The Effect of Iron Oxide Impurities on the Hot Properties of Mullite
Steven Ashlock* and Austin Scheer
Kyanite Mining Corporation
ABSTRACT
Mullite is an important material in the refractory industry due to its exceptional hot properties. Most commercially available mullite materials are non-stoichiometric and therefore have much lower maximum usage temperatures than theoretically possible. It is often thought that the alumina content of the mullite aggregate or grain is the most important property when deciding between two materials. However, the number of impurities is equally important. Inclusions of fluxing materials, such as iron oxide, alkalis, and alkaline earth oxides lower the maximum usage temperatures by reacting with the silica in the refractory and creating glassy phases. To test the importance of impurity level on the refractory, samples of Virginia MulliteTM with static alumina contents and varying iron oxide contents ranging down to 0.1% were tested. Post fired modulus of rupture, creep resistance, and the thermal coefficient of expansion were all examined and showed that decreasing the amount of iron oxide was beneficial for the refractory grain. XRD was also used to compare glassy phase formation at different temperatures with varying iron oxide contents.
INTRODUCTION
Aluminosilicate materials have played an important role in the history of ceramics. Early examples can be seen as far back as 1500-1000 BC, when the Chinese created in-situ mullite during the firing of clay pottery.1 The use of aluminosilicates in refractories began with cutting natural stone into block.2 The depletion of these natural refractories necessitated the technological development of fireclay products round the time of the industrial revolution.1 As technology has progressed, lesser alumina content fire clay has largely been replaced with aggregates of various alumina content, depending on the application.2
One such material is mullite. The term mullite has been used to describe the family of aluminosilicates with an alumina content of 56-79%.3 Mullite has a high refractoriness, is resistant to thermal shock, and exhibits excellent hot properties, such as creep resistance.1 These properties have popularized the use of mullite in refectories around the world. However, pure mullite is rarely found in nature due to its geologically high temperature/low pressure formation criteria. As such, it is not mined for industrial purposes.
Mullite used for industrial purposes must be created using precursor minerals. There are three main methods of formation. The first is extruding or spheredizing clay minerals followed by a calcination step. Another formation method is the calcination of the one of the sillimanite group of minerals. Alternatively, pure silica and alumina can be melted in an electric arc furnace to form fused mullite, which can have a much higher alumina content.
Fig. 1. The Alumina-Silica Phase Diagram shows that lowering the alumina content from stochiometric values lowers the melting point.
Creating mullite via fusion can produce a high-quality product with alumina values close to or exceeding the stochiometric alumina content (72 wt%), depending on the blend of raw materials. The other two formation methods generally create an aluminosilicate material that has a lower alumina content than stochiometric mullite. Lowering the alumina content has a negative effect on the melting point of the created mullite, as shown in Fig. 1.
These different formation methods produce a mullite material that can range greatly in their alumina contents based on the purity of the raw materials. The most important property of the created mullite is the alumina content. Simply stated, higher alumina percentages lead to a higher refractoriness. While this is generally true, the impurities in mullite can have a greater effect on the usage temperature than the overall alumina content if in high enough concentrations. The geology of the raw material deposits, the beneficiation methods of those raw materials, and the processing conditions all impact the final mullite aggregate produced.
Iron oxide is one of the most problematic impurities in refractories. Introductions of iron oxide to a pure alumina-silica system cause the formation of a liquid phase.5 This liquid phase penetrates the porosity of the refractory, allowing for thermochemical reactions to occur and create new phases.6 These phases lower the hot properties of the refractory, such as resistance to creep at high temperatures. Mullite materials with alumina contents lower than the stochiometric 72% intrinsically have an excess of silica in their composition. This silica will react with iron oxide impurities to create a glassy phase, creating an aggregate that will have negative effects on the hot properties of the refractory body.
As lower iron contents are known to be generally beneficial for the hot properties of the refractory, Kyanite Mining Corporation (KMC) began to experiment with ways to remove excess iron oxide from the standard kyanite and mullite products. The question was three-fold:
Does reducing the iron content from the standard of 0.5% to something lower impact the hot properties of a refractory body made with Virginia MulliteTM?
How low can that iron be reduced?
At what point do we see diminishing returns?
TESTING PROCEDURES
Multiple samples of Virginia MulliteTM were produced by varying processing parameters in production to lower the iron content below the standard levels. To create higher than typical values, samples were doped with high iron material that is typically removed from the process. The iron oxide contents tested were as follows: 0.1%, 0.15%, 0.2%, 0.25%, 0.45%, 0.6%, 0.85%, and 1.2%. X-Ray Flouresence (XRF) was used to determine the chemistry. Testing was done on a Panalytical Axios mAX.
Attempts were made to hold the alumina value constant for this test to mitigate any effects variances would have on the outcome of the testing. The average alumina content for these samples was 57.5%. It must be noted that while this was the aim, there was some deviation in the alumina content that may have affected the results to some degree. The other impurities remained relatively constant throughout the samples with TiO2 averaging 1.1% and the alkalis/alkaline earths combined for an average of 0.08%.
35 mesh material was ground, pressed into pellets, and cut into rectangular prisms to make the dilatometer samples. These were tested in the Linseis L75VS1600C Dilatometer in the KMC lab.
Two additional sets of samples were created by pressing 100 mesh material into 6.0 x 1.0 x 1.0 inch (150 x 25 x 25 mm) bars. These bars were fired at 1600°C for 5 hours, with both a heating and cooling rate of 10°C/min.
One set of bars was used to check the post fired Modulus of Rupture (MOR). They were the tested in accordance with ASTM C133-97(21).
Half of the broken bars were used to create samples for X-Ray Diffraction (XRD) These were crushed and ground before being pressed into pellets for loading onto the XRD machine. A Panalytical Cubix3 was used to measure the samples. They were then analyzed using Rietveld refinement on the HighScore program from Panalytical.
The second set of bars was used for a hot load sag test in a box furnace. These bars were supported at the ends on top of a semi-circular alumina rod and allowed to sag under their own weight. The firing cycle was a 10°C/min ramp to 1600°C, a 24-hour hold, and then cooled at 10°C/min to room temp. The bars were measured before and after this test with a depth gauge to determine deflection.
Tests were performed at 1600°C because it is at the solidus line of a pure alumina-silica system within this alumina range. (Refer to Fig. 1)
RESULTS AND DISCUSSION
XRD Results
Analysis of pre-fired samples showed the presence of quartz, kyanite, rutile, and cristobalite phases. After firing, the only mineral phase present in the XRD scans was mullite. The other phases present in the green state melted into the amorphous phase. Noticeable background level differences were observed, indicating varying amounts of amorphous material.
Fig. 2. High amorphous phase content was witnessed in samples with higher iron oxide concentrations.
The K-Factor Method7 was used in HighScore to generate approximations for amorphous material in each post fired sample. These results are shown in Fig. 2. It should be noted that values obtained with this method give an approximation of amorphous phase and are mainly to be used for directional correlation.
The increasing amorphous content with higher iron oxide concentrations can explain many of the other results in this paper. The reduction of amorphous phase in the 0.25%, 0.2%, 0.15%, and 0.1% samples seems to slow, hinting at a point of diminishing return.
Dilatometer Results
Fig. 3. Increases in iron oxide concentration drastically lowered the temperature required to begin sintering.
Tests on the dilatometer showed variances in the curves between each of the samples in an exponential fashion. The onset of sintering temperatures are shown in Fig. 3. Dilatometer results indicate the onset of sintering temperature of the 0.1% material to be 72°C higher than the 1.2% material. The first large decrease in temperature was seen between the 0.25% and 0.3% with a drop of 20°C. Concentrations of 0.25% and below were all tightly grouped, hinting at a point of diminishing return. The higher onset of sintering temperatures for the lower concentrations also suggests a higher max usage temperature for these mullites when compared to the standard mullites made historically by KMC (around 0.5% Fe2O3).
Hot Load Sag
Hot load sag tests did not show as much variance in some of the intermediate samples as expected. However, there was a noticeable difference from the samples with the lowest iron content (0.1%, 0.15, and 0.2%) to the highest (1.2%). The high iron material slumped about 20% more than the lowest iron sample. This shows that lowering the iron oxide content in the raw material should provide a higher degree of creep resistance to the refractory body. This could be useful not only in bricks and monolithics, but in kiln furniture as well.
Fig. 4. Increasing the iron oxide content made bars that were noticeably darker in color after a 24 hold at 1600°C. (Shown in decreasing concentration from 1.2% at top- to 0.1% at bottom)
Another observation from this test was the change in color (Fig. 4). Color is often an indicator of impurities, and that held true in this study. The samples started out white and progressively got darker in color as the concentration of iron oxide increased.
MOR
MOR testing showed increasing strength with higher concentrations of Fe2O3. The increases seem to happen rather linearly except for the sample at 0.6%, which was oddly higher than the 0.85% sample. The samples at 0.2% and 0.3% iron were quite similar. Unfortunately, an equipment error occurred before we were able to test the 0.15% and 0.25% samples. It is hypothesized that these samples would fall in line with the linearity of the rest of the data. Testing will be completed when the machine can be fixed.
These bars were made by pressing together material with a narrow particle size distribution (PSD) with only a small amount of sodium lignosulfonate as the binder. This narrow PSD led to a lack of fines whose large surface area generally help with sintering. Minimal presence of fluxing agents other than the iron oxide also slow the sintering of the bar, lowering the MOR. With these conditions in mind, it can be inferred that the largest effect on the MOR comes from the formation of the amorphous phase. As stated previously, a higher amount of amorphous phase was formed as the iron oxide concentration increased. This created a bar with a higher post fired MOR. (Fig. 5) when the glass solidified upon cooling.
Fig. 5. MOR results show a linear trend with increasing iron oxide concentrations
While the increase of iron oxide provided a higher MOR in the post fired state, this would not be true to the refractory at working temperatures. It is hypothesized that the amorphous phase would cause slip along the grain boundaries, lowering the hot MOR and leading to increased creep. Further work is to be done to confirm this hypothesis.
Determining the Iron Oxide Content of the New Product
The last question to answer was how low KMC could reduce the iron content in the kyanite (and therefore the calcined form, mullite) on a production scale. The lowest achieved sample was 0.04% Fe2O3. However, it was determined the steps required to get to this low level were too difficult to reproduce on a large scale while also being uneconomical. It was decided that the new product, named Premium Grade Virginia MulliteTM, would be held to an iron oxide content of 0.19% or less. This is below the point of diminishing return (0.25%), which should mitigate any effects segregation could have on the performance of the product, while not being so low as to cause the economics of the product to make it-impractical for industrial use.
Future Work
Most of the testing done in this study was done on post fired samples except for the dilatometer work. Additional testing will be conducted to further understand the effects of iron oxide at temperature. Hot MOR and pyrometric cone tests are planned for the next phase of the project. While hot load sag gave directional indicators, high temperature creep testing will also be conducted to give a better understanding of the increase in creep resistance with lowering impurity level.
Additional studies will also be required to compare the new Premium Grade Virginia MulliteTM to other mullites to understand its place in refractory applications.
CONCLUSION
The study had three main objectives. The most important was if lowering the iron oxide content of the mullite would lead to measurable and meaningful improvements in the hot properties of the mullite. The data showed that a reduction in iron oxide improved the products performance in all tests. It was assumed this was due to a lower amount of amorphous phase, which was confirmed via XRD testing.
The second question was how low KMC could reduce the iron oxide concentration. KMC was able to create a product with 0.04% Fe2O3, but this was deemed both impractical and uneconomical on an industrial scale.
Lastly, did there appear to be a point of diminishing return? In several of the tests, there appeared to be minimal gains by lowering the iron oxide content below 0.25%. It was decided that the new Premium Grade Virginia MulliteTM product would be held to a value of 0.19% or less Fe2O3.
While these results are promising, further work still needs to be done in order to determine the usefulness of this product to the industry.
REFERENCES
1. H. Schneider and S. Komarneni, Mullite. Wiley-VCH, Weinheim, Germany, 2005.
2. K. Dana, S. Sinhamahapatra, H. Sekhar Tripathi, and A. Ghosh, “Refractories of Alumina-Silica System,” Trans. Ind. Ceram. Soc., 73 [1], 1-13 (2014)
3. ASTM Standard C467-14 (2018), “Standard Classification of Mullite Refractories,” ASTM International, West Conshohocken, PA, 2014.
4. S. Aramaki and R. Roy, “Revised Phase Diagram for the System Al2O3—SiO2,” Journal of the American Ceramic Society, 45 [5] 229-242 (1962).
5. A. Muan’ “Reactions Between Iron Oxides and Alumina-Silica Refractories,” Journal of the American Ceramic Society, 41 [8] 275-286 (1958)
6. J. Porier and M. Rigaud Corrosion of Refractories: The Fundamentals. Göller Verlag GmbH, Baden-Baden, Germany, 2017.
7. B.H. O’Connor and M.D. Raven, “Application of the Rietveld Refinement Procedure in Assaying Powdered Mixtures,” Powder Diffraction, 3, 2–6 (1988).
