Technical Paper CHEMICAL INTERACTION OF MGO IN SPINEL FORMING CASTABLES

Transcription

1 Reference : TPGBRELAF CHEMICAL INTERACTION OF MGO IN SPINEL FORMING CASTABLES by Thomas A. Bier, Christopher Parr, Catherine Revais, Michel Vialle, Hervé Fryda, presented at the XXVII ALAFAR Congress, Lima, Peru; rue des Graviers 9 Neuilly sur Seine Cedex Tel : () Fax : () 6 7 9

2 Reference : TPGBRELAF Abstract Following the recent work published on MgO containing alumina castables a closer look has been taken at possible chemical interactions between the MgO and the other constituents especially during heatup of the castable. Different MgO grades have been reexamined to asses their interaction with Calcium Aluminate Cement (CAC) and also silica fume. Influence on flow decay, strength and behavior upon heating are examined and discussed. The chemical aspect of the interactions has been followed by conductivity and calorimetry. 8 rue des Graviers 9 Neuilly sur Seine Cedex Tel : () Fax : () 6 7 9

3 Reference : TPGBRELAF Introduction Ladle castables containing spinel having been formed during heatup by sintering magnesia MgO and aluminaal O to MA spinel show reportedly /,,/ advantages over castables with preformed spinels what their slag resistance is concerned. This fact seems to be due to a homogeneous distribution of very fine spinel. This fine spinel is achieved by using a finely ground magnesia as one of the raw materials of the castable. On the other hand can a finely ground spinel disturb workability // and also lead to slaking upon heatup. Using deadburnt MgO helps definitively reduce the negative impact on workability together with adapted admixtures /,/. The slaking effect though, exists sometimes quite pronounced with all types of MgO. Kitamura /6/ has studied this phenomena and explains the slaking phenomena by formation of brucite Mg(OH) on the grain boundaries within a polycrystallin MgO particle. His so called dusting phenomena is schematically shown in Figure. Grain boundary destruction Disintegration into finer grains ( m) Hydrate Further disintegration into single crystals ( 6 m) Hydrate Hydration curve Grain boundary Figure. Hydration model of MgOPolycrystalls according to Kitamura In order to minimize this effect the following suggestions can be found in the literature: Quality of MgO: Few grain boundaries and impurities such as CaO, SiO, B O seem to diminish the tendency to slaking /,/. Surface coating: A surface layer acting as a diffusion barrier is proposed as a method to reduce the negative effects of slaking. Additives: Silica fume or silica flour /,6/ as well as aluminum lactate // have been proposed as additives reducing the slaking tendency of such castables. t It should not be neglected that the addition of or impurities in form from SiO or CaO can reduce the hot strength behavior. In this paper results are presented showing the influence of different MgO s and the impact of admixtures on placing behavior and slaking capacity in a spinel forming castable. 8 rue des Graviers 9 Neuilly sur Seine Cedex Tel : () Fax : () 6 7 9

4 Reference : TPGBRELAF Experimental approach.. Raw materials The raw materials used are given in Table I with details on the MgO grades used in Table II. Table I. Origin of raw materials Raw material Tabular alumina Spinel Reactive alumina Silica fume Cement Additives Code T6 AR78 CTSG 97U Secar 7 Dispersants: TPP, HMP, Darvan 7S, Castament FS Retarders: Boric acid, Citric acid Accelerators: Lithium carbonate, Sodium carbonate.. Systems Investigated Model castables representing spinel castables (SP) as well as spinel forming castables (SPF) have been used in this study. In order to study the Table II. Chemical composition of Magnesia studied M M M M M M6 M7 M8 M9 M MgO (%) CaO (%) SiO (%) 98,7 97, 9,8 99, 96,8 9,8 9, 97, 97, 97,9,69,9,9,,7,,,,9,7,,,,,9,,,, influence of MgO the binder composition (Material < m) without coarse aggregates has been used. All systems are summarized in Table III.,7 Table III. Composition of castables used Castable SP SP binder SPF SPF SPF binder Raw material Tabular alumina ¼ (mesh) , 9 7,8 Spinel (mm,,, Reactive alumina Silica fume Magnesia CAC 6 7, 8 8,,8 6,8 7,8 6,6 Mixing water,,,9 All formulations are used with admixtures. 8 rue des Graviers 9 Neuilly sur Seine Cedex Tel : () Fax : () 6 7 9

5 Reference : TPGBRELAF.. Testing... Placing characteristics Flow value The flow value has been determined using a cone with mm base diameter, mm high and 7 mm top diameter. The cone is placed on a vibrating or shock table (ASTM C), filled with the castable, then taken away and after blows (shock table) or seconds of vibration the diameter of the resulting "cake" is measured and considered as the flow value in mm. To express the flow value in percent we calculate : FV [%] = cake diameter initial diameter initial diameter Working time The time after which the initial cone would not flow at all under vibration is called working time. Flow decay Flow values measured as a function of time and displayed as a curve show with which speed an initial flow decays or can be maintained.... Slaking resistance The slaking resistance is measured by an autoclave test. The cylindric sample of cm is placed in a cylindric, tightly closing container. Some water is added ( ml) in order to insure a humid environment. The whole container is placed in an oven at C for different lengths of time, the standard exposure being hours. After the soak time the container is cooled down in ice. During soak at C a vapor pressure of, Mpa ( bars) is obtained. After cooling down the sample is taken out of the container and inspected visually. A resistance rating ( to ) is then given as detailed in Table IV. Table IV. Resistance ratings in autoclave test State of sample No changes visible Some cracks Cracks become grooves Sample in several pieces Total destruction... Physical and chemical tests Rating Calorimetry Calorimetry has been used for the binder compositions to trace interaction between the MgO, CAC and admixtures. After mixing, the binder is placed into the cell of a calorimeter which has been specially developed to measure the heat after mixing. Heat flow is associated with a chemical reaction occurring within the binder. Heat flow has been reported by other workers to be representative of castable properties /7/. Rheometry After mixing, the binder is placed into a rheometer. The rheometer (Contrives, Rheomat A) is normally composed of two coaxial cylinders, the binder being placed within the narrow space between the cylinders. In the present experiment, the inner cylinder was replaced by a special blade. The binder is placed within the outer cylinder after mixing. The propeller is rotated every minutes at rpm during seconds, and the momentum necessary is monitored. The result of the experiment is a value of the momentum every minutes. An increase of the momentum is associated with a stiffening of the binder. XRD Xray diffraction has been used to detect the formation of brucite Mg(OH) the phase associated with slaking. For data reduction the following peaks have been measured using a CopperK radiation with =,. Al O peak area between = 6, to 8, Mg(OH) peak area between = 8, to 9,98 8 rue des Graviers 9 Neuilly sur Seine Cedex Tel : () Fax : () 6 7 9

6 Reference : TPGBRELAF Results.. Basic findings Earlier investigations showed that MgO interacts with the other raw materials in an AlMg castable. Figure shows the chemical interactions as they can be observed with calorimetry. The heat flow curves of a ULCC castable show the apparition of an additional peak (pi) when MgO is added to the system. The position of this peak depends more from the admixture system employed in the castable than from the MgO fineness used. It s appearance shows that an exothermic reaction takes place which can disturb workability //. The peak representing massive precipitation, however, seems to be influenced by the MgO fineness leading to an earlier hydration for fine MgO. mw/g,8,6,, Package, pi TPP, pi Fine massive precipitation Coarse MgO, , min,6,8 ULCC w/o MgO Package = Darvan 7S, Citric acid, Sodium carbonate Figure. Heat flow curves for binder phase of castables containing MgO.. Workability of castables To better show the effect of fineness of magnesia M rheology curves momentum as a function of time are given in Figure. Momentum (mnm) WT = min WT = 6 min WT = 8 min Coarse MgO TPP Coarse MgO Package Fine MgO TPP Fine MgO Package ULCC w/o MgO Time (min) Figure. Rheometry curves for formulation SPF Working times which are all superior to minutes are also shown. We observe that a finer magnesia grading diminishes the working time of these castables. Also optimized admixture combinations allow to extend the working times /,8/. However, working times for SPF, which does not contain silica fume, did not exceed minutes even with optimized three component admixtures packages. This phenomena shows again that SF containing castables achieve more easily good placing behavior. This effect can also be observed for LCC castables with and without silica fume /9/. 8 rue des Graviers 9 Neuilly sur Seine Cedex Tel : () Fax : () 6 7 9

7 Reference : TPGBRELAF.. Magnesia hydration The brucite formation during autoclaving is taken as a measure of degree of slaking of the magnesias in binder formulation SPF. For magnesia M Figure shows the Mg(OH) /Al O peak ratio as a function of time in the autoclave. Mg(OH)/AlO Time (min) Figure. Brucite formation as a function of exposure in the autoclave With increasing time the MgO hydrates to Mg(OH) thus destroying the microstructure of the sample. After five hours the sample was destroyed corresponding to a rating according to Table IV. Figure shows a comparison of different magnesias after one hour autoclaving each. Mg(OH)/AlO,, M M M M M Figure. Brucite intensities for different magnesia after autoclave treatment Although the test conditions have been very severe with,% MgO in the binder phase it is very obvious that some of these magnesias seem to withstand better the heat treatment. For some selected magnesias some more tests on MgO concentration and autoclave exposure time have been conducted... Magnesia screening For different MgO s, quantities of,% up to % have been tested in the binder composition of SPF. The amount of tabular alumina < plus MgO has been kept constant to %. Table V gives the ratings for those magnesias after hours and hours of autoclave exposure. Additionally the magnesia s M and M8 have been submitted to the test after grinding. M6 M7 M8 M9 M Table V. Resistance rtings from the autoclave test for different MgO grades % M * M ground M M7 M8 M8 ground MgO h ** h h h h h h h h h h h,,,,8, * type of MgO ** exposure time 8 rue des Graviers 9 Neuilly sur Seine Cedex Tel : () Fax : () 6 7 9

8 Reference : TPGBRELAF As compared to Figure we observe that MgO s which failed in the very severe test (M and M8) can show quite a good performance in a composition closer to the real castable. It is MgO concentration and exposure time which will determine for a given MgO if destructive slaking will occur. Grinding of MgO will certainly either destroy a protective coating or introduce more grain boundaries and therefore diminish its resistance to slaking. The results in Table V suggest also that a given MgO should be tested in the final formulation through a heatup test in order to evaluate the impact on slaking. Conclusion an outlook Together with earlier experiences // and the results presented here we would like to outline the following conclusions. Workability Silica fume showed also for AlMg castables its benefits for improving working time and flow. Castables without SF tend to achieve shorter working times and need admixture packages. Only HMP or TPP will not be sufficient. Magnesia will by its hydraulic nature disturb the flow behavior. Slaking Magnesia will hydrate at heatup to brucite and therefore damage the castable. To reduce or suppress these damages different MgO s should be screened and carefully tested in formulations to find the amount allowable. Further studies The impact of SiO as a anti slaking agent at high temperature has to be further investigated. Slaking resistance as well as HMOR and CCS after heatup have to be tested for typical spinel forming castables. Acknowledgements The authors would like to thank all coworkers of Kerneos and Lafarge Central Research Laboratory who contributed through laboratory testing or discussion to this paper. // References A.Kameyasu et al; MgO raw material improved the hydration resistance ; UNITECR 9; Kyoto, 99. // A.Kaneyasu et al; Magnesia raw materials with improved hydration resistance ; Taikabutsu Overseas, Vol. 7, No. // A.Watanabe et al; Magnesia containing basic castables ; UNITECR 9; Kyoto, 99. // Th.A.Bier et al; Workability of calcium aluminate cement based castables containing magnesia ; ALAFAR; 996. // Y.Koga et al; Effects of alumina cement grades and additives on AluminaMagnesia dastables containing Aluminium Lactate ; Taikabutsu Overseas; Vol. 8, No. /6/ A.Kitamura et al; Hydration characteristics of magnesia ; Taikabutsu Overseas; Vol. 6, No. /7/ R.Krebs; Demands on unshaped refractories in the next decade ; UNITECR 9; Kyoto, 99. /8/ Th.A.Bier et al; Chemical interactions in Calcium Aluminate Cement based castables containing magnesia ; UNITECR 97. /9/ Th.A.Bier et al; The roles of fine silica and reactive alumina and their interactions with calcium aluminates in refractory castables ; TARJ; Sept // H. Fryda et al; Relation between setting properties of Low Cement Castables and interactions within the binder system (CAC FillersAdmixturesWater) ; UNITECR rue des Graviers 9 Neuilly sur Seine Cedex Tel : () Fax : () 6 7 9