SEARCH

SEARCH BY CITATION

Abstract

  1. Top of page
  2. Abstract
  3. 1 Introduction
  4. 2 Experimental
  5. 3 Results
  6. 4 Discussion
  7. 5 Conclusions
  8. 6 Acknowledgment
  9. References

The desulfurization abilities of some commonly used agents, namely fluidized CaO, CaC2, commercial-CaO, Mg, MgO, CaO · MgO, and mixtures of commercial-CaO–Mg were studied and compared under the same experimental conditions in a laboratory furnace at 1773 K. The desulfurization mechanisms of CaO · MgO, commercial-CaO, and mixtures of commercial-CaO and Mg were also studied. While fluidized CaO showed the best performance, commercial-CaO mixed with 20 mass% Mg achieved the second best desulfurization. Mg-granules performed slightly better than CaC2 and commercial-CaO, but somewhat less satisfactory compared to fluidized CaO and commercial-CaO–Mg mixtures. Since only the CaO portion in CaO · MgO functioned to take up sulfur, additional 70% mass had to be added to achieve the same desulfurization level. The poor ability of commercial-CaO in comparison to fluidized CaO powder was due to both its less reactive surface and agglomeration of the particles.


1 Introduction

  1. Top of page
  2. Abstract
  3. 1 Introduction
  4. 2 Experimental
  5. 3 Results
  6. 4 Discussion
  7. 5 Conclusions
  8. 6 Acknowledgment
  9. References

In view of the importance of hot metal treatment, a number of studies on the industrial performance of different agents in hot metal desulfurization have been carried out.[1-7] Nölle et al.[1] conducted plant trials to examine the performance of lime, lime–soda mixtures, Mg-coke, and calcium carbide hot metal desufurization. Plant scale hot metal desulfurization using lime was compared with lime–Al mixtures by Leclercq et al.[2] Dyudkin et al.[3] presented desulfurization data using magnesium bearing cored wire. Results of using Mg granules for hot metal desulfurization were presented by Haimi.[4] Brazzoduro showed that hot metal desulfurization with Mg could be improved by simultaneous additions of another agent such as lime, hydrated lime, or calcium fluoride.[5] The performance of lime–magnesium mixtures was studied by Visser and Boom.[6] Laboratory scale tests were compared to industrial trials with calcinated dolomite by Nakai et al.[7]

The results of the above-mentioned studies have provided valuable information to the steel industries. On the other hand, the differences in experimental conditions have led to unavoidable contradictions in the findings. The differences include kinetic conditions in the reactors, experimental temperature, top slag, treatment time, quantity of desulfurization agents, and more importantly the oxygen potential in the system. A reliable comparison of the abilities of the different desulfurization agents would require the experiment being carried out under identical and well-controlled experimental conditions.

In order to provide the steel industry reliable information of the abilities of the different desulfurization agents, a sound theoretical explanation of the different performances of the agents based on mechanism study is essential. In this connection, the desulfurization mechanisms of lime, calcium carbide, and magnesium have recently been investigated.[8, 9] As a continuation of these previous works, the present study aims at (i) an experimental study to compare the abilities of currently employed desulfurization agents under identical and well-controlled conditions, (ii) mechanism study to understand the behaviors of Dolime (calcinated dolomite), commercial-CaO, and lime–Mg mixtures.

2 Experimental

  1. Top of page
  2. Abstract
  3. 1 Introduction
  4. 2 Experimental
  5. 3 Results
  6. 4 Discussion
  7. 5 Conclusions
  8. 6 Acknowledgment
  9. References

2.1 Materials

Fluidized grinded quick lime with the product name Flucal was used. The powder was delivered in small vacuum sealed packages. The particles size was found to be in the range of 50–100 µm, as revealed by light optical microscope examination. The surface of the CaO particles is treated by silicon. The treatment makes the surface more reactive and the same time having less tendency of agglomeration. To protect the interest of the producer (Lhoist), the production of Flucal and its exact chemical composition are not provided in this paper. Note that Flucal powder is usually packed in closed container and used when it is fresh in steel industry. On the other hand, ordinary lime is very often used after storage of a certain period of time, though care is usually taken to avoid contamination of moisture during storage. To gain an understanding of the difference of the two types of lime in desulfurization, commercial lime of reagent grade from Alfa Aesar was also used. It was calcinated at 1173 K and then stored in a desiccator before the experiment. Samples of this type are referred to commercial-CaO to distinguish them from Flucal lime. Examination of light optical microscope showed that most particles were below 300 µm, while some agglomerates were observed in the powder after storage.

Calcium carbide was obtained from SKW metallurgie group and 80% of the particles were smaller than 63 µm according to the manufacturer. It was stored in tightly sealed cans and kept in a desiccator. Calcinated dolomite (CaO · MgO) with the product name Dolime (supplied by Lhoist) was employed. The size ranges of the particles were 100–300 µm, 30–200 µm, and <30 µm, respectively. MgO powders in two size ranges (<57 and <10 µm) were also supplied by Lhoist. The Mg-granules used were sized below 0.5 mm. Pig iron with a sulfur concentration of 460 ± 10 ppm was supplied by Uvån Hagfors Teknologi AB (UHT). The composition of the pig iron is found in Table 1. All graphite parts (crucible, stirrer, and holder) were made from high purity grade graphite IG 110.

Table 1. Chemical composition of the pig iron
C %Si %Mn %P %S %Cr %Ni %Mo %
3.90.210.190.0470.0460.390.100.011
W %Co %V %Ti %Nb %Cu %Sn %Al %
0.0010.0130.20.0210.0010.0080.0010.12

2.2 Setup and Procedure

The experimental setup has been described in detail in previous publications.[8, 9] Figure 1 presents schematically the setup. It enables:

  1. Accurate control of the sample temperature (±3 K).
  2. Control of oxygen potential (in the present case, CO-gas and pure graphite crucible components are used, which results in an oxygen partial pressure: inline image = 4 × 10−16 atm.)
  3. Addition of the desulfurization agent, originally kept in the quenching chamber.
  4. Mechanical stirring of the bath.
  5. Rapid quenching of the sample by lifting it to the quenching chamber and quenching it using Ar gas stream of high flow rate.
image

Figure 1. The experimental setup.

Download figure to PowerPoint

The following procedure was used for all desulfurization experiments:

  1. Heating up the hot metal held in graphite crucible to 1773 K in a controlled oxygen atmosphere (inline image = 4 × 10−16 atm).
  2. Fast insertion of the desulfurization agent, which is wrapped with thin cellophane film to the tip of the graphite impeller (type B arrangement in Figure 1 is used.)
  3. Stirring the hot metal sample at 400 rpm for 5 min.
  4. Lifting out of stirrer from hot metal, followed by 5 min of waiting time for particle floatation.
  5. Lifting the sample rapidly to the quenching chamber (in less than 1 s) and quenching the sample with Ar-gas stream of high flow rate.

The quenched sample was sent for sulfur analysis using LECO (ASTM E1019).

To study the mechanisms of desulfurization of commercial-CaO and Dolime, some additional experiments were conducted. In these experiments, only a few particles of the agent were kept in a cavity at the tip of the impeller (see type A arrangement in Figure 1). The details of the experimental procedure can also be found in the previous publication.[8] Since only a few particles were used, the mass transfer in the liquid metal would not be the controlling step (very little dissolved sulfur was consumed by the particles). Hence, a constant sulfur concentration in the hot metal around the particles could be assumed; and the diffusion of reactants through the product layer would be the rate-determining step. At a predetermined time after addition, the sample was quenched by Ar gas in the quenching chamber. The reacted particles were found in the cross-section, which was carefully prepared across the cavity of the graphite impeller. The cross-section containing the reacted particles was prepared for SEM analysis. A scanning electron microscope (Hitachi S-3700) equipped with EDS was employed. A light optical microscope of (Leica) was also used.

3 Results

  1. Top of page
  2. Abstract
  3. 1 Introduction
  4. 2 Experimental
  5. 3 Results
  6. 4 Discussion
  7. 5 Conclusions
  8. 6 Acknowledgment
  9. References

3.1 Desulfurization

Different amounts of desulfurization agents were added into the hot metal to study their performances in desulfurization. The added amount of agent is expressed as:

  • display math(1)

where nS in Equation (1) stands for the initial moles of sulfur in the hot metal whereas ni represents the added moles of agent i. The subscript, i, in ni denotes CaO, CaC2, Mg, MgO, CaO · MgO, or CaO–Mg mixtures.

Figure 2 shows the results of desulfurization of different materials. All the experiments were conducted under identical experimental conditions. (i) The experimental temperature was 1773 K. (ii) The oxygen partial pressure: inline image = 4 × 10−16 atm. (iii) The hot metal bath was stirred at a speed of 400 rpm for 5 min after addition. (iv) The floatation time after stirring was 5 min.

image

Figure 2. Desulfurization of different agents as function of the amount of addition.

Download figure to PowerPoint

The materials added and the amounts of additions as well the particle sizes are listed in Table 2 along with the analyzed sulfur concentrations after experiments.

Table 2. Sulfur concentration after desulfurization, added quantities of desulfurization agents and particle sizes
Desulfurization agentinline image after [ppm]inline image [moles agent/moles S]Particle size [μm]
Flucal2001.0050–100
Flucal403.5150–100
Flucal206.9950–100
Flucal207.0050–100
Commercial-CaO2402.44<300
Commercial-CaO2003.50<300
Commercial-CaO1605.00<300
Commercial-CaO1306.99<300
CaC2 powder2403.06<63
CaC2 powder2303.51<63
CaC2 powder3301.01<63
CaC2 powder1307.01<63
MgO3803.50<57
MgO3803.50<10
Dolime2603.50100–300
Dolime1803.5130–200
Dolime2003.50<30
Mg-granules3302.32<500
Mg-granules1204.82<500
Mg-granules1107.00<500
Commercial-CaO–10 mass% Mg1903.96<300
Commercial-CaO–20 mass% Mg804.42<300

Figure 2 shows evidently that an increasing amount of addition leads to lower sulfur concentration for all agents. It is clear that the use of Flucal results in the best desulfurization over the whole range of quantities added. For example, an end sulfur concentration of 20 ppm is reached when the Flucal addition is inline image.

Figure 2 indicates that the ability of desulfurization of commercial-CaO powder is considerably less. The end sulfur concentrations are much higher than the ones obtained using Flucal. For example, the addition of commercial-CaO of inline image leads to the end sulfur content of 130 ppm.

It is interesting to see in the figure that the performance of CaC2 (particles size <63 µm) is very similar as the commercial-CaO, over the whole range of added amounts of agents.

The experiments using Dolime were carried out for three different particle size ranges (see Table 2). The amounts of addition were all inline image. The addition having the largest particles (100–300 µm) leads to quite poor desulfurization (260 ppm). The Dolime powders having size 30–200 and <30 µm perform similar. Figure 2 reveals also that the desulfurization performances of the Dolime powder in these two size ranges are comparable with the commercial-CaO (particles size <300 µm) and calcium carbide (particles size <63 µm). All of these samples reached end sulfur concentrations in the range of 180–240 ppm.

When MgO is used, only a slight change in the sulfur level is seen. Addition of MgO of inline image only lowers the sulfur concentration to 380 ppm. As shown in Table 2, the use of very small particles (<10 µm) does not improve the performance of MgO.

The addition of Mg to hot metal resulted in splashing, bubbling sound, and vibrations in the push rod during the first 2 s after addition. These are due to the escape of Mg gas, which has been reported in a previous work.[8] This study[8] reveals also that only a small amount of dissolved Mg remains in the hot metal. The addition of Mg-granules (inline image) leads to an end sulfur concentration of 120 ppm, lower than the case of calcium carbide as well as commercial-CaO, but still higher than the case of Flucal powder.

When a small fraction of the commercial-CaO is replaced by Mg-granules, a significant reduction of the end sulfur concentration can be achieved. When commercial-CaO-20 mass% Mg (inline image) are used, the end sulfur concentration reached 80 ppm. Note that the total mass of the additions in the case of commercial-CaO–Mg additions are all the same as the mass of inline image. It seems that mixing a certain amount of Mg granules with the commercial-CaO make the lime powder more efficient.

3.2 Growth of Reaction Products

In order to understand the difference in the performance of desulfurization, an understanding of the mechanisms of desulfurization using different agents is essential. The mechanisms of desulfurization of Flucal and CaC2 have been reported in another paper.[9] Complementary experiments to study the growth of reaction products around single particles of Dolime and commercial-CaO were carried out to help to understand the different desulfurization abilities.

3.2.1 Dolime

Figure 3 presents the element mappings of a reacted Dolime particle. The size of the particle is about 100 µm. The element mappings indicate that the particle has fully transformed into CaS, after being kept in the hot metal for 5 min. On the other hand, MgO in the particle still remains as oxide without reacting with sulfur. No appreciable MgS is detected. It should also be mentioned that no calcium silicate is found in the particle.

image

Figure 3. Element mappings of Dolime particle fully transformed to CaS with MgO remaining within after 5 min reaction time.

Download figure to PowerPoint

Unfortunately, no reliable measurement of CaS-layer thickness could be presented. The unreacted Dolime is very porous. Measurement of the reaction product layer thickness requires identifying the interface between reaction product and hot metal as well as the interface between reaction product and unreacted agent. Attempt to identify this interface failed due to the difficulties in polishing the sample (the metal substrate is very hard, while the reacted Dolime is too soft).

3.2.2 Commercial-CaO

Figure 4 presents the element mappings of a particle of Flucal after 30 s of reaction time. In comparison, the mappings of a commercial-CaO particle after 30 s of reaction in hot metal are presented in Figure 5.

image

Figure 4. Element mappings showing CaS growth on Flucal particle after 30 s reaction time.

Download figure to PowerPoint

image

Figure 5. Element mappings showing CaS growth on Commercial-CaO particle after 30 s reaction time.

Download figure to PowerPoint

The CaS-growth on Flucal particles is faster than that on particles of commercial-CaO. This aspect is evidently brought forth by the thicknesses of CaS revealed by Figure 4 and 5. While Figure 4 shows a CaS layer having thickness of about 30 µm around a Flucal particle (sized about 100 µm), the layer of CaS on the particle (about 50 µm in size) of commercial-CaO is only about 10–15 µm (Figure 5).

It should be mentioned that no calcium silicate is formed on the single CaO particles in the case of both Flucal and commercial-CaO (when agglomerates are absence). On the other hand, calcium silicate phase is detected in agglomerates. An example of this case is shown in Figure 6. A calcium silicate layer is evidently seen between the CaS layer and the unreacted CaO agglomerate sized about 600 µm. Agglomerates are frequently found in the commercial-CaO, in contrast to the Flucal powder.

image

Figure 6. Di-calcium silicate layer formed around agglomerate of commercial-CaO particles shown by element mapping.

Download figure to PowerPoint

3.2.3 Commercial-CaO–Mg Mixtures

Distribution of commercial-CaO particles in the hot metal is improved when a small amount of Mg is added together with CaO. Since Mg vaporizes at 1363 K, most of it quickly rushes out of the hot metal bath as gas when added at 1773 K. When an optimized amount of Mg is added together with CaO, the Mg-gas is actually helpful in distributing the CaO particles. This aspect is evidently shown by the experimental results. Figure 7 shows the SEM microphotograph of a sample quenched 20 s after the addition of commercial-CaO–20 mass% Mg mixture. It is seen that the commercial-CaO particles (sized <100 µm) spread out by Mg-gas in a cross-section of a hot metal sample. A detailed discussion will be given later in the discussion part.

image

Figure 7. SEM micrograph of commercial-CaO particles spreading out in hot metal by Mg-vapor.

Download figure to PowerPoint

4 Discussion

  1. Top of page
  2. Abstract
  3. 1 Introduction
  4. 2 Experimental
  5. 3 Results
  6. 4 Discussion
  7. 5 Conclusions
  8. 6 Acknowledgment
  9. References

4.1 Dolime

Figure 3 shows the element mappings of a Dolime particle about 100 µm reacted for 5 min. While the CaO has transformed to CaS, MgO still remain in the Dolime particle. No appreciable MgS was found.

Similar observations were reported by Nakai et al. They studied hot metal desulfurization with calcinated dolomite and Al-ash. Their microphotograph shows an outer layer of CaS about 50–100 µm covering a bulk of unreacted MgO and CaO.[7]

Figure 2 indicates that using Dolime of inline image brings the sulfur concentration from 460 to 180 ppm. The result compares very well with the results obtained by Nakai et al.[7] They employed dolomite–Al to study desulfurization in a 70 kg induction furnace at 1673 K. An addition corresponding to inline image reduced the sulfur concentration from initially 420 ppm to about 200 ppm after 10 min of stirring.

The fact that only CaO in the Dolime contributes in desulfurization is further confirmed by the experimental result of MgO. As seen in Figure 2, MgO has very little function in desulfurization. Since CaO is responsible for sulfur removal from hot metal, the inline image is calculated using the moles of CaO · MgO in Dolime, which is the same as the number of moles of CaO. Because of this reason, about 70% more mass of Dolime (in comparison with lime) has to be added to reach the same desulfurization level as with commercial-CaO or calcium carbide.

4.2 Commercial-CaO

It is well known that CaO is sensitive to calcium hydroxide formation upon interaction with water. Even though the commercial-CaO was stored in a desiccator, it is likely that the surface of CaO particles are subjected to some degree of calcium hydroxide formation and possibly other processes, e.g., adsorption of CO2 when the particles meet the humidity and carbon dioxide during the delivery. Note that this kind of contamination is also inevitable in the delivery and even storage of industrial practice.

A comparison of the thicknesses of the CaS layers on the single particles of the Flucal powder and commercial-CaO powder (Figure 4 and 5) reveals that the layer thickness (30 µm) on Flucal is almost twice of the thickness on the commercial-CaO. While Flucal particle might have superior surface properties, the change of the surface of the particles of the commercial-CaO could also be responsible for the lower growth rate of CaS layer. Obviously, the surfaces of commercial-CaO particles are less reactive. The Flucal powder would perform better than the commercial-CaO when they have the same particle size.

Figure 6 shows that the commercial-CaO particles form agglomerates. The formation of agglomerates would not only reduce the total surface area, but also leads to the formation of calcium silicate.[9] The formation of calcium silicate in agglomerates is in accordance with the observations by Oeters et al.[10] and Takahashi et al.,[11] who used CaO rods. The reason for the formation of calcium silicate in agglomerate but not in single Flucal particle is discussed in detail in another paper.[9] This discussion is briefly given below to orient the reader.

The low concentration of dissolved Ca in the metal (<1 ppm) would make the mass transfer of Ca extremely slow, and therefore ruling out practically the possibility of low local calcium activity for reaction (1) to take place

  • display math(2)

Therefore di-calcium silicate is likely formed by the following reaction.

  • display math(3)

When CaO is added as the desulfurization agent into hot metal containing high Si, the limiting factor for the formation of calcium silicate through reaction (3) would be the oxygen supply. It is well known that the oxygen level in hot metal is at ppm level, which will not be enough to generate calcium silicate. This argument is supported by the absence of calcium silicate in the case of single small Flucal particles (Figure 4 and 5) and CaC2.[9] The desulfurization reaction,

  • display math(4)

leads to the formation of oxygen. A tiny CaO particle (usually < 50 µm) would not generate sufficient oxygen locally for reaction (3) to take place, when the particles are well distributed. A particle of less than 80 µm particle is converted to CaS before the calcium silicate layer has formed. This is a direct observation in the experiments as exemplified by Figure 4. The formation of the silicate layer is due to both kinetics and thermodynamics. In the case of bigger particles or CaO agglomerates, the silicate layer is formed when the layer of CaS has become thicker (e.g., 40 µm). The increase of the local oxygen potential would result in the formation of calcium silicate through reaction (3) under the thick CaS layer (see Figure 6).

As seen in Figure 2, an amount of commercial-CaO of inline image brings down the sulfur content to 130 ppm, whereas the same amount of Flucal takes the sulfur concentration down to 20 ppm. The decrease of desulfurization ability is due to three factors: (i) the surfaces of commercial-CaO particles being less reactive (the growth of CaS becomes slow), (ii) the reduced surface area due to the formation of agglomerates, and (iii) the agglomerates leading to the formation of calcium silicate layer, which increases the resistance to mass transfer.

The better performance of Flucal compared to commercial-CaO could be because: (i) the Flucal has been fluidized, which helps to avoid agglomeration of the CaO particles and (ii) the Flucal particles (50–100 µm) has a smaller particles size compared to the commercial-CaO (<300 µm).

While maintaining good surface property of the powder is essential to maintain high reactivity of lime in desulfurization, decreasing the local oxygen potential by addition of other agents would be possible to help even commercial-CaO to improve its performance.

4.3 Lime–Mg Mixtures

One of the alternatives of additives into CaO for desulfurization is Mg granules. To gain an insight of the function of lime–Mg mixtures, it is essential to understand how the two agents work individually. The function of Mg in hot metal desulfurization has been studied and discussed in detail in a previous work.[8] The study shows that when Mg is added into hot metal, most of it rushes out as Mg-gas in less than 2 s. MgS is not formed by homogeneous nucleation but on MgO particles, which originate from the surface of the added Mg-granules. The study[8] indicates that during the Mg addition, 180 ppm of magnesium is dissolved in the sample. This corresponds to 10% of the total addition. Thus, 90% of the Mg-addition went off as gas during the initial 2 s after addition. Because of the escape of large amount of Mg vapor, the use of Mg is not efficient. The study[8] also shows the sulfur content is lowered from 460 to 360 ppm by the formation of MgS on MgO micro particles.

Irons and Guthrie[12] suggested that MgS formation in hot metal takes place at the surface of inclusions by dissolved Mg through reaction (5)

  • display math(5)

Their finding agrees well with our previous publication.[8] The experimental point shown in Figure 2 indicates that addition of inline image of Mg granules followed by stirring bring down the sulfur content to 120 ppm. As mentioned above, the sulfur content is brought down to 360 ppm after 2 s, and 180 ppm of magnesium is dissolved in the sample after this time period. It is reasonable to expect that reaction (5) continues on the surface of the solid particles in the hot metal after the initial period of 2 s.

A simple calculation reveals that the dissolved Mg quantity (180 ppm) can theoretically remove 237 ppm of sulfur according to reaction (5). Hence, the reaction (5) will bring the sulfur level down further from 360 to 123 ppm. This prediction agrees excellently well with the experimental result in Figure 2, viz. 120 ppm.

A steel industry implemented a desulfurization process of injecting 0.7 kg of magnesium granules (sized 1–1.5 µm) per ton hot metal with heated air at 1633 K. The annual average desulfurization achieved was from 650 to 200 ppm.[4] The results are somehow in line with the present findings.

Jin et al.[13] presented data from an industrial test in a Chinese steel plant using only Mg for desulfurization. The data presented were only a few. But it revealed that the sulfur concentration was lowered to about 100 ppm. Lower sulfur levels were only reached when the Mg addition was accompanied by calcium carbide. In line with Jin's results, Voronova pointed out, based on industrial data that a big increase of Mg consumption is required to reach sulfur content below 100 ppm.[14] Again, the results agree with the present work.

Dyudkin et al. used magnesium bearing wire and lowered the sulfur concentration from 510 to 400 ppm during the 3 min when the wire was inserted. The sulfur concentration then fell to 140 ppm during the subsequent 36 min of floatation under the influence of a slag in a 10-ton ladle.[3] Their results agree also very well with the present work.

Brazzoduro et al.[5] studied hot metal desulfurization by Mg injected with nitrogen gas. They noticed Mg burning off at the surface of the melt. They concluded that the hot metal was not desulfurized to the corresponding degree of theoretical calculation.[5] The escaping of Mg gas observed in the present work is confirmed by their industrial trails.

Shevchenko et al.[15] proposed the use of granular Mg (<0.1 mm) for desulfurization claiming it had superior properties compared to both Mg–CaC2 mixtures and Mg–CaO mixtures. They reported the method was capable of reducing sulfur from 1500 to 10–20 ppm in less than 20 min. Unfortunately, they did not provide any information on the experimental conditions, e.g., the amount of addition. Reading the report carefully reveals that the Mg addition was accompanied by undefined slag additions. Thus, it is difficult to rule out the function of the slag in desulfurization.

The present experimental result of Mg reveals that (i) big amount (90% in the present study) Mg escapes as gas in the initial stage of the addition; (ii) the formation of MgS on the oxide particles brings down the sulfur content from 460 to 360 ppm; (iii) 10% Mg dissolves in the hot metal; (iv) reaction (5) continues on the micro particles, thereby leading to further desulfurization.

Nölle et al.[1] did lab scale desulfurization trials with lime in three different atmospheres (methane, argon, and air) and reached the best desulfurization degree (95%) with methane, 47% using argon, and only 30% desulfurization degree with in air. Kikuchi et al.[16] used a 4-ton scale experimental setup with mechanical stirring and showed 20–70% efficiency improvement of the desulfurization agent (lime–5 mass% Flourspar) when 10–58% propane was added to the nitrogen injection gas. The improvements of desulfurization by using methane and propane are expected, since both gas decrease the oxygen potential in the melt during injection. Lower oxygen potential favors the reaction,

  • display math(3)

It is well understood that mixing Mg with CaO will have the same effect, since the following reaction (5) takes place parallel to reaction (4)

  • display math(5)

Lowering the oxygen potential does not only promote CaS formation through reaction (4), but also helps to avoid formation of di-calcium silicate around the lime particles according to reaction (3).[9]

As reported in the previous work,[8] the generation of huge amount of Mg gas stirs the metal bath violently. This study also shows that the Mg gas helps in spreading the lime particles. To help the reader, the main results of lime–Mg mixtures obtained in the previous study[8] are reproduced in Table 3.

Table 3. Distance from the surface of the metal above which reaction products are found
SampleQuenched after [s]Distance from the surface [mm]
Note
  1. The height of the hot metal is 45 mm.

Mg210–15
Mg2010–15
CaO–96% Mg210–15
CaO–96% Mg2010–15
CaO–70% Mg2015–20
CaO–50% Mg2015–20
CaO–30% Mg2035–40
CaO–20% Mg2035–40

Though the spreading effect depends very much on the reactor and addition position, Table 3 still indicates there is an optimal ratio between CaO and Mg, at which the CaO particles are distributed well in the liquid metal. As the Mg part reduced and the lime part increased, particles are found over a larger area of the sample. Under the present laboratory conditions, 20–30 mass% of Mg seems to be optimal. This optimal agrees very well with the results of desulfurization presented in Table 2 and Figure 2. When the commercial-CaO is mixed with 20 mass% Mg (inline image), the desulfurization reached 80 ppm. The small Mg addition helps the commercial-CaO to perform better, but it is still not as good as the Flucal, which reached 40 ppm at inline image. It should also be mentioned that the violent local stirring of the metal bath by Mg gas could even crush the agglomerates formed during storage. This would further enhance the sulfur pickup by the CaO particles.

The present results and reasoning is in good accordance with the industrial study by Visser and Boom.[6] They desulfurized 283 tons of hot metal at 1673 K from 250 ppm sulfur to 7 ppm in 10 min by injecting 3.1 kg ton−1 CaO–15 mass% Mg mixture (3.1 kg ton−1 steel).

Good results of desulfurization of hot metal by combining CaO and Mg are reported also by other researchers.[5, 15, 17] On the other hand, their explanation of the results is somehow different from the present reasoning.

4.4 CaC2

A study on desulfurization of CaC2 has been made in a parallel work.[9] The results reveal that Flucal CaO particles are more efficient in desulfurization in comparison with CaC2 (see Figure 2). This is because of slower growth of CaS on CaC2 particle compared to its growth on Flucal particle. The detailed discussion can be found in the paper.[9]

5 Conclusions

  1. Top of page
  2. Abstract
  3. 1 Introduction
  4. 2 Experimental
  5. 3 Results
  6. 4 Discussion
  7. 5 Conclusions
  8. 6 Acknowledgment
  9. References

The abilities of desulfurization of commonly used agents, CaO, CaC2, commercial-CaO, Mg, MgO, CaO · MgO, and mixtures of commercial-CaO–Mg were studied and compared under the same experimental conditions at 1773 K. The desulfurization mechanisms of CaO · MgO, commercial-CaO, and mixtures of commercial-CaO with Mg were also examined.

The performance of the agents was found to be in the following order: CaO (Flucal), commercial-CaO mixed with 20 mass% Mg, Mg granules, CaC2 and commercial-CaO, CaO · MgO (Dolime), MgO as the worst.

The present study shows that the surface property of CaO particles plays important role in desulfurization. Poor CaO surface makes the CaS growth on single particles slower and the particles becomes sensitive for agglomeration.

Mixing 20 mass% Mg with the commercial-CaO makes it perform better, but still not as well as Flucal. The Mg helps to improve commercial-CaO's functionality by distributing the lime particles and reducing the local oxygen potential.

The reduced oxygen potential not only enhances the desulfurization reaction, but also minimizes the formation of calcium silicate, the formation of which always hinders the mass transfer.

6 Acknowledgment

  1. Top of page
  2. Abstract
  3. 1 Introduction
  4. 2 Experimental
  5. 3 Results
  6. 4 Discussion
  7. 5 Conclusions
  8. 6 Acknowledgment
  9. References

The authors want to thank Professor Patrice Nortier and Dr. Thierry Chopin for the good discussions and valuable suggestions. The financial support from Lhoist is gratefully acknowledged.

References

  1. Top of page
  2. Abstract
  3. 1 Introduction
  4. 2 Experimental
  5. 3 Results
  6. 4 Discussion
  7. 5 Conclusions
  8. 6 Acknowledgment
  9. References