Replacement of fluorspar in the desulfurization of hot metal

In the steel industry, lime is used as a desulfurizing agent and fluorspar as flux in the hot metal desulfurization process. However there are environmental concerns regarding the use of fluorspar, even when its harmful effects are not fully known. In order to reduce such risks and also to anticipate possible future restrictions, which could undermine the sustainability of steel production, it is proposed to replace the fluorspar for alternative materials in the composition of the desulfurizing lime based mixture. For this purpose new mixtures using aluminum dross, ilmenite, sodalite and ulexite were tested, comparing them with a reference mixture containing fluorspar. The impact of these mixtures on the refractories commonly used in hot metal pretreatment was also checked, since this is an important cost parameter for the steel industries. The laboratory tests showed ulexite as a feasible substitute since the lime (84.03%) – ulexite (10.97%) – graphite (5%) mixture presented the highest desulfurization ratio and low refractory wear (less than the standard fluorspar mixture). The sodalite mixture was the worst by comparison. The ilmenite and aluminum dross mixtures showed similar behavior when compared to the fluorspar mixture.


Introduction
Fluorspar is a common raw material used at ironmaking and steelmaking facilities.Although it is a powerful fluxing agent, its benefits are to be weighed against health and environmental concerns.Fluor can be lost from industrial slags due to the reaction 2CaF 2 (slag) + SiO 2 (slag) →SiF 4 (g) + 2CaO (slag) .
Depending on temperature and slag composition compound such as NaF(g), KF(g) and (SiF 4 ), AlF 3 , CaF 2 , BF 3 can be emitted, Figure 1.(adapted from Zaitsev, 1994)  In regard to hot metal pretreatment, it has been suggested using soda ash additions.CHOI et al (2001), NIEKERK and DIPPENAAR (1993) report de-S using CaO-SiO 2 -Al 2 O 3 -Na 2 O slags at 1350ºC.Reasonable degrees of de-S can be achieved but alkaline losses by evaporation/chemical reactions can pose metallurgical and environmental restrictions.
As can be seen most of the references regarding fluorspar replacement, they are related to steelmaking (not the scope of this article and where the thermodynamic environment is less favorable).AGUIAR et al (2011AGUIAR et al ( , 2012) ) did some work on the utilization of marble cutting residues and fluorspar for hot metal desulfurization; their main drive was related to environmental concerns.SILVA et al (2012) investigated the replacement of fluorspar by sodalite in calcium carbide mixtures for hot metal desulfurization.A less common approach is to investigate lime based mixtures used for hot metal pretreatment.Lime is an important reagent due to its unique combination of price and availability.PEZZIN et al (2016) studied the replacement of fluorspar by sodalite nepheline in lime based mixtures and marble waste based mixtures.They tested some mixtures, see Table 3, for de-S efficiency of a hot metal at 1400ºC.The laboratory results have shown the addition of sodalite nepheline decreases the de-S efficiency; in addition it has been observed that most of the reaction occurred in the first 10 minutes.Table 3 de-S efficiency using sodalite nepheline and fluorspar.
Other reasons for fluorspar substitution is refractory wear and availability (it is becoming scarce).According to VOLKMANN (2002), sodalite nepheline has been evaluated for fluorspar substitution in BOF operations.Using a 1:1 replacement, a ratio of 88.8% de-P was achieved for fluorspar and 88.1% for sodalite nepheline.Furthermore, at ladle metallurgy, de-S ratio didn't change.Refractory wear due to sodalite was evaluated under laboratory conditions and deemed to be smaller than that of fluorspar.KEUM et al (2007)

Raw materials
The chemical composition of hot metal and raw materials is shown in

Mistures
Four de-S mixtures, containing ilmenite, sodalite, aluminum dross and ulexite, were designed with the goal of fully replacing the fluorspar and roughly keeping lime partici-pation (75~85%).All the mixtures and their compositions are shown in Tables 7 and 8.  Thus replacement of fluorspar at hot metal pretreatment facilities, mainly regarding de-S purposes, is still a challenge.This work reports laboratory results dealing with lime mixtures containing ilmenite, sodalite, aluminum dross and ulexite for de-S at the Kanbara reactor.At the same shops, a lime fluorspar mixture is the standard agent; thus it is taken as a basis of comparison.
A brief information regarding price, world production and reserves of the fluxes tested in these experiments is shown in Table 4.Where this information is not readily available due to marketing reasons, a similar mineral or material, such as alumina instead of aluminum dross, borate minerals in place of ulexite and nepheline syenite instead of sodalite syenite, is taken into consideration for comparison purposes.It is apparent that in some cases a market needs to be developed, if the new technology is adopted.

De-S experiments
In order to reflect industrial conditions a slag/metal ratio of 7.5kg of reagent/ ton of hot metal was charged to an electrical resistance furnace, Figure 2. The furnace inside was continuously purged with argon at 10~15 lpm flow rate.Agitation was provided by a stirring rod rotating at 800 rpm.A previously dried and pre-heated reagent mixture was added on the top of the metal bath; samples were collected at the beginning of the experiments, 10, 20 and 30 minutes; and temperature was controlled at 1420 ± 5°C.In addition, small amounts of each mixture were pressed into cylinders (3mm high, 5mm diameter) and placed in a furnace at 1400 o C during 30 minutes; heating rate was about 8K/s and the furnace was continuously purged with dry nitrogen.The specimens were then taken to an Automatized Mineralogical Analyzer (TIMA of Tescan, Nanolab-Redemat).

Slag attack experiments
Four typical refractory bricks were tested for slag corrosion.An octagonal arrangement made of these bricks com-posed a crucible where a load of metal was melted.After temperature stabilization, the mixture was added to the bath top and the refractory wear was measured after 30 minutes contact, at the slag line, wall, impeller (according to Figure 4).

Results and discussion
A glimpse of the mineralogical analysis results is shown at Figure 3 for the Ilmenite reagent.Labelled as Calcite is a 44.6% oxygen, 1.39% aluminum, 54.1 % calcium phase; unclassified is a phase with roughly 33% oxygen, 1.9% aluminum, 33% calcium, 28.8% titanium and 3.48% iron; peroviskite as a phase containing 33.37% oxygen, 2.05% aluminum, 27.9% calcium, 30.87% titanium and 5.81% iron.All other phases are below the 0.5% level.

Figure 3
Ilmenite reagent after heat treatment at 1400 o C during 30 minutes according to TIMA scanning microscopy analysis.
The objective was to assess the formation of liquid phases and it can be seen that most of this reagent is untouched lime.The same was found for the other mixtures.This result was anticipated by module PHLAMECH of CEQCSI software (an ArcelorMittal's in-house software); accordingly a dry slag would be produced with exception being made for the fluorspar containing mixture, with 6.87% liquid phase.As a suggestion, the de-S process at Kanbara reactor can be deemed as kinetically controlled, since calcium oxide (with its characteristically high sulfur capacity) is the dominant phase and less than 10% of calcium oxide is converted to calcium sulfide in the process.
Accordingly CEQCSI software forecasts a remarkable degree of de-S (metal sulfur levels at equilibrium below 2 ppm).This software seeks the minimum of the Gibbs free energy of a chemical system under the mass balance constraints.The Gibbs free energy of the system is the sum of the Gibbs free energy of the constituting phases.Therefore CEQCSI employs some thermodynamic models to describe the different phases: steel, slag, gas, solids, etc.For slag, it uses the Generalized Central Atom model (GCA), which is based on the quasichemical approach.Under equilibria Sulfur and other solutes are distributed within the phases in such way to equalize their chemical potential.Average values for metal composition and degree of desulfurization are shown at Table 9 and Figure 5. Data for fluorspar are averages of three runs.All mixtures can lead to Sulfur removal, as predicted by thermodynamic calculations and mineralogical analysis.Ulexite, one of the less corrosive options seems highly feasible as a hot metal desulfurization reagent additive.Desulfurization with ulexite can be even better than with standard fluorspar mixtures, following the same trend presented by KEUM et al (2007).The sodalite mixture is the worst by comparison.This contrast between fluorspar and sodalite was not observed by VOLKMAN.However, it is important to mention that VOLKMAN's work was done for steel (metallurgy ladle process) with a fluorspar replacement ratio of 1:1, while in the present work de-S was studied for hotmetal.Here, the replacement ratio was almost the same, 6:7 (fluorspar:sodalite).The others, fluorspar, ilmenite and aluminum dross mixtures show similar behavior.Results from PEZZIN (2016)  Table 9 Evolution of carbon and sulfur contents during de-S experiments.
Figure 5 Desulfurization results for some specific mixtures.
Refractory wear by slag attack for components at critical spots (impact region of ladle and torpedo car, ladle slag line, wall and KR's impeller) is shown in Figure4.As can be seen, fluorspar potentially brings about the highest level of slag corrosion.Aluminum dross mixture can be especially corrosive to wall refractory and impeller.Sodalite, Ilmenite and Ulexite mixtures seem promising since the slag attack is comparatively lower.The results regarding fluorspar and sodalite are in accordance with those fromVOLKMANN (2002), although his experiments showed a difference not so pronounced as in the present study.

Figure 4
Figure 4Refractory expected wear by slag attack.
conducted de-S experiments under laboratory conditions in order to assess the performance of some lime mixtures, Table1.B 2 O 3 was suggested as a possible replacement for fluorspar.
are not directly comparable since the mixtures components are different.