Depending on the amount of alloying elements, in particular carbon, cast irons are normally subdivided in hypo- eutectic and hypereutectic cast irons. Applications can be found mainly within the hypo- up to mildly hypereutectic irons. Strong hypereutectic compositions are rarely used, due to the high carbon content, at this moment causing more problems than showing advantages. For most foundrymen, the hypereutectic region remains a relative unknown area. Nevertheless, unique events, found in the hypereutectic regions are often rigorously used to support or even to prove solidification events in the hypo- and eutectic region. In this article, a survey will be made on real known facts found in the hypereutectic region, which should it make possible to judge whether it is justified to use them in order to explain solidification events taking place in other cast iron compositions.
General solidification mechanism of gray irons.
Solidification of hypo eutectic cast iron starts with the separation of austenite from the melt as soon as the temperature drops below the liquidus. The melt is enriched in carbon until the eutectic composition is reached. Stable eutectic solidification starts on or below the eutectic temperature, on nuclei in the melt and consists of the formation of more austenite and free graphite. "Solidification of a hypereutectic cast iron is the same process, except that the first phase to form is graphite" (1). Depending on the presence or absence of certain elements, in all compositions carbon can be deposited in a flake, spheroidal or intermediate shape. If, due to whatever reason, graphite formation is inhibited, the excess carbon will be deposited as the chemical compound Fe3C (iron-carbide or cementite). Traditionally the various types of hypereutectic cast iron will be subdivided into: gray cast iron, ductile iron and white cast iron.
Hypereutectic gray iron.
The fact that the amount of carbon which could be retained within solid cast iron is limited was already known in the second half of the 19th. century. How high this maximum was, however, was still unknown (2). It was clear that in the molten state, more carbon could be picked up, especially when cast iron is melted with a high cokes charge, thus produced at high temperatures (German: "gares Eisen") . This higher carbon content, however, tended to separate before the start of solidification. Caused by the low specific mass, the separated graphite floats to the surface of the solidifying metal bath, where it showed up as "Garschaum" (3,4). The development of the iron carbon diagram made it possible to relate carbon content, temperatures and the various liquid and solid phases that were found. The old idea of floating graphite particles fitted perfectly in this concept, as in hypereutectic alloys, graphite should be the first phase to crystallize. Emphasize was laid on the fact that Garschaum or Kish graphite, as it was called in Anglo-Saxon language (7), was identical to primary graphite. Although all these names are still in use today, this resemblance does not show up in the official description of Kish graphite (9): "Kish,[G.kies,gravel,pyrites]. A substance resembling graphite, consisting of carbon and manganese, found in some iron-smelting furnaces." .The name Kish appears to be derived from the German word "Kies", the best fitting description being "sulfur-arsenic or antimony compounds of metallic appearance."(10), which can hardly be regarded as a resemblance to primary graphite. Chemical analysis of Kish graphite (11-14) showed carbon contents ranging from 28 to 99% . These variations are generally thought to be caused by different sampling and cleaning methods.
Solubility of carbon in liquid iron.
The separation of graphite from hypereutectic alloys is closely related to the solubility of carbon at various melt temperatures. Figure 1 shows the main results that were obtained in this field (15-27).
The solubility of carbon in liquid iron has been investigated up till 25 % at 24000 C. The majority of tests, however were performed in the practical region up till 16000 C. The huge spread in the results are thought to be caused by the difficulties raised by such high temperatures. Even in the normal iron carbon diagram, which is normally drawn from 0 to 6.7% carbon, the graphite liquidus is rather steep.
Redrawn on an unusually scale, up till 100 % carbon as is shown in figure 2, the graphite liquidus almost resembles a vertical line. For completion, the most important results on the liquidus line in the hypoeutectic region (28-33) have been added to the picture.
Formation mechanisms of primary graphite.
Iron carbon diagram(s), build up from the above mentioned results, exist for almost a century. It forms a firm well established base to explaining the way cast iron solidifies, including the hypereutectic region. The earlier mentioned idea of carbon particles that float to the surface, caused by its low specific gravity, could be easily visualized and formed in fact the only logic explanation. The fact that little research was done in this field made it possible that this idea was copied for decennia. A quotation of Morrogh can serve as an illustration of this traditional hypothesis, that is adhered to by many, even in the nineties. "That hyper-eutectic graphite is deposited directly from the melt does not appear to be disputed. Such graphite tends to float to the top of the melt before eutectic crystallization occurs." Figure 3 shows this flotation mechanism.
Figure 3 Kish graphite formation according to traditional concepts. a)Hypereutectic gray cast iron, b)Nucleation of primary graphite flakes, c) Caused by its low specific gravity, graphite particles tend to float to the surface whilst increasing in size, d)Primary graphite agglomerates at the bath surface (Kishing), e) Part of the Kish graphite is taken up by the bath-atmosphere and spread into the surrounding area. Various reasons have been given for this last phenomenon: -Flotation causes such an increase in velocity of the graphite particles, that they are thrown out of the bath, - graphite particles are thrown into the atmosphere by exploding gasbubbles, -Graphite particles, lying on the bath surface are taken up by the circulating hot air above the metal bath.
Kish graphite formation used as proof of direct graphite formation.
Kish graphite of big dimensions can also be found within the cast iron structure, meaning that not all primary graphite is able to float to the top. The reason that this occurs is the fact that primary graphite attaches itself to the wall during its flotation (35) or because of intertwining of primary graphite crystals (37). The presence of large primary graphite inclusions was also used, a long time ago, as a proof that in all circumstances graphite was a decomposition product of carbides. According to (40) primary graphite tends to group itself into clusters, which will give rise to the formation of eutectic cells. (It is interesting to note that it is also postulated that primary graphite tends to prevent the formation of eutectic cells (41,42)). Sun and Loper, after studying the microstructure of hypereutectic flake iron, cannot find any tendency to flotation of the primary graphite. They deduct the solidification mechanism shown in fig.4.
Figur 4 Schematic representation of the solidification sequence of hyper-eutectic gray iron, according to (43). a)Kish graphite forms in the melt,b)Kish graphite and eutectic cells, c) Flake graphite dispersed in the matrix after complete solidification.
Kish graphite as a surface phenomenon.
Zakhartchenko et al (44-45) studied the flotation tendency of primary formed graphite, using chemical analysis methods. Samples taken from 80 tons(!) ladles showed during an eight hours testing period, a gradually decreasing carbon content, but no differences in carbon content at various heights within the ladle. They concluded that no graphite flotation had taken place. The crystal structure of the Kish graphite present on the bath surface showed that Kish graphite is exclusively formed on the surface of the liquid iron. Thermal currents provide a continuous supply of fresh liquid iron with maximum carbon content to the bath surface. The presence of oxygen above the bath (46-47) appears to have a big influence on the formation of Kish graphite.
Figure 5 Kish graphite formation as a surface reaction, acc. to (44-47). A)Hypereutectic gray iron, b)Kish formation at the surface, c)Kish graphite taken up into the atmosphere by moving hot air, d) Thermal currents supply high carbon melt to the bath surface.
It is astonishing that so little attention has been given to the above mentioned mechanism, especially because its outcome changes the old concepts on the formation of primary graphite in cast iron completely. The circumstances for Kish graphite to form (or not to form) as well as the many typical events that occur during the solidification of hypereutectic cast iron, hardly justify to explain the formation of Kish graphite with a simple traditional flotation mechanism.
Typical events that can take place in the hyper eutectic region are:
Increase of viscosity of the melt.
At the end of the last century, it was already known that, when pure iron was melted with a surplus of carbon, at a temperature of 20000C, the melt became so pasty that it could not be poured out anymore (16). The fact that with higher temperatures and higher carbon contents the melt became more and more viscous to even a pasty state, was confirmed by others (18,20). Ruer and Biren (18) noted that after superheating at 25000C, the melt became so pasty that the impression of the stirring rod stayed visible after its removal. It should be noted, however, that in all above mentioned cases, intensive stirring was applied. The higher viscosity of the melt could not have been caused by the formation of graphite particles, because the solubility of carbon increases with increasing temperature (18). Gröber and Hanemann (11) on the other hand noticed a high viscosity only when it was accompanied by a visual formation of Kish graphite. The event occurred when a carbon saturated melt was lowered in temperature and disappeared when the temperature was raised again. In all other cases, the melt remained low viscous even at 22000C. The increase of viscosity was thought to be caused by an early separation of graphite particles, resulting from temperature fluctuations in the melt. Graphite particles once separated, tend to float to the top and dissolve very slowly.
-Kish graphite does not always occur.
The very first studies on the hypereutectic region showed events that were hard to explain. Hanemann (17) noted in 1911 that the formation of Kish graphite did not always take place and certainly was not to be regarded as an indication for the amount of primary graphite. Ruff and Goecke (15) made their tests in vacuum and did not find any Kish graphite up till 26000C. Recently Liu,and Loper (46,47) established the fact that no Kish is formed when melted under argon atmosphere and a cooling velocity above 5000C per hour.
-Influence of alloying elements other than carbon.
Kish graphite in hypereutectic alloys is not formed when the material does not contain silicon. (Howe (8) 1907, Liu.Loper(47) 1990). Different views on the influence of sulfur exist. One source states that no kish is formed when sulfur contents are high. The reason is thought to be due to the fact that the surplus carbon is combined to carbides, which will remain in suspension (21). According to (47) however, kish formation is enhanced by an increasing sulfur content, due to a decrease of the surface tension between liquid iron and graphite.
The use of cooling curves on hypereutectic gray iron.
Cooling curves form a useful tool for studying solidification events. Practice shows, however, that for cast irons, this holds only true for hypoeutectic compositions. Cooling curves taken of hypereutectic alloys show many inaccuracies and controversies. According to the general accepted iron-carbon diagram for instance, the primary phase of a hypereutectic cast iron should be graphite. As any separation induces a change in heat content, any primary graphite formation should be detected and shown on a cooling curve. This, however, is not always the case. Research work, using cooling curve techniques, can be divided into two groups: -Those that were not able to detect a liquidus arrest, and -Those that showed a liquidus arrest.
-No liquidus arrest.
Failing liquidus arrest were noted as far back as the beginning of this century (17,18,28,29), and are still confirmed today (43,49). Hanemann (17) thought that the failing liquidus arrest was caused by a strong tendency of the primary graphite to undercool. But the most popular and widest used argument to account for this failing liquidus arrest, dates back to Stansfiel (7). This argument was formulated at a time that the ink of the very first iron-carbon diagram had barely dried: "With regard to the slope of the line BD, it must be remembered that this part of the curve is somewhat difficult to obtain by the ordinary method of taking cooling curves, as the heat evolved by the separation of graphite is very slight, and the extrusion of graphite from the cooling mass renders the carbon percentage of the original fluid metal somewhat uncertain." (Note: line BD marks the graphite liquidus line in the hypereutectic region).
A simple calculation, however shows that the amount of heat involved in primary graphite formation, is more than enough to be shown on a cooling curve (50).
Figure 6 shows a series of cooling curves taken in 1909 by Gutowsky(48).
Sample I,Carbon %= 0.577
Sample II,Carbon%= 1.071
Sample III,Carbon%= 1.632
Sample IV, Carbon%= 1.632
Sample V, Carbon%= 2.165
Sample VI, Carbon%= 2.795
Sample VII, Carbon%= 3.677
Sample VIII, Carbon%= 4.626
Sample IX, Carbon%= 4.940
The fact that no graphite liquidus was found in the hypereutectic region, apparently was not enough to lead to a change in the iron-carbon diagram, at that time being highly disputed.
-The occurrence of a liquidus arrest.
Some researchers claim to have found liquidus arrests, corresponding to graphite precipitation in coolingcurves, taken on hypereutectic alloys. Pontone (26) f.i. found liquidus arrests in pure Fe-C, Fe-C-Si and Fe-C-Si-P alloys. The shape of these liquidus arrests, however, is quite different from those that can be found in hypoeutectic alloys, as indicated in fig.7.
Liquidus arrest were also detected at CE of 4.6% and more (50), or exclusively at high silicon contents (51). The occurrence or not of a liquidus arrest seems to be correlated with the oxygen content of the melt atmosphere and the melt itself (52).
In this case, the liquidus arrest does not present graphite separation, but the formation of austenite dendrites (Fig.8). Heine (53) is of the opinion that such an arrest shows the beginning of the eutectic solidification. Austenite is the leading phase. The variety of cooling curve types in the hypereutectic region can be subdivided in five types (54).
The occurrence of austenite dendrites in hypereutectic gray cast iron.
In hypoeutectic cast iron, austenite should be the first phase to form, according to the Fe-C-equilibrium diagram. In eutectic cast iron, caused by asymmetric coupled growth, austenite is also formed first. In hypereutectic cast iron, graphite should form first. It is, however, unusual to find primary graphite. On the contrary, dendrites strongly resembling those in hypoeutectic compositions can be found, and as it seems are formed first. (55). In the majority of published micro photographs, these dendrites can be distinguished with the naked eye (56,57,58). An example is shown in photo 1. In many cases, however, dendrites become only visible after using special etching agents, heat treatments or a combination of both (59,60,61). Micro-röntgen techniques can also be used to reveal the extensive presence of austenite dendrites in hypereutectic cast iron, as is shown in fig.9.
To explain such structures, which are definitely found, but in fact should not occur, use is made of various highly theoretical hypothesis. Nipper(56) thought that undercooling of austenite, primary graphite or of both constituents is the reason for the existence of the variety of structures that can be found in hypereutectic cast iron. Severe undercooling of the primary phase is thought to be responsible for the formation of austenite dendrites.
Fig. 10 shows the various ways in which a supposed undercooling could effect the solidification structures. Ruff and Wallace (61) found out that the majority of austenite dendrites must have formed before graphite was deposited, in the same way as in hypoeutectic or eutectic compositions. In this case, a modern version of the undercooling hypothesis is used to explain the occurrence of austenite dendrites as shown in fig.11.
Morrogh (63) supposed that the finer eutectic graphite tends to unify with the bigger primary graphite crystals. The primary graphite is thus surrounded with a layer of eutectic austenite, often of dendritic appearance. This austenite layer is explained by Olen and Heine (59) by a different mechanism. Local deficiency of carbon, due to previous exagerated deposition of primary graphite, leads to deposition of austenite directly on the primary graphite.
Hypereutectic Ductile Iron.
At the start of ductile iron production, predominantly hypereutectic compositions were used. The enormous increase in strength that could be achieved by the new material, made it possible to use a carbon equivalent that was as high as possible in order to obtain the most favourable pouring and feeding conditions. This was regarded as the main advantage as compared to the high-strength flake graphite irons that were used up till the invention of ductile iron (64).
It turned out soon, however, that the carbon equivalent could not be raised at will. Especially with heavier wallthicknesses, high carbon concentrations were found in the top of castings when very high CE's were used (65,66). The reason for the occurrence of these graphite agglomerations seemed to be obvious: hypereutectic nodules are formed in the liquid and tend to float to the surface because of their low specific gravity as compared to that of liquid iron. The supposed floating mechanism gave name to this phenomenon: graphite flotation or in short flotation.
Fig. 12 shows this general accepted flotation mechanism. This idea, derived from and equivalent to the formation of Kish graphite in gray cast iron (67) was to be copied and used the next decennia (68,69,70,71,72,73,74,75,76). Even after ideas about Kish formation were fundamentally changed, the original concept for nodular graphite remained unchanged.
The flotation mechanism is frequently used to prove that all nodular graphite forms directly and independently in the melt (67,68,77). If one studies the flotation phenomenon, however, many facts will be found, which hardly can be explained by a simple flotation mechanism alone.-Influence of composition and casting temperature on the occurrence of graphite flotation. Views on the influence of composition on carbon flotation differ greatly. Flotation occurs from the eutectic composition on (68,71), according to (67) from CE=4.55, or from CE>4.65 (78). Different views on the influence of casting temperature also exist: an increase of flotation is noticed with increasing (71) as well as with decreasing casting temperatures (75).
-The shape of the flotation zone. Remarkable, however, is the shape of the flotation region.
If a flotation mechanism would be exclusively responsible for the occurrence of this region, a band of higher carbon concentration in the top part of a casting might be expected, as indicated in fig. 13b. In reality, however, a flotation region, as indicated in fig. 13c is found (79). Identical flotation patterns were found by others (71,75). The symmetry of this flotation zone can be influenced by the casting method (71). Within such a flotation region, a certain internal structure can be found. Extensive graphite concentration can be found in one part of the region, none will be found in an other part. Within 1 mm., (71) found variations of 11 to 20% graphite content. It was explained by the assumption that insufficient mixing had taken place during the nodularizing (inmold) treatment. When a suitable magnification is used, such structures appear to occur consistently within such flotation zones (66,67,80,81). This also indicates a mechanism more complicated than just the collection of free floating nodules. Fig.14 shows such a structure within a flotation region.
-Time of the occurrence of flotation.
According to the traditional flotation mechanism, the time flotation begins should be the moment that the graphite liquidus is reached. Rauch (67), using carbon analysis techniques, measured a carbon loss during the time interval between inoculation and casting, when the CE was above 4.77% and assumes that this was caused by carbon flotation. Fuller (71) did not find any carbon loss at a CE of 4.75% and 4.95% during holding in the pouring ladle, whilst the temperature dropped from 1400 to 13200C. Graphite flotation in the ladle itself, therefore, was eliminated as the main cause of the flotation that was found in the castings (71). When, however, ceramic filters were used, these filters were very rapidly blocked by carbon separations (71). This means that at least the initiation for graphite flotation starts rather early during the pouring process and not just in the moldcavity itself. This would make it possible to explain the typical shape of the flotation zone and the influence of the casting method on it. The fact that often oxides and slag are found within the flotation zone (69,71,76) also points in this direction. Jonas (76) is of the opinion that flotation should only be studied in relationship with oxide formation.
-Graphite detoriation within the flotation zone.
A typical event, often found in flotation zones is known as "exploded" graphite (71,73,75). The name already suggests the possible formation mechanism, i.e. the disintegration of original perfect nodules (80). Fig. 15 shows this mechanism according to (73). Less well known, however, is the fact that such "exploded" nodules are completely surrounded by austenite dendrites (79) and the fact that "exploded" nodules also occur in compositions that are extremely low in CE (82). Photos 2 and 3 show dendrite-encapsulation of "exploded" nodules.
A fundamental difference with gray cast iron.
Although the flotation phenomenon has always been regarded as an equivalent of the formation of Kish graphite in gray cast iron, there exists a fundamental difference. In strongly hypereutectic gray iron, Kish graphite is directly visible on the metal bath itself. With ductile cast iron, however, this is never the case. De Sy(83) noted that clusters of nodules can be build up in the cope, but never Kish formation. This seemed to be very peculiar, then according to Stoke's law, ball-shaped nodules will have the best opportunities to float to the surface. Other experiments to obtain real "Kish" nodules failed (81,84,85). In a single case, a small number of free nodules were found on the casting surface. The circumstances of these test, however, remain to be of a rather extreme nature. For instance, melting underneath a slag-cover consisting of molten glass, the use of extreme high silicon content (81) or a very high sulfur or aluminum content (84). Such free nodules often turned out to consist of just hollow spheres (84). Various reasons are mentioned, why, under normal circumstances, hypereutectic graphite nodules never appear on the surface as Kish graphite in gray iron does. Motz (81) regards the difference in wettability between melt and graphite as the main reason for this difference in behavior. According to (83,84) hypereutectic nodules cannot leave the liquid metal because they are formed within austenite dendrites. The source for further growth of these nodules was assumed to be the surrounding supersaturated austenite. As this theory was abandoned in the sixties, the interest for dendrite formation also disappeared. Engler et al (86) suppose that austenite dendrites encapsulate hypereutectic nodules in an very early state and in doing so prevent floating to the surface. The fact that all nodules in the flotation zone are restricted to sites within austenite dendrites can be easily revealed (79) and is shown schematically in fig.16.
The use of cooling curves.
As was the case in gray cast iron, findings and interpretations of cooling curves in hypereutectic ductile iron differ a lot. Kalvelage (87) f.i. cannot detect any temperature indication that points to the formation of hypereutectic graphite. Basutkar (88) is of the opinion that the samples that are usually used are to small to show the graphite liquidus during solidification. Measurements taken on heavier test samples show many irregularities that are explained as being liquidus arrests, as is shown in fig. 17.
Heine et al (89) note that one cannot speak of a clear liquidus arrest, but rather of change in inclination of the curve. As these arrests lie closer to the eutectic equilibrium temperature, they show up more clearly and sometimes they appear even more then one arrest! Chaudhari (90) detects liquidus arrests only if the CE% is above 4.6. Although the generally held view is that the heat developed by the formation of hypereutectic graphite is too small to be detectable, a simple calculation shows that this is not the case. Liquidus arrests found in this way are assumed to represent the graphite liquidus. As compared to untreated cast iron, however, this liquidus temperature is 500C lower (fig.19).
If cooling curves are combined with quenching tests, taken at various times during the solidification process, then it suddenly appears to make a completely different interpretation possible. Dunphy (91) noted already in 1952 that the liquidus arrest of a hypereutectic cooling curve does not indicate a graphite separation, but the formation of austenite dendrites. As the composition was clearly hypereutectic, this primary austenite formation was completely unexpected and could only be explained by the assumption that before eutectic solidification started, a severe undercooling of the primary graphite occurs. During this undercooling period, primary austenite dendrites can form, followed by eutectic solidification. Morrogh (92) also used quenching tests and showed that the liquidus arrest in hypereutectic nodular cast iron represents the formation of austenite in dendritic form. The same composition (CE=4.75%) can show a liquidus arrest or not (93). Such differences can occur even within one batch, cast shortly after each other. Fig.20 shows this phenomenon. All castings showing type II cooling curves showed a severe dendritic structure. At CE>4.6 the graphite liquidus shows severe variations, caused by production techniques (94).
Austenite dendrites in hypereutectic ductile iron.
Views about the way austenite forms during solidification as well as the influence on graphite formation, has changed dramatically during the last 50 years. At the start, austenite dendrites played a rather important role. Increasing interest for the melt-theory , however, pushed this role further and further into the background. It is not until the eighties, that the importance of austenite dendrites seems to be rediscovered. This development is documented elsewhere (79,95). The existence of austenite dendrites in hypereutectic ductile iron and its relationship with the occurrence of liquidus arrests was shown long time ago. In most cases, however, the presence of a dendritic structure stays unrevealed.
Only when dendrites are evenly distributed and their dimensions are large, they can be seen, sometimes even with the naked eye. In all other cases, a dendritic structure can only be revealed by special etching techniques (79,98,99). The occurrence of austenite dendrites seems to be a normal event (86). Successive grindings showed that, at first sight, free nodules together with their austenite halo, actually are attached to a dendrite (100,79). The common explanation for the presence of austenite dendrites is, as is the case for gray iron, the occurrence of undercooling.
Hypereutectic Vermicular cast iron.
Flotation regions also found in vermicular cast iron (101). Remarkable is the fact that the graphite shape is not vermicular but nodular or irregular. Below this flotation zone, vermicular graphite is found. A strong austenite formation, resembling those found in hypoeutectic compositions can be found (102). A thermal signal indicating the formation of hypereutectic vermicular graphite was not found (102,103). It was assumed that the heat developed during graphite separation was too low (103).
Hyper-eutectic white cast iron.
To explain the various structures that can occur in cast iron, the iron carbon double diagram was introduced as far back as 1905. Cooling curves showed that the formation of cementite was always associated with a lower eutectic temperature as compared to the formation of graphite that took place at about 100 C higher in temperature. Although in daily foundry practice the tendency of a cast iron to solidify according to either one of the systems is usually treated as a matter of available nuclei, there appear to exist many different views as to the actual chilling tendency. Fras  mentions at least six different mechanisms. Two types of cementite eutectic exist. The original eutectic, named Ledeburite, in which small islands of austenite are dispersed in the carbide phase, and the platly, divorced or massive carbide eutectic. Typical of the last is the fact that the second eutectic phase, the austenite is lacking. To account for the disappearance of 50% of the eutectic, it is assumed that this eutectic austenite settles on the existing primary austenite. The author (105) showed, however, that the austenite islands within the ledeburite do not form the second phase of a eutectic, but are in fact sections through small compact dendrite arms that belong to the neighboring austenite. In all cases, the carbide "eutectic" consist of only one phase! This phase takes the space that is left over by the primary austenite. So there is no reason to make a distinction between the two types of carbide eutectic. It can even be questioned as to whether one can really speak of a eutectic. As the carbide "eutectic" does not show a single melting point, but rather a temperature region that is strongly influenced by elements that do not dissolve in the carbide phase itself (106), the only reason to speak of a eutectic remains the existence of primary carbides!
The hypereutectic carbide region.
As cast iron compositions in this region are of no practical interest, knowledge of these alloys is generally limited to scarcely published material, including micro photographs. Photo 4 shows an example of the structure of a hypereutectic manganese alloyed white cast iron. Clearly visible are the needle shaped primary carbides in an ledeburite matrix. The formation of such primary carbides is normally explained by the solubility curve for cementite in the hypereutectic region of the iron carbon diagram. The needleshaped cementite crystallize as the primary phase, increasing the carbon content of the rest-smelt until eutectic composition is reached, followed by the formation of ledeburite eutectic. An example of the German version of the iron-carbon diagram is shown in fig.21.
In the Anglo-American version, shown in fig.22, the liquidus carbide line is left out.
Reason for this, is the fact that de solubility of iron-carbide in liquid iron is insufficiently known (107), and moreover, the melting point of Fe3C is still not established (108). This is demonstrated in fig. 23, in which all known liquidus lines and assumed melting points of Fe3C as proposed during the last century (109-125), are collected and visualized. Values are based on measurements, theoretical calculations as well assumptions. The CD curve turns out to be so hypothetical, that it has also been proposed to leave it out in the German version of the iron-carbon diagram (126).
The shape of the primary carbide structure.
The most obvious component within the structure of hypereutectic white iron, are without any doubt, the big needle shaped carbide crystals. Benedicks (127) was the first to investigate this structure.
His interpretation is shown in fig. 24, in which the main component is formed by the carbide crystals. The remaining space is filled with ledeburite eutectic. This structure resembles the one that is shown in photo 4. Apart from needle-shaped carbides, also massive carbides, strongly resembling those which are found in hypoeutectic alloys, occur. So a wrong interpretation of such carbides is quite possible (128). The danger to make a wrong interpretation increases with phosphorous containing alloys, where an increasing phosphorous content alters the original ledeburitic structure in needle-shaped carbides, resembling primary carbides (129). Structure judgement without a reliable chemical analysis can easily lead to misinterpretation. The dimensions as well as the shape of primary carbides can also be influenced by external factors. A lower casting temperature refines primary cementite needles (130). When during solidification the melt is vibrated, the orientation of primary carbide separations becomes more at random, moreover, part of the needle structure is lost and changed to massive shapes (131).
Results of research work done by the author .
To obtain hypereutectic carbide structures, use was made of a high manganese containing pig iron. Analysis: C 4.6%, Mn 7.5 %, Si 1%. A typical structure of this material is shown in photo's 5 and 6. Long carbide needles, massive carbides and ledeburite eutectic can be distinguished. This ledeburite is mapped out in a dendritic pattern (photo 7). By using a primary etching agent, the structure is revealed extremely clearly. If one examines the primary structure at a higher magnification however, it appears that there exists still another needle-shaped structure, as is shown by photo 8. Next to the primary carbides, needle-shaped austenite is found, which were noticed earlier by Htun (128). Also the first individual primary carbides, isolated by Benedicks (35) were always covered with austenite dendrite-needles. In fact this is the only real needle shape within the structure! As soon as the austenite needle is interrupted, the carbide just follows the contours and becomes indistinguishable from the ledeburitic carbide phase. The needle shaped primary carbides are thus an optical illusion, caused by the delineating effect of the surrounding austenite dendrite needles. This is clearly shown in photos 9 and 10, where a carbide phase fills up the space which was left over by the austenite. Photo 11 shows that the primary austenite needle is continuous with the austenite phase of the ledeburite eutectic. Exactly as was the case in the carbide "eutectic" in hypo-eutectic alloys, the primary carbide phase is just a space filling component, indicating that it was the last phase to solidify in a final shape forced upon by the austenite. Olen and Heine (59) suggested already 30 years ago that carbides might best be regarded as ordered solid solutions rather than as separate entities which precipitate directly from the liquid. The formation temperature of the carbide phase was found to be close to the stable eutectic temperature. Hecht and Margerie (133) also observed that primary carbides are formed close to the eutectic temperature. How simple this observation may be, accepting this statement has a wide bearing, then it not only means that the existing iron-carbon double diagram is in error, but that its concept is fundamentally wrong!