- Email: [email protected]
A new steel slag for cement manufacture: Mineralogy and hydraulicity
CEMENT and CONCRETERESEARCH. Vol. I I , pp. 85-102, 1981. Printed in the USA. 0008-8846/81/010085-18502.00/0 Copyright (c) 1981 Pergamon Press, Ltd.
A NEW STEEL SLAG FOR CEMENT MANUFACTURE MINERALOGY AND HYDRAULICITY
:
M. Conjeaud LAFARGE, Trappes, France C.M. George LAFARGE FONDU INTERNATIONAL,
Neuilly-sur-Seine,
F.P. Sorrentino LAFARGE Laboratoire Central, Viviers,
France
France
(Communicated by B. Cottin) (Received Sept. I0, 1980)
ABSTRACT This study presents a method of developing useful hydraulic properties in Basic Oxygen Steel Slags, by the addition of a synthetic slag forming flux during the normal steel making process. The transformation of a waste slag into a by-product at least equivalent to the best quality Blast Furnace Slags is achieved without change of technology or sacrifice in steel quality. The flux contains CaO - AI203 - MgO - Fe203 in proportions calculated from phase diagrams. Theoretical predictions are confirmed by laboratory experiments and by semi-industrial trials and show that the introduction of alumina in this manner leads to the formation of hydraulic calcium-alumino-ferrites and homogeneous slags containing < 4 % free lime. The mineralogy of the slags is described and the results of calorimetry, X-ray diffraction, optical microscopy, electron microprobe analysis, and strength tests on non quenched ground slags are given to demonstrate hydraulicity. Des propri~t~s hydrauliques pourraient ~tre d~velopp~es dans les scories d'aci~rie ~ l'oxyggne, par addition, pendant le proc~dg classique d'affinage, d'un produit synth~tique facilitant la formation dulaitier. Le produit d'addition contient CaO-AI203-MgO-Fe203 (abr~gg ~ CAMFIux) en proportions calcul~es d'aprgs les diagrammes de phase. Les prgvisions th~oriques, confirm~es par les experiences de laboratoire et des essais semi-industriels montrent que l'introduction d'alumine sous cette forme conduit, sans changer la technologie de l'affinage, ni compromettre la qualit~ de l'acier, ~ la formation d'une scorie homog~ne contenant moins de 4 % de chaux libre et une alumino-ferrite de calcium hydraulique. La min~ralogie des scories a ~t~ ~tudi~e par microscopic optique, diffraction X et microsonde. Les tests de microcalorim~tre isotherme et de r~sistances effectu~s sur des scories broy~es, mais non tremp~es permettent d'~valuer leurs propri~t~s hydrauliques. 85
86
Vol. I I ,
No. 1
M. Conjeaud, et al.
Introduction 50 million tonnes per year, world wide, of LD slags are produced as byproducts of the Basic Oxygen Process (BOP). Few applications for these slags exist at present since their phosphorus content is too low for fertilisers while their heterogeneity, particularly their free lime content, and the absence of hydraulic phases severly limits their use in the construction industry. Considerable importance thus attaches to methods of valorising these slags especially as the BOP will remain the major steel making process in the forseeable future. This paper describes the successful development of a new LD slag compatible both with the BOP and the needs of the cement manufacturer. The new slag is less variable than classical slags, contains very little free lime and possesses good hydraulic properties. The paper is in three parts
:
- theoretical and experimental studies demonstrating efficient refining of iron into steel and defining the composition of the resulting slags -
structural characterisation of the slag obtained under semi-industrial conditions
- definition of the hydraulic behaviour of these slags. Part I : The BOP and flux additions In the BOP, which is a batch process, gaseous oxygen is blown into liquid Blast Furnace iron in a special converter vessel. Impurities in the iron are oxidised (C ~ C0/CO2, P ÷ P205, Si ÷ Si02, etc.) The non volatile products, including some of the iron oxidised to iron oxide, form, with an added basic material, lime, a slag, which contains the impurities separate from the liquid metal. The iron is thus purified and becomes steel to be decanted separately from the slag, at the end of the process. No fuel is needed in this process since the oxidation reactions are themselves exothermic. At the beginning of the refining process (the blow) carbon, silica and iron are preferentially oxidised. The carbon oxides escape as gas, whereas FeO and SiO 2 combine with solid lime to form the initial slag. This is only partially liquid since some of the silica forms an insoluble coating of calcium silicate on the surface of the lime, inhibiting further reaction. The effect is shown in a pseudo-ternary diagram, figure L, where the development of the slag composition can be seen to traverse the precipitation zone of solid C2 S, (2CaO.Si02). In order to prevent the formation of this passivating layer of silicate, many different fluxes such as fluorspar, borates or titanates, bauxite, etc, have been tried in the BOP. These fluxes act by lowering the viscosity of the slag or enlarging the composition zone of liquid formation. Their effectiveness is, however, compromised by a certain number of disadvantages : - increasing the acidity of the slag which, combined with greater fluidity becomes more aggressive to the basic converter refractories ; -
-
producing harmful exhaust gases (HF) or contaminating the steel (Boron) ; preventing the development of useful properties in the final slags.
Vol. I I ,
No. 1
87
(s) *~
STEEL SLAG, CEMENT, HYDRAULICITY, CHARACTERIZATION
c Tx -
(c)
(~A) FIG.
I
Pseudo-ternary phase diagram CaO-SiO2-FeO showing LD slag composition path during steel manufacture X - without alumina Y - with alumina
In addition, certain materials, particularly fluorspar (the most widely used at present) are becoming of increasingly limited availability. The present work describes the results obtained from substituting such fluxes by a prefired synthetic material (CAMFIux) ~ developed from phase diagram studies reported by J. WHITE (I, 2, 3). In a simplified system such as CaO-SiO2-FeO, in order that a slag in the course of formation should be in equilibrium with lime as the only solid phase, it is necessary that the ratio Si02/FeO should not exceed the value shown by point K in figure I. By extending this arguement to the multicomponent system representing a steel making slag, limiting values for the ratio Si02/(FeO + MgO + A1203) can be defined. This ratio is greater than the ratio Si02/FeO of the simplified system. J. WHITE's studies show that the maximum value of this ratio - representing the greatest capacity of the slag to absorb silica without forming insoluble silicates - is to be found in the system CaO, MgO, A1203, iron oxide, SiO 2. This has led us to develop a prefired material to replace classical BOP fluxes, of such composition that the slag developed in the course of refining achieves this optimum limiting value of the ratio Si02/(FeO + MgO + A1203). Such a slag should be more efficient than classical slags for absorbing P205 due to its undersaturation in SiO 2. Using this material, the final slag contains CaO, MgO, A1203, iron oxide, SiO2, P205, MnO. The phase diagram of this system has only been very partially explored. In particular, the optimum ratio of Al203/iron oxide has not been determined. The composition of the prefired synthetic material which we have developed has therefore been chosen to provide anAl203/iron oxide ratio which confers hydraulic properties to the final slag. Theoretical
study of the hydraulic potential of LD slags
Mineralogical examination shows that classical LD slags contain ~-calcium orthosilicate (2CaO.SiO2, stabilised by dissolved phosphoric oxide), ferrites, typically 2CaO.Fe203, calcium phosphate 3CaO.P205, wustite, FeO and solid solutions of the (Ca, Fe, Mn, Mg)O type, and since they contain only contains llme (C), alumina name CAMFIux.
(A), magnesia
(M), and iron oxide (F). Hence the
88
Vol. I I ,
No. l
M. Conjeaud, et a l . negligible quantities of vitreous material, it is possible to calculate the potential mineralogical composition from a knowledge of the easily analysed chemical composition. In a typical case we find : Chemical composition (by analysis) CaO MgO A1203 Fe oxides SiO 2 P205 MnO
%
Mineralogical composition % (by calculation) 2CaO.SiO2-B 43 3CaO.P205 4 2CaO.Fe203 ]7 (Ca. Fe.Mn.Mg)O 31 Free CaO 5
50 6 0 22 ]5 2 5
The free lime value has been chosen arbitrarily near the lower limit found industrially. Such slags show very little hydraulic activity (5) as their potential mineralogical composition indeed suggests : B.2CaO.SiO 2 hydrates very slowly and calcium ferrite, 2CaO.Fe203 is inert as are the other phases. However, by incorporating alumina in the slag the calcium ferrite can be transformed into calcium alumino-ferrite which possesses substantial hydraulic activity. We find that this activity increases with the AI203/Fe203 ratio. In figure 2 we show the degree of hydration of calcium-alumino-ferrite, measured by X-ray diffraction, as a function of the AI203/Fe203 ratio : the results confirm those reported by Carlson, (6). In order to have a useful level of hydraulic activity - represented by a relatively rapid development of mechanical properties, - an AI203 content corresponding to the formation of 4CaO.AI203.Fe203 is desirable. Referred to a typical LD slag as described above, this implies an AI203 content of 6 to ]0 % from which the mineralogical composition can be calculated to be : 2CaO.SiO2.B 3CaO.P205 4CaO.AI203.Fe203 (Ca.Fe.Mn.Mg)O %RESIDUAL
ANHYDROUS
43 4 29 23
% % % %
MATERIAL
C6A2 F AF
FIG. 2 Rate of Hydration of calcium-alumino-ferrites as a function of alumina content
5(;
C2F 100
1 10mns
l "'' 21=rs
i 1day
, TIME lodays " -
Vol. I I ,
No. 1
89 STEEL SLAG, CEMENT, HYDRAULICITY, CHARACTERIZATION
This primarily theoretical analysis thus leads to the conclusion that useful hydraulic behavlour may be expected of an LD slag by the introduction of 6 % or more AI203. Our initial laboratory study thus aimed at verifying this conclusion by a study of synthetic slags of varied AI203 content. Laboratory study of the hydraulicity of LD slags containing Alumina Two series of slags were studied, obtained by adding 5 - I 0 - 1 5 - 20 and 30 % AI203 to two industrial LD slags. This approach was considered more practical than the study of entirely synthetic materials, particularly as concerns the nature of the solid solutions present, (Ca.Fe.Mn.Mg)O and the 2CaO.SiO 2 and 3CaO.P205 phases. The compositions of each series are shown in tables I and 2. (LD I and LD 2 are the original slags, LD 105 contains approximately 5 % A1203, etc.). The slag-Al203 mixtures were melted between 1450 and 1550°C in a reducing atmosphere of C0/C02, calculated, for each temperature to produce an FeO content of approximately 10 %, and quenched. -
For AI203 contents below 15 % the phases detected were 2CaO.SiO2. ~ (Ca.Fe.Mn.Mg)O solid solution ferrlte in the composition range 2CaO.Fe203 to 2CaO.AI203 negligible vitreous material despite quenching.
:
The alumina is thus combined in the ferrlte phase as anticipated. TABLE I Chemical composition (%) of laboratory LD slags with alumina additions
CaO
SiO 2
FeO
Fe203
AI203
TiO 2
45.6
12.3
]3.0
]2.1
].0
0.7
8.5
1.7
1.2
2.1
LD ]05 44.8
12.1
8.8
I].9
5.9
0.7
8.4
].7
1.2
2.0
LD |10 41.9
11.4
10.6
]0.2
11.2
0.6
7.8
1.6
1.1
1.9
LD 115 40.0
10.7
I1.1
7.6
16.3
0.6
7.4
1.5
1.0
LD 120 3 7 . 7
10.2
10. I
7.1
21.5
0.6
7.0
1.5
1.0 ]
LD I
MgO
P205
MnO
Loss
l
I
1.8
1.7
TABLE 2 Chemical composition (%) of laboratory LD slags with alumina additions
CaO
SiO 2
FeO
AI203
TiO 2
47.1
12.4
13.3
11
1.2
0.5
4.9
2.0
5.4
1.1
LD 205 45.8
12.6
8.7
14.2
6.0
0.5
4.7
2.0
5.2
0.3
LD 210 4 2 . 3
11.2
6.1
]7.6
11.1
0.5
4.4
1.8
4.9
1.0
LD 215 40.5
10.6
lO.l
10.9
16.2
0.4
) 4.2
1.7
4.6
0.9
LD 230 32.9
8.7
II.2
6.6
30.8
0.4
3.4
1.4
3.8
].0
LD 2
Fe203
MgO
P205
MnO
Loss
90
Vol. I I ,
No. 1
M. Conjeaud, et a l .
Above 15 % alumina there is extensive vitrification and only glass and (Ca.Fe.Mn.Mg)O solid solutions were found. Owing to the small quantities of slag obtained under these laboratory conditions, we chose to study their hydraulicity by isothermal calorimetry, a technique widely used in the study of Portland cements. Furthermore, whereas normal cements generate considerable heat of hydration within 24 hours, the slags do not and for this reason their hydration was accelerated by treating them with solutions of caustic soda, (N, 2N and 5N) rather than pure water. The quenched slags, ground to less laser - table 3) were positioned in 500 mg ve the caustic soda solution. Once thermal ned, each sample was precipitated into the time registered at constant temperature.
than 100 ~ granulometry measured by quantities in the calorimeter, aboequilibrium at 21°C had been obtaisolution and the heat evolved with
TABLE 3 Granulometry and heat evolution of ground, alumina doped LD slags
Percentage less than 2 ~
4 B
8 B
LD 1
18
31
46
63
78
95
7.9
LD 105
12
20
32
51
72
96
15.2
LD 110
8
17
30
50
70
93
17.3
LD 115
6
II
21
38
60
88
22.9
LD 120
16
29
47
70
89
100
LD 2
21
34
49
65
80
96
16.5
6
16
20
30
55
80
17.6
LD 2 0 5
16 ~ 32
AH total, Kcal g-l with a 5 N 64 I~ NaOH solution
35
LD 2 1 0
ll
22
37
56
74
91
15.6
LD 2 1 5
14
28
46
65
81
95
25
LD 2 3 0
ll
23
39
61
79
95
32
[
I
The heat evolved in contact with water (or in this case, more rapidly in contact with a caustic solution) depends on the quantity and nature of the hydrates formed. Since these hydrates are the binding agents which give rise to cohesive forces it is thus reasonable to expect a correlation between the heat evolved and the mechanical strengths concomitantly developed. However, morphology and packing factors also influence mechanical properties significantly without necessarily being reflected in heat evolution. It will therefore be understood that our measurement of the compressive strength of the hydrated slags provide a useful but only partial basis for the interpretation of their behaviour measured in the calorimeter. Figure 3 relates the compressive strength observed on neat slag 5 N NaOH samples at 24 hours to the heat evolved under equivalent conditions. We have concluded, on the basis of these results, that the heat evol-
Vol. I I , No. 1
91 STEEL SLAG, CEMENT, HYDRAULICITY, CHARACTERIZATION A H(Kcal/g)
t
40 ÷
R C (MPa)
10
.AH(Kcal/g )
,0 p
2p
30 ~AI203...~
FIG. 4
FIG. 3 Relationship between compressive strength and heat evolution of alumina doped LD slag pastes
I
Relationship between hear evolution and alumina content of laboratory LD slags
ved at 24 hours must be at least 20 kcal/g if the slag is to display useful hydraulic behaviour. Table 3 and figure 4 show that an alumina content of about 15 % provided such a level of heat evolution and this result is to be compared with the minimum theoretical value of 6 % calculated previously. From the preliminary studies the following conclusions may therefore be drawn : - the addition of AI203 to classical LD slags enables hydraulic activity to be obtained ; - the minimum quantity of AI203 necessary for this hydraulic activity to give useful cementitious propertzes is of the order of 6 to 15 % by weight of the slag. Part II : Verification of the hydraulic behaviour of aluminous LD slags obtained semi-industrially Preparation The next step in our investigations was to prepare sufficient quantities of our synthetic slagging agent (CAMFIux) to enable the results of the laboratory study to be tested under more realistic conditions. Two different flux compositions were prepared in a gas-fired intermittent kiln. The compositions are shown in table 4. The use of natural raw materials occasioned certain impurities, notably silica. Slags were then obtained by refining 5 tonne batches of hot metal to steel in a pilot converter at I.R.S.I.D. (~) research station, the CAMFIux being added with lime to give the slag forming charge. (~) Institut de Recherches de la Sid~rurgie FranGaise.
92
Vol. I I ,
No. l
M. Conjeaud, et a l .
Six t e s t s w e r e c a r r i e d out. The f i r s t e m p l o y e d a c h a r g e of lime only, the s e c o n d h a d I0 % of f l u o r s p a r m i x e d w i t h the lime. In the o t h e r four tests, two i n v o l v e d a m i x t u r e of lime w i t h C A M F I u x 1 and the o t h e r two a m i x t u r e of l i m e w i t h C A M F I u x 2. The
chemical
composition
of e a c h
charge
is g i v e n
in T a b l e
5.
TABLE 4 Chemical compositions of CAMFIux used in this study CAMFIux I CAMFIux 2 31,.9 49
Ca5
%
A1203
%
29
MgO
%
12.3
9.1
Fe203
%
18.9
13.8
SiO 2
%
5.6
4.7
22.4
TABLE 5 Charge compositions used for semi-industrial preparation of LD slags Test N °
3
4
5
6
66.9
65.7
66.8
67.0
14.1
14.6
14.6
14.5
%
5.9
6.2
5.9
5.9
Fe203 %
8.9
9.5
8.9
8.9
2.7
2.8
3.1
3.0
F2A
F2B
CaO
%
I
2
100
90
AI203 % MgO
SiO 2
%
CaF 2
%
0
I0 TABLE 6 Compositions of semi-industrial LD slags
Slag :
TI
T2
FIA
FIB
10.25
10.7
43.7
43.4
45.8
40.6
42.65
0.95
1.3
10.2
11.4
6.5
9.15
MgO
10.45
6.25
5.8
7.2
14.5
10.85
MnO
4.36
4.99
3.10
3.82
3.91
P205
2.40
3.95
2.15
2.25
2.5
1.8
sio 2 CaO A1203
9.55 44.4
10.75
7.75
9.0
3.91
Fe ++
19.7
18.35
17.9
12.6
16.25
14.75
Fe +++
8.6
10.7
10.15
8.7
4.8
6.15
Fe++/Fe+++
2.3
1.7
1.75
1 .45
3.4
2.4
Free CaO TiO 2 TOTAL From test N °
10.8 0.21 100.6
3.3
4.2
3.8
4.55
1.8
0.19
0.54
0.60
0.39
0.53
100.18
I00.99
I01.37
99.6
100.49
Vol. I I ,
No. 1
93 STEEL SLAG, CEMENT, HYDRAULICITY, CHARACTERIZATION
The composition of the hot metal was fixed to reflect typical blast furnace quality and the refining operation was conducted to produce acceptable quality steel. At the end of each blow the slags were decanted into normal ladles and allowed to cool freely in contact with air. The compositions of the six slags obtained are shown in table 6. Mineralogical study The interpretation of the mineralogical analysis (optical microscopy on polished sections and X-ray diffraction on powder samples) is difficult due to the complexity of the system. We attempted once again to calculate the potential composition with the results shown in table 7, assuming the calcium alumino ferrite phase to have the composition 4CaO.AI203.Fe203. This assumption leads to the implication that 3CaO.AI203 will also be present, which would not necessarily be the case had we adopted the formulation 6CaO.2AI203. Fe203. However, the quantity of 3CaO.AI203 predicted is relatively small and would be difficult to detect by the analytical techniques. Accordingly it appeared necessary to determine the composition of the calcium alumino-ferrite phases as accurately as possible. For this purpose, electron micro-probe analysis was used. TABLE 7 Calculated mineralogical composition of semi-industrial LD slags Slag : C2S - C3P % Ferrite
%
(Ca-Fe-Mn-Mg) 0 % C3A
TI
T2
31 - 6
32 - 9
16
19
47
40
%
i F]A
FI B
F2A
F2B
22 - 5
27 - 5
30 - 5
32 - 4
32
27
f5
]9
24
26
42
34
17
15
8
If
II
40
30
40
30
C = CaO F = calcium-alumino-ferrite M = MgO mw= magnesio w~stite b = ~-C2S
FIG. 5 X-ray diffraction diagrams of semiindustrial LD slags (TI,T2,FIAF2A)
94
Vol. I I ,
No. 1
M. Conjeaud, et a l . The most extensive slags TI, T2, FIA and F2A.
study was confined
The X-ray diffraction obtained with the micro-probe
~-~1
to four samples
diagrams are given in figure in table 8.
representing
5 and the results
contained:
a) Periclase grains of variable composition, - centre, 1 % CaO, 4 % MnO, 20 % FeO surface, 8 % CaO, 1 1 % MnO, 62 % FeO and containing small quantities of SiO 2 in some cases. -
b) Magnesio wustite with an MgO/FeO ratio varying from 0.18 to 0.51 and contai ning MnO and CaO in quantities from 8 to 36 % and from 4 to 16 % respectively. TABLE 8 Chemical composition of semi-industrial LD slags analysed by electron microprobe Sample TI Pure lime charge
Phase analysis Periclase (centre of grains) (edge of grains Magnesio wustite
Lime Dicalcium silicate Alumino ferrite T2
Magnesio wustite
Dicalcium silicate Lime + Alumino ferrite fluorspar Calcium phosphate charge
Average molar composition +÷ Mg0.84 Ca0.0; Mno.02 Fe
Magnesio wustite
Dicalcium silicate Lime + CAMFIux 1 Alumino ferrite charge F2A
Magnesio wustite
Wustite Lime + Lime CAMFIux 2 Dicalcium silicate charge Alumino ferrite
0.]20
O 0 0 0
Ca2 Si0.91P0.1Fe++0.01 04 +++ Ca2. 2 Si0.03 AI0. 4 Fe
1.4 05 ++ Mg0.35 Ca0.06 Mn0.]; Fe 0.48 0 Ca2 Si0.84 P0.12 AIo.40 +++ Ca2. 3 AI0. 3 Fe ].5 05 Ca3 P].8 Sio.2 08
Tricalcium silicate Ca2.5 Sio.9 P0.09 Fe FIA
++
Mg0.56 Ca0.06 Nn0.04 Fe 0.27 ++ Mg0.12 Ca0.14 Nn0.11Fe ++0.63 Mg0.21 Ca0.09 Nn0.12 Fe 0.57 ++ Ca0o83 Mg0.08 Nn0.04 Fe 0.05
++
0.07 05 ++ Mg0.47 Ca0.05 Mn0.07 Fe 0.43 O ++ Ca2 Si0.89 P0.1AI0.02 Fe 0.010
Cal. 8 AIi.32 Feo.79 05 (Ca].8 Fe++0.2) AII.32 Fe +++ 0.59 05 ++ Mgo.41 Ca0.05 Mn0.08 Fe 0.46 O FeO Ca0. 9 Feo.05 Mn0.05 O
Ca2 Si0.86 P0.10 A10.10 O Cal.86 All.34 Fe0.75 05 (Cai.86 Fe++0.14) A11.34 Fe
+++ 0.61 05
Vol. I I , No. 1
95 STEEL SLAG, CEMENT, HYDRAULICITY, CHARACTERIZATION
c) Lime in large grains several tens of microns across containing FeO (~ 6 %), MnO (~ 4 %) in the apparently homogeneous areas and more FeO and less CaO in apparently heterogeneous regions. d) Dicalcium silicate with 4 to 8 % dissolved P205 . e) Calcium alumino-ferrite, rich in iron and containing practically all the alumina (~ 1%) present in the slag. A photomicrograph
of this sample is shown in figure 6.
O
Magnitude
~c2s
600
FIG. 6 Photomicrograph
of semi-industrial
slag T 1 (no alumina)
a) Magnesio wustite of relatively constant composition - ratio MgO/FeO between 0.53 and 0.73. b) Dicalcium silicate with a more variable P205 content c) Calcium alumino-ferrite
containing
(3 to 16 %).
less alumina than in sample T I .
d) Tricalcium phosphate. e) Some tricalcium silicate containing P205 .
S_am_~e_FlA contained : a) Magnesio wustite
(MgO/FeO = I).
b) Dicalcium silicate with about 8 % dissolved P205 and about c) Calcium alumino-ferrite
1.2 % Fe203.
with a composition close to 6CaO.2AI203.Fe203.
d) No other phases were detected and this sample was considerably more homogeneous than the preeceding ones. SamE~e_F2AContained a) Magnesio wustite
: (MgO/FeO = I).
b) Dicalcium silicate with about 8 % dissolved P205 and 2.5 % AI203.
96
Vol. I I ,
No. 1
M. Conjeaud, et al. c) Calcium alumino-ferrite with a composition close to 6CaO.2AI203oFe203. d) Pure wustite. e) Small quantities of lime containing Mn and Fe in solid solution. Thus five phases were detected in this sample, of which a photomicrograph is shown in figure 7. MW = magnesiowustite
AF = aluminoferrite
Magnitude
3 0 0
FIG. 7 Photomicrograph of semi-industrial
slag F2A (6.5 % alumina)
The formulae given in table 8 have been calculated for an integer number of oxygen atoms. Of principal interest here are the compositions of the alumino-ferrite phases, the potential seat of hydraulic activity. In samples FIA and F2A one notes a deficiency in Ca and an excess of Fe with respect to the formula 6CaO.2AI203.Fe203. Since not all the iron will be present in the trivalent state we have attempted to group a proportion in the ferrous state with the Ca ++ ions. This is a purely theoretical calculation since the microprobe does not distinguish valence states. However, it is interesting to note that this calculation leads to a formula very close to the phase 6CaO.2AI203. Fe203 • The overall result of this mineralogical analysis is sum~arised qualitatively in table 9. From these results it is evident that the three different types of LD slag (based on a pure lime charge, a lime-fluorspar charge, and a lime-CAMFlux charge) all contain : - B-dicalcium silicate calcium ferrite or calcium alumino-ferrite (Ca-Mn-Mg-Fe)O solid solution. -
-
The most important difference is the presence of calcium ferrite in the classical slag compositions and calcium alumino-ferrite in the CAM~Iux based slag composition. Further differences are the nature and quantities of other elements dissolved in these phases. This is demonstrated in table I0. From our mineralogical fore conclude that :
study of these semi-industrial
slags we there-
Vol. I I ,
No. 1
97 STEEL SLAG, CEMENT, HYDRAULICITY, CHARACTERIZATION TABLE 9 Mineralogical composition of semi-industrial detected by various methods
Sample
T1
T2
CaO
~
~
E
E
C2S
T
~
~
~
Ferrite C2F - C~A?F
~
LD slags
F lA
~
~
~
~
~
~
# T
~
~
F2A ~
~
~
~
~
T
~
~
•
MgO
~
Magnesio wustite
~
~ ~
FeO
~
~
~
~
~
~
~
~
~
~
Detected by :
Y
electron microprobe X-ray diffraction optical microscopy calculation
~
~
~
~
Trlealcium phosphate C3S
- the slags r e s u l t i n g homegeneous ;
from
the u s e
of an a l u m i n o u s
- the a l u m i n a is u s e d (dicalcium silicate
to f o r m c a l c i u m a l u m i n o - f e r r i t e , the o t h e r m a j o r p h a s e s and (Ca, Mn, Mg, Fe) 0 s o l i d s o l u t i o n s b e i n g u n c h a n g e d )
- the u s e of an a l u m i n o u s f l u x leads to a l o w lime c o n t e n t in the r e s u l t a n t slag.
charge
(and f i n e l y
(CAMFIux)
are more
distributed)
TABLE 10 I m p u r i t y c o n t e n t s of magnesio w u s t i t e , B-C2S and c a l c i u m (alumlno) f e r r i t e p h a s e s i n t h e s e m i - l n d u s t r i a l LD s l a g s S~mple MgO/FeO Magnesio wustite
Calcium aluminoferrite
0.18 - 0.55
MnO %
I0
T2 0.5 -
0.7
FIA
F2A
I
I
12
15
I0
I0
-16
4
- 29
3
3
P205 %
6
3
16
AI203 %
0
0
3
Fe203 %
0
0
1
0
AI203/Fe203
0.5
0.2
2
2
TiO 2 %
0.5
SiO 2 %
0
CaO %
B-C2S
TI
4
0
0
]
7 - 10
8
1 -
2.5
0 0
4
3
~ 3 0
free
;
98
Vol. I I ,
No. 1
M. Conjeaud, et a l . PART III : Study of Hydraulicity Classical techniques were used in this study : isothermal micro-calorimetry, X-ray diffraction identification of hydration products, quantitative analysis using DTA, and strength determination by cube crushing. Slags were reacted both with pure water, and, by analogy with Blast Furnace Slag practice, using activating agents, caustic soda or gypsum. a) Slag reaction with water The slags TI, T2, FIA and F2A as well as a sample of typical industrial LD slag, were gauged with water to form neat pastes at a water/cement ratio of I. At successive ages (6 hours, I, 2, 7 and 28 days) hydration was arrested by washing the specimens with acetone, and drying with ether. The phases present were identified by X-ray diffraction and the combined water was then determined by loss on ignition at IO00°C. The results for loss on ignition (combined water) are given in table II. TABLE II Combined water in semi-industrial slags after various periods of hydration Sample T1 T2 FIA F2A Industrial LD slag
6 hours 7.6 % 5.2 % 12.3 % 10.4 % 6.3 %
I day 2 days 7 days 28 days 11.8 % 11.8 % 13.0 % 1 7 . 1 % 8.3 % 9.4 % II.I % 14.9 % 12.9 % 14.3 % 14.6 % 17.5 % 10.2 % 10.7 % 13.2 % 17.2 % 7.7 % I0.0 % 13.6 % 17.0 %
It can be seen that the aluminous slags react much more rapidly in the first instance. Then follows a somewhat dormant period before further hydration ensues, leading to a final fixation of water after 28 days similar to that obtained in the absence of alumina. Only in the case of the fluorspar fluxed slag the total degree of reaction remains significantly lower. The nature of the hydrates formed, detected by X-ray diffraction, shown for slags T 1 and FIA in figures 8 and 9.
is
In general we found that slags T I and T2, and the industrial LD slag formed calcium hydroxide and a calcium ferrite hydrate of the form 4CaO (Fe203).H2Ox , which reacted rapidly with CO 2 from the atmosphere to give calcium mono-carbo ferrite hydrate analogous in structure to calcium monocarbo-aluminate hydrate. The aluminous slags gave rise to calcium hydroxide and cubic calcium aluminate hydrate (3CaO.AI203.6H20), which also reacts with CO 2 to give calcium mono-carbo aluminate hydrate : the anhydrous calcium alumino-ferrite phase was no longer detectable after the first 6 hours of reaction. In interpreting these results we recall that the mineralogical analysis showed that slags based on fluorspar flux continued a lower AI203 content in the calcium alumino-ferrite phase than in the case of slags produced from a pure lime charge. (The alumina content was very low in both cases, < 2 %). Also, although from our experiments it was difficult to know whether the AI203/Fe203 ratio in the hydrates was the same as that of the anhydrous reactants, the published literature suggests that it will be lower in the hydrate, (7). These two considerations help to explain the lower reactivity of the
Vol. I I ,
No. 1
99 STEEL SLAG, CEMENT, HYDRAULICITY, CHARACTERIZATION
-t 20
llllB
20
10
15
FIG. 8 Hydrate formation with semi-industrial LD slag T 1 (no alumina)
15
10
FIG. 9 Hydrate formation with semi-industrial LD slag F]A (10.2 % alumina)
fluorspar slag (T2) and the likelihood that reactivity will increase with the AI203/Fe203 ratio of the alumino-ferrite phase. (However we did not detect Fe(OH)3 from the aluminous slags, but if present it may well have been amorphous). In figure 10, the heats of hydration of slag T I (pure lime charge) and FIA (lime-CAMFlux charge) are compared. The much greater reactivity of the latter is confirmed. The reference slag and the industrial LD slags reacted much less rapidly than the aluminous slags and the Blast Furnace slag. Furthermore, the aluminous slags are more reactive than the Blast Furnace slag at early ages - the slower reaction of the latter is characteristic of its predominantly vitreous composition.
b) Slag reaction with NaOH solution Isothermal micro-calorimetry was used to study the reactivity of two aluminous slags, a reference slag (based on pure lime), two industrial LD slags and one good quality granulated Blast Furnace slag, with 5 N NaOH solution. The experimental conditions were the same as those described previously. The results are shown in figure
11.
I00
Vol. I I ,
No. 1
M. Conjeaud, et al.
AH '~~1A (ALUMINAI0,2%)
FIA , F2A , T I - semi-industrial slags BF - Blast Furnace Slag LDI, LD 2 - industrial LD slag
AH(Kcil/g
2
t
..~
10
IY"
i'o
~o
L D2
TIME|
~-~'o T''~
FIG. i0 The Effect of alumina content on the rate of heat evolution of semi-industrial LD slags in contact with water.
~BFF
FIG. ii Heat evolution from various slags in contact with NaOH solution
These results are encouraging in view of the relationship established earlier (laboratory study) between the development of useful mechanical properties and the rate of evolution of heat of hydration. c) Slag reaction with gypsum The reporting of the present study (being continued) would be incomplete without including preliminary results obtained by activating the aluminous slags with gypsum. Hydration in the presence of CaSO 4 results in the formation from the alumino-ferrite of calcium (mono-or-tri) sulpho-aluminate hydrate. In traditional LD slags (containing little A1203) this compound may be replaced by a sulpho compound of equivalent structure in which ferric ions replace the majority oF the trivalent aluminium. Our experiments detected the formation of traces of such compounds but the reaction of low alumina ferrites with CaSO 4 is extremely slow. By contrast, the aluminous slags, in which the alumina was combined as calcium alumino-ferrite reacted rapidly with gypsum to form ettringite (trisulpho compound) in substantial quantities after 24 hours. This is shown in figure 12 and it is to be noted that ettringite formation occured without the presence of 3CaO.AI203. The reaction may therefore be written
:
C6A2F + 3CH + 9CS--H2 + 75H ÷ 3 (C6A0.67F0.33S3H32)
Vol. I I ,
No. 1
I01 STEEL SLAG, CEMENT, HYDRAULICITY, CHARACTERIZATION g
E cn
f
|y
I d
cl
et = ettrzngzte
t
~J
= gypse
h.
ch = lime
f
(hydrated)
= ferrite
I a
FIG. 12 i
f
Formation of ettringite by activating an aluminous semi-industrial LD slag with gypsum
Mr 20
3o
d) Aluminous LD slags as additions to Portland cement Our investigations remain to be completed but the initial results are positive. Neat cement pastes were prepared at a water/cement ratio of 0.4 from mixtures of ordinary Portland cement and the ground aluminous slag obtained from the semi-industrial scale trials. The overall chemical composition of the mix is given in table 14 and the strength development observed is compared with similar mixes based on the pure lime + fluorspar reference slags, in figure 13. TABLE 14 Chemical composition of 50 % Portland cement, 50 % semi-industrlal aluminous LD slag mixture
CaO
SiO 2
AI203
Fe203
%
%
% 4.4
65.5
20.9
K20
%
SO 3 total %
2.8
2.7
0.51
Na20
%
% 0.11
TI, T$ semi-zndustrial LD slags without alumina
FIA, F2 A semi-industrial LD slags with alumina FIG.
]3
Development of compressive strength of Portland cement (50 %) - slag (50 %) pastes
C3S
C2S
C3A
C4AF
%
%
%
%
7.0
8.5
56.3
17.6
R C (M PII)
PORTLAND CEMENT
.////' ;
'
;.
oAY*,ME
102
Vol. I I ,
No. 1
M. Conjeaud, et al. The reference slags (T I and T2) - as well as a sample of typical LD slag (not shown) give unacceptable results - major crack development in the first 28 days. This effect, due to expansion forces, results from the presence of reactive free lime. The aluminous slags, mixed with Portland cement remained structurally intact (even under standardised autoclave conditions) and gave rise to long term strengths at least as good as these obtained with good quality granulated Blast Furnace slag. At early ages the strengths of the aluminous slag mixtures were less advanced due to an effect analogous to that of C3A in Portland cement clinkers. By adding gypsum, again analogously with Portland cements, we were able to overcome this retarding action and obtain good early strengths. Conclusions Laboratory experiments, confirmed by semi-industrial trials demonstrate, in accordance with theoretical predictions, that the addition of alumina to steel (LD) slags enables useful hydraulic properties to be obtained, provided the alumina is combined in the form of calcium-alumino-ferrite. This can be achieved by the use of a prefired synthetic flux material added to oxygen steel converters during the refining process. The composition of this material is chosen on the basis of the appropriate phase diagrams and contains CaO, A1203, MgO and Fe203, (CAMFIux). This process enables good quality steels to be produced without change of technology, and leads to the formation of homogeneous slags containing less than 4 % of free lime. The quantity of alumina needed in the final slag to obtain useful hydraulic activity is of the order of 6 to 15 %. The slags are fully crystallised and their hydraulic activity is not enhanced by quenching. References I. J. White, "Slag control in basic steelmaking process : an examination of the possibility of eliminating fluorspar", Iron and Steelmaking, N ° 2, 115 (1974). 2. H. Gaye, C.M. George, P. Riboud, F.P. Sorrentino, J. White, "A new aluminous product for B.O.P steelmaking and slag utilisation'~ International Conference on the Physical Chemistry of Iron and Steelmaking, Versailles, France, October 23-25, 1978. 3. J. White, "Process for steelmaking by oxygen refining of iron", US Patent N ° 4.010.027 - March I, 1977, British Patent N ° 1 508 024 - May 15, 1976. 4. C.M. George, F.P. Sorrentino, "Nouvelle m~thode de valorisation des scories d'affinage B.O.P", Colloque international sur l'utilisation des sousproduits et d~chets dans le Ggnie Civil, Paris 28-29 Novembre 1978. 5. C.M. George, F.P. Sorrentino, "Etude de l'hydratation des laitiers d'aci~rie LD", Silicates Industriels, 4-5, 71-76, (1977). 6. E.T. Carlson, "Action de l'eau sur les alumino-ferrites de calcium", J. Res. Nat. Bur. Stand. - A. Physics and Chemistry - USA, 68 A, 453-63, (Sept.-Oct. 1964). 7. D.E. Rogers, L.P. Aldridge, "Hydrates of calcium aluminate and calcium alumino-ferrite", Cement and Concrete Research, 7, 399-410, (1977).
A NEW STEEL SLAG FOR CEMENT MANUFACTURE MINERALOGY AND HYDRAULICITY
:
M. Conjeaud LAFARGE, Trappes, France C.M. George LAFARGE FONDU INTERNATIONAL,
Neuilly-sur-Seine,
F.P. Sorrentino LAFARGE Laboratoire Central, Viviers,
France
France
(Communicated by B. Cottin) (Received Sept. I0, 1980)
ABSTRACT This study presents a method of developing useful hydraulic properties in Basic Oxygen Steel Slags, by the addition of a synthetic slag forming flux during the normal steel making process. The transformation of a waste slag into a by-product at least equivalent to the best quality Blast Furnace Slags is achieved without change of technology or sacrifice in steel quality. The flux contains CaO - AI203 - MgO - Fe203 in proportions calculated from phase diagrams. Theoretical predictions are confirmed by laboratory experiments and by semi-industrial trials and show that the introduction of alumina in this manner leads to the formation of hydraulic calcium-alumino-ferrites and homogeneous slags containing < 4 % free lime. The mineralogy of the slags is described and the results of calorimetry, X-ray diffraction, optical microscopy, electron microprobe analysis, and strength tests on non quenched ground slags are given to demonstrate hydraulicity. Des propri~t~s hydrauliques pourraient ~tre d~velopp~es dans les scories d'aci~rie ~ l'oxyggne, par addition, pendant le proc~dg classique d'affinage, d'un produit synth~tique facilitant la formation dulaitier. Le produit d'addition contient CaO-AI203-MgO-Fe203 (abr~gg ~ CAMFIux) en proportions calcul~es d'aprgs les diagrammes de phase. Les prgvisions th~oriques, confirm~es par les experiences de laboratoire et des essais semi-industriels montrent que l'introduction d'alumine sous cette forme conduit, sans changer la technologie de l'affinage, ni compromettre la qualit~ de l'acier, ~ la formation d'une scorie homog~ne contenant moins de 4 % de chaux libre et une alumino-ferrite de calcium hydraulique. La min~ralogie des scories a ~t~ ~tudi~e par microscopic optique, diffraction X et microsonde. Les tests de microcalorim~tre isotherme et de r~sistances effectu~s sur des scories broy~es, mais non tremp~es permettent d'~valuer leurs propri~t~s hydrauliques. 85
86
Vol. I I ,
No. 1
M. Conjeaud, et al.
Introduction 50 million tonnes per year, world wide, of LD slags are produced as byproducts of the Basic Oxygen Process (BOP). Few applications for these slags exist at present since their phosphorus content is too low for fertilisers while their heterogeneity, particularly their free lime content, and the absence of hydraulic phases severly limits their use in the construction industry. Considerable importance thus attaches to methods of valorising these slags especially as the BOP will remain the major steel making process in the forseeable future. This paper describes the successful development of a new LD slag compatible both with the BOP and the needs of the cement manufacturer. The new slag is less variable than classical slags, contains very little free lime and possesses good hydraulic properties. The paper is in three parts
:
- theoretical and experimental studies demonstrating efficient refining of iron into steel and defining the composition of the resulting slags -
structural characterisation of the slag obtained under semi-industrial conditions
- definition of the hydraulic behaviour of these slags. Part I : The BOP and flux additions In the BOP, which is a batch process, gaseous oxygen is blown into liquid Blast Furnace iron in a special converter vessel. Impurities in the iron are oxidised (C ~ C0/CO2, P ÷ P205, Si ÷ Si02, etc.) The non volatile products, including some of the iron oxidised to iron oxide, form, with an added basic material, lime, a slag, which contains the impurities separate from the liquid metal. The iron is thus purified and becomes steel to be decanted separately from the slag, at the end of the process. No fuel is needed in this process since the oxidation reactions are themselves exothermic. At the beginning of the refining process (the blow) carbon, silica and iron are preferentially oxidised. The carbon oxides escape as gas, whereas FeO and SiO 2 combine with solid lime to form the initial slag. This is only partially liquid since some of the silica forms an insoluble coating of calcium silicate on the surface of the lime, inhibiting further reaction. The effect is shown in a pseudo-ternary diagram, figure L, where the development of the slag composition can be seen to traverse the precipitation zone of solid C2 S, (2CaO.Si02). In order to prevent the formation of this passivating layer of silicate, many different fluxes such as fluorspar, borates or titanates, bauxite, etc, have been tried in the BOP. These fluxes act by lowering the viscosity of the slag or enlarging the composition zone of liquid formation. Their effectiveness is, however, compromised by a certain number of disadvantages : - increasing the acidity of the slag which, combined with greater fluidity becomes more aggressive to the basic converter refractories ; -
-
producing harmful exhaust gases (HF) or contaminating the steel (Boron) ; preventing the development of useful properties in the final slags.
Vol. I I ,
No. 1
87
(s) *~
STEEL SLAG, CEMENT, HYDRAULICITY, CHARACTERIZATION
c Tx -
(c)
(~A) FIG.
I
Pseudo-ternary phase diagram CaO-SiO2-FeO showing LD slag composition path during steel manufacture X - without alumina Y - with alumina
In addition, certain materials, particularly fluorspar (the most widely used at present) are becoming of increasingly limited availability. The present work describes the results obtained from substituting such fluxes by a prefired synthetic material (CAMFIux) ~ developed from phase diagram studies reported by J. WHITE (I, 2, 3). In a simplified system such as CaO-SiO2-FeO, in order that a slag in the course of formation should be in equilibrium with lime as the only solid phase, it is necessary that the ratio Si02/FeO should not exceed the value shown by point K in figure I. By extending this arguement to the multicomponent system representing a steel making slag, limiting values for the ratio Si02/(FeO + MgO + A1203) can be defined. This ratio is greater than the ratio Si02/FeO of the simplified system. J. WHITE's studies show that the maximum value of this ratio - representing the greatest capacity of the slag to absorb silica without forming insoluble silicates - is to be found in the system CaO, MgO, A1203, iron oxide, SiO 2. This has led us to develop a prefired material to replace classical BOP fluxes, of such composition that the slag developed in the course of refining achieves this optimum limiting value of the ratio Si02/(FeO + MgO + A1203). Such a slag should be more efficient than classical slags for absorbing P205 due to its undersaturation in SiO 2. Using this material, the final slag contains CaO, MgO, A1203, iron oxide, SiO2, P205, MnO. The phase diagram of this system has only been very partially explored. In particular, the optimum ratio of Al203/iron oxide has not been determined. The composition of the prefired synthetic material which we have developed has therefore been chosen to provide anAl203/iron oxide ratio which confers hydraulic properties to the final slag. Theoretical
study of the hydraulic potential of LD slags
Mineralogical examination shows that classical LD slags contain ~-calcium orthosilicate (2CaO.SiO2, stabilised by dissolved phosphoric oxide), ferrites, typically 2CaO.Fe203, calcium phosphate 3CaO.P205, wustite, FeO and solid solutions of the (Ca, Fe, Mn, Mg)O type, and since they contain only contains llme (C), alumina name CAMFIux.
(A), magnesia
(M), and iron oxide (F). Hence the
88
Vol. I I ,
No. l
M. Conjeaud, et a l . negligible quantities of vitreous material, it is possible to calculate the potential mineralogical composition from a knowledge of the easily analysed chemical composition. In a typical case we find : Chemical composition (by analysis) CaO MgO A1203 Fe oxides SiO 2 P205 MnO
%
Mineralogical composition % (by calculation) 2CaO.SiO2-B 43 3CaO.P205 4 2CaO.Fe203 ]7 (Ca. Fe.Mn.Mg)O 31 Free CaO 5
50 6 0 22 ]5 2 5
The free lime value has been chosen arbitrarily near the lower limit found industrially. Such slags show very little hydraulic activity (5) as their potential mineralogical composition indeed suggests : B.2CaO.SiO 2 hydrates very slowly and calcium ferrite, 2CaO.Fe203 is inert as are the other phases. However, by incorporating alumina in the slag the calcium ferrite can be transformed into calcium alumino-ferrite which possesses substantial hydraulic activity. We find that this activity increases with the AI203/Fe203 ratio. In figure 2 we show the degree of hydration of calcium-alumino-ferrite, measured by X-ray diffraction, as a function of the AI203/Fe203 ratio : the results confirm those reported by Carlson, (6). In order to have a useful level of hydraulic activity - represented by a relatively rapid development of mechanical properties, - an AI203 content corresponding to the formation of 4CaO.AI203.Fe203 is desirable. Referred to a typical LD slag as described above, this implies an AI203 content of 6 to ]0 % from which the mineralogical composition can be calculated to be : 2CaO.SiO2.B 3CaO.P205 4CaO.AI203.Fe203 (Ca.Fe.Mn.Mg)O %RESIDUAL
ANHYDROUS
43 4 29 23
% % % %
MATERIAL
C6A2 F AF
FIG. 2 Rate of Hydration of calcium-alumino-ferrites as a function of alumina content
5(;
C2F 100
1 10mns
l "'' 21=rs
i 1day
, TIME lodays " -
Vol. I I ,
No. 1
89 STEEL SLAG, CEMENT, HYDRAULICITY, CHARACTERIZATION
This primarily theoretical analysis thus leads to the conclusion that useful hydraulic behavlour may be expected of an LD slag by the introduction of 6 % or more AI203. Our initial laboratory study thus aimed at verifying this conclusion by a study of synthetic slags of varied AI203 content. Laboratory study of the hydraulicity of LD slags containing Alumina Two series of slags were studied, obtained by adding 5 - I 0 - 1 5 - 20 and 30 % AI203 to two industrial LD slags. This approach was considered more practical than the study of entirely synthetic materials, particularly as concerns the nature of the solid solutions present, (Ca.Fe.Mn.Mg)O and the 2CaO.SiO 2 and 3CaO.P205 phases. The compositions of each series are shown in tables I and 2. (LD I and LD 2 are the original slags, LD 105 contains approximately 5 % A1203, etc.). The slag-Al203 mixtures were melted between 1450 and 1550°C in a reducing atmosphere of C0/C02, calculated, for each temperature to produce an FeO content of approximately 10 %, and quenched. -
For AI203 contents below 15 % the phases detected were 2CaO.SiO2. ~ (Ca.Fe.Mn.Mg)O solid solution ferrlte in the composition range 2CaO.Fe203 to 2CaO.AI203 negligible vitreous material despite quenching.
:
The alumina is thus combined in the ferrlte phase as anticipated. TABLE I Chemical composition (%) of laboratory LD slags with alumina additions
CaO
SiO 2
FeO
Fe203
AI203
TiO 2
45.6
12.3
]3.0
]2.1
].0
0.7
8.5
1.7
1.2
2.1
LD ]05 44.8
12.1
8.8
I].9
5.9
0.7
8.4
].7
1.2
2.0
LD |10 41.9
11.4
10.6
]0.2
11.2
0.6
7.8
1.6
1.1
1.9
LD 115 40.0
10.7
I1.1
7.6
16.3
0.6
7.4
1.5
1.0
LD 120 3 7 . 7
10.2
10. I
7.1
21.5
0.6
7.0
1.5
1.0 ]
LD I
MgO
P205
MnO
Loss
l
I
1.8
1.7
TABLE 2 Chemical composition (%) of laboratory LD slags with alumina additions
CaO
SiO 2
FeO
AI203
TiO 2
47.1
12.4
13.3
11
1.2
0.5
4.9
2.0
5.4
1.1
LD 205 45.8
12.6
8.7
14.2
6.0
0.5
4.7
2.0
5.2
0.3
LD 210 4 2 . 3
11.2
6.1
]7.6
11.1
0.5
4.4
1.8
4.9
1.0
LD 215 40.5
10.6
lO.l
10.9
16.2
0.4
) 4.2
1.7
4.6
0.9
LD 230 32.9
8.7
II.2
6.6
30.8
0.4
3.4
1.4
3.8
].0
LD 2
Fe203
MgO
P205
MnO
Loss
90
Vol. I I ,
No. 1
M. Conjeaud, et a l .
Above 15 % alumina there is extensive vitrification and only glass and (Ca.Fe.Mn.Mg)O solid solutions were found. Owing to the small quantities of slag obtained under these laboratory conditions, we chose to study their hydraulicity by isothermal calorimetry, a technique widely used in the study of Portland cements. Furthermore, whereas normal cements generate considerable heat of hydration within 24 hours, the slags do not and for this reason their hydration was accelerated by treating them with solutions of caustic soda, (N, 2N and 5N) rather than pure water. The quenched slags, ground to less laser - table 3) were positioned in 500 mg ve the caustic soda solution. Once thermal ned, each sample was precipitated into the time registered at constant temperature.
than 100 ~ granulometry measured by quantities in the calorimeter, aboequilibrium at 21°C had been obtaisolution and the heat evolved with
TABLE 3 Granulometry and heat evolution of ground, alumina doped LD slags
Percentage less than 2 ~
4 B
8 B
LD 1
18
31
46
63
78
95
7.9
LD 105
12
20
32
51
72
96
15.2
LD 110
8
17
30
50
70
93
17.3
LD 115
6
II
21
38
60
88
22.9
LD 120
16
29
47
70
89
100
LD 2
21
34
49
65
80
96
16.5
6
16
20
30
55
80
17.6
LD 2 0 5
16 ~ 32
AH total, Kcal g-l with a 5 N 64 I~ NaOH solution
35
LD 2 1 0
ll
22
37
56
74
91
15.6
LD 2 1 5
14
28
46
65
81
95
25
LD 2 3 0
ll
23
39
61
79
95
32
[
I
The heat evolved in contact with water (or in this case, more rapidly in contact with a caustic solution) depends on the quantity and nature of the hydrates formed. Since these hydrates are the binding agents which give rise to cohesive forces it is thus reasonable to expect a correlation between the heat evolved and the mechanical strengths concomitantly developed. However, morphology and packing factors also influence mechanical properties significantly without necessarily being reflected in heat evolution. It will therefore be understood that our measurement of the compressive strength of the hydrated slags provide a useful but only partial basis for the interpretation of their behaviour measured in the calorimeter. Figure 3 relates the compressive strength observed on neat slag 5 N NaOH samples at 24 hours to the heat evolved under equivalent conditions. We have concluded, on the basis of these results, that the heat evol-
Vol. I I , No. 1
91 STEEL SLAG, CEMENT, HYDRAULICITY, CHARACTERIZATION A H(Kcal/g)
t
40 ÷
R C (MPa)
10
.AH(Kcal/g )
,0 p
2p
30 ~AI203...~
FIG. 4
FIG. 3 Relationship between compressive strength and heat evolution of alumina doped LD slag pastes
I
Relationship between hear evolution and alumina content of laboratory LD slags
ved at 24 hours must be at least 20 kcal/g if the slag is to display useful hydraulic behaviour. Table 3 and figure 4 show that an alumina content of about 15 % provided such a level of heat evolution and this result is to be compared with the minimum theoretical value of 6 % calculated previously. From the preliminary studies the following conclusions may therefore be drawn : - the addition of AI203 to classical LD slags enables hydraulic activity to be obtained ; - the minimum quantity of AI203 necessary for this hydraulic activity to give useful cementitious propertzes is of the order of 6 to 15 % by weight of the slag. Part II : Verification of the hydraulic behaviour of aluminous LD slags obtained semi-industrially Preparation The next step in our investigations was to prepare sufficient quantities of our synthetic slagging agent (CAMFIux) to enable the results of the laboratory study to be tested under more realistic conditions. Two different flux compositions were prepared in a gas-fired intermittent kiln. The compositions are shown in table 4. The use of natural raw materials occasioned certain impurities, notably silica. Slags were then obtained by refining 5 tonne batches of hot metal to steel in a pilot converter at I.R.S.I.D. (~) research station, the CAMFIux being added with lime to give the slag forming charge. (~) Institut de Recherches de la Sid~rurgie FranGaise.
92
Vol. I I ,
No. l
M. Conjeaud, et a l .
Six t e s t s w e r e c a r r i e d out. The f i r s t e m p l o y e d a c h a r g e of lime only, the s e c o n d h a d I0 % of f l u o r s p a r m i x e d w i t h the lime. In the o t h e r four tests, two i n v o l v e d a m i x t u r e of lime w i t h C A M F I u x 1 and the o t h e r two a m i x t u r e of l i m e w i t h C A M F I u x 2. The
chemical
composition
of e a c h
charge
is g i v e n
in T a b l e
5.
TABLE 4 Chemical compositions of CAMFIux used in this study CAMFIux I CAMFIux 2 31,.9 49
Ca5
%
A1203
%
29
MgO
%
12.3
9.1
Fe203
%
18.9
13.8
SiO 2
%
5.6
4.7
22.4
TABLE 5 Charge compositions used for semi-industrial preparation of LD slags Test N °
3
4
5
6
66.9
65.7
66.8
67.0
14.1
14.6
14.6
14.5
%
5.9
6.2
5.9
5.9
Fe203 %
8.9
9.5
8.9
8.9
2.7
2.8
3.1
3.0
F2A
F2B
CaO
%
I
2
100
90
AI203 % MgO
SiO 2
%
CaF 2
%
0
I0 TABLE 6 Compositions of semi-industrial LD slags
Slag :
TI
T2
FIA
FIB
10.25
10.7
43.7
43.4
45.8
40.6
42.65
0.95
1.3
10.2
11.4
6.5
9.15
MgO
10.45
6.25
5.8
7.2
14.5
10.85
MnO
4.36
4.99
3.10
3.82
3.91
P205
2.40
3.95
2.15
2.25
2.5
1.8
sio 2 CaO A1203
9.55 44.4
10.75
7.75
9.0
3.91
Fe ++
19.7
18.35
17.9
12.6
16.25
14.75
Fe +++
8.6
10.7
10.15
8.7
4.8
6.15
Fe++/Fe+++
2.3
1.7
1.75
1 .45
3.4
2.4
Free CaO TiO 2 TOTAL From test N °
10.8 0.21 100.6
3.3
4.2
3.8
4.55
1.8
0.19
0.54
0.60
0.39
0.53
100.18
I00.99
I01.37
99.6
100.49
Vol. I I ,
No. 1
93 STEEL SLAG, CEMENT, HYDRAULICITY, CHARACTERIZATION
The composition of the hot metal was fixed to reflect typical blast furnace quality and the refining operation was conducted to produce acceptable quality steel. At the end of each blow the slags were decanted into normal ladles and allowed to cool freely in contact with air. The compositions of the six slags obtained are shown in table 6. Mineralogical study The interpretation of the mineralogical analysis (optical microscopy on polished sections and X-ray diffraction on powder samples) is difficult due to the complexity of the system. We attempted once again to calculate the potential composition with the results shown in table 7, assuming the calcium alumino ferrite phase to have the composition 4CaO.AI203.Fe203. This assumption leads to the implication that 3CaO.AI203 will also be present, which would not necessarily be the case had we adopted the formulation 6CaO.2AI203. Fe203. However, the quantity of 3CaO.AI203 predicted is relatively small and would be difficult to detect by the analytical techniques. Accordingly it appeared necessary to determine the composition of the calcium alumino-ferrite phases as accurately as possible. For this purpose, electron micro-probe analysis was used. TABLE 7 Calculated mineralogical composition of semi-industrial LD slags Slag : C2S - C3P % Ferrite
%
(Ca-Fe-Mn-Mg) 0 % C3A
TI
T2
31 - 6
32 - 9
16
19
47
40
%
i F]A
FI B
F2A
F2B
22 - 5
27 - 5
30 - 5
32 - 4
32
27
f5
]9
24
26
42
34
17
15
8
If
II
40
30
40
30
C = CaO F = calcium-alumino-ferrite M = MgO mw= magnesio w~stite b = ~-C2S
FIG. 5 X-ray diffraction diagrams of semiindustrial LD slags (TI,T2,FIAF2A)
94
Vol. I I ,
No. 1
M. Conjeaud, et a l . The most extensive slags TI, T2, FIA and F2A.
study was confined
The X-ray diffraction obtained with the micro-probe
~-~1
to four samples
diagrams are given in figure in table 8.
representing
5 and the results
contained:
a) Periclase grains of variable composition, - centre, 1 % CaO, 4 % MnO, 20 % FeO surface, 8 % CaO, 1 1 % MnO, 62 % FeO and containing small quantities of SiO 2 in some cases. -
b) Magnesio wustite with an MgO/FeO ratio varying from 0.18 to 0.51 and contai ning MnO and CaO in quantities from 8 to 36 % and from 4 to 16 % respectively. TABLE 8 Chemical composition of semi-industrial LD slags analysed by electron microprobe Sample TI Pure lime charge
Phase analysis Periclase (centre of grains) (edge of grains Magnesio wustite
Lime Dicalcium silicate Alumino ferrite T2
Magnesio wustite
Dicalcium silicate Lime + Alumino ferrite fluorspar Calcium phosphate charge
Average molar composition +÷ Mg0.84 Ca0.0; Mno.02 Fe
Magnesio wustite
Dicalcium silicate Lime + CAMFIux 1 Alumino ferrite charge F2A
Magnesio wustite
Wustite Lime + Lime CAMFIux 2 Dicalcium silicate charge Alumino ferrite
0.]20
O 0 0 0
Ca2 Si0.91P0.1Fe++0.01 04 +++ Ca2. 2 Si0.03 AI0. 4 Fe
1.4 05 ++ Mg0.35 Ca0.06 Mn0.]; Fe 0.48 0 Ca2 Si0.84 P0.12 AIo.40 +++ Ca2. 3 AI0. 3 Fe ].5 05 Ca3 P].8 Sio.2 08
Tricalcium silicate Ca2.5 Sio.9 P0.09 Fe FIA
++
Mg0.56 Ca0.06 Nn0.04 Fe 0.27 ++ Mg0.12 Ca0.14 Nn0.11Fe ++0.63 Mg0.21 Ca0.09 Nn0.12 Fe 0.57 ++ Ca0o83 Mg0.08 Nn0.04 Fe 0.05
++
0.07 05 ++ Mg0.47 Ca0.05 Mn0.07 Fe 0.43 O ++ Ca2 Si0.89 P0.1AI0.02 Fe 0.010
Cal. 8 AIi.32 Feo.79 05 (Ca].8 Fe++0.2) AII.32 Fe +++ 0.59 05 ++ Mgo.41 Ca0.05 Mn0.08 Fe 0.46 O FeO Ca0. 9 Feo.05 Mn0.05 O
Ca2 Si0.86 P0.10 A10.10 O Cal.86 All.34 Fe0.75 05 (Cai.86 Fe++0.14) A11.34 Fe
+++ 0.61 05
Vol. I I , No. 1
95 STEEL SLAG, CEMENT, HYDRAULICITY, CHARACTERIZATION
c) Lime in large grains several tens of microns across containing FeO (~ 6 %), MnO (~ 4 %) in the apparently homogeneous areas and more FeO and less CaO in apparently heterogeneous regions. d) Dicalcium silicate with 4 to 8 % dissolved P205 . e) Calcium alumino-ferrite, rich in iron and containing practically all the alumina (~ 1%) present in the slag. A photomicrograph
of this sample is shown in figure 6.
O
Magnitude
~c2s
600
FIG. 6 Photomicrograph
of semi-industrial
slag T 1 (no alumina)
a) Magnesio wustite of relatively constant composition - ratio MgO/FeO between 0.53 and 0.73. b) Dicalcium silicate with a more variable P205 content c) Calcium alumino-ferrite
containing
(3 to 16 %).
less alumina than in sample T I .
d) Tricalcium phosphate. e) Some tricalcium silicate containing P205 .
S_am_~e_FlA contained : a) Magnesio wustite
(MgO/FeO = I).
b) Dicalcium silicate with about 8 % dissolved P205 and about c) Calcium alumino-ferrite
1.2 % Fe203.
with a composition close to 6CaO.2AI203.Fe203.
d) No other phases were detected and this sample was considerably more homogeneous than the preeceding ones. SamE~e_F2AContained a) Magnesio wustite
: (MgO/FeO = I).
b) Dicalcium silicate with about 8 % dissolved P205 and 2.5 % AI203.
96
Vol. I I ,
No. 1
M. Conjeaud, et al. c) Calcium alumino-ferrite with a composition close to 6CaO.2AI203oFe203. d) Pure wustite. e) Small quantities of lime containing Mn and Fe in solid solution. Thus five phases were detected in this sample, of which a photomicrograph is shown in figure 7. MW = magnesiowustite
AF = aluminoferrite
Magnitude
3 0 0
FIG. 7 Photomicrograph of semi-industrial
slag F2A (6.5 % alumina)
The formulae given in table 8 have been calculated for an integer number of oxygen atoms. Of principal interest here are the compositions of the alumino-ferrite phases, the potential seat of hydraulic activity. In samples FIA and F2A one notes a deficiency in Ca and an excess of Fe with respect to the formula 6CaO.2AI203.Fe203. Since not all the iron will be present in the trivalent state we have attempted to group a proportion in the ferrous state with the Ca ++ ions. This is a purely theoretical calculation since the microprobe does not distinguish valence states. However, it is interesting to note that this calculation leads to a formula very close to the phase 6CaO.2AI203. Fe203 • The overall result of this mineralogical analysis is sum~arised qualitatively in table 9. From these results it is evident that the three different types of LD slag (based on a pure lime charge, a lime-fluorspar charge, and a lime-CAMFlux charge) all contain : - B-dicalcium silicate calcium ferrite or calcium alumino-ferrite (Ca-Mn-Mg-Fe)O solid solution. -
-
The most important difference is the presence of calcium ferrite in the classical slag compositions and calcium alumino-ferrite in the CAM~Iux based slag composition. Further differences are the nature and quantities of other elements dissolved in these phases. This is demonstrated in table I0. From our mineralogical fore conclude that :
study of these semi-industrial
slags we there-
Vol. I I ,
No. 1
97 STEEL SLAG, CEMENT, HYDRAULICITY, CHARACTERIZATION TABLE 9 Mineralogical composition of semi-industrial detected by various methods
Sample
T1
T2
CaO
~
~
E
E
C2S
T
~
~
~
Ferrite C2F - C~A?F
~
LD slags
F lA
~
~
~
~
~
~
# T
~
~
F2A ~
~
~
~
~
T
~
~
•
MgO
~
Magnesio wustite
~
~ ~
FeO
~
~
~
~
~
~
~
~
~
~
Detected by :
Y
electron microprobe X-ray diffraction optical microscopy calculation
~
~
~
~
Trlealcium phosphate C3S
- the slags r e s u l t i n g homegeneous ;
from
the u s e
of an a l u m i n o u s
- the a l u m i n a is u s e d (dicalcium silicate
to f o r m c a l c i u m a l u m i n o - f e r r i t e , the o t h e r m a j o r p h a s e s and (Ca, Mn, Mg, Fe) 0 s o l i d s o l u t i o n s b e i n g u n c h a n g e d )
- the u s e of an a l u m i n o u s f l u x leads to a l o w lime c o n t e n t in the r e s u l t a n t slag.
charge
(and f i n e l y
(CAMFIux)
are more
distributed)
TABLE 10 I m p u r i t y c o n t e n t s of magnesio w u s t i t e , B-C2S and c a l c i u m (alumlno) f e r r i t e p h a s e s i n t h e s e m i - l n d u s t r i a l LD s l a g s S~mple MgO/FeO Magnesio wustite
Calcium aluminoferrite
0.18 - 0.55
MnO %
I0
T2 0.5 -
0.7
FIA
F2A
I
I
12
15
I0
I0
-16
4
- 29
3
3
P205 %
6
3
16
AI203 %
0
0
3
Fe203 %
0
0
1
0
AI203/Fe203
0.5
0.2
2
2
TiO 2 %
0.5
SiO 2 %
0
CaO %
B-C2S
TI
4
0
0
]
7 - 10
8
1 -
2.5
0 0
4
3
~ 3 0
free
;
98
Vol. I I ,
No. 1
M. Conjeaud, et a l . PART III : Study of Hydraulicity Classical techniques were used in this study : isothermal micro-calorimetry, X-ray diffraction identification of hydration products, quantitative analysis using DTA, and strength determination by cube crushing. Slags were reacted both with pure water, and, by analogy with Blast Furnace Slag practice, using activating agents, caustic soda or gypsum. a) Slag reaction with water The slags TI, T2, FIA and F2A as well as a sample of typical industrial LD slag, were gauged with water to form neat pastes at a water/cement ratio of I. At successive ages (6 hours, I, 2, 7 and 28 days) hydration was arrested by washing the specimens with acetone, and drying with ether. The phases present were identified by X-ray diffraction and the combined water was then determined by loss on ignition at IO00°C. The results for loss on ignition (combined water) are given in table II. TABLE II Combined water in semi-industrial slags after various periods of hydration Sample T1 T2 FIA F2A Industrial LD slag
6 hours 7.6 % 5.2 % 12.3 % 10.4 % 6.3 %
I day 2 days 7 days 28 days 11.8 % 11.8 % 13.0 % 1 7 . 1 % 8.3 % 9.4 % II.I % 14.9 % 12.9 % 14.3 % 14.6 % 17.5 % 10.2 % 10.7 % 13.2 % 17.2 % 7.7 % I0.0 % 13.6 % 17.0 %
It can be seen that the aluminous slags react much more rapidly in the first instance. Then follows a somewhat dormant period before further hydration ensues, leading to a final fixation of water after 28 days similar to that obtained in the absence of alumina. Only in the case of the fluorspar fluxed slag the total degree of reaction remains significantly lower. The nature of the hydrates formed, detected by X-ray diffraction, shown for slags T 1 and FIA in figures 8 and 9.
is
In general we found that slags T I and T2, and the industrial LD slag formed calcium hydroxide and a calcium ferrite hydrate of the form 4CaO (Fe203).H2Ox , which reacted rapidly with CO 2 from the atmosphere to give calcium mono-carbo ferrite hydrate analogous in structure to calcium monocarbo-aluminate hydrate. The aluminous slags gave rise to calcium hydroxide and cubic calcium aluminate hydrate (3CaO.AI203.6H20), which also reacts with CO 2 to give calcium mono-carbo aluminate hydrate : the anhydrous calcium alumino-ferrite phase was no longer detectable after the first 6 hours of reaction. In interpreting these results we recall that the mineralogical analysis showed that slags based on fluorspar flux continued a lower AI203 content in the calcium alumino-ferrite phase than in the case of slags produced from a pure lime charge. (The alumina content was very low in both cases, < 2 %). Also, although from our experiments it was difficult to know whether the AI203/Fe203 ratio in the hydrates was the same as that of the anhydrous reactants, the published literature suggests that it will be lower in the hydrate, (7). These two considerations help to explain the lower reactivity of the
Vol. I I ,
No. 1
99 STEEL SLAG, CEMENT, HYDRAULICITY, CHARACTERIZATION
-t 20
llllB
20
10
15
FIG. 8 Hydrate formation with semi-industrial LD slag T 1 (no alumina)
15
10
FIG. 9 Hydrate formation with semi-industrial LD slag F]A (10.2 % alumina)
fluorspar slag (T2) and the likelihood that reactivity will increase with the AI203/Fe203 ratio of the alumino-ferrite phase. (However we did not detect Fe(OH)3 from the aluminous slags, but if present it may well have been amorphous). In figure 10, the heats of hydration of slag T I (pure lime charge) and FIA (lime-CAMFlux charge) are compared. The much greater reactivity of the latter is confirmed. The reference slag and the industrial LD slags reacted much less rapidly than the aluminous slags and the Blast Furnace slag. Furthermore, the aluminous slags are more reactive than the Blast Furnace slag at early ages - the slower reaction of the latter is characteristic of its predominantly vitreous composition.
b) Slag reaction with NaOH solution Isothermal micro-calorimetry was used to study the reactivity of two aluminous slags, a reference slag (based on pure lime), two industrial LD slags and one good quality granulated Blast Furnace slag, with 5 N NaOH solution. The experimental conditions were the same as those described previously. The results are shown in figure
11.
I00
Vol. I I ,
No. 1
M. Conjeaud, et al.
AH '~~1A (ALUMINAI0,2%)
FIA , F2A , T I - semi-industrial slags BF - Blast Furnace Slag LDI, LD 2 - industrial LD slag
AH(Kcil/g
2
t
..~
10
IY"
i'o
~o
L D2
TIME|
~-~'o T''~
FIG. i0 The Effect of alumina content on the rate of heat evolution of semi-industrial LD slags in contact with water.
~BFF
FIG. ii Heat evolution from various slags in contact with NaOH solution
These results are encouraging in view of the relationship established earlier (laboratory study) between the development of useful mechanical properties and the rate of evolution of heat of hydration. c) Slag reaction with gypsum The reporting of the present study (being continued) would be incomplete without including preliminary results obtained by activating the aluminous slags with gypsum. Hydration in the presence of CaSO 4 results in the formation from the alumino-ferrite of calcium (mono-or-tri) sulpho-aluminate hydrate. In traditional LD slags (containing little A1203) this compound may be replaced by a sulpho compound of equivalent structure in which ferric ions replace the majority oF the trivalent aluminium. Our experiments detected the formation of traces of such compounds but the reaction of low alumina ferrites with CaSO 4 is extremely slow. By contrast, the aluminous slags, in which the alumina was combined as calcium alumino-ferrite reacted rapidly with gypsum to form ettringite (trisulpho compound) in substantial quantities after 24 hours. This is shown in figure 12 and it is to be noted that ettringite formation occured without the presence of 3CaO.AI203. The reaction may therefore be written
:
C6A2F + 3CH + 9CS--H2 + 75H ÷ 3 (C6A0.67F0.33S3H32)
Vol. I I ,
No. 1
I01 STEEL SLAG, CEMENT, HYDRAULICITY, CHARACTERIZATION g
E cn
f
|y
I d
cl
et = ettrzngzte
t
~J
= gypse
h.
ch = lime
f
(hydrated)
= ferrite
I a
FIG. 12 i
f
Formation of ettringite by activating an aluminous semi-industrial LD slag with gypsum
Mr 20
3o
d) Aluminous LD slags as additions to Portland cement Our investigations remain to be completed but the initial results are positive. Neat cement pastes were prepared at a water/cement ratio of 0.4 from mixtures of ordinary Portland cement and the ground aluminous slag obtained from the semi-industrial scale trials. The overall chemical composition of the mix is given in table 14 and the strength development observed is compared with similar mixes based on the pure lime + fluorspar reference slags, in figure 13. TABLE 14 Chemical composition of 50 % Portland cement, 50 % semi-industrlal aluminous LD slag mixture
CaO
SiO 2
AI203
Fe203
%
%
% 4.4
65.5
20.9
K20
%
SO 3 total %
2.8
2.7
0.51
Na20
%
% 0.11
TI, T$ semi-zndustrial LD slags without alumina
FIA, F2 A semi-industrial LD slags with alumina FIG.
]3
Development of compressive strength of Portland cement (50 %) - slag (50 %) pastes
C3S
C2S
C3A
C4AF
%
%
%
%
7.0
8.5
56.3
17.6
R C (M PII)
PORTLAND CEMENT
.////' ;
'
;.
oAY*,ME
102
Vol. I I ,
No. 1
M. Conjeaud, et al. The reference slags (T I and T2) - as well as a sample of typical LD slag (not shown) give unacceptable results - major crack development in the first 28 days. This effect, due to expansion forces, results from the presence of reactive free lime. The aluminous slags, mixed with Portland cement remained structurally intact (even under standardised autoclave conditions) and gave rise to long term strengths at least as good as these obtained with good quality granulated Blast Furnace slag. At early ages the strengths of the aluminous slag mixtures were less advanced due to an effect analogous to that of C3A in Portland cement clinkers. By adding gypsum, again analogously with Portland cements, we were able to overcome this retarding action and obtain good early strengths. Conclusions Laboratory experiments, confirmed by semi-industrial trials demonstrate, in accordance with theoretical predictions, that the addition of alumina to steel (LD) slags enables useful hydraulic properties to be obtained, provided the alumina is combined in the form of calcium-alumino-ferrite. This can be achieved by the use of a prefired synthetic flux material added to oxygen steel converters during the refining process. The composition of this material is chosen on the basis of the appropriate phase diagrams and contains CaO, A1203, MgO and Fe203, (CAMFIux). This process enables good quality steels to be produced without change of technology, and leads to the formation of homogeneous slags containing less than 4 % of free lime. The quantity of alumina needed in the final slag to obtain useful hydraulic activity is of the order of 6 to 15 %. The slags are fully crystallised and their hydraulic activity is not enhanced by quenching. References I. J. White, "Slag control in basic steelmaking process : an examination of the possibility of eliminating fluorspar", Iron and Steelmaking, N ° 2, 115 (1974). 2. H. Gaye, C.M. George, P. Riboud, F.P. Sorrentino, J. White, "A new aluminous product for B.O.P steelmaking and slag utilisation'~ International Conference on the Physical Chemistry of Iron and Steelmaking, Versailles, France, October 23-25, 1978. 3. J. White, "Process for steelmaking by oxygen refining of iron", US Patent N ° 4.010.027 - March I, 1977, British Patent N ° 1 508 024 - May 15, 1976. 4. C.M. George, F.P. Sorrentino, "Nouvelle m~thode de valorisation des scories d'affinage B.O.P", Colloque international sur l'utilisation des sousproduits et d~chets dans le Ggnie Civil, Paris 28-29 Novembre 1978. 5. C.M. George, F.P. Sorrentino, "Etude de l'hydratation des laitiers d'aci~rie LD", Silicates Industriels, 4-5, 71-76, (1977). 6. E.T. Carlson, "Action de l'eau sur les alumino-ferrites de calcium", J. Res. Nat. Bur. Stand. - A. Physics and Chemistry - USA, 68 A, 453-63, (Sept.-Oct. 1964). 7. D.E. Rogers, L.P. Aldridge, "Hydrates of calcium aluminate and calcium alumino-ferrite", Cement and Concrete Research, 7, 399-410, (1977).