Los que se quedan Paulina, la madre niña.

Slags with carbon in one form or another is undoubtedly one of the most important reactions. In open hearth, steelmaking the interaction is apparent from the slag layer covering the

molten metallic hath.. The oxygen steelmaking processes with a metal-s:lag-gas emulsion depend, to some extent, upon the decarburization of iron b y reacting with the slag.

During the preliminary work attempts were made to deter­ mine the paths followed by the synthetic slags used during decarburization of Fe-C alloys. Although the quantity of carbon removed by the slag was small (2 to 3% from a one gram d r o p ) , a change in the chemical composition would indicate the most likely reaction taking place.

The slag path studies carried out under the present experi­ mental conditions show that a reduction of iron oxide from the slag by carbon tends to move the composition toi'rards: the lime- silica join. In practical oxygen steelmaking the same tendency is found but, due to lime additions which, are made during a given cycle, the compositions tend to move towards the CaO corner.

This effect is shown in figures:15a andl5b, which, were determined from plant operations (10,107).•

The changes in chemical composition wit h time for CaO-FeO- Si02 decarburization reactions are given graphically in figures 29 to 32. All the results clearly show a reduction of FeO and ]?e2^3 content and therefore the silica and lime contents tend to increase. Since the actual amount of silica and lime

remain constant throughout the experiments their actual percent­ ages in the slag increases. The appearance of free iron in most of the experiments also confirms the reduction of iron' oxide from the slags. The reactions taking place m a y thus he represented in the most simple way by:

*

C + (FeO) 5 .2.1

or C + 2 (FeO) 5.2 .2

The FeO may well participate in prior reactions as it acts as the oxygen donor, according to:

The first three reactions taking place at the metal/slag inter-

The free iron was found to increase with time for all slags but was greater when high, lime contents were involved. This

can be expected as it has been shown (54) that with high silica contents anc no lime (i.e. Fayalite) some FeO remains combined as silicate. On adding small quantities of l i m e ,displacement of FeO from the silicate network takes place and some FeO is set free (i.e. effectively increasing yFeO) (43,36,52,54). '

From the ternary diagram in figure 12 the liquidus iso­ therm is seen to decrease and consequently a less viscous slag is present which in turn favours a higher degree of turbulence to be generated within the slag bringing "fresh” FeO in

contact with the Fe-C droplet. On further increasing the lime content (e.g. to 26%CaO) a small increase in the amount of

free iron was obtained as compared to previous slags. These results may be accounted for when considering that the CaFeSiO^ composition (2 9.78; 38.3 and 31.92 wt% respectively) is closely approached. As found by Johnson and Muan 0>4} a pronounced negative deviation from ideality is found within the vicinity

(FeO) Fe(liq) + 0 5.2.3

followed by C _{+ 0 = CO gas} 5.2 .4

of this composition, indicating a strong tendency for the

CaFeSiO^ formation. At the highest lime contents used in the present investigation (32%CaO) a rapid increase in free iron was found. The free iron was again mainly from FeO although Fe2 0g was greatly reduced. The.rates obtained for decarburization with slags containing 32% CaO were in fact lower than those found for slags with 261 CaO. The effect may be due to appearance of dicalcium silicate precipitates. Such precipitation has been found by Trentini (.83) in

oxygen steelmaking and by ICootz (90) in open hearth steel- making. The latter states that the slag penetrates very soon into the heterogeneous zone. This leads to a reduction in the rate of decarburization. The penetration into the zone is shown in figure 15c.

With regards to surface nucleation effects on the reacted droplets, it was apparent that high lime contents may have had an effect simultaneously with carbon content. Evidence was found that only in high lime slags and at about 2.2%C contents did surface nucleation take place. The faster decarburization rates obtained with these slags may have caused depletion of carbon in the outer region, and therefore CO evolution was subsequently retarded. A lesser degree of turbulence can be expected due to this effect and detachment of CO bubbles is facilitated with less ease. The possibilities of bubble nucleation must therefore be considered as discussed in the following section.

Attempts were made to quantify the results of the slag path, analyses obtained b y reactions such as

FeO + . C ---- »• Fe + CO 5.2.5

and ^ e2^3 + ^ ^ 5.2,6

The slag path followed by one of the slags studied, tbat which, is shown in figure 29., will be discussed in detail having in mind that the slag paths given in figures 31 and 32 were similarly examined for mass balances.

The decrease of FeO and F ^ O ^ contents from the initxal slag were found to be 2.641 and 2.1% respectively after five minutes of reaction. The weight of slag used was 30 grams and therefore the Fe generated from 2,64 %FeO reduction is

calculated as: .

wt. of slag x wt % FeO x Fe ^molecular vrt0 - 30x2.64x56 100 -x FeO (m°3-ecular wt} 7200

= 0 . 616g Fe

Assuming that 0.616 g of Fe react with by reaction 5.2.6, the decrease of Fe2 0^ is calculated as:

0.616x160 n r v n

56--- = 1>76g 0f 2 3

- The weight of Fe^O^-which the slag actually lost by reaction \ S Z0 * 0«63g of Fe2 0g. These last two values was

differ by a factor of 2.8 and therefore it must he assumed that not all the free iron reacted wit h ferric oxide. Since, from chemical analysis, only 0.63g of ferric oxide were reduced, the quantity of free iron

required is calculated as wt%Fe = .0.63x56/160 = 0 . 22g of Fe. The free iron generated by ferrous oxide would then he

decreased by the above quantity (0.616 - 0.2 2) leaving 0.396g of Fe free from the slag. .. The weight of analysed free iron is less than 0.15g (i.e. 0.5%), which in turn differs from the calculated value of 0.396g by a factor of more than 2.6.

Slag paths given in figure 29, 31 and 32 were all analysed in a similar manner at '.he times corresponding to 5, 10 and 15 minutes. To facilitate comparison between the calculated values and those given in the corresponding figures in wt %, the calculated results have all been con­ verted to wt%. Figure- 30 was not used in the present calculations because the free iron present was not

analysed as described in section 3.5.5, The results are given in table III-A together w i t h the ratio of Fe analysed/ Fe calculated which is a measure of the disagreement,

It can be seen from table III-A that only in two cases does the Fe analysed/Fe calculated approach unity. The results indicate a wide disagreement and therefore a low p r o b ­

ability for a correlation would be obtained. The me thod described above clearly shows that the slag p a t h analysis

data are not sensitive as means of quantifying, through mass balances, the reaction (or extent of reaction's) which are taking place in the metal phase.

One other approach to quantify slag path analysis is to use the known carbon loss, from the iron carbon .sample during reaction, and calculate the extent to which

ferrous and ferric oxides in the slag would be reduced by this known carbon drop.

The mass of slag is 30 grams and the mass of iron- carbon is one gram. The carbon lost after 5 minutes of reaction (figure 32) wit h the slag given in figure 29 is 0.81%C. The reaction considered is 5.2.5;

FeO + C *■ Fe + CO ,

The mass of ferrous oxide reduced by 0.0081g of carbon is thus 0.0081x72/12 = 0.0486 g of ferrous oxide. The value obtained from chemical analysis is 0.792g of ferrous oxide

(or 2.64% FeO in figure 29).

The slag paths given in figure 29 and 32 (figure 31 was not used - no comparable, decarburization curve) were analysed in the above manner and the results are given in

table III-B. The calculated values in table II1-B are clearly much smaller than the analysed values. This pronounced dis­ agreement substantiates the previous disagreement between FeO, Fe2 0^ and free Fe contents from the slag path analysis and mass balances. It is quite obvious that decarburization rates within the metal drop cannot be followed by sampling and measurements of the slag phase. As explained in section 3.5.5 the errors resulting from the chemical analysis of slags_{i - >} t h e .sampling of slags,. the slag mass/metal mass ratio and the

JLI1 Lex ItJI eil^e U1 ci:? a L-UclLJLIlg, ail COIIirXQUie LO the wide range of scatter obtained.