Hydroxylation Studies on High-Solid Load Magnesia Aqueous Suspensions

Abstract The magnesia (MgO) hydroxylation behavior in dilute suspensions (below 50% volumetric solid loads) has been extensively studied over the past decades due to its role in refractory castables. However, its equivalent effects on concentrated systems have not been analyzed in such a systemic way, although they are known to be as or more deleterious than those observed in dilute systems. This study focuses on the hydroxylation behavior of different sources of magnesia (sinter and caustic magnesia) in aqueous suspensions prepared at various solids concentrations (17-96 vol%) and shaped by distinct methods. They were analyzed by thermogravimetry, apparent volumetric expansion measurements, X-ray diffraction, scanning electron microscopy, and in situ temperature measurements during curing. The ratio between experimental and theoretical extents of the hydroxylation degree resulted in the reaction yield. A comparison between samples containing the same water amount revealed those with caustic magnesia showed a faster evolution of hydroxylation degree, apparent volumetric expansion, and higher maximum internal temperature during curing. In both systems, the yield levels of compositions of heavier solid loads were higher, despite the small quantity of hydroxylation products formed. Significant differences in the products’ microstructure were observed and related to the ions' mobility toward crystallization nuclei.

Previous studies on the impact of intrinsic variables on MgO hydroxylation have reported a significant temperature increase during the process 2,26,28,29 , which is related to the excess of energy contained in the dissolved MgOH + ion released as heat after the precipitation of Mg(OH) 2 particles 17,33 .In some cases, such an effect accelerates the reaction and is more intense at high curing temperatures and in samples of larger volume 29 .Other reports have suggested MgO suspensions prepared with distinct solid loads may display different hydroxylation behaviors 28 .In such systems, the balance between MgO being dissolved and the precipitation of Mg(OH) 2 may be affected by the large quantity of water to be heated in the spaces amongst solid particles.Although such aspects were deeply studied in dilute aqueous suspensions, such as refractory castables (10-50 vol% of solids for selfflow 1,3,4 and 50-80 vol% for vibrate ones 49,50 ), they remain unexplored for more concentrated systems such as refractory mortars (80-95 vol% of solids) 6,8,12 and pressed pellets and bricks (above 95 vol% of solids) 1,2 .Even though these classes of pre-shaped refractories contain little or no water in their original formulations, hydroxylation reactions can occur due to contact with atmospheric moisture 14,15 , hydraulic cementing agents 9,10 , or layers of spray-applied refractory concrete for maintenance repair 1,2 .Because such situations are frequent in steelmaking industries and their potential mechanical damage can lead to long equipment idle time, the study of MgO hydroxylation in highly concentrated suspensions can have a significant technological impact.
The present study analyzed the hydroxylation behavior of two sources of magnesia, namely magnesia sinter (or hard-burnt magnesia) and caustic magnesia (or dead-burnt magnesia) of similar chemical composition and average diameter in aqueous suspensions prepared at various solids concentrations (from 17 vol% up to 96 vol%).MS is typically obtained by sintering pellets of magnesium hydroxide (Mg(OH) 2 ) or carbonate (MgCO 3 ) at temperatures as high as 1800ºC, which generate very dense structures with large and well-built crystals and practically no significant surface contamination by (CO 3 ) 2-and (OH) -ions (Figure 1a) 1,2,5 .After milling and sieving, the particles attained show a low specific surface area and practically contain no internal grain boundaries 28 .Consequently, their chemical reactivity and hydroxylation rate are low under testing temperatures below 100ºC.On the other hand, CM is attained as a by-product of sinter production at sleeve filters that retain fines at furnaces' overflow 5 .Because of the lower temperatures involved (700-1000ºC), the crystalline structure of the MgO attained after the MgCO 3 decarbonation remains highly defective and its particles show a much higher specific surface area and reactivity due to the large fraction of mesopores and numerous cracks formed during gas evolution (Figure 1b) 23,27,28,40 .
Such raw materials were selected because they also largely differ in their ability to form castable self-flow suspensions or thick pastes that require pressing to be shaped 28,29 .Their hydroxylation behaviors were investigated by thermogravimetric analysis (TGA), X-ray diffraction (XRD), scanning electron microscopy (SEM), and in situ temperature measurements during curing tests.Concomitantly, the volumetric expansion that follows such reactions was evaluated by apparent volumetric expansion (AVE) measurements 6,7,28,29,50 .
Mixtures of MS or CM and twice-distilled water (ionic conductivity of 0.07 µS/cm, at 25 ± 0.5ºC) were prepared with different solid loads and shaped by uniaxial pressing or direct casting (Table 1).To prepare pressed samples, water was slowly sprayed by a peristaltic pump (Masterflex, LS77201-60, USA), working under 0.01 cm 3 .s - constant flow, and connected to an ultrasonic nozzle, inside a closed-vessel propeller blender (operating at 500 rpm for 5 min) containing magnesia particles.After mixing, wet particles were sieved (D Part < 100 μm) to ensure optimum homogenization of the mixture, and uniaxially pressed (40 MPa, 60 s) as 40 mm diameter per 40 mm height cylinders.For the directly cast samples, magnesia particles and water were mixed in a paddle mixer (PowerVisc, Ika, Germany) at 1000 rpm for 5 minutes.The suspensions attained were cast in thin nonadherent polymeric molds (Figure 2a; pressed samples were placed in similar molds after extraction and demolding).
After the samples had been shaped or cast, a thin K-type thermocouple was inserted at their centers and half-heights for monitoring the inner temperature during hydroxylation tests (Figure 2b) 7,28,29 .They remained in sealed flasks in an environment of close to 100% relative humidity and The AVE parameter indicates the level of damage caused by hydroxylation expansion to ceramic structure and a close relationship with hydroxylation degree (W H Exp , described ahead) and loss of mechanical strength and rigidity.AVE's most important characteristic is it is continuously measured for the same sample at any time interval required.A detailed explanation of such a technique and its uses can be found elsewhere 6,50 .
Equivalent samples were removed from hydroxylation tests every 24 h, crushed, sieved (D Part < 100 μm), and dried overnight at 120ºC under vacuum for the removal of unreacted water.After weighting (M H , g), they were calcined at 1000ºC for 5 h to fully dehydroxylate Mg(OH) 2 and weighed again (M C , g).Equation 5provided, respectively, experimental (W H Exp ) and maximum theoretical (W H Theor ) values of hydroxylation degree (W H , wt%) attained for each combination of magnesia source and water.The 0.447 therm is a numerical adjustment, based on the molar mass values of MgO (40.303 g.mol -1 ) and Mg(OH) 2 (58.318 g.mol -1 ), for making W H vary from 0, when there is no reaction, up to 100%, for stoichiometric hydroxylation reactions 1,7,28,29,40 .
( ) W H Exp indicates the extension of the hydroxylation reaction, whereas the W H Theor represents the maximum hydroxylation degree to be attained if the reaction occurs stoichiometrically for each particular formulation.As an example, in a system containing 1 mol of MgO and 0.5 mol of water, the maximum theoretical hydroxylation degree that can be attained is 50% because there is not enough water to fully consume MgO.Experimentally, on the other hand, hydroxylation degree levels lower than 50% can be observed for the same system during the first hours of testing, for low-reactivity MgO sources and low-temperature testing conditions.Therefore, the W H Exp / W H Theor ratio (ranging from 0 up to 100%) can be adopted for the evaluation of the reaction yield.

Results and Discussion
Caustic magnesia (CM) and magnesia sinter (MS) exhibited significantly different hydroxylation behaviors for the same testing time in all concentrations tested.In general, CM-containing samples showed a faster evolution of hydroxylation degree (W H Exp ), apparent volumetric expansion (AVE) (Figure 3), and more intense heating (T Max ) above testing temperature (Figure 4) in comparison to MS-containing ones.Previous studies have reported similar results, explained through the microstructure of the MgO particles 28,29,40 , as discussed in the first section.
Regarding the yield of hydroxylation reactions (Figure 5), all CM-containing samples showed W Exp /W H Theor ratios above 94%, indicating the reactions consumed practically all water or all MgO available towards forming Mg(OH) 2 .On the other hand, only concentrated suspensions (81-99 vol% of MgO) provided an above 90% yield for the MS-containing samples.The understanding of such differences requires analyses of MgO hydroxylation as a two-sequential-step mechanism, i.e., dissolution of MgO and precipitation of Mg(OH) 2 [20][21][22][23]26 . Aftr the initial stages of hydroxylation, the precipitation of Mg(OH) 2 particles tends to block the unreacted MgO surfaces reducing their dissolution rate, hence the overall speed of the process 20,21,23 .Such behavior occurs more intensely in diluted suspensions due to the large space for accommodating the hydroxylation products.In high solid load suspensions, on the other hand, the water available is located mainly at the particles' surface rather than in the empty spaces amongst them.Consequently, the dissolution step can occur with no kinetic barriers imposed by precipitated products 31,42 .After all water in the mixture has been consumed, the reaction stops, and a high yield is attained despite the small amount of Mg(OH) 2 produced.
Another important aspect of the results is the maximum temperature achieved by the samples during the tests (Figure 4).Whereas, the maximum temperature for MS-containing samples was 5ºC above the testing one, over 40ºC increases were observed for CM-containing samples, particularly for castable suspensions.Such an exothermic event is related to the excess of free energy contained in the dissolved MgOH + ions during the saturation period and which is released after the precipitation of Mg(OH) 2 particles 7,26,33 .Because CM intense dissolution takes place at the first moments of contact with water, saturation and precipitation steps also occur in a short time 26,2829 .Since the heat evolution rate is faster than its withdrawal from the system, the samples' inner temperature rises and increases the speed of the whole reaction in a self-catalytic process.Interestingly, for both systems, samples of high (MS: 1.1-8.2wt%; CM: 4.3-18.3wt%)or low (MS: 40.1-57.3wt%; CM: 57.3 wt%) MgO content showed less intense temperature increases than the stoichiometric composition.In the first case, the short extension of the MgO hydroxylation reaction released small energy content, whereas, in the latter, the excess water consumed a significant part of the energy to be heated a few degrees above the testing temperature.
According to strong linear trends between the experimental hydroxylation degree (W H Exp ) and the mass ratio between crystalline phases present (MgO and Mg(OH) 2 , determined by the Rietveld method) (Figure 7), the higher the W H Exp values, the smaller the quantity of unreacted MgO after the hydroxylation test.Nevertheless, the morphology of precipitated Mg(OH) 2 particles varied significantly in the function of the MgO source and solid load in the suspensions (Figure 8).For MS-containing samples prepared with small amounts of water (Figures 8a-b), Mg(OH) 2 precipitation occurs initially over MS surfaces due to its very low solubility 20,22,31 .The tensile efforts from the density mismatch between the materials led to the exposure of unreacted MgO surfaces as elongated rods of triangular cross-sections 28 .The Mg(OH) 2 particles formed detached and remained as thin irregular clusters amongst the prismatic rods.Regarding suspensions prepared with higher water content (Figures 8c-d), ions showed greater mobility during the tests, and more hydroxylation products were formed.Therefore, the microstructure of hydroxylation products consists of large plate-like hexagonal crystals of    Mg(OH) 2 grown over each other and that cover a significant fraction of unreacted and fractured MS particles.
On the other hand, CM suspensions produced hexagonal plate-like particles of Mg(OH) 2 whose average diameter and shape regularity increased from concentrated suspensions (Figures 8e-f) to more diluted ones (Figures 8g-h).Such an effect is typically observed in particles attained from controlled precipitation of dissolved ions 33,34,37,40 .In those processes and for concentrated solutions, nucleation is the main free-energy-lowering mechanism, leading to a rapid formation of clusters of thin particles.Oppositely, diluted ion solutions display a lower driving force for nucleation and, consequently, each crystallization nucleus grows intensively and shows regular geometry, following a most stable habit.Similar to the behavior of MS suspensions, the higher the water content, the larger the Mg(OH) 2 particles formed and the more regular their geometry.
From a technological point of view, the results somehow explain why MgO hydroxylation produces such different effects in the function of the ceramic structure tested 1,2,5 .For sintered MS bricks, for instance, a small quantity of water from air humidity can produce high-yield hydroxylation reactions at exposed surfaces, resulting in severe cracks and dusting, even if a small portion of Mg(OH) 2 has been formed 1 .On the other hand, a large amount of water in MgO-containing selfflowing castables prevents the effects of heat release and the spaces amongst particles are wide enough to accommodate hydroxylation products 50 .Therefore, even if a significantly larger quantity of Mg(OH) 2 is formed, its macroscopic effects (e.g., AVE and cracks) are less intense than those on bricks.
Nevertheless, CM can be used in applications that require its full hydroxylation even under low availability of water such as soil corrector 5,26 .

Conclusions
The MgO sources (caustic magnesia, CM, or magnesia sinter, MS) exhibited distinct hydroxylation behaviors when tested as aqueous suspensions of different solid loads shaped by uniaxial pressing or direct casting.In general, the high chemical reactivity of caustic magnesia particles resulted in hydroxylation reactions of higher-yielding and greater apparent volumetric expansion (AVE) in comparison to equivalent suspensions containing MgO sinter.
For the same source of MgO, those compositions with higher solid content showed hydroxylation reactions with higher yields despite the lower total amount of Mg(OH) 2 formed and the lower levels of AVE observed.Such suspensions exhibited a self-catalytic behavior according to which hydroxylation reactions occurred directly at the surfaces of MgO particles, exposing unreacted material and promoting significantly higher heat evolution.On the other hand, in diluted compositions, the precipitation of Mg(OH) 2 at the surfaces of unreacted MgO particles reduced their dissolutionprecipitation and heat evolution rates, hence the overall reaction speed and yield.
Although all hydroxylation tests resulted in mixtures of different proportions of MgO and Mg(OH) 2 after drying, the microstructure of their products significantly differed.Diluted castable suspensions of CM produced Mg(OH) 2 particles of the highest regularity in size and shape with a small amount of unreacted MgO, whereas pressed MS led to compacts of fragmented MgO particles surrounded by Mg(OH) 2 clusters.

Figure 1 .
Figure 1.Microstructure of as-received a) magnesia sinter (MS) and b) caustic magnesia (CM) particles employed in this study.

Figure 2 .
Figure 2. Schematic representation of a) samples employed for hydroxylation tests with details on b) inner temperature monitoring and analysis during the curing period and c) molds for apparent volumetric expansion (AVE) measurements.

Figure 3 .
Figure 3. Evolution of apparent volumetric expansion (AVE) and experimental hydroxylation degree (W H Exp ) for samples prepared with different contents of a-b) magnesia sinter (MS) and c-d) caustic magnesia (CM) during hydroxylation tests (up to 7 days at 60ºC).

Figure 4 .
Figure 4. Combined effects of varying solid load and shaping process on the maximum temperature observed during hydroxylation tests for samples prepared with a) magnesia sinter, MS, or b) caustic magnesia, CM (after 7 days at 60ºC).

Figure 5 .
Figure 5. Combined effects of varying solid load and shaping process on hydroxylation degree and yield during hydroxylation tests for samples prepared with a) magnesia sinter, MS, or b) caustic magnesia, CM, (after 7 days at 60ºC).

Figure 7 .
Figure 7. Relationship between the amount of unreacted MgO (quantified from XRD results, Figure 6) and experimental hydroxylation degree observed after hydroxylation tests (W H Exp , Figure 3, after 7 days at 60ºC).

Figure 8 .
Figure 8. Combined effects of varying solid load and shaping process on magnesia sinter (MS) or caustic magnesia (CM) particles' microstructures attained after hydroxylation tests (after 7 days at 60ºC).

Table 1 .
6,50acteristics of the magnesia sources and compositions tested.Materials Research 60ºC ± 0.5ºC for up to 168 h6,50.During that period, their external dimensions (height, H i , and diameter, D i , in mm) were measured every 24 h (Figure2c-d and their external volume (V i ) and apparent volumetric expansion (AVE, %) were calculated by Equations 3 and 4, where lower indices 0 and E indicate, respectively, the initial condition and the condition after a certain hydroxylation period and t is the thickness of each mold (in mm).