The ferrite phase

V Unit cell parameters (nm) 0 h c Spaee group X-ray density (kgm ) References 

Plots of unit cell parameters or XRD powder spacings against a for the pure Ca,(Al,Fe)205 series show changes in slope near a = 0..1, attributable to the structural change (S4). Such plots may be used to determine composition provided that oxide components other than CaO, A1,0, and Fe,0, are absent. They have often been applied to ferrites in clinkers, but this gives seriously inaccurate results because of the effects of other substituents (Sections 1,5.2 and 1.5.. ). 

Biissem (B12) determined the approximate crystal structure of C4AF. Subsequent determinations or refinements were reported for preparations with. Y = 0 (C F) (B13,C4,B11),. v = 0.285 and 0..36 (C2) and, v = 0.5 (C3). Fig. 1.8 shows the structure for the compositional range with 0.33 . Y 0.7. It is built from layers of corner-sharing octahedra similar to those in perovskite (CaTiOj), alternating with layers composed of chains of tetrahedra, together with Ca ions. The layers are perpendicular to the h axis and the chains are parallel to c. The composition of an individual octahedral layer in the ac cross-section of the unit cell is MiO. and that of an individual chain is where M and T denote octahedral and 

Each Ca ion in C4AF has 7 oxygen neighbours at 0.23-0.26 nm (C3). The aluminium and iron atoms are both distributed between octahedral and tetrahedral sites, the fraction of the aluminium entering tetrahedral sites under equilibrium conditions decreasing with temperature. For the three preparations with. y = 0.285, 0.36 and 0.5 on which X-ray structure determinations were made, 75-76% of the total content of aluminium was found to be in tetrahedral sites. These preparations were shown to have been in equilibrium at about 750°C (C3) for a C4AF preparation quenched from 

1290°C, the Mossbauer spectrum indicated that only 68% oftlic aluminium was in tetrahedral sites (G8). It was suggested that, v does not exceed the observed limit of about 0.7 because at this eomposition the tetrahedral sites are all occupied by aluminium (C2). There is evidence of ckistering of aluminium and iron atoms to an extent depending on composition and conditions of formation (Zl). 

---

The hydration of the ferrite phase (C AF) is of greatest interest in mixtures containing lime and other cement compounds because of the strong tendency to form soHd solutions. When the sulfate in solution is very low, soHd solutions are formed between the cubic C AH and analogous iron hydrate C FHg. In the presence of water and siUca, soHd solutions such as C3 ASH4-C3FSH4 may be formed (33). Table 7 Hsts some of the important phases formed in the hydration of mixtures of pure compounds. 

Additional high temperature changes cause decarburization, wherein carbon in the ferrite phase of carbon steel can be oxidized to carbon dioxide. 

There may be a chelating effect whereby TEA reacts with the ferrite phase of Portland cement [10], as illustrated in Fig. 5.2. 

Step cooling will not simulate embrittlement of 1 WCM Mb, though it occurs (e.g., a 100°F [38°C] increase in transition temperature was reported after 8 y at 930° F [500° C]). This is because the embrittlement in 11/iCr-1 Mo is caused by precipitation of carbides in the ferrite phase rather than segregation of impurities to the grain boundaries. Temper embrittlement can be reversed by heating at 1,150°F (620°C) for 2 h per inch of thickness. 

The ferrite phase makes up 5 15% of normal Portland cement clinkers. It is tetracalcium aluminoferrile (CajAIFeOj) substantially modified in composition by variation in Al/Fe ratio and incorporation of foreign ions. The rale at which it reacts with water appears to be somewhat variable, perhaps due to differences in composition or other characteristics, but in general is high initially and intermediate between those of alite and belite at later ages. 

In many clinkers, the ferrite phase is closely mixed with aluminate due to a similarity in cell parameters, oriented intergrowth can occur (MIS). The close admixture often renders X-ray microanalysis difficult or unreliable. For ordinary Portland cement clinkers, the compositions found in dilferent laboratories are nevertheless remarkably consistent. Table 1.2 includes an average value based on the results of investigations using X-ray microanalysis (H8,K1,B2,U1,H3,B4) or chemical analysis of separated material (Yl). Table 1.3 includes suggested site occupancies corresponding to these data. 

Sulphate-resisting Portland cements have relatively high ratios of iron to aluminium, and the ferrite phase cannot have the composition given above if it contains most of the iron. Tables 1.2 and 1.3 include a tentative composition and atomic ratios corresponding to it, based on scanty data for the interstitial material as a whole (G3,G4) and the requirement of reasonable site occupancies. 

Regourd and Guinier (Rl) reported unit cell parameters for the ferrite phase in five clinkers. The ranges observed were a = 0.5517-0.5555 nm, h = 1.455-1.462 nm, c = 0.5335-0.5350 nm. Boikova (B4) reported XRD powder spacings for clinker ferrites which indicate similar values. The similarity of these cell parameters to those of the laboratory preparations... 

Fig. 2.5 shows part of the ternary system CaQ-Al203-Fe203. C3A, C12A7 and CA can all accommodate some Fe for C3A under equilibrium conditions at 1325°C, the limit is about 4.5%, expressed as Fe203 (M20). The ferrite phase in equilibrium with iron-substituted C3A can have compositions with. V between 0.48 and 0.7 in the formula Ca2(Al,jFej jj)205 if CaO is also present,. v is fixed at 0.48, i.e. the composition is close to C4AF. Some reduction of Fe to Fe occurs when the ferrite phase is prepared from mixes with compositions in the Ca2(AljjFej - j)205 series in air it leads to the formation of minor amounts of other phases, which are not observed when similar experiments are carried out in oxygen (M20). 

Swayze (S8) noted that equilibrium was often difficult to achieve in this system. One effect was the tendency for crystals of the ferrite phase to be zoned. For bulk compositions in the Ca2(Al Fei J2O5 series, the liquid is of higher Al/Fe ratio than the ferrite phase with which it is in equilibrium the crystals that are initially deposited on cooling such a liquid therefore have a lower Al/Fe ratio than the bulk composition of the mix. On further cooling, the Al/Fe ratio of the material deposited progressively increases. Equilibrium within the ferrite crystals is difficult to attain, causing them to remain zoned, with cores richer in Fe and outer regions richer in AF" than the mean composition. 

A second effect was the tendency for protected phases to be formed. If a liquid having a composition somewhat on the CaO-rich side of the boundary between the CaO and CjA primary phase fields (Fig. 2.5) is cooled calcium oxide is initially deposited and the liquid composition moves away from CaO and towards that boundary. When the latter is reached, and assuming that equilibrium were to be maintained, calcium oxide would redissolve, C3A would be deposited, and the liquid composition would move along the boundary. In reality, the C3A quickly surrounds the particles of calcium oxide, which thus form a protected phase, effectively removed from the system. This can markedly affect the composition of the ferrite phase which is subsequently formed on further cooling. 

CjA is larger, and that of the ferrite phase smaller, than was shown in Lea and Parker s diagrams, which in this region partly rested on early results for the CaO-C,2A7 C4AF system. To simplify the diagram, the volumes of phases other than CjS are only indicated in general terms on Fig. 2.7. 

Table 2.3 lists some phases containing MgO that are in varying degrees relevant to cement chemistry. It is not a complete list of phases with essential MgO in the CaO-MgO-AljOj-SiOj system. As seen in Chapter 1, some MgO is also taken up by all four of the major clinker phases, typical contents being 0.5-2.0% for alite, 0.5% for belite, 1.4% for the aluminate phase, and 3.0% for the ferrite phase. Magnesium oxide (periclase), like calcium oxide, has the sodium chloride structure it is cubic, with a = 0.4213 nm, space group Fm3m, Z = 4, = 3581 kgm (S5) and refrac-... 

Fig. 2.8 The pseudosystem CaO-C2S-C]2A7-C2p modified by the presence of 5% of MgO, showing the phase volume of CjS and tie lines for the ferrite phase, the compositional range of which is represented by the hatched line. For details of invariant points P1-P8, see Table 2.1. After Swayze (SIO), with later modifications.

Physical methods are of limited effectiveness for separating the phases in cement clinker because of the intimate scale on which the latter are mixed however, Yamaguchi and Takagi (Yl) had some success with the use of dense liquids. Some concentration of the ferrite phase can be effected by magnetic separation (M35,Y1). 

Transmission electron microscopy of ion-thinned sections provides data at higher resolution than can be obtained with polished sections. Rodger and Groves (R24) described regions which had probably formed in situ from the ferrite phase, and which consisted of C-S-H, a hydrotalcite-type phase and a poorly crystalline phase containing iron that could have been the precursor of a hydrogarnet. The particles of this last constituent were almost spherical and some 200 nm in diameter. The same investigation also showed that much of the product formed in situ from alite or belite was essentially pure calcium silicate hydrate. 

Towards the end of the middle period, a renewed growth of AFt crystals takes place (D27,P3I,S4I) (Fig. 7.6d). They are markedly more acicular than those formed earlier their lengths are typically 1-2 pm, but sometimes up to 10 pm. Their formation is associated with a shoulder on the heat evolution curve (Section 7.5.1). Their formation implies an increase in the rate of reaction of the aluminate, or less probably the ferrite phase, which is probably related to the reaction of the alite (S68). 

In Ciment Fondu, the ferrite phase seems to play no significant part in early hydration at 20 C, but at 30-38 C over 80% was found to have reacted by 2 months (C47). The melilite and pleochroite seem to be unreactive. When belite is present, silicate ions can be detected in the solution within a few minutes, but then disappear it seems that precipitation occurs and further dissolution is inhibited. Among the minor oxide components. TiO, and MgO mainly occur in the unreactive phases. Na,0 and K,0 scarcely affect the solution equilibria at early ages, as their concentrations are very low (M88). 

---