Qualitative and quantitative investigation of post–mortem cement refractory: The case of magnesia–spinel bricks

Author’s Accepted Manuscript Qualitative and quantitative investigation of post – mortem cement refractory: The case of magnesia – spinel bricks Sahar Belgacem, Haykel Galai, Houcine Tiss www.elsevier.com/locate/ceri

PII: DOI: Reference:

S0272-8842(16)31609-1 http://dx.doi.org/10.1016/j.ceramint.2016.09.077 CERI13737

To appear in: Ceramics International Received date: 25 July 2016 Accepted date: 12 September 2016 Cite this article as: Sahar Belgacem, Haykel Galai and Houcine Tiss, Qualitative and quantitative investigation of post –mortem cement refractory: The case of magnesia –spinel bricks, Ceramics International, http://dx.doi.org/10.1016/j.ceramint.2016.09.077 This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting galley proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

Qualitative and quantitative investigation of post –mortem cement refractory: The case of magnesia –spinel bricks Sahar BELGACEMa, Haykel GALAIb, Houcine TISSc a

Application Laboratory of Chemical & Natural resources and environmental substances, Department of Chemistry, Faculty of Sciences Bizerte, 7021, Jarzouna -Bizerte –Tunisia b Laboratory for Analysis Methods and Techniques, INRAP, 2020 Sidi Thabet c Laboratoire Production, 7000. Les Ciments de Bizerte

Abstract Post –mortem investigation of damaged refractory bricks remains an essential source of information to predict the reactions responsible for corrosion phenomena. In this context, the corrosion of two post–mortem magnesia spinel bricks, taking from the sintering zone of a rotary cement kiln, was investigated qualitatively and quantitatively. The results of the X-rays diffraction, combined with the Rietveld method supported by heating microscope and elemental analysis, revealed that the sulphur and the high temperature (>1600°C) were the main factors which caused the degradation of the refractory matrix of magnesia – spinel bricks by the infiltrated mineralogical phases of the clinker melt. In fact, the appearance of the merwinite C3MS2, the akremanite Ca2Mg (SiO7), mayenite C12A7 and Yeelimite CA ̅ was the warning of the refractory matrix corrosion. Also, the deterioration rate of MgAl2O4 was more important in the hottest zone and it was deeper in brick 2 than in brick 1. As a result, the elasticity loss in brick 2 was more severe than in brick 1, which explains the length difference after a short duration of service (40 days). Keywords: Post –mortem; Magnesia –spinel brick; Corrosion; High temperature; sulphur; Rietveled method

1. Introduction The stability of the refractory lining in the sintering zone of rotary cement kiln is a key parameter to ensure a high-quality clinker and a good functioning of the process. In view of that, magnesia -spinel MgO -MgAl2O4 bricks are now frequently used to protect the burning zone because of their high thermal and chemical resistance. In fact, in this zone, the refractory lining is exposed to the thermo-chemical stresses resulting from a variety of corrosive agents: The change in the flame temperature, the material and gases, the amount and viscosity of the liquid phase, the contents of FeO, Fe2O3, Fe3O4, SiO2, the contents of alkalis and sulfur [1-8]. The kiln alignment, as well as his ovality, represents the mechanical stresses. In practice, different degradation modes are often associated.

Yet, under these conditions, basic refractory bricks can be degraded, thus, weakening their performance. Their refractory matrix is generally corroded by the clinker phases enriched with iron, calcium and aluminum, which diffuse into the structure of the brick and react with the spinel and the magnesia [1, 2]. Some studies have concluded that the periclase and the spinel do not behave similarly. For instance, the study of Jacek Szczerba [3] indicated that there is no reaction between the periclase MgO and different clinker phases. This is due to the high solubility of MgAl2O4 compared to that of MgO in CaO/Al2O3 slag at 1600 ° C [9]. Indeed, J.Poirier [9] reported that there is no dissolution of MgO when the liquid phase CaO/Al2O3 becomes saturated (≈ 19 wt% of MgO and T = 1630 ° C). Corrosion of refractories is a very complex phenomenon considering the number of parameters that are included. Hence, there is a multitude of corrosion mechanisms that involve different phases (solid, liquid and gas). There are two types of reactions: the elementary reactions [9] (dissociation, condensation, dissolution, precipitation) and complex reactions in polyphasic systems whose corrosion is via a liquid or gaseous phase and the formation of novel compounds [9]. As a result, many tests (the sessile drop, the finger, crucible tests, etc.) have been established to reproduce and simulate, as much as possible, the real furnace conditions in order to better understand the mechanisms corrosion of this materials [10]. However, post –mortem examinations of corroded and failed furnaces are still the most important source of knowledge. This is due to the importance of the local stress and equilibrium specific to a particular region. Thereby, several modules have been established to control the contents of each mineralogical phase clinker by setting the chemical composition of the raw material such as: silica ratio SR=SiO2/ (Al2O3+Fe2O3) and the alumina module AM =Al2O3/Fe2O3) [11]. Precisely, this tow ratio offer data about the liquid phase. Moreover, alkalis (Na, K) behave in a harmful manner to the refractory lining [7, 12], for the finished product and for the proper operation [13]. They affect directly the percentage of the liquid phase [3,14]. The content limit is expressed in alkalis Na2O equivalent (Na2O equivalent = Na2O + 0.659*K2O) [15, 16]. Besides, the alkali Sulphate Ratio (ASR) is used to predict the behavior of volatile components according to this equation: ASR= (Na+ K)/ (2S/ Cl) [1, 17]. The variation of one of the values of these modules outside the recommended limits (Tab.III) affects the quality of the clinker on the one hand, and engendered an additional stress on the refractory bricks on the other. In this context, two samples of a magnesia -spinel MgO -MgAl2O4 bricks with different length, taken from the burning zone of a new rotary cement kiln (of the company "Les Ciments de Bizerte", Tunisia) was investigated to highlight the relationship between the thermal and chemical stresses and their effect on the brick structure. Also, X –Rays powder diffraction combined with

Rietveld method were used for the mineralogical characterization of the refractory as a tool to evaluate the damage resulting from the corrosion.

2. Experimental procedure 2.1. Samples and preparation In this study, two magnesia – spinel bricks have been taken from the burning zone of a rotary cement kiln (4000 t/day) after 40 days of service. The position of the samples was under the flame between 3.6 m from the out of the kiln until 15 m. The length of bricks was reduced from 220 mm to 125.2 mm for the brick 1and to 84.2 mm for the brick 2. The cross section of samples (Fig.1) shows regions with different colors; based on this distinction, samples were divided from the hot side to the cold one as follows: The first brick was divided into three zones (noticed S1a, S1b and S1c) and the second was divided into four zones (noticed S2a, S2b, S2c and S2d) 2.2. Analytical techniques The chemical composition of the crusting (CR1and CR 2), bonded to the hot side of the brick, was obtained by X-rays fluorescence. Chemical analysis of damaged bricks was carried out by ICP -AES. Sulphur (S) and Carbone (C) were determined by C/S analyzer. The amount of silicon was obtained by gravimetric method. All samples were analyzed by X-ray powder diffraction using an ARL 9900 diffractometre operating with monochromated Co Kα1 radiation (1.788996 Ǻ) and patterns were recorded between 8° and 80° (2Ɵ). FTIR spectra were recorded over the 400-4000 cm-1 spectral range using Vertex 70 ATR Bruker Diament spectrophotometer with a resolution of 4 cm-1. The softening behavior of the crusting (CR1) was investigated by heating microscope at 10 ° C / min. 2.3. Rietveld analysis Rietveld refinement was done using the X’Pert High Score plus from PANalytical. Peak shape was modeled by a pseudo –Voigt and the background was fitted with a polynomial function. Parameters varied in refinements are: Scale factor, background coefficients, cell parameters, profile parameters. For all samples, periclase MgO presented a preferred orientation effect along [0 0 1] direction which was corrected by March –Dollase algorithm [18]. Table 1 collects phases used for Rietveld analysis and their Inorganic Crystal-Structure Database (ICSD) code.

3. Results 3.1. Analysis of crusting by heating microscope

It is important to know nearly the firing temperature during the functioning of the kiln for 40 days. This provides information about the endurance of the thermal stress that the refractory lining has supported. The softening behavior of crusting (CR 1) was studied by a heating microscope (fig.2). Results showed that the sintering started between 1300 ° C (fig.2-16- ) and 1400 ° C (fig.2-19- ). The softening point was determined at a temperature of 1606 ° C (fig.2-25- ). At 1650 ° C, the material is partially molten (fig.2-28- ). According to the previous results and the appearance of dense crusting, similar to lava and adhering strongly to the hot side of the brick, the temperature was higher than that required for the clinkerization (> 1450°C). 3.2. Chemical composition Crusting CR1 and CR2 of the brick 1 and 2, respectively, show similar composition (Table 2) of the four main oxides (CaO, SiO2, Al2O3 and Fe2O3). However, the amount of MgO is too high especially in the crusting CR2.The total of alkalis (K2O and Na2O) resides below 1%; and the sulphur was between 3.5% and 4.5%. Table 3 shows the chemical composition in weight percentages of the different zones of the two post –mortem bricks and of the new one. Compared to the chemical composition of the original brick, the contents of CaO and SiO2 slightly augmented expect in zone Sa1 and Sa2 where a significant increase was detected because of the interaction between the hot sides of the bricks and the cement clinker at elevated temperature. A small amount of alkalis (K2O, Na2O) and sulphur was detected in the other zones. However, in zone S1a of the brick 1, the contents of K2O, generated from the raw meal, and SO3, generated essentially from the fuel, significantly increased; compared to the new brick. The EDX coupled to MEB confirm the presence of sulphur and potassium in area S1a (fig.3). 3.3. Phases identification Analysis by XRD (fig.4) of crusting revealed the presence of main phases constituting the clinker which are C3S, C2S, C3A and C4AF. Also, the periclase MgO and Ca2Mg0.2AlFe0.6Si0.2O5 were observed as well as the trace of calcite CaCO3 in CR1. The XRD pattern for the original brick shows its mineragical composition: the major phases are periclase MgO and magnesia – spinel MgAl2O4 while the forsterite Mg2SiO4 and sodium calcium silicate Na2Ca6 (Si2O7)(SiO4)2 are the minor phases. All zones of the damaged bricks include the larnite C2S among their composition, but the abundance of this phase is much higher in zones S1a, S2a and S2b than in the others. The areas in direct contact with the melt of the clinker which are S1a and S2a of bricks 1 and 2, respectively, present

diffractogrammes (Fig. 5, 6) extremely different compared to the original brick. Yet, they contain the same crystalline phases newly formed in binary systems CaO–Al2O3, ternary CaO-Al2O3-SO3 and CaO-MgO-SiO2 as well as the presence of the arcanite K2SO4.The most important phases are mayenite C12A7 and yeelimite CA ̅. The similarity of the mineralogical composition can be observed throughout the brown color of the areas S1a and S2a. The diffractogram of the area S2b of the brick 2 reveals the absence of arcanite. The area S1b of the brick 1 has the same mineralogy as S1a, S2c and S2d. 3.4. Infrared spectroscopy FTIR spectrum of all zones (Fig. 7,8) relative to damaged bricks and the original one present absorption peaks at 488 cm-1 and 670 cm-1 resulting from the vibration of Mg –O in the spinel structure, the tow bands correspond, respectively, to the bending and asymmetric stretching modes [19]. The infrared spectrum (Fig.6) of the area S1a and S2a present a band at 840.73 cm-1 and 839.13 cm-1 corresponds to the vibration of the AlO4 mayenite [20], and one more at 1114.82 cm-1 related to the stretching mode of the [SO4] 2 - in Yeelimite [20].

4. Discussion 4.1. Study of crusting Calculating crusting clinker data (Table 4) gave a low alumina module for CR1 and CR2 (1.4 and 1.52, respectively) and a silica module (3.77 and 1.9, respectively). Furthermore, alkali equivalent Na2Oeq is less than 0.60 % for CR1, which means that the liquid phase is poor in alkalis (K2O, Na2O). In fact, the presence of these elements affects the fluidity of the liquid phase [21]. There is no guarantee that these values are representative, but they indicate that the clinker needs higher temperatures than that required for the clinkerization and forms a low viscosity liquid phase. Also, coating clinker data, especially those of CR2, marked the difficulty in layer formation causing instable coating. These results explain why the length of the tow bricks was not equal after 40 days of service. The penetration of the clinker melt through the pores of the brick gave rise to strong adhesion of the crusting with the hot surface. The strength of the bond is due to clinker and brick minerals as well as their reaction product. Consequently, the presence of MgO periclase and Ca2Mg0.2AlFe0.6Si0.2O5 in the crusting may be due to a direct dissolution of the refractory matrix by the liquid phase at a superior temperature. Practically, the temperature at the burning zone exceeds 900 ° C, which implies that the calcite should be fully decarbonated. But, the XRD indicated the presence of traces of calcite, probably coming from ash circulating through the kiln during its empting.

4.2. Study of the damaged brick 4.2.1. Characterization of the damage Corrosion of refractory matrix by the "liquid" silicate is intimately related to the change in composition / viscosity with time/ temperature and system atmosphere [21]. In our case, the service life of the bricks, object of this study, did not exceed 40 days. During this time, the refractory coating has been applied by temperatures more than 1450 ° C. The amount of the liquid phase increases with temperature [22], indeed, its viscosity is a function of temperature and over heat [23]. Also, the existence of molten salts and mineralizing cations essentially iron and aluminum increases the fluidity on the one hand [21], and affects the kinetic and the stability of reactions by generating a low eutectic point on the other hand [3, 23]. Thus, the penetration of the liquid phase through the open porosity of the bricks increases. In this case, the mayenite and yeelemite could be infiltrated in the bricks after their formation in the clinker liquid phase. Or, their emergence was the result of the interaction between the minerals of the brick and those coming from the clinker. Indeed, the dissolution of alite and spinel lead to the mayenite formation between 1000°C and 1350°C [24], probably, following the reaction (I) [4]. MgO.Al2O3 + 3CaO. SiO2  2CaO.SiO2 + 3CaO.Al2O3 + 12CaO.7Al2O3 + MgO (I) The mayenite could be generated also from the free lime according to the reaction (II) [4]. The XRD of crusting (Fig.3), which surmounts the hot side of the bricks, revealed the presence of the uncombined lime. MgO. Al2O3 + CaO  3CaO.Al2O3 + 12CaO.7Al2O3 + MgO (II) Sulphur liberated from the combustion of the petroleum coke. It’s moved with the hot gases up stream of the furnace where it condenses partially or totally in the form of anhydrite [13, 25]. Then, it is introduced with the raw meal in the kiln. Upon reaching high temperatures (1450 ° C), anhydrite dissociates [26]. Indeed, the chemical analysis of crusting (Table 2) revealed a content of sulfur equal to 3.81% in CR1 and 4.22% in CR2. The equations (III) and (IV) illustrate the possible reactions that have given rise to the formation of yeelimite Ca4Al6O12.SO4 from sulfur SO3, spinel MgAl2O4, the alite C3S and celite C3A [3]. The CA ̅ appears below 1200 ° C [22] and is totally disintegrated at temperatures above 1350 ° C [3, 22, 27]. 4Ca3SiO5 + 3MgAl2O4 + SO3  Ca4Al6O12.SO4+4Ca2SiO4 +3MgO (III) Ca3Al2O6 + Ca3SiO5 + 2MgAl2O4 + SO3  Ca4Al6O12.SO4 + Ca2SiO4 + MgO (IV)

In addition, the CA ̅ can be formed not only from gaseous sulfur but also from anhydrite CaSO4, free lime and spinel [3]. However, crusting does not contain anhydrite according to the result of XRD. In that case, there are two possibilities: either the CaSO4 was not involved in the formation of the yeelimite because of its volatilization, or it was present in the crusting as an intermediate phase, and it was totally disintegrated via the appearance of CA ̅. In fact, the atmosphere is rich with sulphur in the burning zone; also, the partial pressure of SO2 controls the dissociation of sulfate in the system (A + CaSO4). The equilibrium partial pressure of SO2 / CA ̅ is higher than that of SO2 / CaSO4. Thus, the volatilization of the sulphat will be delayed by the formation of yeelimite [22] according to the reaction illustrated by equation (V). CaSO4 + 3CaO + 3MgAl2O4  Ca4Al6O12.SO4 + 3MgO (V) It’s obvious that the presence of C2S is due to the infiltration of liquid phase. Yet, the reactions (I), (III) and (IV) have larnite C2S among their products, which explains the richness of the area S1a, S2a and S2b of the used bricks with this phase. The magnesium oxide MgO was released during the formation of the mayenite or yeelimite independently of the reaction. As a result, the interface between the hot zone of the bricks and the crusting is formed by C2S, MgO and the secondary liquid phase (mayenite + yeelimite) newly formed subsequent to the corrosion of the spinel MgAl2O4. The alteration of MgAl2O4 means a loss of elasticity of the brick. Considering that the layer is not stable and the refractory coating has been exposed to high temperatures, the hot sides (S1a and S2a) have lost their strength. Consequently, they have been eroded continuously by the burning material and the bricks tend to burst under mechanical loads and / or thermal shock. The interaction zones S1a, S2a contain three new phases formed in the ternary system CaO-SiO2-MgO which are merwinite Ca3Mg(SiO4)2, the akermanite Ca2Mg(SiO2)7 and monticellite CaMgSiO4. Furthermore, the MgO dissolved form with CaO-SiO2-Al2O3-Fe2O3 an eutectic less than 1350 ° C [4], which promoted the reactions illustrated by equations (IV), (IIV) and (IIIV) [28]. Liquid + MgO + Ca2SiO4  Ca3Mg (SiO4)2

(Merwinite)

Liquid + MgO + Ca3Mg (SiO4)2  CaMgSiO4

(Monticellite) (IIV)

(IV)

Liquid + Ca3Mg (SiO4)2+ CaMgSiO4  Ca2Mg (SiO2)7 (Akermanite) (IIIV) 4.2.2. Rietveld quantitative phase analysis The quantitative phase analysis by XRD coupled with Rietveld method was carried out for the original brick and the different zones of the damaged one. The observed, calculated and residual profiles of patterns related to the new brick and the damaged bricks (not shown) present a good match between the

measured and predicted diffractogrammes. Agreement factors: R expected, R profile, Weighted R profile, Goodness of Fit and R Bragg (Table 5) of all analysis samples confirm the good quality of the fitting procedure. March Dollase algorithm was used to correct the preferred orientation of the periclase along the [0 0 1] direction and the refined parameter was around 0.85. The results from Rietveld quantification match with those from the other techniques considering the errors introduced by each analyses technique (XRF, gravimetric method, sulphur analyser, ICP-AES) and all refinement parameters which can influence the final results. Fig 9 confirms the good agreement between ICP –AES and Rietveld analysis for the magnesium oxide. These results (Table 6) show a significant decrease in the amounts of the major phases (MgO and MgAl2O4) particularly in the area S1a, S2a and S2b, compared to those of the new one. Subsequently, the secondary liquid phase formed becomes rich in MgO from, both, the dissolution of periclase (MgO) and principally of spinel (MgAl2O4). However, the percentage of periclase in zone S1b (78.2%) and S2a (79.1%) were higher than that in the new brick (76.5%). This could be the result of diffusion and/or condensation of MgO into the cold sides. Rietveld analysis revealed that the amount of larnite Ca2SiO4, found in zones S1a, S2a and S2b, is equal to 5.1%, 6.6 and 5% respectively, which are much higher than that which exists in the other zones. This indicates that the rate of corrosion in the hottest areas was much important than that in the cold sides. Also, the mayenite and yeelimite are present with considerable amounts. Yet, the amount of spinel MgAl2O4 decreased to 10.2% in the area S1a and to 9.7% in S2a against 21.2% in the original brick. So, the presence of this phase was reduced almost to the half in the zones which are in direct contact with the hot material. In the quasi absence of sulphur into zone S2b, the percentage of spinel was reduced to 15%. This may provide evidence that the corrosion reactions of MgAl2O4 in S2b were only those producing the mayenite (1.1%). As a result the performance was decreased and the bricks were enabled to support the thermo –chemical overload. Therefore, controlling the amount of sulphur means the manage of corrosion reactions. In the same area, the amount of periclase remains almost unaffected and the merwinite, akermanite and montecellite are present with a small quantity. In reality, the values obtained by Reitveld method are standardized to 100% of crystalline fraction. They could be affected by errors introduced via many factors, particularly, the absence of standard, the preferred orientation (Periclase case) and the presence of amorphous phase. In fact, the crystallinity amount of MgO, introduced by the liquid phase, is small, which affects the value of periclase. The results make evident that the knowledge of the different mineralogical phase’s contents is an important parameter to determine and control the microstructure of the refractory. Thus, the performance of a refractory material

will be better evaluated and controlled. This is due to the fact that the desired characteristics of a given refractory reverse mainly to the properties of crystalline phases that recognized. 4. Conclusion The corrosion behavior of two post –mortem magnesia –spinel brick by cement Portland clinker was studied quantitatively and quantitatively. The lengths of the samples were different after 40 days of service in the burning zone of cement kiln. Indeed, the analysis by heating microscope showed softening and melting clinker crusting temperatures beyond 1600 ° C. According to this result and to the look of the crusting (very dense and similar to lava), the burning zone was exposed to temperatures above that required for clinkerization reactions. Under high temperatures, the presence of sulfur in the interaction zone of the bricks and the crusting has, probably, accelerated the corrosion phenomena which have involved the constituent phases of the clinker and of the refractory matrix brick. The X-rays diffraction combined with the Rietveld method showed that the appearance of merwinite C3MS2, akermanite Ca2Mg (SiO7) was probably related to the dissolution of periclase while the larnite, mayenite and Yeelimite was an indication of magnesium aluminum oxide degradation. These results showed the imperative role of the Rietveld quantification, as an efficient tool, to have command over the refractories corrosion. Acknowledgments “Les Ciments de Bizerte” is acknowledged for their support. References [1] Guangping Liu, Nan Li, Wen Yan, Changhe Gao, WeiZ hou, Yuanyuan Lia, Composition and microstructure of a periclase–composite spinel brick used in the burning zone of a cement rotary kiln, Ceram. Inter (2014) 8149 8149. [2] A.G.M.Othman, M.A. Serry, Attack of Magnesite–Chrome Refractories by Portland Cement Clinker, Am. Ceram. Soc. Bull, July 2005. [3] Jacek Szczerba, Chemical corrosion of basic refractories by cement kiln materials, Ceram. Inter. 36 (2010) 1877–1885. [4] Zongqi Guo, Stefan Palco, and Michel Rigaud, Reaction Characteristics of Magnesia–Spinel Refractories with Cement Clinker, Int. J. Appl. Ceram. Technol. 2 [4] 327–335 (2005). [5] Sara Serena, M. Antonia Sainz, Angel Caballero, Corrosion behavior of MgO/CaZrO3 refractory matrix by clinker, J. Eur. Ceram. Soc. 24 (2004) 2399–2406.

[6] V. V. Slovikovskii, Rotary kiln, corrosion-erosion-resistant linings, Refract. Ind. Ceram. Vol. 49. No. 2. 2008. [7] J.L. Rodr´ıguez-Galicia, A.H. de Aza, J.C. Rend´on-Angeles, P. Pena, The Mechanism of corrosion of MgO CaZrO3–calcium silicate materials by cement clinker, J. Eur. Ceram. Soc. 27 (2007) 79–89. [8] V. I. Shubin, The effect of temperature on the lining of rotary cement kilns, Refract. Ind. Ceram. UDC 66.041.571.043.1:666.94, Vol. 42, Nos. 5 – 6, 2001, pp. 216-22. [9] J. Poirier. M.L. Bouchetou. P. Prigent. J. Berjonneau, An Overview of Refractory corrosion: Observations, Mechanisms and Thermodynamic Modeling, Refract. Appl. Trans. 071007 (2007). [10] W. E. Lee and S.Zhang, Direct and direct slag corrosion of oxides and oxide-c refractories, VII International Conference on Molten Slags Fluxes and Salts, The South African Institute of Mining and Metallurgy, 2004, Pp. 309 319. [11] F. Sorrentino, Chemistry and engineering of the production process: State of the art, Cem. Con. Res. 41 (2011) 616–623. [12] J.H. Potgieter, R.H.M. Godoi, and R. van Grieken, A case study of hightemperature corrosion in rotary cement kilns, The Journal of The South African Institute of Mining and Metallurgy. November 2004. [13] Anjan K.Chartterjee. Chemistry and engineering of the clinkerization process-Incremental advances and lack of break throughs. Cem. Con. Res. 41 (2011) 624-641. [14] Jacek Szczerba, Ilona Jastrzębska, Zbigniew Pędzich, Mirosław M. Bućko, Corrosion of Basic Refractories in Contact with Cement Clinker and Kiln Hot Meal, J. Mat. Sc. Chem. Eng. 2014, 2, 16-25. [15] S. Nasrazadani. T. Springfield, Application of Fourier transform infrared spectroscopy in cement Alkali quantification, Materials and Structures. (2014) 47:1607–1615. [16] Inam Jawed and Jan Skalny, ALKALIES IN CEMENT: A REVIEW. Forms of Alkalies and Their Effect on Clinker Formation, Cem. Con. Res. Vol. 7. pp. 719-730. 1977. [17] Terry F. Newkirk, Effect of SO3 on the Alkali Compounds of Portland Cement Clinker, J. Res. National Bureau of Standards, Vol. 47, No.5, November 1951. [18] W. A. W. A. DOLLASE DOL, Correction of Intensities for Preferred Orientation in Powder Diffractometry: Application of the March Model, J. Appl. Cryst. (1986). 19, 267-272. [19] C.T. Mathew, Sam Solomon , J.K.Thomas, Structural, optical and vibrational characterization of infrared - transparent nanostructured MgAl2O4 synthesized by a modified combustion technique, 5th International Conference on Perspectives in Vibrational Spectroscopy, Materials Today: 2 ( 2015 ) 954 – 958. [20] L. Fernández-Carrasco, D. Torrens-Martín, L.M. Morales and Sagrario Martínez- Ramírez, Infrared Spectroscopy in the Analysis of Building and Construction Materials, Infrared Spectroscopy – Materials Science, Engineering and Technology, 2012, pp. 371 –382.

[21] W.E. Lee, D.D. Jayaseelan, S. Zhang, Solid–liquid interactions: The key to microstructural evolution in ceramics, J. Eur. Ceram. Soc. 28 (2008) 1517–1525. [22] Bruno Touzo, Karen L. Scrivener, Frederic P. Glasser, Phase composition and equilibria in the CaO –Al2O3 –Fe2O3 –SO3 system, for assemblages containing ye’elimite and ferrite Ca2 (Al, Fe)O5, Cem. Con. Res. 54 (2013) 77 86. [23]. M.RIGAUD, Corrosion of refractories and ceramics, Third Edition, Uhlig’s Corrosion Handbook, 2011, pp. 387 -398. [24] G. E. Gonçaves, G. R. Cota Pacheco,M.A.deMoura Brito, S. L. Cabral de Silva, V.de Freitas Cunha Lins, Influence of magnesia in the infiltration of magnesia –spinel refractory bricks by different clinkers, Metallurgy and materials, REM: R. Esc, Mina Ouro Preto, 68 (4), 2015. 409 - 415. [25] Gang –Soon Choiand F. P. Glasser, The sulphur cycle in cement kilns: Vapour pressures and solid –Phase stability of the sulphate phases, Cem. Con. Res.Vol.18,pp 367 -374,1988. [26] A. K. Chatterjee, Role of Volatiles in Cement Manufacture and in the Use of Cement, Advances in Cement Technology - Critical Reviews and Case Studies on Manufacturing, Quality Control, Optimization and Use, 2011, pp.203 -236. [27] J. Szczerba, Changes in basic bricks from preheater cement kilns using secondary fuels, Ind. Ceram.Vol.29.1/2009. [28] In-Ho Jung, Sergei A. Decterov, Arthur D. Pelton, Critical thermodynamic evaluation and optimization of the CaO–MgO–SiO2 system, J. Eur. Ceram. Soc. 25 (2005) 313–333. Figure captions

Figure. 1. Two cross-sections of the two post - mortem bricks taken from the burning zone: A / Brick 1, B/ Brick 2. Figure. 2. Crusting analysis (CR1) by heating microscope. Figure. 3. EDS analysis confirming the presence of K2SO4 phase in zone S1a of the brick 1. Figure .4. XRD pattern of crusting CR1 and CR2/ (*) Besswax was mixed with the sample as a binder. Figure. 5. XRD patterns of the zones S1a, S1b, S1c and the new brick. Figure.6. XRD patterns of the zones S2a, S2b, S2c, S2d and the new brick. Figure.7. Infrared spectrum of zones S1a, S1b, S1c and the original brick. Figure.8. Infrared spectrum of zones S2a, S2b, S2c, S2d and the original brick. Figure.9. Comparison of the results achieved by Rietveld analysis and ICP –AES elemental composition for the MgO. Table 1 Some structural details of investigated phases used for Rietveld analysis

Phases

Formula

Crystal system

ICSD Codes

PDF Codes

Year

Periclase

MgO

Cubic

60692

01-077-2364

1999

Spinel

MgAl2O4

Cubic

171879

01-075-1796

2006

Forsterite

Mg2SiO4

Orthorhombic

27536

01-074-1685

1984

(Mg1.84 Fe0.16)SiO4

Orthorhombic

66491

01-079-1491

1994

(Mg0.54 Fe0.46)SiO3

Monoclinic

86468

01-089-1595

1999

(Mg1.014Fe0.986)Si2O6

Orthorhombic

40650

01-088-1914

2007

Montecellite

CaMgSiO4

Orthorhombic

202284

01-084-1323

1998

Merwinite

Ca3Mg(SiO4)2

Monoclinic

26002

00-026-1064

2007

Akermanite

Ca2Mg(SiO7)

Tetragonal

67691

01-079-2425

1994

Larnite

Ca2SiO4

Monoclinic

79552

01-083-0462

2006

Mayenite

(CaO) 12(Al2O3)7

Cubic

6287

01-070-2144

2007

Yeelimite

Ca3Al6O12.CaSO4

Cubic

80361

00-016-0440

2006

K2Mg(SiO4)

Orthorhombic

83226

01-087-1487

1998

K2SO4

Orthorhombic

79778

01-083-0682

2006

Na2Ca(Si2O7)(SiO4)2

Monoclinic

69122

01-080-1296

1992

Arcanite

Oxides

Table 2 Chemical analysis of crusting CR1 and CR2 (Wt %) PF CaO SiO2 Al2O3 Fe2O3 P2O5 MgO TiO2

K2O

Na2O

SO3

CO2

Cl-

CR 1

2.65

50.80

17.15

4.55

3.25

0.56

13.33

0.25

0.42

0.15

3.81

1.25

0.02

CR 2

1.56

48.52

16.16

5.13

3.36

0.56

18.43

0.20

0.56

0.30

4.22

0.6

0.01

Table 3 Chemical composition of damaged bricks and the original one B New Brick 1 Brick 2 S1c

S1b

S1a

S2d

S2c

S2b

S2a

MgO

85.54

84.06

83.96

78.98

84.73

82.91

81

79.7

Al2O3

10.92

9.68

8.85

8.72

8.76

9.52

9.38

8.38

Fe2O3

0.36

0.56

0.88

0.58

0.83

0.72

CaO

0.92

1.53

1.37

4.04

0.68

0.56

2.69

4.63

SiO2

0.88

1.22

1.57

3.87

2.61

1.83

3.02

4.26

Na2O

0.13

0.36

0.30

0.44

0.94

0.48

0.4

0.39

K2O

0.21

0.44

0.58

0.79

0.39

0.59

0.57

0.61

SO3

0.042

0.052

0.096

1.32

0.42

0.47

0.41

0.95

Table 4 Some clinker data of crusting CR1 and CR2 Modules Croûtage Module

CR1

1.40

0.54

0.69

Marges de valeurs optimales pour ciment Portland normal 1.5 - 2.5

d’alumine (AM)

CR2

1.52

Module d’acide silicique (SR)

CR1

3.77

CR2

1.90

Equivalent alcalin (EA) (%)

CR1

0.42

CR2

0.66

2.5 - 3.5 < 0.60

Table 5 Agreement indices related to analysis of original brick, brick 1 and brick 2 Samples R Exp Rp R wp GOF RBragg New brick Brick 1

Brick 2

2.55 2.57 2.60 2.64 2.64 2.56 2.72 2.60

S1c S1b S1a S2d S2c S2b S2a

5.87 5.07 5.07 4.64 5.89 5.34 5.49 4.90

7.71 6.95 6.99 6.23 7.58 7.58 7.12 6.41

9.12 7.30 7.21 5.56 8.19 8.76 6.84 6.04

Min

Max

1 1.04 0.89 1.01 1.18 0.60 1.07 0.59

3.87 3.79 4.80 3.33 4.76 4.77 4.67 3.95

Table 6 Mineralogical composition of new brick, Sa, Sb and Sc, as determined by Rietveld Method Mineralogical Composition Brick 1 Brick 2 New Brick 76.5

S1c 75.9

S1b 78.1

S1a 71.7

S2d 79.8

S2c 78

S2b 76

S2a 72.9

MgAl2O4

21.2

19

17

10.2

13.9

18.4

15

9.7

Mg2SiO4

0.4

1.1

1.2

1.5

0.9

0.2

0.8

0.8

(Mg1.84 Fe0.16)SiO4

-

0.9

0.4

-

0.9

0.2

-

-

(Mg0.54 Fe0.46)SiO3

-

2.1

2

-

2.9

2.4

-

-

(Mg1.014Fe0.986)Si2O6

-

-

-

1.7

-

-

0.2

0.8

CaMgSiO4

-

-

-

0.4

-

-

0.3

0.4

Ca3Mg(SiO4)2

-

-

-

1.5

-

-

0.5

1.7

Ca2Mg(SiO7)

-

-

-

2.2

-

-

0.7

2

Ca2SiO4

-

0.6

0.8

5.1

0.8

0.5

5

6.6

(CaO) 12(Al2O3)7

-

-

-

1.8

-

-

1.1

1.9

Ca3Al6O12.CaSO4

-

-

-

2.5

-

-

0.4

2 .4

K2Mg(SiO4)

-

0.4

0.5

-

0.8

0.3

-

-

K2SO4

-

-

-

1.5

-

-

-

0.8

1.9

-

-

-

-

-

-

-

Phases MgO

Na2Ca(Si2O7)(SiO4)2

Figure 1

Figure 2

New brick

S1a

S2a

Figure 3

C S

CR1 CR2

2

C A

C AF

MgO

CaO

3

4

4

1,6 10



C S

3

 K SO 2

CaCO

4

Ca Mg 2

AlFe

0.2

Si

0.6

3

O

0.2

   

5

Beeswax (*) 4

1,2 10

   

8000

  









    









      



      

 





4000

  16.23

24.36

32.49

40.62 Pos. [°2Th.]

48.75

65.01

56.88

73.14

Figure 4 4

1,2 10

MgO 2

2

2

1 10

Ca SiO 2



2

3

4

4

2

4 2

12

6

2

7

Fe

1.014

7

Ca Al Si SO 4 2 12 4 (CaO) (Al O )

Na Ca (Si O )(SiO )

 (Mg

2

Ca Mg(SiO )

4

Mg SiO 4

 CaMgSiO  4  K MgSiO 

 Ca Mg(Si O )

MgAl O

4 2

)Si O

0.986

2

6

2



New brick S1c S1b S1a



4

K SO 2



4

3

(Mg

(Mg

Fe

)SiO

Fe

)SiO

0.54

1.84

0.46

0.16

3

 4



8000





6000



   

4000

 



2000



 

  



 











    



16

24

32

40

Pos. [°2Th.]

48

Figure 5

56

64

72

80

4

1,2 10 MgO 2

4

Mg SiO 2

Ca SiO 1 10

2

4

4

4

2

2

2

 (Mg

2

7

Fe

1.014

(Mg

0.54

(Mg 8000

6

0.46

Fe

1.84





4

6

3

)SiO

 

4





6000

2

)SiO

0.16



4 2

)Si O

0.986

Fe

New brick S2d S2c S2b S2a



4

K SO

4

Na Ca (Si O )(SiO )





Ca Al Si SO  2 2 7 4 2 12 4 Ca Mg(SiO ) (CaO) (Al O ) 3 4 2 12 2 3  CaMgSiO  K MgSiO

 Ca Mg(Si O )

MgAl O

 





    

 



  



  





4000

2000 

16

24

32

40

Pos. [°2Th.]

48

56

64

72

80

Figure 6 140

S1a 1114.76 S1b

120

840.73 S1c 1433.62 New brick

100

Transmittance

996.5

80

670 60

40

488.76 20

500

1000

1500

2000

W num (cm-1)

Figure 7

2500

3000

S2a 1114.76

1,2

1433.39 S2b

839.18

995.98

S2c S2d New brick

Transmittance

1

0,8

670

0,6

0,4

500

1000

1500

W num (cm-1)

2000

2500

3000

Figure 8 100 Rietveld analysis ICP-AES analysis

MgO (% Wt)

80

60

40

20

0

New brick

S1c

S1b

S1a

Figure 9

S2d

S2c

S2b

S2a