Rankin and Wright — Ternary System CaO-Al 2 3 -Si0. 2 . 61 



From r, to 4, C 2 S will, wholly or in part, disappear and GJ3 % 

 will crystallize. On reaching 4 the concentration of the solid 

 is a. The temperature will now remain constant while C. 2 AS 

 crystallizes with GJS % \ the solid C 2 S \s used up in forming C 3 S. 2 

 before the CS in solution disappears, since the concentration 

 of CS in the solution is high. Consequently crystallization 

 will proceed along boundary 4-6, C 2 AS and C 3 S 2 separating. 

 At 6, CS will crystallize with the C 2 AS and C 3 S 2 and the crys- 

 tallization curve ends at that point. 



In the case just discussed a phase which had crystallized 

 partially disappeared as the crystallization proceeded along a 

 boundary curve. There are types of solutions in which a phase 

 which lias crystallized will entirely disappear as the crystalliza- 

 tion curve follows a boundary curve. Such a solution is that 

 represented by point t (fig. 18) within the triangle CS-C 3 S 2 - 

 C 2 AS. Since solution t lies within the field for C 2 S, C 2 S will be 

 the first solid to appear as the solution is cooled. As in the former 

 cases the composition of the solution changes along a straight 

 line and intercepts the boundary B'-k at t v This solution is 

 saturated with respect to both CJS and C 3 S 2 and hence when 

 the temperature corresponding to t 2 is reached, the solid con- 

 sists of both C 2 S and C 3 S 2 , and its mean composition is a 2 . On 

 continued cooling a point t 3 is reached at which the composi- 

 tion of the solid is that of the compound C 3 S 2 . During the 

 cooling of the solution t through the points t„ l 2 , £,, extrapola- 

 tion to the line joining the compositions of the pure phases 

 separating, shows that the percentage of CJS in the crystal 

 mass has been steadily decreasing. Since the solid phase 

 which originally consisted of pure CJS changes continuously to 

 that called for by the composition of the compound C 3 S 2 , while 

 C 3 S 2 and C 2 S separate together, it must follow that this change 

 means that pure C 2 S disappears faster than it crystallizes ; in 

 other words, the compound CJS, which has separated in cooling 

 along the curve from t to t l9 has disappeared in cooling from 

 t x to t 3 , while the compound C 3 S 2 has crystallized. With the 

 solution £ 3 , then, there exists in equilibrium only the compound 



c&. 



When the solution t 3 is cooled again there seems to be 

 one of two courses for the crystallization curve. It might fol- 

 low the boundary curve, as in the preceding cases, to the quin- 

 tuple point 4; in which case the mean composition of the solid 

 in equilibrium with the solution would be given by points 

 along C 2 S-CS, but on the side of C 3 S 2 away from C. 2 S. Such 

 a course, however, is impossible, because there is no mixture of 

 C^S and C 3 S 2 which, by any possible combination, can be made 

 to contain an amount of C 2 S less than that required for the 

 composition of C 3 S 2 . Therefore, the crystallization curve can- 



