in accordance with Table  1. X-ray analysis of crystalline phases, slags of Lúčka, Jakubany, and Henckovce, shows  Samples were cut by cooled diamond saw ground with dry SiC sandpaper (220 → 3000ANSI/CAMI grit) and polished with diamond paste (1 µm). The rest of the sample was pulverised and analysed by AAS PERKIN ELMER 3100, and by spectrometer Niton XL3 Goldd, the results are in Table 2. The presence of secondary (1 -0.001 %, bold type) and trace (0.01-0.00001 %) elements was determined by spectrograph PGS-2. The crystalline phases were identified by X-ray powder diffractometer The Vickers micro-hardness was measured by Hanemann tester, type Mod D32 fitted to microscope Neophot-32. A reference block with specified hardness Hc = 519 HV0.05 and standard uncertainty uCRM = 6.75 HV0.05 was used for the calibration of the tester according to the standard ISO 6507-2 [23]; the tester has met its requirements (rrel = 4.65 %, Erel = -1.31 % and Urel = 6.32 %).

Table 2. Composition of analysed slag and compared specimens (in wt. %), and estimated production of the slag in the ironworks.
As regards the measurement of the micro-hardness, five indentations were done at each load P between 0.09807 N and 0.9807 N in step 0.09807 N with the load duration 15 s and the average speed of the indenter's penetration 1.0 μm/s. The "measurement" is a file of 50 values. The values of micro-hardness are in Fig. 3  The normality and outliers of individual "measurements" were calculated by Grubbs' test (significance level α = 0.05) and by Anderson -Darling test, p > 0.05 for the "measurement" with a normal distribution. The average micro-hardness HV of the "measurement", micro-hardness HV0.05, normality (pvalue), outliers, and relative expanded uncertainty of micro-hardness HV0.05 Urel calculated according to standard, ISO 6507-1 [24], are in Table 3.

Theory and calculation
Meyer's power law, proportional specimen resistance (PSR), and Hays -Kendall approach were used for the quantitative description of the ISE. The procedure for calculating ISE characteristics using references of Sangwal et al. [18], Li & Bradt [25] and Michels & Frischat [26], and in particular, the procedure of the measuring the ceramic micro-hardness by Kim & Kim [27] presented in more detail Petrík [20].  (1), is the simplest way to describe the ISE: Meyer's index n or work hardening coefficient is the slope, and Aln is the yintercept of the straight line graph of ln (d) versus ln (P); d (mm) is the diagonal of indentation, and P (N) is applied load. For "normal "ISE n < 2, for reverse ISE n > 2 and n = 2 is given by Kick's Law for the load independent micro-hardness.

Fig. 4. The relationship between the micro-hardness HV and Meyer's index n.
Values of indices n and Aln are listed in Table 4. As can be seen in Fig. 4, the value of n decreases (with a tendency towards a more pronounced ISE) with increasing microhardness of the slag. The proportional specimen resistance model of Li & Bradt (PSR) is practically a modified Hays/Kendall approach. Several authors, as Gong et al. [21] and Li & Bradt [25] have proposed that the "normal" ISE behaviour can be described by equation (2): Indices a1 (N/mm) and a2 (N/mm) are related to the elastic and plastic properties of the material, respectively [27]. Index a1 is connected with the dependence of microhardness on the load; harder materials as a slag generally have higher a1 values. Gong et al. [21] presented that the index a2 is associated with load-independent "true hardness "HPSR, calculated by equation (3).
Indices A1 or c2 also can be used; values of "true hardness "are in Table 4. Equation (4) is a modified form of the PSR model.
The index c0 (N) is associated with residual surface stress, and indices c1 ~ a1 and c2 ~ a2 are related, respectively, with the elastic and plastic properties, Table 4.
The ratio c1/c2 is a measure of the stresses due to machining and polishing of the sample. The expected relationship between c0 and c1/c2, was achieved only after the elimination of the hardest slag (No. 6), Fig. 5.
Hays and Kendall proposed the existence of minimum test load W (N) necessary to initiate plastic deformationvisible indentation. Below it, only elastic deformation occurs. It is expressed by equation (5).
Where A1 (N/mm 2 ) is an index independent of load, Table 4.

Results and discussion
Pure metals (Al, Zn, Cu, Fe, Ni, Co) tested at loads between 0.09807 N and 0.9807 N show "reverse "ISE [28]. Particles of mullite (3Al2O3.2SiO2) may occur in blast furnace slags, and their hardness is about 7 GPa (700 HV) with low dependence on an applied load (n = 1.979); according to Gong et al. [21], the ISE in mullite is negligible. The microhardness of the slag with chemical composition listed in Table 2 (sample No. 12) ranges between 3.75 -7.25 GPa (375 -725 HV) depending up a final heating temperature and soaking period [29]. As presented Fredericci [30], the micro-hardness of the slag (sample No. 10 in Table 2) is 5.2 ± 0.4 GPa (520 ± 40 HV) at load 0.4903 N and of the slag (sample No. 11 in Table 2) is 8.4 GPa (840 HV) at load 1 N. According to what Ostrowski & Rzechula [31] and Liu et al. [15] reported the blast furnace slag with basicity CaO/SiO2 = 0.4 and ~ 10 wt. % Al2O3 has micro-hardness of 6.2 GPa (620 HV) at load 1.96 N. These examples illustrate the measurement of micro-hardness of slags in practice. However, the results are rarely published, and the study focusing on the load influence on the micro-hardness, is scarce.
The basicity B of the slags was calculated by equation (6).
With increasing of basicity, the micro-hardness and the value of Meyer's index n moderately decreases (toward to "normal" values). The drop in "true hardness" is more pronounced, Fig. 6.
The present blast furnace slag is basic. Greater porosity and inhomogeneity of basic slag (more or less homogeneous fayalite is replaced by a mixture of wollastonite, gehlenite, melilite, merwinite, rankinite, or monticellite) will complicate the measurement.
Microstructure (also with an indication of the presence of further phases in the glass matrix of samples No. 2 and 8) and crystalline phases (wollastonite in sample No. 1 and diopside in samples No. 2 and 8) have not a significant impact on the microhardness and ISE. Because there are differences in micro-hardness of wollastonite (5.6 GPa at load 2.495 N), stated by Teixeira et al. [32], diopside (7.0 -8.8 GPa at load 0.5 N) stated by Smedskjaer et al. [33], and fayalite (6.0 GPa at 0.495 N) stated by Takeda et al. [34], the more pronounced effect of these crystalline phases on micro-hardness was expected. Fig. 6. The relationship between basicity and "true hardness" of the slag.
In addition to the difference in chemical composition and consequential crystalline texture, the morphology, different cooling rate, the age of samples, the uncertainty in measurement could be the reason for the variance in the measured values of microhardness and ISE.
In the next, the research would be appropriate to focus on the relationship between micro-hardness (and in particular true hardness) and grindability (for example, the influence of the micro-hardness on the grindability index or grinding resistance).

Conclusion
1. The influence of the load on the micro-hardness of the slag is statistically significant. 2. The relationship between the applied load and micro-hardness manifests "normal" ISE for all tested samples. 3. The relationship between c0 (the residual stress) and c1/c2 ratio (a measure of the stress due to grinding and polishing) is proportional except the hardest slag (sample No. 6). 4. The presence of wollastonite and diopside does not have a significant effect on the micro-hardness and ISE. 5. The increasing of the basicity moderate decreases Meyer's index n (towards to "normal" ISE). 6. Due to the significant increase in hardness at lower loads, the authors recommend not to use a load lower than 0.6 N if the porosity of the sample allows.