2007:224 CIV

M ASTER’S TH E SI S

Study of the reactions between iron ore slag and refractories

Julien Zimmermann

MSc PrograMMeS in engineering Materials Technology (eeigM)
Luleå University of Technology Department of Applied Physics and Mechanical Engineering Division of Material Mechanics
2007:224 CIV • ISSN: 1402 - 1617 • ISRN: LTU - EX - - 07/224 - - SE

Summary:
For the fabrication of iron ore pellets at LKAB (Luossavaara-Kiirunavaara AB), one of the methods used is to sinter the pellets in a rotary kiln. This Kiln has a drift temperature of 1250°C in average. During this process a coating, causing the wear of the refractories materials, is formed, inducing expensive maintenance periods. The aim of this work is to find the reaction that takes place between the Slag and the bricks, in order to find the parameters that have to be changed to improve the bricks lifetime in the kiln. The use of bricks with higher alumina content is one of the solutions, but it is expensive. The influence of Alkalis and especially Na2O has been studied before in inert atmosphere. In this work it will be studied in air as well as the influence of the alumina content of the brick. Experiments were performed in air atmosphere, using powder samples that have been compressed into pellets. The devices used to characterize the reaction and the product of the reaction were: Kanthal furnace, X-ray diffraction (XRD), dilatometer, differential scanning calorimetry and thermogravimetry (DSC/TG) in-situ mass spectrometer, Scanning electron microscope (SEM) and Energy dispersive spectrometry (EDS). The XRD experiments permitted to show the presence of a glassy phase after a heat treatment of the samples containing alkalis at 1350°C. Dilatometer experiments gave more information about the influence of alkalis; it catalyzed the sintering reaction. The DSC/TG/spectrometer permitted to show three main reactions. One at 1175°C for the samples containing a low content of alkalis and a low content of Alumina. For the samples with addition of 10wt% of Na2CO3, and 20wt% and more of alumina to the brick, a reaction is observed at 1235°C and one with a maximum at 1300°C. The SEM pictures and EDS show the formation of different phases. Unfortunately it was only possible to identify with certitude the Hematite, the Corundum and a glassy phase. The reliability off the EDS measurements is low, due to the oxides used and a special correction of the data had to be done. The conclusions of the study are: For the samples with a small amount of alkalis, hematite dissolves mullite at 1175°C and forms Fe2O3Al2O3 and a phase with a nepheline composition (Na2O.Al2O3.2SiO2) that melts at 1235°C. For the samples with high content of alkalis, all the mullite is dissolved. Silica coming form the decomposition of mullite enters completely the melt and needles crystallize in the melt around 1300°C. Brick with high alumina content start to react with the slag at 1235°C to form needles around 1300°C.

-

-

In a future work, WDS or TEM will give a higher reliability than EDS, according detection of the phases.

0

Table of contents:

Summary: ................................................................................................................................... 0 Table of contents: ....................................................................................................................... 1 I Theoretical background ........................................................................................................... 3 I.1) Introduction ..................................................................................................................... 3 I.2) Processing of iron ore pellets at LKAB........................................................................... 3 I.3) Materials involved ........................................................................................................... 6 I.3.1. Bricks ........................................................................................................................ 6 I.3.2. Slag ......................................................................................................................... 10 I.3.3. Na2CO3 .................................................................................................................. 10 I.3.4. Alumina (Al2O3) ..................................................................................................... 11 I.4) Previous work ................................................................................................................ 12 I.4.1 Microstructural studies on the growth of FeOx on Al2O3-SiO2 based brick insulation materials .......................................................................................................... 12 I.4.2 Slag fixation on the Svappavaara’s rotary Kiln ....................................................... 13 I.4.3 Infiltration of iron slag into refractory bricks .......................................................... 13 I.4.4 Literature review ...................................................................................................... 14 II Experiments .......................................................................................................................... 16 II.1) Experimental procedure ............................................................................................... 16 II.2) Devices used................................................................................................................. 17 II.1.1 Kanthal furnace....................................................................................................... 17 II.1.2 XRD (X-ray diffraction) ......................................................................................... 18 II.1.3 Dilatometer ............................................................................................................. 20 II.1.4 Differential Scanning Calorimetry / Thermogravimetry / Mass spectrometry:...... 21 II.1.5 Scanning Electron Microscope (EDX) ................................................................... 23 II. 3) Sample Preparation ..................................................................................................... 25 III.2.1 Powders preparation .............................................................................................. 25 III.2.2 Pellets preparation ................................................................................................. 27 III.2.3 Sample preparation for XRD: ............................................................................... 27 III.2.4 Sample preparation for SEM:................................................................................ 28 III Results and discussion......................................................................................................... 29 III.1) Structure and phases analysis: .................................................................................... 29 III.1.1. X-ray diffraction................................................................................................... 29 a) Results of the X-rays diffraction analysis: ............................................................... 29 b) Discussion: ............................................................................................................... 29 III.1.2. SEM pictures ........................................................................................................ 30 a) Samples with different Na2O content:...................................................................... 30 b) Samples with different alumina content of the brick: .............................................. 33 III.1.4. results of the EDS analysis:.................................................................................. 37 III.1.5. Discussion about the results of the ESD and SEM analysis: ............................... 37 III.2) Thermal analysis ......................................................................................................... 40 III.2.1. Kanthal furnace .................................................................................................... 40 a) Results: ..................................................................................................................... 40 b) Discussion: ............................................................................................................... 40 III.2.2. Dilatometer results: .............................................................................................. 40 a) Effect of the Na2CO3 content: .................................................................................. 40 1

b) Test of the reversibility of the reactions: ................................................................. 41 III.2.3. Results of the DSC/TG and mass spectroscopy ................................................... 42 a) Effect of Na2O:......................................................................................................... 42 b) Effect of the brick content in Alumina (Al2O3) : ..................................................... 43 c) Test of the reversibility of the reactions:.................................................................. 44 III.2.3. Discussion for the DSC/TG and dilatometer measurements:............................... 47 III.5) Simulation ................................................................................................................... 49 IV Conclusion .......................................................................................................................... 51 Acknowledgements .................................................................................................................. 52 References ................................................................................................................................ 53 VI Annex .................................................................................................................................. 56 Annex I: XRD ...................................................................................................................... 56 Annex II: EDS ...................................................................................................................... 58 AnnexIII: Building of the correction for the data: ............................................................... 75

2

I Theoretical background
I.1) Introduction
In the production of iron ore pellets at LKAB (Luossavaara-Kiirunavaara AB), the process includes the sintering of milled ore formed to green balls in a rotary kiln. The average working temperature is1250°C but temperatures close to 1350°C can be reached locally. During this procedure a coating is formed on the refractory material. This coating is made of slag coming from pellet’s dust and impurities. As the thickness of the reaction layer increase during time, the weight of the layer involve a chipping off of big peaces of slag causing damage to the refractory (peaces of brick crack and stay fixed to the slag blocks). Due to this phenomenon the rotary kiln has to be stopped every year for a maintenance period. The kiln is then cleaned and the damaged bricks are replaced. The replacement of the bricks is an expensive operation and it involves the complete stop of pellet production that induces a big overall cost. The aim of this work is to identify the reactions that take place between the Slag and the refractory brick that decrease the mechanical proprieties. This will permit to find solutions in order to improve the materials or the process.

I.2) Processing of iron ore pellets at LKAB
Iron ores are natural resources found in the ground. They are usually bended in rocks composed of minerals and iron oxides. There are different compositions of iron ores that are used. The two mains are hematite (Fe2O3) and magnetite (Fe3O4). The one that is processing at LKAB Kiruna, is magnetite rich iron ore. The magnetite is mixed with other impurities and has to be extracted from the ground.1 There are different types of iron ore used to make steel: fines (fine ore), lump, and pellets. Fines is an iron ore powder that have grain size under 4,75mm in diameter and needs a further process of sintering by the steel producer before being used. This sintering is necessary to give a good permeability to air in the blast furnace (otherwise air cannot be blast through the ore in the furnace). Lump have iron ore particles with a diameter higher than 4,75mm. Lump is the sintered version of the fines. It can be directly used in the blast furnace, but the mechanical strength is low, it increases the mechanical wears problems in the blast furnace. Both fines and lump are the less expensive ores. Pellets are made of fines-grained concentrate and sintered into balls that have a high strength and a diameter from 9,5 to 16 millimetres. This type of ore offers the highest purity and permeability in the blast furnace, but is more expensive.2 The pellets permit also to design an ore that will give a good slag in the blast furnace, in order to get the most pure iron as possible after the reduction.

-

-

Facts about LKAB, Presentation brochure march 2007 http://minerals.usgs.gov/minerals/pubs/metal_prices/ Metal Prices in the United States through 1998 Iron ore, by William S. Kirk
2

1

3

LKAB sell their ore principally in the pellet form. Because it has high iron content and consists mainly of magnetite. It requires then lower energy input during processing of pellets, which results in lower carbon dioxide emissions. It is therefore more environmentally friendly than other ores. The reaction of oxidation from magnetite to hematite is a high exothermal reaction, which permit to sinter the ore without addition of a lot of energy. LKAB can because of this reason stay competitive compare to much bigger ore producers such as the Brazilian producers that extract hematite directly at the surface and need much more energy to produce hematite pellets. LKAB supplies 2 different types of iron ore pellets products: one for blast furnaces (BF) and the other for direct reduction (DR). Blast furnace pellets are made for the blast furnace process using coke as reactant, which is the most common method of producing iron. DR pellets are used in the direct reduction process. This process permits to produce sponge iron. The DR process is based on the use of natural gas and has become increasingly common in countries with access to inexpensive natural gas.3 The composition of those two types of pellets is given in the following table:Figure1
Fe2O3 % 94,6 96,3 FeO % 0,4 0,3 MnO % 0,08 0,08 CaO % 0,50 1,05 MgO % 1,55 0,75 Al2O3 % 0,22 0,19 SiO2 % 2,05 0,85 TiO2 % 0,24 0,14 V2O5 % 0,20 0,21 P2O5 % 0,06 0,06 Na2O % 0,04 0,04 K 2O % 0,04 0,03

BF DR

Figure1: Chemical composition of KPBO and MPRD4

The first step of the process is the mining of crude ore. In Kiruna the ore is extracted from the ground at around 1000m under the peak of the mountain. The ore is then transported to a crushing plant where it is crushed to a fine powder in several stages. After being crushed, the magnetite is separated from the other impurities using magnetic separators, as magnetite is a natural magnetic component. After this step the ore is transported to the pelletizing plant using water. The second step is the pelletizing of iron ore. After filtration, some additives are added to the ore, such as: limestone, olivine and bentonite. Olivine and limestone are added in order to get a good slag in the blast furnace. Then an addition of bentonite into the filter cake act like a binder. In direct reduction furnace dolomite is used instead of olivine. Once the mixture is ready, the ore is transported to the balling drums. They are cylinders that rotate in order to make balls of iron ore, called “green balls” in the case of magnetite ore. A drum can be seen in Figure2.
F i Figure 2: Balling Drum Mixer.5

If the size of the green balls are not appropriate at the end of the drum, they go one more time in the drum, if they are two big they are crushed and insert in the process again. The separation is made by cylinders separated by different spaces. Once the right sized is reached, the green balls are transported to the first step of the grate: the drying part. The schematic outline of the warm part of the pellet production can be seen in Figure 3. www.LKAB.com, Products, iron ore Products Niiniskorpi V., LKAB (2001): Phases and Microstructures in LKAB’s Olivine and Dolmite fluxed pellets. P.768 5 http://www.energysolutionscenter.org/HeatTreat/MetalsAdvisor , agglomeration, equipment.
4 3

4

Materialflöde Returer Luftflöden
6 Rullkretsar Bindemedel

Magnetitslig från anrikningsverket

2A

Kolkvarn Våtskrubber Föravskiljare

3C
Multicyklon

3A
Avgaspanna

3B
Kollager

41FL33 2B 2C
Filter Avgaspanna

Figure 3: Scheme of the pelletizing process at LKAB4

The first zone in the grate is called UDD (up draught drying) where air enters from the bottom of the bed with a temperature of 140°C. This zone permits a first drying of the green balls. The second step is the DDD (down draught drying). In this zone they finish to be dried, at an average temperature of 400°C. Then the balls are preheated and heated in the Temperature preheating zone and preheating zone with respectively a temperature of 1000°C and 1120°C. After the great, the green balls enter the rotary kiln (where the reaction between the slag and the refractory material takes place). Finally the sintered green balls, pellets, enter a cooling system. In order to reduce the energy consumption of the process, the warmed air used for cooling is reused in the system. The main reaction that takes place is the oxidation of magnetite into hematite: 2 Fe3O4 + ½ O2 → 3 Fe2O3 (1)

Most of the green balls oxidize already in the grate and only a few percent in the Kiln. The rest of the reactions take place in the cooler. The oxidation rate is only limited by the diffusion of oxygen in the green ball. The moment when the magnetite is oxidizing to hematite is influencing a lot on the mechanical behavior of the pellets. That is why it is important to control this parameter, in order to have the same mechanical proprieties for all the pellets.6

Grate-kiln-cooler. Where to oxidize and why?, Niiniskorpi V., LKAB 2002 Ironmaking Conference Proceedings

6

5

The rotary Kiln is the part where the pellets get their high mechanical proprieties by sintering. It is a rotating cylindrical furnace. In Svappavaara it has a length of 43m and a diameter of 6m. Figure 4 shows the Svappavaara´s Kiln.

Figure 4: Rotary kiln.7

The working temperature of the kiln is 1250°C in average, but in the hot part it can reach a maximal temperature of 1350°C. It is constituted of a metallic body protected with refractory bricks inside, directly in contact with the pellets. Those brick are made principally of mullite. The Kiln is mainly self-heated by the reaction of oxidation of magnetite into hematite. However an additional heater using oil permits to increase the temperature and to maintain it constant. The gas exhaust of the heater is introducing some more impurities in the kiln, such as alkalis.

I.3) Materials involved
This part will present the materials that have been used in this study and give their main properties: Bricks, Slag, Na2CO3 and Al2O3. I.3.1. Bricks The Bricks used in the kiln are made of refractory materials. The main proprieties needed for refractory materials are their high heat resistance, low thermal conductivity, mechanical resistance, thermal stresses resistance, resistance to corrosion, resistance to erosion, liquid and gas permeability. Most of the refractory materials are made of ceramics. Ceramics are inorganic, non-metallic materials. Ceramic materials are usually ionic or covalent bounded materials, and can be crystalline or amorphous. Due to those strong bounds, the material has a low toughness and tends to break before deforming plastically. Those properties do not permit an easy machining of the material and the sintering technique is then, in most of the case, necessary. 8

http://www.metsominerals.com/ , Iron ore pelletizing Handbook of Ceramics and Composites, by Nicholas P. Cheremisinoff, Published 1990 CRC Press, ISBN 0824780051
8

7

6

Scintering: Sintering is the treatment of fined grained materials, that are heated up at a temperature below the melting point of the main constituent, in order to bound the particles and reduce the porosity. Due to the atomic diffusion at high temperature, close to the melting point, the atoms diffuse from one particle to the other, forming bridges between them. The energy of the surface of those bridges is higher then the one of the particle. That induces a preferential diffusion from the surface of the particle, to the bridge and an increase in the bridges size. A schematic explanation of the phenomena is shown on the Figure5. The presence of a liquid phase between the particles permits to high up the diffusion and the Sintering speed. After sintering, recrystallisation and grain growth takes place.

Particle

Diffusion

Particle

Low energy

Sintering

High energy Bridge

Figure5: schematic representation of the sintering process

Sintering is used to produce most of the ceramics objects. Refractory bricks also use this method. The ceramic used in the refractory bricks for the steel industry is mainly mullite.

Mullite: Mullite is a minearal with appropriate refractory propeties. It is used in the industry as high technology ceramic. Mullite is an alumino-silica component. It got is name from Bowen and Greig in 1924 when they discovered that the stable phase in the alumina-silica system was a 3:2 (3Al2O3.2SiO2) crystal. They found this crystal in a natural rock from the island Mull. They decided to name this phase mullite. Mullite is forming at high temperature and low pressure. Which explain why it is rare in nature. This refractory material is found as main constituent since years in pottery and porcelains. Nowadays it is used as advanced structural and functional ceramic in wide application field.9 The formation of pure mullite with a powder of silica and alumina is a reaction that happened above 1400°C. The processing of this materials and the raw materials used are costly and make mullite expensive.10 Mullite can have variation in it composition, which corresponds to a different substitution of silicon by aluminium in an orthorhombic lattice. The composition is described by the general
Mullite and Mullite Ceramics, by H.Schneider, K. Okada, J.A. Pask. Published 1994 by John Wiley & Sons Ltd. ISBN:0471942499 10 Mullitisation behaviour of south Indian clays Ceramics International, Volume 28, Issue 5, 2002, Pages 557564 V. Viswabaskaran, F. D. Gnanam and M. Balasubramanian
9

7

formula: Al2(Al2+2x.Si2-2x)O10-x, where x is the number of oxygen vacancies and can be between 0,17 and 0,59.9 The binary phase diagram of alumina silica shows the major phases stable depending on the temperature (Figure6)

Figure 6: Alumina-Silica binary phase diagram. Liq.: liquide, Cor. :corrundum, ss : solid solution.11

In the phase diagram it can be seen that the stable form of mullite is 3:2. For a higher content of alumina, the mullite is in solid solution in corundum. For a lower content of alumina, we have a mixture of mullite and SiO2 (that is in glassy state with a melting temperature at 1595°C). Mullite has a high melting temperature of 1850°C, close to pure alumina (2050°C). Mullite is nowadays one of the best refractory materials, due to its good proprieties: high melting point, small deformation under load (good creep resistance), low thermal expansion, good thermal shock resistance, high shear modulus and good corrosion resistance.9 (p.233)

The three types of brick: There are 3 different type of brick used in the Svappavaara’s Kiln: The Victor brick, the Alex brick and the Silox 60 brick. The Victor brick is used in the zone working at high temperatures. The Alex brick is only used in the first zone of the kiln; working at lower temperatures, due to a lower price and lower resistance. The Silox brick is a new type of brick tested insitu in Svappavaara. The composition of the bricks is given in the table: Figure7.

Ceramic Microstructures, by William E. Lee and W. Mark Rainforth, published in 1994 by Chapman & Hall.P.

11

8

VICTOR HWM High alumina brick based on bauxite Bulk density Apparent porosity Cold crushing strength Refract. Under load -T05 -T5 Thermal shock resistance Linear thermal expansion at 1000°C Thermal conductivity W/(m.K) 500°C 800°C 1000°C 1200°C Chemical analysis w% Al2O3 SiO2 CaO TiO2 Fe2O3 Alkalis 2700 kg/m³ 18 % 75 MPa

ALEX Dry pressed bauxite boosted brick based on chamotte 2300 – 2400 kg/m³ 17-20 % 45-65 MPa

Silox 60 Dry pressed brick based on andalusite 2400-2500 kg/m³ 14-16 % 60-64 MPa

1400 °C 1590 °C > 30 cycles 0,6 %

1380 °C 1470 °C > 30 cycles 0,6 %

1410°C 1620°C > 30 cycles -

1,8 1,9 2,1 2,0 70,7 25,1 0,2 2,6 1,1 0,3

1,8 1,6 1,5 1,6 58 36 0,3 2,1 1,4 1,3

1,2 1,4 1,9 2,0 60 37 0,1 1,5 0,9 0,5

Figure7: Table resuming main properties of the three types of refractory bricks used in the kiln12

The main difference between the brick composition is their alumina content. The Victor brick is a high alumina content brick with 71% of Al2O3 and 25% of SiO2 and 1% of impurities. This brick is manufactured from bauxite. Bauxite is an aluminium ore. It is a heterogeneous material containing: silica, iron oxide, titanium, aluminium silicates and some other impurities in low quantities. Bauxite is then fired at high temperature. The main phase after firing are mullite, corundum and a small amount of glassy phase containing vitreous SiO2 and impurities such as titanium oxide, hematite and alkalis. Alex brick consist of a dry pressed

12

Technical informations by Höganäs Refractories

9

bauxite boosted brick based on chamotte. Silox60 brick is a dry pressed brick based on andalusite. I.3.2. Slag The slag, which forms on the walls of the Kiln, is mainly constituted of disintegrated pellets and impurities. Pellets are constituted mainly of hematite and the rest is impurities. Their composition is given in Figure1. During the sintering process this magnetite is transformed into hematite. Hematite is one of the stable oxide form of iron that is find in nature (the other one is magnetite). It got its name from a Greek word meaning blood-like, because of its red color. The coal and oil that are burned in the burner release some dust, mainly sodium and potassium.13The slag is then constituted of hematite, impurities coming from the iron ore, and alkalis introduced by the burner. Its composition is given in the following table: Figure8.

SiO2 TiO2 Al2O3 FeO Fe2O3 MnO MgO CaO Na2O K2O V2O5 P2O5

2,07 % 0,26 % 0,54 % 0,71 % 92,18 % 0,08 % 1,44 % 0,58 % 0,1 % 0,059 % 0,19 % 0,037 %

Figure8: Analyze made in Malmberget of Chemical composition of the Svappavaara Kiln14

We can see in this chemical analysis that the main impurities in the Slag after Fe2O3 are SiO2 and MgO. SiO2 is forming a glassy phase that bounds the dust together and permits the slag to enter into the pores of the refractory materials and be fixed. I.3.3. Na2CO3 In this work sodium carbonate (Na2CO3) has been used in order to study the influence of this alkali on the reaction. Na2CO3 is a white crystalline solid. It is obtained from the Solvay reaction using salt (NaCl) and chalk (CaCO3).

Action of Heavy Oil Combustion Products on Aluminosilicate Refractories. Material Science and Engineering, 88 (1987) p.75-80. G. Mascolo and O. Marlino 14 Research project: Infiltration of iron slag into refractory bricks, Mariya Chestnyeysha, LTU 2006, supervisor: Magnus Oden and Jesper Stjernberg.

13

10

Solvay reaction: The first step is the reaction of see salt with CO2 using: a solution of water, salt and ammoniac. CO2 passes through this solution and react to form a precipitate of NaHCO3: NaCl + CO2 + NH3 + H2O → NaHCO3 + NH4Cl Ammoniac is used to maintain the solution with a basic pH. Then NaHCO3is burned to obtain Na2CO3: 2 NaHCO3 → Na2CO3 + H2O + CO2 The carbon dioxyde that is used for the reaction is obtained by heating up the chalk: CaCO3 → CO2 + CaO And the Calcium oxide is then used to form Ca(OH)2: CaO + H2O → Ca(OH)2 The ammoniac can then be regenerated from (Ca(OH)2) and NH4Cl: 2 NH4Cl + Ca(OH)2 → 2 NH3 + CaCl2 + 2H2O (reference)15 (6) (5) (4) (3) (2)

Sodium carbonate is a hygroscopic material. That means that it absorbs likely water. Different forms of crystals exist16: Na2CO3 · H2O and Na2CO3 · 10H2O. When a dehydrated sodium carbonate powder is placed in the air, it becomes hydrated after a short time. To be dehydrated it has to be heated up. The sodium carbonate decomposes into Na2O before melting, at a temperature ranged between 500°C and 800°C: Na2CO3(s) → CO2(g) + Na2O(s) The theoretical melting point of the constituent is 851°C17. I.3.4. Alumina (Al2O3) Alumina is a ceramic material. There are three stable crystalline form of alumina: α-alumina, β-alumina and ɣ-alumina. The most common form is α-alumina, is known as corundum and has a hexagonal close packed (HCP) lattice structure. It is a good isolating material, due to its high melting point of 2054°C. The main disadvantage of this material is his price; in order to obtain pure alumina the process is complex and the sintering of the alumina powder has to be http://fr.ekopedia.org/Carbonate_de_sodium Thermodynamic properties and phase transitions of salt hydrates between 270 and 400 K II. Na2CO3 · H2O and Na2CO3 · 10H2O The Journal of Chemical Thermodynamics, Volume 15, Issue 9, September 1983, Pages 881-889 Fredrik Grønvold and Knut K. Meisingset 17 Y. Otsubo, K. Yamaguchi, J. Chem. Soc. Japan, 82, 557-560 (1961)
16 15

(7)

11

realized at high temperature. The powder is obtained by purifying bauxite, using the Bayer process: Al2O3 + 3H2O + 2NaOH + heat → 2NaAl(OH)4 (8)

Fe2O3 that is present in Bauxite does not dissolve in the base and can be then removed by filtering the solution. SiO2 dissolves as silicate Si(OH)62-. When the Bayer liquor is cooled, Al(OH)3 precipitates and the silicate remains in the solution. Then by deshydratation the reaction is 11 18: 2Al(OH)3 + heat → Al2O3 + 3H2O (9)

I.4) Previous work
The reaction between iron ore slag and refractory material has been studied in several different works. First the work realised by The Division of Engineering Materials at Luleå University of Technology will be briefly summarised and then the main conclusion of different publications on the subject will be given. I.4.1 Microstructural studies on the growth of FeOx on Al2O3-SiO2 based brick insulation materials In this project Antje Schlieter43 studied the microstructure and the composition of the slag coating of the brick. In a second part she studied the slag attack in different places in the Kiln and its depth in the brick. She made different analyses of the samples that she collected during maintenance at Svappavaara, using optical microscopy, SEM (scanning electron microscopy) and XRD (X-ray diffraction). Schliter’s conclusions were that: -The optical and SEM analyses showed a penetration zone where the slag interacted with the brick, affecting the mechanical and refractory properties of the brick on a thickness of 5cm. In this zone there is crack formation in a parallel direction to the surface of the brick. -She also detected the presence of a glassy phase close the brick/slag interface. -The chemical analyze permitted to say that the bulk of the slag derive from the pellets with a high hematite content. -The main process that is observed during the attack of the slag is: the slag penetration, dissolution and the wear of the brick. -Victor brick is more slag resistant than Alex brick. -In the Alex brick the iron content is lower at the surface, because the brick crack faster so the iron has less time to penetrate the surface (each time the brick crack it forms a “new” surface) -With the presence of slag, a growth of secondary mullite that gives appreciable physicomechanical properties have been detected.

The transfomation of aluminium oxide from the beta to the alpha form, John GalluP 1RCA RADIOTRON DIVISION RCA MANUFACTURING CO/ INC. HARRISON, NEW JERS RCA RADIOTRON DIVISION RCA MANUFACTURING CO/ INC. HARRISON, NEW JERSEY, Presented at the Annual Meeting, American Ceramic Society, Buffalo, N. Y., February, 1935 (Refractories Division). Examensarbete: Microstructural studies on the growth of FeOx on Al2O3-SiO2 based brick insulation materials, Antje Schlieter, LTU 2006, supervisor: Magnus Oden.
43

18

12

I.4.2 Slag fixation on the Svappavaara’s rotary Kiln Jesper Stjernberg worked on the mechanism responsible for the penetration of slag into the brick and the different parameters that may influence the degradation of the refractory. Using XRD, SEM and optical microscopy the influence of alkali contnent on the brick has been studied and the resistance of the three different bricks using different dwell temperatures and holding times. Stjernberg concluded that a combination between diffusion and infiltration of the slag was responsible of the slag attack. With the study of the alkalis content he could see that the brick with more alumina where the more resistant. The study clearly showed that alkalis affected the wear of the brick, and that the dwell time had a smaller influence than the temperature used 44. I.4.3 Infiltration of iron slag into refractory bricks Mariya Chestnyeysha worked on the high-temperature mechanisms responsible for the wear of the bricks and to determine the influence of alkalis on those reactions. Differential scanning calorimetry (DSC), SEM and XRD were used in the study with powders from milled slag, refractory brick and of Na2CO3. The studies were carried out in Helium atmosphere from 20°C to 1300°C. XRD were used to analyze the changes in composition of her sample, before and after heating them at 1300°C. First different mixtures of brick and slag were prepared: using 25% of brick powder and 75% of slag powder, 50% brick and 50% slag, and 75%brick and 25% slag. She concluded that the 50% brick, 50% slag mix was the most interesting because the reaction was optimal compare to the two others mixtures. Further experiments have been run in the furnace and DSC in order to determine the influence of alkalis, as well as the reaction happening between the slag and the brick. DSC/TG were used in order to characterize the reactions that occurred and then used XRD to compare her samples before and after heat treatment. 3 different powder mixtures have been used: -50% brick + 50%slag -50% brick + 50%slag with an addition of 2% of Na2CO3 -50% brick + 50%slag with an addition of 5% of Na2CO3 Conclusion were drown that: -XRD did not detect the presence of any new phase. -DSC measurements showed some phenomena, which probably correspond to the melting of alkali, that take place in a temperature range between 700°C and 1000°C - SEM pictures permitted to see differences in microstructure between the samples having different content of alkalis. She also pointed out the presence of a glassy phase. - Suppositions were made on the influence of alkalis on the diffusion and infiltration of the slag into the brick. Those conclusions are made on experiments performed in helium. Due to the atmosphere used she observed the reduction of hematite into magnetite at low temperature: 1125°C instead of 1400°C in air. In the conclusions it was pointed out that further work in an air atmosphere would be a good complement, in order to approach industrial conditions.14

13

I.4.4 Literature review In this project, the most interesting parameters are the interaction between the mullite and the iron slag and the influence of alkalis in this reaction. The majority of the studies that treats the attack of slag on refractory materials do not explain clearly the reactions that appear between the slag and the brick. Only a few studies give some details on some specific reactions happening between the constituent present in this study. Interaction between mullite and hematite: The slag is mainly constituted of hematite. The solid solubility limit of hematite into mullite can reach 11 wt% depending on the temperature. But this solubility decrease with the temperature (10,5 wt% at 1300°C and 2,5 wt% at 1670°C). Titanium is also following the same solubility tendency. Schneider also reported that with the increase of the Temperature the mullite composition equilibrium is displaced and moves toward higher alumina content. Due to the penetration of metallic ions in the lattice, silica leaves the mullite and forms a glassy phase between the mullite grains.10 Interaction between mullite and alkalis: N. V. Pitak worked on the interaction between a mullite-corundum refractory and alkalis under variable gas media. They reported that under an oxidation atmosphere, and thermal cycles at 1580°C, there were a formation of vitreous substance with a nepheline composition (Na2O.Al2O3.2SiO2) and the corundum crystal in the glass are recrystallized into flake hexagonal, rhombohedral, and prismatic crystals. They reported that mullite disappears almost completely. But this firing lead to a decrease in the open porosity and a densification as well as a strengthening. They have shown that the microstructure was constituted of corundum crystals and acicular mullite, with a matrix that consists of prismatic and isometric corundum crystals bind in a glass with a nepheline composition. It has been shown that Na2O passes almost completely into the melt. It lead to a decreased viscosity and increased its amount. This improved the recrystallisation of corundum that had a flake hexagonal and rhombohedral shape.19 Effect of alumina content on the attack of mullite by impurities: In others literature review it has also be find that the Silica formed during the formation of the refractory material partially react with impurities to from a vitreous phase that permit a higher diffusion and a better sintering, but at high temperature it lows down the mechanical and refractory proprieties, as well as the thermal shock resistance of the material. It is shown that the addition of an appropriate amount of alumina displaces this equilibrium, and the silica available for the glassy phase will decrease. That will also increase the sintering temperature and decrease the effect of the impurities that will be found in solid solution in the mullite. That study also showed that the mullite that form in a high silica content will have an acicular form.20 H. Schneider also reported this acicular form of mullite that recrystallize from a silicate glassy phase.10 (p.165)

Interaction of mullite-corundum refractories and alkalis under conditions of variable gas media Translated from: Ogneupory, No.7, July 1992, Pages 24-26. N. V. Pitak, Ukrainian Research Institute of Refractories, Kharkov, Ukraine. 20 Synthesis and mechanical properties of mullite from beach sand sillimanite: effect of TiO2 Journal of the European Ceramic Society, Volume 18, Issue 14, December 1998, Pages 2081-2087 H. S. Tripathi and G. Banerjee

19

14

Decrease of the mechanical properties and porosity: The glassy phase that is formed at rather low temperatures, close to 1000°C for Na2O, due to the impurities decrease the refractory properties of the material and seems to be one of the reasons on the wear of the brick. It increases the diffusion and the reactions between the brick and the slag21 22. But it is also reported by W. Rainforth that the porosity is a major factor of slag attack. The Slag enters the open porosity and then reacts with the brick. The porosity increases the surface in contact with the slag12.

High temperature mullite dissolution in ceramic bodies derived from Al-rich sludge Journal of the European Ceramic Society, Volume 25, Issue 5, February 2005, Pages 703-710 M. J. Ribeiro, D. U. Tulyagavov, J. M. Ferreira and J. A. Labrincha 22 Influence of thermal aging on Microstructural Development of Mullite Containing Alkalis American Ceramic Society: Volume 81, Issues [10], 1998, Pages 2741-2745 C. Baudìn and M. Pilar Villar.

21

15

II Experiments
In a first part the experimental procedure will be given. A second part will present the devices that have been used in the experiments and a third part will give more details about the samples prepared and their method of preparation.

II.1) Experimental procedure
This work treats two main phenomena influencing the slag attack. The first one is the influence of the alkalis on the reactions between the brick and the slag. The second phenomenon is the higher slag resistance of bricks containing more alumina, as reported in previous works13 43 44. This tendency has been studied in the specific system, concentrating the research on the influence of the addition of alumina on the reactions causing wear of the brick. In this study all the experiments have been carried out in an air atmosphere in order to work in the industrial conditions (the kiln is running in air). Previous tests in Kanthal furnace In the first part of the work it was important to have an idea about the reactivity of the samples. Indeed the materials involved in the reaction can be highly reactive. If the reaction is too violent (boiling, explosions, reaction with the crucible) it can cause the destruction of the measurement devices such as DSC and dilatometer. In order to test the reactivity of the samples, first tests have been done in the Kanthal furnace following the same heating program as the one used in DSC. This temperature program is given in Figure9. Temperature (°C) 1350°C

600°C 40°C 131 206 Time (min)
Figure9: Temperature program for Kanthal tests and thermal analysis

The heating step from 40°C to 1350°C and the cooling from 1350°C to 600°C are made with a rate of 10°C/min. If the test shows that the samples are not too reactive, they can be used for the thermal analysis.

16

Thermal analysis: The thermal analysis were carried out to characterize the reactions in the samples. Differential scanning calorimetry and thermogravimetry (DSC/TG) and in-situ mass spectrometry were used to characterize the thermodynamical and chemical transformations, dilatometer were used to study the volume modifications. The combination of those measurements will permit to have multiple information on each transformations and give conclusion on the nature of each modification in the system. For the dilatometer such as the DSC/TG, the same temperature program has been used. It can be seen in Figure9. Test of the reversibility of the reactions have also been made by running several times the same sample. In order to have more way of characterization, a second step of the experimental part is the structure and phase characterization after the reaction. Structure and phase characterization: The phase and structure characterization after reaction has been done in two steps. The first step was the use of XRD.XRD permitted to compare the phases present before and after the transformations in order to find the nature of the new phases that have possibly been formed. The study of the influence of Na2O on the reaction, has been realized with a sample containing a high concentration in Na2CO3, the 50/50+10p sample (see Figure20). It permits to find the nature of the phases containing Na2O. The second step was the study of the combination of Scanning Electron Microscopy (SEM) pictures with Energy dispersive Spectrometry (EDS). The aim of this step was to have more data about the formation of new phases, and at the same time permitted to study the influence of alkalis and alumina content of the brick by comparing the samples structures after reaction. Four samples have also been prepared just before the slag attack begins in order to compare the structures before and after reaction. Samples: In order to study the two main phenomena, different sample compositions have been used. One set of samples with different Na2CO3 content for the influence of the alkalis and another set with different alumina content for the study of the influence of the brick’s alumina content. A simulation program called “Equilibrium” part of the program “factsage” has also been used. This program is based on thermodynamical calculus, in order to give the possible reactions between constituent. The main advantage of this program is to give an idea about the reaction that might occur, and the phases that are stable at which temperature. The thermodynamically equilibriums for all the samples from 0 to 1350°C has been calculated.

II.2) Devices used
II.1.1 Kanthal furnace The Kanthal furnace is a programmable furnace that can reach 1600°C, using electrical resistance heating element. The refractory material inside the furnace is made of fiber insulation bricks. In order not to pollute the refractory of the furnace, the samples are usually put in alumina boats, which can be seen in Figure10.
Figure10: Alumina boat with a sample

17

II.1.2 XRD (X-ray diffraction) History: In 1895 Wilhelm Conrad Röntgen discovered the X-rays. He made an X-ray picture of his wife's hand. A few months later, Carl Muller started the production of X-ray tubes. Marx von Laue was the first one who discovered the diffraction of X-rays by crystals. He was awarded the Nobel Prize in physics in 1914. The father and son Bragg developed its theoretical explanation: the Bragg Law: Figure11

n = integer – 1, 2, 3…, etc λ = wavelength in Angströms d = interatomic spacing in Angströms. θ = diffraction angle in degrees.

Figure11: Explanation of the Bragg Law23

The Netherlands, Philips, a leading manufacturer of incandescent lamps, sets the first step towards turning X-ray research into functional applications and produced the first commercialized X-ray difractometers.

X-ray powder diffraction spectrometer: The spectrometer used in the work is an X-ray powder diffraction spectrometer. This device is using powder samples and is one of the most widely used X-ray diffraction technique for characterizing materials. The samples used are grind to a small particle size, in order to get a homogeneous surface with crystallites that are randomly distributed. It involves that crystalline domains are randomly oriented in the sample. When a X-ray beam is sent on the surface of the sample (with an angle theta from the surface), the intensity of the beam that is reflected (also with an angle theta from the surface) is measured (see configuration Figure12). In this case only crystallites having reflecting planes (h, k, l) parallel to the specimen surface will contribute to the reflected intensities.

23

http://cxpi.spme.monash.edu.au/xray_history.htm

18

Figure12: Bragg Brentano Theta:Theta Setup24

In order to get all the angles for which the crystallographic plane diffract the beam, the diffracted intensity for a range of theta that can go from nearly 0° to 150° is measured. Peaks appear for the different planes that reflect the beam. And a diffractogram of the sample is gotten such as in Figure13. This type of diffractogram is a fingerprint of the material. Each material have a different diffractogram: - The d-spacing of lattice planes depend on the size of the elementary cell and determine the position of the peaks.
Figure13: XRD diffractogram of K2Ta2O6 sample25

- The intensity of each peak is caused by the crystallographic structure, the position of the atoms within the elementary cell and their thermal vibration.

24 25

http://epswww.unm.edu/xrd/xrdbasics.pdf http://www.mrl.ucsb.edu/mrl/centralfacilities/xray/xray-basics/index.html

19

A database containing most of the known materials is available. When a sample is analyzed, each substance present in the sample gives its peaks. Using the database it is possible to analyze the composition of the sample.24 25 26 II.1.3 Dilatometer The dilatometer is an apparatus that measure the dilatation of the sample. This dilatation takes place due to the change of the temperature. When the temperature is modified the sample expand ore shrink. Some reactions or phase transformation can also be the cause of shrinkage or expansion. The size measured is the thermal expansion coefficient α of the sample. It is defined as the rate of change of sample length with respect to the change in its temperature. Thermal expansion is a measurement of the entropy when a body reacts to changes its volume. It can be assimilated to the vibration of the atoms due to thermal excitation. In a dilatometer measurement, α correspond to the derivation of the curve. For a given phase α can be assimilated as constant. That means that each slope modification correspond to some phase transition or phase formation. The dilatometer is made of a heating source (furnace) and a measurement device. There are different types of measurement devices: The capacity dilatometer is constituted of a parallel plate capacitor with a mobile plate that is a spacer sensor. The precision of this apparatus can reach the range of the picometer. The connecting rod dilatometer is constituted of a connecting rod that is in contact with the surface of the sample. This connection transfers the thermal expansion to a sensor which measures the displacement. This sensor is situated out of the warm zone. The rod permits to protect the sensor from the heat. The sensor can be a strain gauge, or nowadays an inductive displacement transducer. A connectingrod dilatometer can be seen in Figure 14.27
Figure14: Dilatometer

-

-

The optical dilatometer is an instrument that measures the dilatation, using a digital camera. This technique is a non contact technique that allow to measure ductile samples such as polymers or fragile samples such as ceramics powders.28

The chamber in which the sample is placed can be purged with a gas (inert, oxidizing, reducing, vacuum et al.) and permit to measure the dilatation under different atmosphere. There are different furnaces and each can only been used under certain conditions. For instance the graphite furnace cannot be used in air (oxidation of the refractory of the furnace).

http://lsu.epfl.ch/cours/ultra/Bas-XRD.pdf http://www.netzsch-thermal-analysis.com/ 28 Dilatometry, University of Cambridge, Materials Science & Metallurgy 2002, H. K. D. H. Bhadeshia. http://www.msm.cam.ac.uk/phase-trans/2002/Thermal3.pdf
27

26

20

In this project a connecting rod furnace have been used. The connecting rod is made of alumina. Alumina has a good thermal resistance, a minimal thermal dilatation and can be used in air. Before the measurements, a calibration run is necessary in order to build a correction file (suppression of the dilatation of the measurement device: connecting rod, sample holder et al.). The dilatometer is controlled by a program that permits to give the heating route and take into account the data from the correction. II.1.4 Differential Scanning Calorimetry / Thermogravimetry / Mass spectrometry: A differential thermal analysis apparatus is used to study the heat that have to be delivered to a sample in order to correlate its temperature compared to a reference. This study gives information about the modification of heat capacity (phase transition, vitreous transition, crystallization, melting et al.) and also reactions that occurs in the sample. The study of the DSC curve gives then the transition temperatures such as the enthalpy of reactions. There are different thermal analysis (DTA) apparatus. The first that has been developed in 1964 was the DTA calorimeter: Differential Thermal Analysis calorimeter. It consists of a chamber that contains a sample and a reference. The chamber is heated with a constant heating rate. A thermocouple is used to measure the difference in temperature ∆T between the reference and the sample. ∆T is plot as a function of T. In order to get the enthalpy, ∆T needs to be converted into an energy unit.29 30 31 Nowadays DSC (Differential scanning Calorimeter) instruments are the most common. There are two sorts of apparatus: heat flux DSC and power compensation DSC. A schematic view of the two systems is represented in Figure15. Those instruments are different from DTA, because it measures directly the difference in energy that is delivered to the sample compared to the reference. The enthalpy is directly gotten, after having a short calculus to change the units (generally electrical current that has to be transformed into heat flow). The power compensation DSC is using two separated furnaces that are separately controlled and identical. The temperatures of the samples and the reference are maintained constant by Figure15 : A: heat flux DSC, B: Power compensation DSC 32 controlling the power input of each furnace. The energy necessary for the two furnaces are compared and it gives directly the enthalpy of reaction or heat capacity changes of the sample compare to the reference. The heat flux DSC
29 30

Thermal analysis of materials, by Robert F. Speyer, Published in New York : Dekker, cop. in 1994, ISBN: 0-

8247-8963-6

Handbook of thermal analysis of construction materials, by V.S. Ramachandran et al., Published by Norwich, N.Y. : Noyes Publications : William Andrew Pub., in 2002. 31 Thermal Analysis Technique, University of Cambridge, Materials Science & Metallurgy 2002, H. K. D. H. Bhadeshia. http://www.msm.cam.ac.uk/phase-trans/2002/Thermal1.pdf

21

is the same principle as a DTA instrument, but the reference and the sample are connected by a material with high heat transfert. The sample and reference are heated by a single furnace. The temperature difference between the sample and the reference are low, due to the high thermal conductivity between the two. This temperature difference is correlated to enthalpy change in the sample using calibration experiments (with pure metals).32 For the experiments it was a heat flux DSC that has been used. It is the “Netsch STA 449 C Jupiter®”. Those instruments can be combined with a mass spectrometer. In this study the “Netsch QMS 403 Aëolos®” that analyses the gases that escape the DSC chamber has been used. This type of apparatus permits to measure the masses of the elements that escape the chamber. For example if the chamber will emit CO2 then it will detect a mass of: 12, 16, 28 and 44. This apparatus can detect the masses from 1 to 300 amu.33 Both apparatus can be seen on Figure16. The DSC apparatus also contain a Thermogravimetry (TG) apparatus. This is a high precision balance that records the mass of the sample in function of the temperature. It gives information about the mass loss during a reaction and the quantity of gases that have left the sample can be calculated. Coupling TG with the mass spectrometer permits to give the quantity and nature of gases that leave the sample.34 Atmosphere: In this type of apparatus (“Netsch STA 449 C Jupiter®”) when a corrosive atmosphere (air) is used, the furnace and the TG chamber have to be protected with inert gases (argon, helium et al.). The flow rate of the inert gas has to be calculated.

STA 449 C Jupiter®

QMS 403 Aëolos®

Figure16 : DSC apparatus with mass spectrometer31

The example of this study’s experiments conditions is going to be taken in order to give the procedure of calculation. The air flow rate in the chamber is set to 20ml/min in nitrogen (reference). The furnace chamber is protected with argon with a flow rate of 35ml/min (gas inlet furnace). The Balance is protected with nitrogen with a flow rate of 25ml/min (protective). In order to know the flow rate of the air in the chamber, a formula given by the manual of the DSC apparatus is used:

32

Differential Scanning Calorimetry, University of Cambridge, Materials Science & Metallurgy 2002, H. K. D. H. Bhadeshia. http://www.msm.cam.ac.uk/phase-trans/2002/Thermal2.pdf

Netzsch dilatometers : http://www.dilatometers.com/ Thermogravimetry, University of Cambridge, Materials Science & Metallurgy 2002, H. K. D. H. Bhadeshia. http://www.msm.cam.ac.uk/phase-trans/2002/Thermal4.pdf
34

33

22

QSG = K × Q N 2

With: K =

1

ϕ SG ϕN
2

(10)

QSG : Flow rate of the sample atmosphere Q N 2 : Flow rate of N2 calibrated normally for the sample

ϕ SG : density of the selected gas ϕ N : density of N2
2

In the case of air: Kair is equal to 0,98. It gives then a flow rate that is almost equivalent to the nitrogen reference flow rate.

Calibration: DSC apparatus have to be calibrated with reference samples made of pure metals. This calibration has to be done for each atmosphere used. The exact melting point of the reference samples is known. Five reference samples are run in the DSC in order to get the measured melting points for different pure metals in the atmosphere used. The program built then a calibration file for this atmosphere. Once the apparatus is calibrated, a correction file has to be done such as in the dilatometer. This correction permits to have a 0 difference between the reference crucible and the sample crucible that are used. A new correction file should be made after each modification of a parameter in the measurement.

II.1.5 Scanning Electron Microscope (EDX)
Scanning electron microscopy is a method that uses a fine focused electron beam that hit the surface of the sample. This electron beam interacts with the surface of the sample and loose energy. This energy is absorbed by a zone at the surface called the interaction volume and can be from less than 100nm to 5 µm depending on the beam accelerating voltage. This volume has a teardrop-shape and emits electrons and electromagnetic radiation. A schema of this interaction volume can be seen in Figure17. The signals emitted are captured by detectors and give a lot of information about the surface of the sample such as: topography of the surface (Back scattered electron and secondary electrons), structure of the surface and chemical difference (Back scattered electrons), elements analysis (X-rays), compositional information (Auger electron), electrical information (Cathodoluminescence and specimen courrent) et al35
36

http://www.culture.gouv.fr/culture/conservation/fr/methodes/meb_01.htm Danilatos, G,D (1988). "Foundations of environmental scanning electron microscopy". Advances in Electronics and Electron Physics 71: 109-250
36

35

23

Elektron Beam E0=30keV Backscattered electrons ZRE ca. 2,5 µm Auger electrons 1-2 nm Secondary electrons 10-100nm Sample’s surface X-rays ca. 5 µm Specimen currant X-rays continuum ca. 5µm heat Cathodoluminescence ca. 10µm

Figure 17: Schematic interaction volume with respective emission and size.

The SEM pictures, in this project, were taken using back scattered electrons. Back scattered electron have a high energy and they come out of the surface in all the directions. With their high energy they can come from quite deep in the surface, that is why the resolution is not as good as secondary electron resolution. But the number of electrons that hit the detector is a function of the atomic number of the material analyzed. That is why by scattering the surface, for each measured point a composition can be gotten, and a map of the composition of the surface. With this method the different phases present in the samples can be seen. Energy dispersive spectrometry: Energy dispersive X-ray spectrometry is the study of the number of X-rays photons with the same energy that hit the detector, in function of the energy of the photons. Then each atom has some emitting peaks corresponding to certain photon energy. A database permit to find which atom corresponds to which peak, and the area of the peak gives information on the quantity of the atoms. With that information the chemical analysis of a point on the surface of the sample can be measured. By measuring several points of an area, an average in this area of the chemical composition can be obtained. This method is fast, because the detector can treat simultaneously the different energies of the photons. But the resolution is limited because some peaks are superposed and the signal of some elements is reacting with others. Another method exists to analyze the X-rays coming from the interaction volume. It is the WDX (wavelength dispersive X-rays spectrometry). This method is separating the wavelength from the X-ray signal, using a single crystal. The single crystal and the detector are mounted on a goniometer, in order to measure the number of photons for each wavelength. There are different systems and different detectors, but in all the case, the advantage of this method is that the precision is high, but the disadvantage is that it takes more time. Some systems are capable to go faster using for example more than one crystal and detector.37

37

http://en.wikipedia.org/wiki/Wavelength_dispersive_X-ray_spectroscopy

24

II. 3) Sample Preparation
In this study, different methods have been used to characterize the samples and to provide some reactions. For each study, a special sample preparation has to be done and different types of samples have been used. But the first steps of the sample preparation are the same and can be seen on the Figure18 that summarize the process.

Mixture and milling Powder of raw constituent Shaker Powder mixture First compression

CIP preparation DSC/TG XRD Dilatometer

Kanthal
Figure18: Process of sample preparation.

CIP

The steps of the samples preparation are the following: - Powders preparation - Pellets preparation Samples in pellets form can be directly used in the dilatometer and the DSC/TG. For XRD and SEM some more steps of preparation are needed.

III.2.1 Powders preparation
The first step is the preparation of powders of the raw constituents: brick, slag, Na2O3 and Al2O3. Brick powder: The brick powder was taken directly from a brick that have been first roughly crushed with a hammer and then milled with a ball mill shaker. The ball mill shaker is a jar made of hard ceramic material that contains ceramic ball. The powder that had to be milled is introduced in the recipient with one ore more ceramic balls. The jar is then shaken. The movement of the balls against the particles breaks them. The biggest particles are more likely crushed. It permits to make the powder finer, to have a smaller particle size distribution and a good homogenization. The device can be seen on Figure19.

25

Figure19: Ball mill shaker and schematic view of the inside of a jar38.

Na2O3 and Al2O3 powders: Na2CO3 and Al2O3 were already in powder form. The powder had to be refined by using a ball mill shaker. Slag powder: The slag was also already in powder form, but it was coated in a wax in order to protect it from humidity. The wax has fist been burned out in the Kanthal furnace at 1000°C. The powder has also been milled. The second step was to prepare the powders mixtures. It was said earlier (II.1 page18) that two sets of samples were prepared. A recapitulative table of the sample composition is shown Figure20.

Study of Na2O effect

Study of Al2O3 effect

Figure20: Composition of the samples

The sample 20Al+0p has been made to be able to have the reaction in this type of mixture, without the addition of Na2CO3; in order to make a differential study.

38

Mill fabricant : http://www.retsch.com

26

An example of the powder sample preparation is given: the 20Al+5p sample (20Al represent the content of the brick in pure Al2O3) 0,4g of brick with 0,1g of Alumina have been mixed, and then 0,5g of slag powder has been added. The total weight of the mixture was 1g. To this mixture 0,05g of Na2CO3 has been added. After the preparation of the powder samples, the powders have been mixed in the ball mill shaker in order to homogenize them.

III.2.2 Pellets preparation

In this project samples in pellet form (compressed powders) were used. The advantage of pellets is that quasi solid samples can be handled, the reaction is more efficient (the particles are closer one to the other), the reaction with the crucibles is lower (the surface of contact is lower), and the samples can be used in the dilatometer. In order to make pellets from powders it is important to crush the powder as fine as possible, then the particles will more likely be fixed together with the Van-der-Waals forces. First compression: Once the powder is finely milled it is placed in a plastic cylinder in order to shape a semisolid pellet. The force has to be adjusted according to the powder grain size and to the powder nature so that the pellets can be removed from the plastic shape without braking. Too much force will make that the pellet stay too strong in the shape and it will break because of the compression with the cylinder walls. Too low force will make that the particles are not compressed enough to stay together. Some previous tests have to be made to obtain the right force. The next step is the Cold isostatic pressing. Cold isostatic pressing (CIP): The semi-solid pellet is introduced in a hermetic plastic bag (plastic glove). Once the pellet is inside vacuum is applied and the bag is sealed with a string. Several bags can be introduced in the CIP machine using a cylindrical plastic holder that is shown on Figure21. This holder is then introduced in the CIP machine. In the autoclave a pressure of 2000bars is applied. The pellet is then quite solid and can be handled with a pair Figure21 : CIP holder of tweezers.

III.2.3 Sample preparation for XRD:
As earlier mentioned, XRD experiments were performed with powders. In order to get a random crystallographic orientation, the powder has to be milled as fine as possible. The powder is prepared from a pellet that is heat treated in the Kanthal furnace. The temperature cycle followed is the one from Figure9.

27

After being crushed, the powder has to be placed in a sample holder. There are different sample holder size and sample holder materials. Two different samples holder have been used: The plastic sample holder and a sample holder made of aluminum. Plastic sample holder: Plastic sample holders do not produce peaks in the XRD spectrum due to their amorphous state. Aluminum sample holder: When a crystalline sample holder is used (such as aluminum) the peaks coming from it do not have to be taken in account. Both sample holder types have been used during these experiments. Once the powder is in the sample holder it is pressed with a plane plate (peace of glass) in order to get a plane surface. It is important to press perpendicularly to the surface in order to avoid a preferential orientation of the grains.

III.2.4 Sample preparation for SEM:
In order to observe the different phases present in the samples and to compare the structures of the different sample, the best way is to have smooth and plane surfaces; mechanical polishing has been used. The first step is to put the sample into a resin. A two component epoxy resin is used. The component are mixed and put into vacuum in order to avoid air bubbles. Then the sample is placed in the bottom of a plastic shape and the shape is filled with the resin. After 24 hours the resin is polymerized and can be handled. The next step is the polishing. A polishing table is used (rotating plate with sand paper on it). Different silica papers with different grain sizes are used for each step. The sample has to be holed on the paper in one direction and rotated on the paper. In order to get a good result the sample has to be turned 90° after each polishing step, and the polishing lines from the previous step has to disappear before going to the next step. Between each step the sample has to be washed with water in order to rinse the particles present on the surface. The papers used had grit sizes of: 800, 1200, 2500 and 4000. For the fine polishing with diamond powder, another plate is necessary. The diamond polishing steps are: 6µm, 3µm, 1µm and 1/4µm Between each step the sample has to be washed with water, put in the ultrasonic bath for 5 minutes and rinsed with ethanol. The plates were also washed before and after use with soap and water, in order to get a good result, and to avoid any contaminations.

28

III Results and discussion
In a first part the results of the structure and phase analysis will be presented and discussed. This includes the results of the XRD experiments, the results of the SEM and EDS analysis. A second part will give the results of the thermal analysis and discuss them. The thermal analyses consist of the first tests in the Kanthal furnace, the dilatometer experiments and the DSC/TG experiments. The last part will shortly explain the work that has been done with a simulation program.

III.1) Structure and phases analysis:
In a first part the results and discussion of the XRD measurement will be given. A second part will give the results of the SEM and EDS analysis. The last part will be the discussions for both analyses.

III.1.1. X-ray diffraction
a) Results of the X-ray diffraction analysis: The spectrum of the experiment realized with the plastic holder can be seen in annexI.1 and the spectrum realized with the aluminum sample holder can be seen in annexI.2. On both spectrums the peak has been correlated using data sheets. The phases that could be clearly identified were: hematite, alumina, silica and mullite, even if the peaks were shifted from 0,2° for the plastic holder and 0,1° for the aluminum holder. By using the XRD analysis made on pure constituent (brick, slag) in a previous work14, all the peaks were identified either corresponding to the brick or to the slag. A strong background noise could also be detected in both the spectrum. One more observation that was made is that both spectrum showed the same peaks, except for the one situated at 35,87°; the intensity of this peak was much higher in the measurement made with the aluminum sample holder.

b) Discussion: The first observation is that both constituents, slag and brick, are still present after the reaction. The peaks are shifted because the height of the sample was not exactly set to the origin. But this does not affect the reliability of the measurements. The two measurements with different sample holders permitted to compare the difference, the advantages and disadvantages of both sample holders. The plastic sample holder is made of amorphous phase and gave a strong background noise between 5° and 20°. The aluminum sample holder showed a peak at 35,87° due to the crystalline aluminum. It is the main disadvantage of this sample holder, because the peak is close to the ones induced by SiO2, Al2O3 and Fe2O3. The measurement with the aluminum holder has been made with a bigger amount of powder and a longer counting time. The advantage is that the peaks appear more clearly, but the background noise is not reduced. It has been reported in a previous work14 that with the addition of Na2CO3 the background noise increased. Furthermore the presence of phase containing Na2O can not be detected even

29

if the concentration of the sample in Na2O is higher than 5% (5% is the limit of detection of the XRD equipment in the conditions of work used). That means that the background noise detected is coming from an amorphous phase and Na2O is contained in this phase. The more Na2CO3 added to the sample, the more amorphous phase is get.

III.1.2. SEM pictures
The first samples that have been observed were the samples with a different Na2O content. The observations permitted to see the structures of the samples, with the different phases and compare the samples.

a) Samples with different Na2O content: Different phases are present in the pictures. Each phase correspond to one color. The nature of the phases has been studied in the EDS analysis. The different phases that are found in the pictures have been named: - Grey phase - White grains - Grey grains - Black grains - White particles - Grey particles - Grey needles - Matrix - Glassy phase

50/50: First the sample 50/50 has been observed, in order to get an idea about the microstructure after the heat treatment. Its microstructure can be seen on the following picture: Figure22.

White particle Grey phase Matrix White grain Grey grain

Figure22 : 50/50 sample observed in SEM, with BE

30

In this sample, containing a low amount of alkalis, the structure is mainly composed of white grains, grey grains, grey phase and matrix. The grey phase has a needles-like shape, is compact and is situated in-between white and grey grains that have a triangular shape and straight grain boundaries. On those grain boundaries a matrix is present.

50/50+2p:
This sample has a higher concentration in alkalis. Its structure can be seen in the following SEM picture: Figure23.

White particle Grey particles Matrix White grain Black grain Grey grain

Figure23: 50/50+2p sample observed in SEM, with BE

The structure of this sample is different the 50/50 sample. New phases are present. The white and grey grains are still present, but their grain boundaries are more smooth. Black grains are also present. They have in average the same size then the white and grey grains. In this sample, instead of having the grey phase, there are grey particles, with a more circular shape, which are bended in a matrix. The quantity of matrix is much higher then in the 50/50 sample.

50/50+5p:
In the sample 50/50+5p, 5% of Na2CO3 have been added. Its SEM picture can be seen in Figure24. The structure in this sample is again different then the two others. The white grains, grey grains and black grains are still present but two new phases were formed. One of these phases is grey needles that are present in the Matrix. The other one is glassy phase (this phase has smooth boundaries and has the appearance of a melt).

31

Glassy phase White particle Grey particles Matrix White grain Black grain Grey grain
Figure24: 50/50+5p sample observed in SEM, with BE

50/50+10p:
This sample is the one containing the highest amount of alkalis (10% has been added). Its structure can be seen in the following SEM picture: Figure25. In this sample the structure is made only of white grains, grey grains, and grey needles in the matrix or glassy phase. The quantity of matrix is really high and the black grains and grey particles were no more present.

Glassy phase/Matrix

White particle Grey needles White grain

Grey grain
Figure25 : 50/50+10p sample observed in SEM, with BE

32

b) Samples with different alumina content of the brick: In this part the samples having different Al2O3 content of the brick have been compared.

5Al+5p:
The sample 50/50+5p can be seen on the following picture: Figure26. Glassy phase/Matrix White grain

White particle Grey needles Black grain

Grey grain Grey particles
Figure26 : 5Al+5p sample observed in SEM, with BE

This sample has a structure that is similar to the 50/50+5p sample. The major difference is that the porosity is higher and there is less matrix. The grey needles are more elongated and there are also more black grains with a smaller size.

10Al+5p:
The SEM picture of the 10Al+5p can be seen in Figure27. The structure of this sample is almost the same then the sample 5Al+5p except that there are more black grains.

33

Glassy phase/Matrix White grain

Grey needles Black grain

Grey grain Grey particles
Figure27 : 10Al+5p sample observed in SEM, with BE

20Al+5p:
The structure of the 20Al+5p sample is shown in Figure28. This sample has a different structure then the 10Al+5p sample. The grey particles are no more present and instead only grey needles are present. The number of black grain is also much higher then in the previous sample and there is only a small amount of matrix.

White grain

Grey needles

Black grain

Grey grain

Figure28 : 20Al+5p sample observed in SEM, with BE

34

50Al+5p:
This sample can be seen in the Figure 29.

White grain

Grey needles Black grain

Figure29: 50Al+5p sample observed in SEM, with BE

This sample has a structure close to the 20Al+5p sample, with more black grains and less matrix.

20Al+0p:
The 20Al+0p is the sample has a low alkalis content and a high aluminum content of the brick. It structure can be seen in Figure30. In this sample the structure is different then in the 20Al+5p sample (sample containing Na2O). The structure has the same type then the one of the 50/50 sample, except that the porosity is higher. That means that the sintering is lower.

Figure30: 20Al+0p sample observed in SEM

35

Samples before the transformation: The sample 50/50+5p and 10Al+5p that have been heated up at 1160°C can be seen respectively in Figure31 and Figure32. The structure of the sample 50/50+10p and 50Al+5p after having been heated up at 1220°C can be seen respectively on Figure33 and Figure34.

Figure31: 50/50+5p sample heated up at 1160°C

Figure32: 10Al+5p sample heated up at 1160°C

Figure33: 50/50+10p sample heated up at 1220°C

Figure34: 50Al+5p sample heated up at 1220°C

The grains in those samples are still distinct one from the others and did not begin to sinter.

36

III.1.4. results of the EDS analysis:
The study of the structure has showed the presence of different phases. The EDS have been done on the different phases listed previously. The sheets containing the data and the places where the data has been taken for all the measurement are in the annex files: AnnexII. On each data sheet it can be seen to what correspond which phase by using the pictures. A special correction of the data had to be done in order to get results closer to the real composition. The detail of the building of this program is given in AnnexIII. Each data sheet contain a table with only the average value of the elements, a table with the average value of each phases, and a table with the calculated value of Fe2O3, Al2O3, SiO2, and Na2O in weight percent and moles percent. Those tables have been used in order to get a better idea of the phases present. Using the compositions of the phase of all the samples, an average value of the composition of each phase has been calculated. The results can be seen in Figure35.
Phase White grain Grey grain Black grain Matrix Grey in matrix Grey needls Glassy O 59,81 59,80 59,88 60,74 60,08 56,09 55,44 Na 1,41 1,46 1,40 1,75 1,76 11,70 15,88 Mg 0,29 0,50 0,05 0,10 0,07 0,03 0,05 Al 8,22 11,46 27,64 14,72 20,53 17,82 10,79 Si 2,03 2,14 2,15 7,49 4,04 4,02 9,31 K 0,06 0,03 0,03 0,15 0,08 0,10 0,16 Ca 0,04 0,05 0,05 0,25 0,11 0,07 0,28 Ti 0,29 0,37 0,20 0,36 0,22 0,16 0,29 Fe 27,85 24,19 8,60 14,44 13,10 10,00 7,80

Figure35: average of the composition of the different phases in all the samples

III.1.5. Discussion about the results of the ESD and SEM analysis:
First the EDS analysis results will be discussed and then the results of the SEM analysis.

EDS:
Using Figure35 it is possible to determine the nature of the different phases present in our samples. Grey phase: The measurements realized on this phase did not permit to conclude on the nature of the phase, but regarding the composition of the sample and according to the literature21, this phase is secondary mullite (3Al2O3.2SiO2). White and grey grains: The composition of the white grains and the grey grains is close to the one measured for slag. Those grains are made of hematite, with some mullite in solution and other impurities.

37

The grey grains are the ones containing more Alumina. The darker color come from the substitution of Fe3+ ions by Al3+ ions in the hematite (Al is lighter then Fe). The more Alumina is present in substitution, the darker get the grain. Black grain: The composition of the black area does not change a lot between the different samples. It is always a grain that has high alumina content. The black grain corresponds to corundum. The grain contain 8mol% Fe in average. This can come from an error of the measurement and the correction, or it can come from some hematite that enters the alumina crystal by substitution, such as alumina in hematite. Matrix: The matrix and the glassy phase are the same phases. The composition is different from one sample to another and that makes the average value from the Figure35 not relevant, but this is due to the different Na2CO3 and alumina content that has been added to the sample. Grey particles in matrix: The grey phase in the matrix shows a composition that is close to an iron-aluminum oxide (Fe2O3Al2O3). Its composition does not change a lot from one sample to another. Grey needles: The composition of the long grey needles also change a lot from one sample to another, and even in the same sample. Those needles have generally a high content of Na2O. This phase has also in average 10mol% of Fe and the rest is Alumina, silica and impurities. The ratio alumina to silica is corresponding to the one of mullite (3/2). Ribeiro21 and Pitak19reported the recrystallisation of tertiary mullite into the liquid phase. This has an elongated needles shape. This explanation fits to the observations in SEM but the composition measured is not corresponding to mullite. Glassy phase: The measurement of the glassy phase has shown in all the samples a high content of SiO2 and Na2O. But also the same amount of alumina then silica, and 7,8mol% of iron. This type of glass has a nepheline composition (Na2O.Al2O3.2SiO2). More: White small particles or needles are also observed. The Ti content in those particles is high. Those particles correspond to titanium oxide (TiO2). The EDS measurement has to be completed with WDS measurements concerning the matrix, the grey particles, the grey needles and the glassy phase.

38

SEM:
Influence of hematite: In the 50/50 sample the hematite grains are surrounded by matrix, and with addition of 2p of Na2CO3 the shape of the hematite grain is more “round”. That means that hematite and mullite react to form iron-aluminum oxide (Na2CO3.Al2O3). Influence of Na2CO3 on the structure: The addition of Na2CO3 in the sample has a large influence on the structure after heat treatment. The more Na2CO3 is added, the more matrix is formed. This matrix corresponds to the amorphous phase that was found in XRD. Under the action of Na2CO3, the mullite is dissolved and new phases are formed: - For a small addition of Na2CO3 (between 2% and 5%) grey particles (iron-aluminum oxide: Na2CO3.Al2O3) and black grains (corundum: Al2O3) are formed. The silica that comes from the dissolution of mullite enter the melt, and the alumina contained in the mullite react with hematite to form Na2CO3.Al2O3 and the rest of alumina form corundum. - With a high addition of alkalis (>10%), no iron-aluminum oxide and no currundum are formed, but instead needles (tertiary mullite) form in the melt. The addition of alumina is favorable for the formation of those needles. This has also previously been reported 19 21 . When Alkalis content is high or the alumina content is high, the melt concentration in alumina is higher, and needles crystallize into the melt. Influence of addition of alumina on the structure: The samples with a high addition of alumina, superior at 20% of the brick, do not form ironaluminum oxide anymore, the quantity of amorphous phase is low and the quantity of needles formed is high. The samples with addition of aluminum also contain more pores then the samples without addition of aluminum, and the more aluminum is added the more pore are present. This can be seen easily if the Figure30 and the Figure22 are compared. This indicates that the addition of aluminum is reducing the sintering. This comes from the fact that the melt present between the grain catalysis the sintering, and alumina is reducing the quantity of melt in the samples by the crystallization of needles. Study of the samples before reaction: In the samples heated up before their main reaction, no big interaction or no reaction between the particles had happened. But some of the mullite, contains “holes”, it correspond to the beginning of the dissolution of the mullite. Figure33 shows the beginning of the formation of nepheline where the spot “spectrum1” is situated. But the main reaction is happening after 1220°C. For this part a larger study should be done, and samples should be prepared at different temperatures after and before each main peaks of the DSC. That would permit to identify more clearly the modification in the samples, and see to which reaction corresponds which peak. For example for the 50/50+5p composition a sample should be heated up to 1220°C and see the modification in structure. For the 50/50+5p, a sample should be heated up at 1300°C in order to see which reaction correspond to the first DSC peak and which to the second.

39

III.2) Thermal analysis
A first part will give the results and discussions of the tests in the Kanthal furnace, a second part will present the results of the Dilatometer. Then the results of the DSC/TG will be given and finally a discussion about the results of the thermal analysis will be made.

III.2.1. Kanthal furnace
a) Results: The tests in the Kanthal furnace have shown the following results: - All of the samples kept their initial shape, they only shrunk due to the sintering. - None of the samples were fixed to the alumina boat - Samples with a high content of Na2CO3 shrank more then the ones with low content. - The samples let brown traces in the bottom of the crucible. A 50/50+10p sample in powder form has reacted with the crucible after a firing at 1350°C. The powder in the boat has shrunk and was fixed to the alumina walls. The fixation was so strong that the boat broke during the extraction of the sintered powder. The powder in the alumina boat after sintering can be seen in Figure36.
Figure36: reaction of the 50/50+10p powder and the alumina boat

b) Discussion: The brown traces that were observed in the crucible are coming from the diffusion of slag in the alumina boat. These phenomena cannot be avoided. The reaction of the sample in powder form shows the advantage of using sample in pellets form compare to using samples in powder form. The reaction with the crucible is reduced. The conclusions for the test in the Kanthal furnace is that all the samples, except the powder samples, are suitable to be used in the Dillatometer and DSC/TG. None of them is too reactive, explosive, becoming liquid or else.

III.2.2. Dilatometer results:
a) Effect of the Na2CO3 content: The dilatometer curves of the samples having different Na2CO3 content can be seen in the following graph: Figure37.

40

50/50 50/50+2p 50/50+5p 50/50+10p

Figure37: Dilatometric curves of samples: 50/50 (1.1), 50/50+2p (2.1), 50/50+5p (3.1), 50/50+10p (4.1)

In the Figure37, it can be seen that between 50°C and 200°C the curves show a shrinkage of all the samples. The difference between the dilatometric curves appears at 600°C. The 50/50 sample does not show any transformation at this temperature. For the other curves (2.1, 3.1 and 4.1), between 550°C and 800°C, a major modification in the curves slope has happened. The 50/50+2p tend to shrink and the 50/50+5p and 50/50+10p tend to expand. The more Na2CO3 is added, the more the samples expend. At 900°C the sample 50/50 has begun to shrink. This corresponds to the sintering of the powder. At 1030°C the 50/50+2p sample begins to shrink quite fast until 1100°C when the shrinkage slow down again. At 1170°C all the samples except the 50/50 sample are shrinking fast due to a fast sintering (sintering in liquid). At 1255°C for the 50/50+5p and at 1335°C for the 50/50+10p sample the slope of the curve have changed. This shows the end of a reaction, the apparition of a new phase or a phase transformation. b) Test of the reversibility of the reactions: The results of reversibility test on the 50/50+10p sample are given in the Figure38.

41

50/50+10p first run 50/50+10p second run

Figure38: Dilatometer curve of 50/50+10p and second time for 50/50+10p (same sample)

The green curve corresponds to the dilatometric curve of the 50/50+10p sample during the first temperature cycle. The blue curve corresponds to dilatometric curve of the same sample during the second temperature cycle. Those two curves permit to see that the main reaction happening between 550°C and 800°C is a non reversible reaction. And that the reaction happening at 1235°C is a reversible reaction.

III.2.3. Results of the DSC/TG and mass spectroscopy

In a first part the result of the study of the influence of the addition of Na2CO3 will be given and in a second part the results of the influence of the alumina content of the brick. The last part will give the results of the test of the reversibility of the reactions and also the results of a test of reproducibility of the measurements. a) Effect of Na2O: The result of the DSC/TG measurement of the samples 50/50, 50/50+2p, 50/50+5p, 50/50+10p can be seen on Figure39. Different peaks are present for the different samples, but the reactions happened at the same temperature for all the samples. - The first reaction that has been detected happened between 70°C and 170°C, for all the samples except for the 50/50. It corresponds to an endothermic reaction with lost of mass. The more Na2CO3 was added in the sample, the stronger is the reaction, and bigger is the lost of mass. The mass spectrometer shows peaks for almost all the mass (28, 14, 32, et. al).

42

- The second reaction that happened in all the samples except the 50/50 sample, is an exothermic reaction followed directly by an endothermic reaction. Those reactions started at 450°C. The TG curve shows that those reactions also induced a loss of mass. The mass spectrometer detected the lost of CO2 (peaks appear for the mass: 14, 44, 12, 22 and 46). The more Na2CO3 is added, the stronger is the reaction, the more mass is lost, and the longer does the reaction last.

- The third reaction happened only in the samples with an addition of Na2CO3 inferior at 10% (in the 50/50+2p and in the 50/50+5p). It is an endothermic reaction that starts at 1175°C. This reaction does not induce any mass lost.
- The next reaction start at 1235°C in the 50/50+10p sample. - The last reaction takes place at 1300°C in all the samples with addition of Na2CO3. A further observation is a lost of mass that began at 1000°C, for all the samples. No peak could have been found in the mass spectrometer curves. That means that the spectrometer was not measuring the mass that is lost.

Figure39: DSC/TG curves of 50/50, 50/50+2p, 50/50+5p,

b) Effect of the brick content in Alumina (Al2O3) : The DSC/TG curves of the samples 5Al+5p, 10Al+5p, 20Al+5p and 50Al+5p are shown in Figure40. It can be seen that for all the samples the reactions that happened before 1100°C

43

are the same then previously. For all the samples, the addition of 5p of Na2CO3 induced the same lost of masses than for the reaction at 450°C. The influence of the addition of alumina can be seen by comparing the reactions taking place after 1100°C: The samples with an addition of alumina to the brick lower then 20%(5Al+5p and 10Al+5p) begins to reacts at 1175°C and the samples with an addition of alumina higher then 20% begins to react at a 1235°C.

Figure40: DSC/TG curves of 5Al+5p, 10Al+5p, 20Al+5p, 50Al+5p

c) Test of the reversibility of the reactions: Three test of reversibility of the reaction have been performed on three different types of samples: a sample that reacted at 1235°C with a high content of alkalis (50/50+10p), a sample that reacted at 1175°C (50/50+5p) and one sample that reacted at 1175°C with a high content of alumina (10Al+5p). The three runs of the 50/50+10p sample can be seen in the following graph: Figure41.

44

50/50+10p first run 50/50+10p second run 50/50+10p third run

Figure41: 3 runs of 50/50+10p (same sample)

The only reaction that is reversible is the reaction happening at 1175°C. All the others reactions are irreversible for this sample. For the second run a small peak appeared at 1300°C. It corresponds to the same reaction that happened at 1300°C during the first run, which was not completely finished. There is also a difference between the second and the third run that appeared at 670°C. The 2nd run shows an exothermic peak and the third run an endothermic peak. A new correction for the third run “correction2” has been used. A difference between the first and the second correction at this temperature can be seen and it is not possible to conclude if there is an endothermic or exothermic reaction at this temperature. The test of the reversibility of the reaction that happen at 1175°C in the sample 50/50+5p is shown in Figure42 and it permits to see that this reaction is irreversible.

45

50/50+5p first run 50/50+5p second run

Figure42: The two runs of the sample 50/50+5p (same sample)

Finally the test of reversibility of the reaction that started at 1175°C with a sample with addition of alumina (10Al+5p) can be seen in Figure43 and shows that the reaction happening at this temperature for samples with high content of alumina is also not reversible.

10Al+5p first run 10Al+5p second run 10Al+5p third run

Figure43: The 3 runs of the sample 10Al+5p (same sample)

46

In this sample it can be seen such as previously a difference between the two last runs at 670°C. An endothermic peak appeared for the second run and none for the first run. For the second run, this time, the “correction2” have been used. The same error then previously is seen: It can be concluded that the second correction contains an error, because it introduced an exothermic peak at 670°C. Study of the reproducibility of the measurements: The run of the second sample of 50/50+10p, is shown in Figure44. The measurement has been made with using the same correction file. The peaks appear at the same temperature and the energies of transformation are similar. The measurements have a good reproducibility.

50/50+10p 1st sample 50/50+10p 2nd sample

Figure44: 50/50+10p run with two different samples and the same heating program

III.2.3. Discussion for the DSC/TG and dilatometer measurements:
The combination of the data from the dilatometer, the DSC/TG + mass spectrometer and the test of reversibility give information that permits to identify the reactions taking place during the temperature cycle. The Figure45 is a recapitulative figure that shows the different reactions and phenomena in the DSC/TG and dilatometer curves. Reaction between 70°C and 170°C: The first reaction taking place between 70°C and 170°C is an irreversible reaction, inducing a lost of mass, with an energy and a mass lost dependent to the content of Na2CO3. This reaction corresponds to the deshydratation of the Na2CO3 powder. Deshpande made thermal analysis on Na2CO3.H2O and showed from dynamic TG that the single molecule of water of crystallization is lost in two steps of 0,3mole and 0,7mole at temperatures 153 ± 5°C and 181 ± 5°C, respectively39. In the experiments the reaction is happened at lower temperature. It also
Thermal analysis of Na2CO3 · H2O crystals Thermochimica Acta, Volume 60, Issue 3, 1 February 1983, Pages 295-302 D. A. Deshpande. R. Ghormare, V. L. Jawadekar and N. D. Deshpande
39

47

seams that the mass spectrometer was not detecting well this lost of water, and that the signal of the water vapor interfered with the other signals. The TG curve permits to approximate the water content of Na2CO3 in each sample. But this measurement is not precise, because the sample itself, also contain water, not only Na2CO3 contains water. The water content for the sample 50/50+2p, 50/50+5p, 50/50+10p was respectively: 0,5%, 0,4% and 2%. The dilatometer curve shows a shrinkage between 50°C and 200°C. This shrinkage do not comes from the lost of water but from a bad contact of the connecting rod when the correction file has been done.

70

450

1175

1300 1135

Figure45: Dilatometric curve, TG curve and DSC curve for the sample 50/50+10p

Reaction between 450°C and 700°C: This reaction is irreversible, with an exothermic reaction and an endothermic reaction that induce the dilatation of the sample. The dilatation, the mass lost, the reaction energy are proportional to the amount of Na2CO3 added. The mass spectrometer detected the lost of CO2. This reaction corresponds to the decomposition of Na The lost of CO2 induces a lost of material and Na2CO3 is a bigger molecule then Na2O. When CO2 is lost, Na2O can then diffuse, and this induces the shrinkage of the sample. But CO2 at this temperature is in a gas form. That means that it expend. If there is not enough space and opened porosity for CO2 to escape that will involve an expansion of the sample in order to liberate CO2. That means that there are three simultaneous phenomena (2 that induce shrinkage and one expansion). In the low content Na2CO3 samples the shrinkage is stronger, and in the high Na2CO3 content samples the expansion is stronger. It seams that the equilibrium is situated between the content of the 50/50+2p sample and the 50/50+5p sample.

48

The exothermic peak at 450°C corresponds to the reaction of decomposition and the endothermic peak around 600°C corresponds to the evaporation of CO2. Reaction at 1175°C: This reaction is also irreversible and according to the study of the structure, the samples reacting at this temperature are the samples forming iron-aluminum oxide (Fe2O3Al2O3). This reaction corresponds to the dissolution of mullite due to the impurities. The peak observed correspond to the formation of iron-aluminum oxide (Fe2O3Al2O3), corundum and a nepheline (Na2O.Al2O3.2SiO2) phase. Reaction at 1235°C: The test of reversibility of the reaction prove that this reaction correspond to the melting of the nepheline phase. Reaction at 1300°C: The starting temperature of this reaction cannot be known, but the maximum of the DSC peak is situated at 1300°C. The dilatometer tests permits to say that this reaction is the formation of a new phase. It is the formation of the grey needles that has been observed in the SEM analysis (tertiary mullite). Lost of mass after 1000°C: The lost of mass detected by the TG correspond to the lost of Na2O. Different work mentions that Na2O can evaporate from the silica melts containing Na for temperature below 800°C.40 41 19

III.5) Simulation
The thermodynamical equilibriums for all the samples from 0 to 1350°C have been calculated, but the results are similar. Only one example will be seen; the 50Al+5p sample. The results of the calculation for this sample can be seen in the following graph: Figure46. The first reaction is happening at 1140°C, it is the transformation of hematite and Alumina into Fe2O3Al2O3. Then at 1220°C, NaAlSiO4, a nepheline melt is forming. The limitation in the use of this program is the library. In the study different libraries have been used it was observed that the result change completely from one library to another.

Evaporation rates of Na from Na2O-SiO2 melt at 1Atm, Departement of Earth an Space Science, Graduate School of Science, Osaka University, 1-1 Machikaneyama-cho, Toyonaka, 560-0043, JAPAN, shirai@ess.sci.osaka-u.ac.jp, T. Shirai, S. Tachibana and A. Tsuchiyama. 41 Evaporation rates of hot sodium clusters, Zeitschrift für Physik D Atoms, Molecules and Clusters Volume 20, Numbers 1-4 / March, 1991,pages 119-122, A. Mañanes, M. P. Iñiguez, J. A. Alonso and M. J. López.

40

49

Figure46: results of Thermodynamical calculus with « Equilibrium » for 50Al+5p

50

IV Conclusion
The goal of the project was to find the reactions happening between iron ore slag and refractories. The conclusions are the following: The slag attack is strongly influenced by the concentration of alkalis and the Alumina content of the brick. Not the same reactions happen when the concentrations of alumina and soda are changed: For a low content of alkalis (20%): The mullite is also dissolved at 1235°C and due to the high concentration of alumina, the melt crystallize almost completely into needles. The addition of 20% or more of alumina has a positiv effect on the resistance of the brick. It reduces the sintering and the destruction of mullite happened at a temperature 60°C superior then without addition of alumina.

Further work: - In order to get more precise compositions of the phases present after the reaction, a further study on the phases should be performed, using WDS or TEM. - The composition of the needles could be find by using XRD measurement on the sample with a high content of alumina. The quantity of the glassy phase is low, that mean that the background noise will allow the indexation of the peaks. - In order to use the slag resistance property of the brick containing more alumina, it would be interesting to try to design and test a brick having a higher content of alumina in first centimeters of the surface in contact with the slag. For example: to design a brick with a gradient in alumina concentration. - DSC runs should be made for each steps of the transformation in order to characterize better each transformation.

51

Acknowledgements
I would like to thank: My advisors: - Jesper Stjernberg for always being here when I needed him, for his great help and teaching and for being greatly involved in my project. - Marta-Lena Antti who gave me a lot of advice and permitted to conduct this project in really good conditions. - Lars-Olof Nordin for the visit at the LKAB mine in Kiruna and his good advices. Axel Knutsson, Johanne Mouzon, Jonny Grahn, Peter Hedström and Ida Kero for their precious help and good working conditions. LKAB and LTU, for giving me the opportunity to carry out this project.

52

References
1

Facts about LKAB, Presentation brochure march 2007 Metal Prices in the United States through 1998 Iron

2

http://minerals.usgs.gov/minerals/pubs/metal_prices/ ore, by William S. Kirk www.LKAB.com, Products, iron ore Products

3

Niiniskorpi V., LKAB (2001): Phases and Microstructures in LKAB’s Olivine and Dolmite fluxed pellets. P.768
5

4

http://www.energysolutionscenter.org/HeatTreat/MetalsAdvisor , agglomeration, equipment.

Grate-kiln-cooler. Where to oxidize and why?, Niiniskorpi V., LKAB 2002 Ironmaking Conference Proceedings
7 8

6

http://www.metsominerals.com/ , Iron ore pelletizing

Handbook of Ceramics and Composites, by Nicholas P. Cheremisinoff, Published 1990 CRC Press, ISBN 0824780051
9

Technical informations by Höganäs Refractories

10

Mullite and Mullite Ceramics, by H.Schneider, K. Okada, J.A. Pask. Published 1994 by John Wiley & Sons Ltd. ISBN:0471942499

11

Mullitisation behaviour of south Indian clays Ceramics International, Volume 28, Issue 5, 2002, Pages 557564 V. Viswabaskaran, F. D. Gnanam and M. Balasubramanian

Ceramic Microstructures, by William E. Lee and W. Mark Rainforth, published in 1994 by Chapman & Hall.P.
13

12

Action of Heavy Oil Combustion Products on Aluminosilicate Refractories. Material Science and Engineering, 88 (1987) p.75-80. G. Mascolo and O. Marlino Research project: Infiltration of iron slag into refractory bricks, Mariya Chestnyeysha, LTU 2006, supervisor: Magnus Oden and Jesper Stjernberg. P7. http://fr.ekopedia.org/Carbonate_de_sodium

14

15

Thermodynamic properties and phase transitions of salt hydrates between 270 and 400 K II. Na2CO3 · H2O and Na2CO3 · 10H2O The Journal of Chemical Thermodynamics, Volume 15, Issue 9, September 1983, Pages 881-889 Fredrik Grønvold and Knut K. Meisingset
17 18

16

Y. Otsubo, K. Yamaguchi, J. Chem. Soc. Japan, 82, 557-560 (1961)

The transfomation of aluminium oxide from the beta to the alpha form, John GalluP 1RCA RADIOTRON DIVISION RCA MANUFACTURING CO/ INC. HARRISON, NEW JERS RCA RADIOTRON DIVISION RCA MANUFACTURING CO/ INC. HARRISON, NEW JERSEY, Presented at the Annual Meeting, American Ceramic Society, Buffalo, N. Y., February, 1935 (Refractories Division). Interaction of mullite-corundum refractories and alkalis under conditions of variable gas media Translated from: Ogneupory, No.7, July 1992, Pages 24-26. N. V. Pitak, Ukrainian Research Institute of Refractories, Kharkov, Ukraine.
19

53

Synthesis and mechanical properties of mullite from beach sand sillimanite: effect of TiO2 Journal of the European Ceramic Society, Volume 18, Issue 14, December 1998, Pages 2081-2087 H. S. Tripathi and G. Banerjee 21 High temperature mullite dissolution in ceramic bodies derived from Al-rich sludge Journal of the European Ceramic Society, Volume 25, Issue 5, February 2005, Pages 703-710 M. J. Ribeiro, D. U. Tulyagavov, J. M. Ferreira and J. A. Labrincha Influence of thermal aging on Microstructural Development of Mullite Containing Alkalis American Ceramic Society: Volume 81, Issues [10], 1998, Pages 2741-2745 C. Baudìn and M. Pilar Villar.
23 22

20

http://cxpi.spme.monash.edu.au/xray_history.htm http://epswww.unm.edu/xrd/xrdbasics.pdf http://www.mrl.ucsb.edu/mrl/centralfacilities/xray/xray-basics/index.html http://lsu.epfl.ch/cours/ultra/Bas-XRD.pdf http://www.netzsch-thermal-analysis.com/

24

25

26

27

Dilatometry, University of Cambridge, Materials Science & Metallurgy 2002, H. K. D. H. Bhadeshia. http://www.msm.cam.ac.uk/phase-trans/2002/Thermal3.pdf
29

28

Thermal analysis of materials, by Robert F. Speyer, Published in New York : Dekker, cop. in 1994, ISBN: 0-

8247-8963-6
30

Handbook of thermal analysis of construction materials, by V.S. Ramachandran et al., Published by Norwich, N.Y. : Noyes Publications : William Andrew Pub., in 2002.

31

Thermal Analysis Technique, University of Cambridge, Materials Science & Metallurgy 2002, H. K. D. H. Bhadeshia. http://www.msm.cam.ac.uk/phase-trans/2002/Thermal1.pdf Differential Scanning Calorimetry, University of Cambridge, Materials Science & Metallurgy 2002, H. K. D. H. Bhadeshia. http://www.msm.cam.ac.uk/phase-trans/2002/Thermal2.pdf Netzsch dilatometers : http://www.dilatometers.com/

32

33

Thermogravimetry, University of Cambridge, Materials Science & Metallurgy 2002, H. K. D. H. Bhadeshia. http://www.msm.cam.ac.uk/phase-trans/2002/Thermal4.pdf
35

34

http://www.culture.gouv.fr/culture/conservation/fr/methodes/meb_01.htm

Danilatos, G,D (1988). "Foundations of environmental scanning electron microscopy". Advances in Electronics and Electron Physics 71: 109-250
37

36

http://en.wikipedia.org/wiki/Wavelength_dispersive_X-ray_spectroscopy Mill fabricant : http://www.retsch.com

38

39 Thermal analysis of Na2CO3 · H2O crystals Thermochimica Acta, Volume 60, Issue 3, 1 February 1983, Pages 295-302 D. A. Deshpande. R. Ghormare, V. L. Jawadekar and N. D. Deshpande 40

Evaporation rates of Na from Na2O-SiO2 melt at 1Atm, Departement of Earth an Space Science, Graduate School of Science, Osaka University, 1-1 Machikaneyama-cho, Toyonaka, 560-0043, JAPAN, shirai@ess.sci.osaka-u.ac.jp, T. Shirai, S. Tachibana and A. Tsuchiyama.

Evaporation rates of hot sodium clusters, Zeitschrift für Physik D Atoms, Molecules and Clusters Volume 20, Numbers 1-4 / March, 1991,pages 119-122, A. Mañanes, M. P. Iñiguez, J. A. Alonso and M. J. López.

41

54

42

Elin Eliasson, Institutionen för geovetenskaper, Berggrundsgeologi, Villav. 16, 752 36 UPPSALA, Elin.Eliasson@geo.uu.se, Phone: 0046702-50 51 04

Examensarbete: Microstructural studies on the growth of FeOx on Al2O3-SiO2 based brick insulation materials, Antje Schlieter, LTU 2006, supervisor: Magnus Oden.
44

43

Examensarbete: Slaggangrepp på teglet I Svappavaaras roterugn, Jesper Stjernberg, LTU 2006, supervisor: Magnus Oden.

55

80

Plastic holder
33,425

Fe2O3 Al2O3

70

60

SiO2 Mullite

50

35,895

AnnexI.1: 50/50+10p with plastic holder:

40
21,135 24,355

30
26,595

29,705 41,155 49,705 54,445 62,885 64,545 57,585 66,125 72,615 76,065

20

43,265

Annex I: XRD

VI Annex

10

0 20 30 40 50 60 70 80

56

aluminium holder
100 35,865

Fe2O3 Al2O3
80

SiO2
33,365 60 29,605

Mullite

AnnexI.2: 50/50+10p with aluminium holder:

23,085 15,645 40 21,185 24,235 38,365 41,315 43,645 20 49,915 54,325 56,935 62,765 65,975 67,485 76,135 89,915 27,185 30,685

0 15 25 35 45 55 65 75 85 95

57

Annex II: EDS
AnnexII.1: 50/50+0p:

58

59

AnnexII.2: 50/50+2p:

60

61

AnnexII.3: 50/50+5p:

62

63

AnnexII.4: 50/50+10p:

64

65

66

AnnexII.5: 5Al+5p:

67

68

AnnexII.6: 10Al+5p:

69

70

AnnexII.7: 20Al+5p:

71

72

AnnexII.8: 50Al+5p

73

74

AnnexIII: Building of the correction for the data:
The first measurements that have been down showed the presence showed a dispersion of the results compare to the reality, a program have been done to correct the data. First measurements: The results of the 50/50+0p sample are given in the following table: Figure40.
0p White grain Point A1-2 A1-5 A2-2 A2-6 A2-8 Avrage A1-1 A1-6 A2-1 A2-5 Avrage A1-3 A1-8 A2-3 Avrage A1-4 A1-7 A2-4 C 3,97 0 0 5,1 1,88 2,19 3,97 0 0 7,31 2,82 0 6,55 0 2,18 0 0 0 O 16,31 13,97 13,64 12,92 13,33 14,03 16,31 16,31 15,68 14,03 15,58 24,73 21,12 23,49 23,11 17,17 20,21 22,67 23,64 20,92 27,78 27,78 Na 0 0,11 0 0,09 0,18 0,08 0 0 0 0 0,00 0,23 0 0 0,08 0,2 0,34 0,14 0 0,17 0 0,00 Mg 0 0,41 0,25 0,65 0 0,26 0 0 0,14 0,3 0,11 0 0,29 0,57 0,29 0 0 0 0,4 0,10 0 0,00 Al 10,38 9,17 7,86 7,68 7,58 8,53 10,38 10,93 11,53 9,94 10,70 18,44 14,99 15,98 16,47 10,49 13,79 18,38 14,75 14,35 34,81 34,81 Si 3,18 2,9 2,36 2,41 2,35 2,64 3,18 2,8 2,47 1,94 2,60 8,28 7,87 9,76 8,64 3,55 6,14 8,52 8,89 6,78 2,16 2,16 K 0,31 0,33 0,34 0 0,17 0,23 0,31 0 0,33 0 0,16 0 0 0,47 0,16 0,19 0,3 0,26 0 0,19 0 0,00 Ca 0 0,32 0,45 0,24 0 0,20 0 0,29 0 0 0,07 0,56 1,02 1,16 0,91 0,58 0,48 0,54 1,44 0,76 0,31 0,31 Ti 2,12 0,86 0,81 1,05 2,53 1,47 2,12 1,6 1,82 2,13 1,92 1,34 1,18 1,23 1,25 2 2,3 1,32 1,48 1,78 1,33 1,33 Fe 63,78 71,92 74,34 69,85 71,97 70,37 63,76 68,12 68,03 64,36 66,07 46,42 47,01 47,49 46,97 65,82 56,43 48,16 42,54 53,24 33,61 33,61

Grey Grain

Grey grain in matrix

Matrix

A2-7 7,12 Avrage 1,78 Black area SI1 0

Avrage 0,00

Figure40: average composition in mol% of the phases present in the sample

The first observation is that the content of carbon is changing a lot between the measurements in the same phases, such as oxygen do. This elemental analysis gives high iron content and low oxygen content in all the phases. The lower content measured is in the black area with 33mol% and the highest is 70,37mol%. Pure hematite has an iron content of 40mol%. That means that the higher content that a phase could have is 40mol% (no lost of oxygen have been detected in the mass spectrometer). The iron content measured is too high compare to the reality and the oxygen content too low. The explanation of this phenomenon is that the energy coming from the lighter element is partially absorbed by the heavier elements. That means that the heavy elements will be

75

detected with a higher concentration then they have and the light elements with a lower concentration, when light and heavy elements are simultaneously measured. EDS apparatus uses a correction in order to correct this phenomena. Anyway, this correction is not made to work with oxides. If the measurements are made on oxides with the same range of mass, for example only metallic oxides, then the solution is not to take the oxygen content into account. Then to get the percentage of oxygen a simple calculation is made. This is what has been done in projects working with oxides, such as in Elin Eliasson’s PHD thesis.42 In this project it is important to detect elements with a big difference in mass, for example iron oxide and sodium oxide. Iron has a mass of 54 and sodium a mass of 23, oxygen is 16. That means that iron will absorb much more the signal of oxygen then sodium. If hematite and soda are mixed, the iron content will be too high compare to what was put in the sample. In order to get results closer to the exact composition, a special correction has been build. But this correction can only be used with the same elements present and with almost the same composition then the samples of this study.

Correction of the data:
The steps realized to build the correction are the following: - Calculate the composition of each of our samples - Measure the global composition of each of the samples - Calculate a coefficient of correction for each element using the ratio between the calculated compositions and the measured compositions - Use this coefficient of correction and make a final correction on the oxygen content.

a) Composition of each sample:
The composition of the brick and the composition of the slag are used to calculate the composition of the samples. The water content the Na2CO3 powder has also to be found. The Figure6 and Figure8 give the composition of the brick and the composition of the slag respectively. To calculate the water content of the Na2CO3 powder, a sample containing 50 wt% of Na2CO3 powder has been weighted. The mass of Na2CO3 powder was so known. Then the sample has been dried at 200°C and weighted again. The mass lost correspond to the water lost. This permitted to find that Na2CO3 powder was containing 16wt% of water. It has also to be taken into account that Na2CO3 decompose in Na2O and CO2. The content in mass of Na2CO3 in CO2 is 41,5%. Only 49,2wt% of the Na2CO3 powder is Na2O. From the weight percent of brick, of slag of Na2CO3 and of Al2O, the weight percent of Fe2O3, Al2O3, SiO2, Na2O, TiO2, CaO, MgO, K2O in the sample have been calculated and converted it in mol% of O, Na, Mg, Al, Si, K, Ca, Ti, Fe.

42

Elin Eliasson, Institutionen för geovetenskaper, Berggrundsgeologi, Villav. 16, 752 36 UPPSALA, Elin.Eliasson@geo.uu.se, Phone: 0046702-50 51 04

76

The so calculated composition of each the samples are given in the following table: Figure36.

Calculated 50/50 50/50+2p 50/50+5p 50/50+10p 5Al+5p 10Al+5p 20Al+5p 50Al+5p 20Al+0p 50Slag+50Na Brick Slag

O 61,0 60,4 59,5 58,1 59,4 59,4 59,3 59,0 60,8 45,2 61,8 59,9

Na 0,2 1,6 3,8 7,0 3,8 3,8 3,7 3,7 0,1 37,0 0,2 0,1

Mg 0,4 0,4 0,4 0,4 0,4 0,4 0,4 0,4 0,4 0,5 0,0 1,0

Al 16,6 16,2 15,7 14,9 16,0 16,3 17,0 18,9 17,9 0,1 28,4 0,3

Si 5,4 5,3 5,1 4,8 4,9 4,6 4,2 2,7 4,4 0,4 8,6 1,0

K 0,0 0,0 0,0 0,0 0,0 0,0 0,0 0,0 0,0 0,0 0,0 0,0

Ca 0,2 0,2 0,2 0,2 0,2 0,2 0,2 0,1 0,2 0,1 0,1 0,3

Ti 0,4 0,4 0,4 0,4 0,4 0,4 0,3 0,2 0,4 0,0 0,7 0,1

Fe 15,8 15,5 15,0 14,2 15,0 15,0 14,9 14,9 15,8 16,6 0,3 37,2

Figure36: Calculated composition of the samples.

Samples with pure brick, pure slag, pure alumina and a mix with 50wt% of Na2CO3 powder and 50wt% of Slag have been realized in order to determine in which domain our calculated factors are valid.

b) EDS global measurement of each sample:
The second step was to measure a global composition of each sample, using EDS. A measuring window of approximately 1600µm² has been used. The compositions measured are given in Figure37.
Measured 50/50 50/50+2p 50/50+5p 50/50+10p 5Al+5p 10Al+5p 20Al+5p 50Al+5p 20Al+0p 50Slag+50Na Brick Slag O 17,135 17,84 17,355 21,2 19,3 17,86 18,42 19,455 15,62 7,95 50,57 12,59 Na 0 0,875 2,625 5,32 2,65 1,89 2,13 2,55 0,07 13,61 0 0 Mg 0 0,21 0,085 0 0,21 0 0 0,00 0,30 0 0,44 Al 12,385 12,335 12,005 13,05 13,58 11,51 14,6 16,905 10,42 0,00 30,76 0,25 Si 4,38 4,93 4,45 4,87 4,41 3,88 3,7 2,615 2,41 0,26 9,79 0,66 K 0,125 0,17 0,29 0,33 0,41 0,27 0,23 0,16 0,25 0,55 0 Ca 0,515 0,495 0,435 0,41 0,47 0,6 0,45 0,295 0,42 0,37 0,44 0,18 Ti 1,735 2,115 1,42 1,25 1,31 1,38 1,34 0,68 0,99 0,00 4,76 0,28 Fe 63,715 61,035 61,33 53,56 57,65 62,61 59,14 57,495 69,85 77,57 3,12 85,61

Figure37: Measured compositions of the samples.

c) Correction factors:

77

In order to simplify the problem, the error had a linear dependence with the composition for each element. Example for Fe: Real content of Fe = xFe*measured content of Fe , where “xFe” is the correction factor for Fe. From the measurement and the calculated compositions, a correction factor is calculated. This factor is the division of the calculated data by the measured data. The results are given in Figure45.
Factors 100/0/50/0 100/0/50/2 100/0/50/5 100/0/50/10 95/5/50/5 90/10/50/5 80/20/50/5 50/50/50/5 80/20/50/0 0/0/50/50 Brick Slag Average O 3,559 3,383 3,427 2,742 3,079 3,325 3,218 3,032 3,891 5,682 1,221 4,758 3,534 Na 1,880 1,434 1,317 1,419 1,986 1,757 1,454 1,833 2,715 Mg 2,031 4,854 1,965 Al 1,339 1,314 1,307 1,140 1,179 1,419 1,162 1,117 1,723 0,924 1,230 1,300 Si 1,232 1,070 1,146 0,994 1,103 1,193 1,124 1,050 1,825 1,716 0,877 1,515 1,245 K 0,122 0,088 0,050 0,042 0,035 0,054 0,063 0,099 0,065 0,000 0,069 Ca 0,325 0,330 0,364 0,366 0,332 0,257 0,334 0,470 0,383 0,362 0,161 1,667 0,352 Ti 0,247 0,198 0,285 0,307 0,295 0,267 0,247 0,325 0,356 0,141 0,337 0,281 Fe 0,249 0,254 0,244 0,265 0,260 0,239 0,253 0,259 0,226 0,214 0,097 0,435 0,246

1,569 2,357

1,755

2,605

Figure45: Correction factors for each samples and element and average

An average correction factor is calculated on all the samples, except the pure slag and pure brick. The pure slag and pure brick only permitted to have an idea about the linearity, for high content, of the dependence between the measured data and the real content. The linearity is not respected. That means that in the case of alumina and Silica, with a high content of those elements, the content after correction will be to high. With the slag, the content of iron, after correction is going to be too low compared to the reality. The average correction factors give results closer to the reality then without correction, even if those results do not have a high precision.

d) Correction of the Oxygen content:
After the use of the corrections factors, there is an error reminding on the oxygen content. A further correction is made such as in the work of Elin Eliasson42.

78