Examining the Effect of the Partial Replacement of Calcined Alumina with Kyanite in a 60% Low Cement Castable

Steven Ashlock
Kyanite Mining Corporation
Dillwyn, VA, USA

Dilip Jain
Kyanite Mining Corporation
Dillwyn, VA, USA

---

ABSTRACT

The partial replacement of calcined alumina with Virginia KyaniteTM in a 60% LCC in an attempt

to enrich the matrix with mullite was examined. Previous work has shown that calcined alumina

does not react as readily as believed to form secondary mullite. Bars of three different castable

mixes were prepared and tested for MOR, CCS, density, and other properties. Chemistry and

phase composition were evaluated using XRF and XRD respectively. Despite the lowering of the

overall alumina content the majority of the castable properties were improved. The addition of -

325m kyanite created mullite in the matrix after sintering to 1480°C and higher. In all three mixes

some calcined alumina remained unreacted even after firing to 1600°C. The density is also

increased as the -325m kyanite expands into and reduces the porosity of the refractory.

INTRODUCTION

The demands on refractories in the modern market are constantly changing and becoming more

challenging.1 Requests for longevity increases are commonplace from the buyer to the refractory

supplier. In order to achieve this goal companies are increasingly enhancing the overall alumina

content. This requires using more expensive higher end raw materials. One such material that is

a staple of many high end alumina refractories is calcined alumina. Calcined alumina is added to

improve refractoriness to the matrix.2,3 It is also said to react with silica fume to form interstitial

mullite.4,5 It has long been known that mullite in the matrix greatly enhances the thermal shock

capabilities of the refractory. However, research shows that the reaction between silica fume

and/or silica and calcined alumina takes place at fairly high temperatures.6,7 Most refractories

would need to be heated to well above usage temperature in order to see this beneficial reaction.

Virginia KyaniteTM is a member of the sillimanite group of minerals.8 The minerals that make up

this group are kyanite, sillimanite, and andalusite. These minerals are used in a wide variety of

refractory applications. One unique property of the sillimanite group minerals is conversion to

mullite after calcination via the following reaction:

Kyanite has the lowest temperature of conversion of the group (1430°C) as well as the largest

amount of expansion upon conversion at 17 volume percent. This expansion is used throughout

the industry as a means of offsetting shrinkage of the other minerals in the refractory recipe. To

achieve this goal the larger commercially available sizes of kyanite, such as -35, -48, and -100

mesh (425, 300, and 150 μm), are used. The finer mesh sizes such as -200m (75 μm) and -325m

(45 μm) are used to increase the density of the refractory. Most of the expansion of this finer mesh

kyanite is absorbed by the porosity thus increasing the bulk density. Fine mesh kyanite has

another attribute than can be exploited: mullite formation in the matrix of the refractory.

TEST PROCEDURE

Three 60% alumina low cement castables (LCCs) were created to observe the change in castable

properties as calcined alumina (A35-325) is replaced with Virginia Kyanite. The formulations,

based off of mullite aggregates, are listed in Table 1. As the amount of 325m kyanite is increased

the amount of 48m kyanite needs to be decreased. Too much kyanite in the refractory mix will

cause an overexpansion that is detrimental. After mixing and water additions the mixes were

vibration cast in 25x25x152 mm molds. The bars were allowed to cure overnight before

demolding. The bars were then dried at 110°C before firing in an electric furnace. Firing conditions

were a ramp of 5°C per minute with five hour hold times at either 1370°C, 1480°C, or 1600°C with

a natural cool down.

Tab. 1: LCC mix recipes

Various castable properties were measured for each series of bars. Permanent linear change

(PLC) was measured using calipers before and after firing. Bulk density, apparent porosity, and

water absorption were measured using ASTM C20-00 (2015). ASTM C133-97 (2015) was used

to test modulus of rupture (MOR) via three point bend and cold crushing strength (CCS) on

25x25x25 mm cubes. Chemical analysis was performed via X-Ray fluorescence (XRF) on a

Panalytical PW2400 as well as phase analysis by X-ray diffraction (XRD) on a Panalytical Cubix3

utilizing the Rietveld method.

RESULTS AND DISCUSSION

Chemistry

The chemistry results obtained via XRF are shown in Table 2. The most noticeable chemistry

change is the decrease in alumina with increasing amount of -325m kyanite. This increase in

kyanite, as a replacement for the pure calcined alumina, also slightly increases the iron and titania.

Tab. 2: Chemistry data of the castables in oxide percentages

Physical Properties

Bars of all three recipes were then fired and had their physical properties tested. The bulk density

data is shown in Figure 2. The bulk density of all three castables was the same after firing to

1370°C. As the firing temperature increases the differences become apparent. Mix 3, the mix with

the most -325 mesh kyanite, has the highest bulk density of the three mixes at both 1480°C and

1600°C. This is due to the high amount of expansion into the porosity of the refractory. This mix

is also the only mix to maintain or even increase it’s density with a higher firing temperature. Mix

1 and Mix 2 both had a lower density after firing to 1480°C but regained some of this value when

firing to 1600°C. This decrease at 1480°C is likely due to the expansion of the 48m kyanite. Mix

1 contained more 48m kyanite thus the larger decrease in density. Evidence of this expansion

into the pores was verified by looking at the apparent porosity. As the -325 kyanite increased the

apparent porosity was lowered.

Fig. 2: Bulk Density

The permanent linear change (PLC) of the bars was examined. A graphical display of this data is

shown in Figure 3. At 1370°C little to no expansion is seen. This is to be expected as this is below

the temperature of expansion for kyanite. At 1480°C the expansion becomes more apparent due

to the phase transformation and expansion of kyanite. All three mixes experienced further

expansion when fired to 1600°C as amorphous is formed and bloating occurs. As seen in the

XRD data examined later in this paper, a small additional amount of mullite is also formed after

the 1600° firing. The secondary mullite formation contributes to the expansion at 1600°C. The

mixes with higher amounts of 48m kyanite (Mix 1 and 2) showed the largest expansion at both of

the higher temperatures.

Fig. 3: Permanent Linear Change

The cold crushing strength (CCS) of the mixes was then tested (Figure 4). The mixes containing

-325m kyanite showed greater strength than the mix without at both of the lower temperatures.

At 1600°C Mix 1 (89.5 MPa) actually had an almost equal CCS value to Mix 3 (89.4). Mix 3 had

a lower CCS value than Mix 2 at every temperature. Mix 3 has the lowest alumina percentage

and likely created the most amorphous phase when fired thus weakening of the mix. Mix 2 had

the highest CCS values at all three temperatures. Previous studies on this system of mixes

showed increasing strength with increasing 325m kyanite meaning Mix 3 was expected to exhibit

the highest CCS values.7 Results in this test were contradictory to our previous work and another

study needs to be done to clarify the discrepancy.

Fig. 4: Cold Crushing Strength

The modulus of rupture (MOR) was the last of the physical properties tested. Results (Figure 5)

showed higher results with an increase in the -325m kyanite. This trend is seen at all three firing

temperatures with the exception of Mix 3 at 1600°C in comparison to mix 2. Gains in strength,

while minimal, were seen as the mixes were fired to higher temperatures. The lower MOR values

of Mix 3 at 1600°C are likely a result of the mix having the lowest overall alumina content and

thus the highest amount of amorphous material present.

Fig. 5: Modulus of Rupture (3-point bend)

Phase Analysis

Each of the three mixes was examined using XRD for the presence of kyanite and corundum at

all three temperatures. The goal was to determine that a) the kyanite had converted to mullite and

b) to determine if there was any residual corundum or if it had been converted to mullite as well.

To examine our first goal we look at Figure 6. This shows Mix 2 after firing to the three different

temperatures. Here we can see the kyanite being converted at 1480°C. The kyanite peak at 28.0

has been highlighted for ease of viewing. Similar trends are shown in all three mixes.

We also can see from the XRD values that there is still some residual corundum in all three of the

mixes even after firing to 1600°C. This can also be seen when looking at Figure 7. The corundum

peak at 43.34 has been enlarged to show it is still present in the sample. The scan showing the

highest percentage of corundum is Mix 1 with the least amount being shown in Mix 3.

Fig. 6: The kyanite peak is present at 1370°C but is gone upon conversion after 1480°C.

Fig. 7: XRD shows that corundum is still present even after firing to 1600°C.

All three of the mixes showed an increase in mullite as the temperature increased. This can be

seen in the scans peak height and in Rietveld analysis. The largest gain in mullite percentage

was seen between 1370 and 1480°C. This is to be expected as the kyanite is converting to mullite

between these temperatures. A smaller gain is seen between 1480 and 1600°C. This is a clear

indication that some secondary mullite is being formed in the firing process. However, as stated

earlier, there is still corundum present even at 1600°C. Unfortunately, the secondary mullite

formation did not completely consume all of the calcined alumina intended for that purpose. Mix

3 had the highest amount of mullite present after each of the higher firing temperatures. This is to

be expected as Mix 3 contained the highest percentage of kyanite and thus the highest amount

of mullite formation not associated with secondary mullite formation involving calcined alumina.

CONCLUSION

The partial replacement of calcined alumina with -325 mesh kyanite showed several advantages.

Density increases were seen as -325m kyanite was introduced and allowed to expand to fill the

porosity of the castable. The addition of -325 mesh kyanite also improved both MOR and CCS

when the amount of calcined alumina and- 325 mesh kyanite were equal. At a higher percentage

of kyanite these values lowered, likely due to the increasing amount of impurities in the castable.

Further testing would need to be done in order to determine the ratio of calcined alumina to kyanite

at which the properties begin to decline. Although the overall alumina content is reduced with

kyanite addition, the replacement of calcined alumina with -325 mesh kyanite appears to be

beneficial. Lastly, despite popular belief, XRD analysis showed that even when firing to 1600°C

calcined alumina did not completely react with silica to form secondary mullite. Replacing some

of the calcined alumina with -325m kyanite guarantees the formation of mullite in the matrix.

---

REFERENCES

[1] Lee, William E., and Robert E. Moore. “Evolution of in Situ Refractories in the 20th

Century.” Journal of the American Ceramic Society, vol. 81, no. 6, 1998. pp. 1385–1410.

[2] Buhr, Andreas., et al. “Review of Matrix Alumina for Refractory Formulations.” China’s

Refractories, vol. 24, 2015. pp. 57-62.

[3] McConnell, Robert W., and F. Andrew Fullington. “Responding to the Refractory

Industry’s Need for Fully Ground Matrix Aluminas.” Proceedings of the United

International Technical Conference on Refractories, 4-7 Nov. 2001. pp. 768-80.

[4] McGee, Thomas D., and C. M. Dodd. “Mechanism of Secondary Expansion of High-

Alumina Refractories Containing Calcined Bauxite”- Journal of the American Ceramic

Society, vol. 44, no. 6, p 277-83 1961.

[5] Myhre, Bjorn. “Let’s Make a Mullite Matrix” Refractories Applications and News, vol. 3,

no. 3-4, 2008. pp. 16-24.

[6] Jain, Dilip. “Mullite Formation: A Myth or Reality?.” St. Louis Section Meeting of the

American Ceramic Society, 29 Mar. 2007.

[7] Jain, Dilip. “Substitution of Kyanite for Calcined Alumina in a Low Cement Castable.”

IREFCON, 1-4 Feb. 2012

[8] Brown, Jesse. “Kyanite.” Report. Materials Engineering. Virginia Polytechnic Institute and

State University. 1982.