l llllllllHHlHllll I l HI II 114 882 HTHS i‘EfiL-‘UENCE C??- {5.52 30% CCINTEHT ON THE Dififlfii OF l‘éUCLEfiR‘SQi‘é (3-? THE EUTEECTW FREEZING 3N GREY CAST IRON ‘f‘hmis ‘m' fiw film-awe of M S, WCHI'IEAR ESTATE UNP‘JERSETY flékshixw Pmsacfi Laiftsis'f WM This is to certify that the thesis entitled Influence of Carbon Content on the Degree of Nucleation of the Eutectic Freezing in Grey Cast Iron. presented by Mr. Dakshina P. Lahiri has been accepted towards fulfillment of the requirements for M-S- degree in Metallurgical Engineering ‘4 // :l l ,r ,‘ T l’ I I I / (23’! , ,' Major professor Date July 21+, 1961 0-169 LIBRARY Michigan State University INFLUENCE OF CARBON CONTENT ON THE DEGREE OF NUCLEATION OF THE EUTECTIC FREEZING IN GREY CAST IRON BY DAKSHINA PRASAD LAHIRI A THESIS Submitted to Michigan State University in partial fulfillment of the requirements for the degree of MASTER OF SCIENCE Department of Metallurgical Engineering 1961 \ ABSTRACT INFLUENCE OF CARBON CONTENT ON THE DEGREE OF NUCLEATION OF THE EUTECTIC FREEZING IN GREY CAST IRON By Dakshina Prasad Lahiri ! A study was made to determine the correlation between the carbon content of the iron melts and the degree of nucleation of the eutectic freezing. The technique of quenching partly solidified wedge—shaped specimens, when the eutectic solidi— fication had progressed to some extent, was used. The experi- mental heats were melted in an induction furnace. The amounts of Si, Mn, P and S in the melts were kept constant within reasonable limits. A limited amount of data was also obtained to determine the effect of superheating on the degree of nucleation of the eutectic freezing. The following conclusions were drawn from this study: 1. The number of the eutectic cells decreased sharply as the carbon content was lowered from 3.56% to 3.29%. 2. The microstructures of the eutectic cells changed sharply from random to interdentritic graphite also in the same range of variation in carbon content. 3. The microstructures at the surface of the slow cooled pieces also changed abruptly from random to interdentritic iii graphite when the carbon content was lowered from 3.56% to 3.29%. 4. The eutectic start temperature was lowered with lowering of carbon content in the melt. 5. Superheating of the melt increased the degree of nuclea- tion of the eutectic, produced less interdendritic graphite and decreased the depth of clear chill. On the other hand, it lowered the eutectic temperature. ACKNOWLEDGEMENT The author wishes to acknowledge gratefully the encouragement and aid extended by Dr. H. L. WOmochel and Dr. D. D. McGrady during the course of this study. Assistance contributed by Mr. B. D. Curtis is also acknowledged. iv TABLE OF CONTENTS Page I. INTRODUCTION . . . . . . . . . . . . . . . . . . 1 II. SCOPE OF INVESTIGATION . . . . . . . . . . . . . 11 III. EXPERIMENTAL PROCEDURE . . . . . . . . . . . . . 12 IV. DATA AND RESULTS . . . . . . . . . . . . . . . . 18 Part A. Chemical Compositions 18 Part B. Macrostructures of Quenched wedges 22 Part C. Microstructures of Quenched wedges 25 Part D. Microstructures of Slow Cooled wedges 30 Part E. Chilling Tendency 38 Part F. Cooling Curve Data 41 Part G. Effect of Superheating 44 V. DISCUSSION . . . . . . . . . . . . . . . . . . . 47 VI. CONCLUSION . . . . . . . . . . . . . . . . . . . 53 VII. BIBLIOGRAPHY . . . . . . . . . . . . . . . . . . 55 vi LIST OF TABLES Table Page I. Chemical Compositions of Charge Constituents . . . 19 II. Constituents of Charge for Various Heats . . . . . 20 III. Chemical Compositions of Experimental Heats . . . 21 IV. Chill Test Data . . . . . . . . . . . . . . . . . 39 V. Cooling Curve Data . . . . . . . . . . . . . . . . 43 10-15 16-17 18 19 20 21 LIST OF FIGURES Macrostructures of quenched wedges Microstructures of quenched wedges of heats D1 to D5 and D7 . . . . . . . . . . . . . Microstructures of slow cooled wedges of heats D1 to D5 and D7, at surface . Microstructures of slow cooled wedges of heats D3 and D4, 1.5 mm below surface . . Microstructure of slow cooled wedge of heat D7, in the interior . . . . . . . Photograph of the fractured chill specimens of heats D1, D3 and D7 . . . . . . . . . Microstructure of quenched wedge of heat D6 0 O O O O O O O O O I O O O O 0 O O O O Microstructure of slow cooled wedge of heat D6, at surface . . . . . . . . . . . vii Page 24 27—29 33-35 36 37 4O 46 46 I INTRODUCTION Many methods of producing different forms of graphite in cast iron are known, but no generally acceptable explan- ation has been offered for the mechanism of graphite forma- tion, nor are the reasons for obtaining different cast iron structures under certain conditions fully understood. Among other things, there is still disagreement on whether graphite forms directly from the eutectic liquid or is a product of decomposition of cementite. Theoretically, it appears that either graphite can form straight from the liquid state or, when graphite is not easily nucleated, cementite may be formed first, subsequently decomposing into graphite. It is generally agreed at present that coarse, randomly oriented (type A or B) graphite forms directly from the melt, but the origin of fine, interdentritic under-cooled (type D or E) graphite is still disputed. The solidification of a hypo-eutectic grey cast iron under conditions resembling those of stable Fe-Si-C equilibrium diagram involves first the formation of primary dendrites of austenite followed by the eutectic. The solidification of the latter takes place in a cellular form in which the graphite is present as thin flakes whose surfaces are roughly parallel to the close—packed plane of carbon atoms. These eutectic cells may be regardedzfisgrowing outwards until they fill the inter- dendritic space. Whentie carbon content is low, the primary dendrites of austenite grow to considerable size before the separation of eutectic begins, and consequently the volume occupied by the flakes of graphite is severely limited; the longest graphite flakes lie in the directions of the inter— dendritic regions and may be bent and deformed. In contrast to this, when the carbon content approaches that of the eutectic, the primary dendrites occupy a relatively small proportion of the volume, and the cells of the eutectic have greater freedom to grow until they touch one another. When the stable eutectic in hypo—eutectic grey cast irons crystallizes, the flake form of graphite (type A or B) is usually formed except when the cooling is rapid. Very pure Fe-C-Si alloys tend to give a finely divided form of graphite variously referred to as super-cooled, eutectiform and under- cooled (type D or E). This form of graphite is promoted in commercial irons by superheating the melt, by rapid cooling and by the presence of titanium (1). The sulfur of commer— cial irons assists in the formation of flake graphite. Much doubt existed as to the origin of the under—cooled structure, and in particular whether it represented the true iron-graphite eutectic. It was suggested by Bash (2) that under-cooled graphite resulted from the decomposition of what was originally a metastable (austenite + cementite) eutectic. This con— clusion was at first thought to be confirmed by Morrogh and Williams (3, 4), but in a later paper Morrogh (5) agreed that in many cases under-cooled graphite forms directly from the melt, as was suggested by the work of Hultgren, Lindblom and Rudberg (6). On the other hand, Igarashi, Ohira, Ikawa and Horigoma found that under—cooled graphite can form directly from the melt as well as from the decomposition of ledeburite (10). So, it is seen that the mode of formation of under-cooled graphite is still disputed. In the hyper-eutectic cast irons, it is well known that when the carbon content is high, graphite known as "kish" readily separates, and floats to the top of the melt if the temperature is not kept sufficiently high. This graphite is in the form of flakes whose flat surfaces are parallel to the hexagonal layers of carbon atoms. The pro-eutectic graphite flakes in hyper eutectic cast irons is believed to form directly from the melt. It is now known that grey cast irons with randomly oriented graphite flakes (type A or B) have superior mechanical properties than corresponding irons with under- cooled inter-dendritic (type D or E) graphite structure. It is also known that by inoculation with suitable agents, it is possible to produce randomly oriented graphite in low carbon hypo—eutectic irons which ordinarily produce under— cooled structure. The technique of inoculation is at present commercially exploited to produce high strength cast irons of hypoeutectic composition. Until recently, inoculation has been regarded as primarily a process of addition of silicon or various high silicon alloys. Some of the most important alloys are ferro—silicon, calcium-silicon and silicon-manganese- zirconium alloys. It was pointed out by McClure, Khan, McGrady and Wmmochel (7) as well as by McGrady, Langenberg, Harvey and WOmochel (8) that only those grades of Fe—Si, Fe-Mn and Si—Mn-Zr alloys which contained some amount of active metal like calcium had the ability to inoculate irons successfully. The inoculating power of the various alloys was found to be dependent on the amount of calcium present in them. It was found that inoculation with metallic calcium produced a marked improvement in graphite distribution and mechanical properties in hypo-eutectic grey cast irons. It was concluded that calcium and high calcium silicon alloys were the most effective inoculants (8). McGrady et al. (8) also found that inoculation by ladle addition of graphite produced an improvement in mechanical properties and in graphite distribution. Graphite addition in ladle was effective in nucleating the eutectic solidification. Various hypotheses were proposed from time to time to explain the mechanism of inoculation. Most of them were based on a nucleation mechanism as it is implied in the term "inoculation." Some of the more important hypotheses are as follows: 1. Gas theory: Since silicon and other elements contained in the addition alloys are deoxidizing agents, it is argued that the reduction of the oxygen content produced by the addition effect the change in microstructure. Nitrogen and hydrogen are also thought to be involved in changing the graphite shape and distribution. 2. Undercooling theory: According to this theory, type D or E graphite irons (the so-called “abnormal irons") are formed as a result of under- cooling. Addition of inoculating agents prevents this under— cooling by providing nuclei which produce the desirable ran— dom or type A graphite distribution. 3. Graphite nucleus theory: This theory advances the idea that addition of silicon causes localized precipitation of graphite which nucleates flake formation. In support of this idea, it is pointed out that direct graphite additions serve to inoculate iron. 4. Carbide stability theory: It is argued that changes in carbide stability affect the availability of carbon to the growing graphite flakes and thus influence flake size and shape. Since inoculation is accompanied by a reduction in chilling tendency, it is contended that this fact supports the carbide stability theory. 5. Surface tension or surface energy theory: According to this theory, the inoculant influences the size and shape of graphite particles by supplying or re- moving adsorbed substances from the graphite-metal interface thereby changing the interfacial tension and thus promoting the formation of certain graphite forms. None of these theories seems to be capable of explaining all of the observed facts relating to the inoculation pheno- menon (8). McClure et al. (7) noted that inoculation of grey irons by calcium produced decarburization of the iron melts by the formation of calcium carbide of limited solubility in the melts. They proposed that calcium carbide might have played some role in the formation of type A graphite, either by nucleating eutectic solidification, or by changing the graphite-austenite interfacial energy as a result of its adsorption at the interface. It was pointed by Langenberg (9) that late addition of graphite to the melt in the ladle causes the nucleation of many small centers of eutectic crystallization as opposed to a small number of large centers in the untreated iron. This observation lends support to the idea of the presence of minute graphite particles in the melt which may serve as nuclei for the crystallization of graphite-austenite eutectic. It can be argued that the late addition of graphite may not give a homogeneous solution. The graphite may first be dispersed non-homogeneously in the melt and when part of this dispersed phase goes in solution, that may cause local saturation of liquid with respect to graphite. Those localized regions which are thus saturated can throw out graphite as precipitate quite easily when the temperature of the liquid is lowered. Depending upon the degree of local saturation, it is quite conceivable that precipitation of graphite by this mechanism may take place over a range of temperature as the melt approaches the eutectic temperature during cooling in the mold. Only those graphite particles thus precipitated, which will live long enough without re- solution in the melt as to reach the eutectic temperature, can take part in nucleating the eutectic freezing. It is known that the effect of addition of nucleating agent wears off completely if the melt is held at high temperature for some time after adding the inoculant. This observation can be explained by pointing out that the long holding will allow the nucleating agent to go in uniform solution, thus depriving it the chance to cause local saturation. It is also known that to be most effective, the inoculant should be of optimum size for a given amount of addition (9). If it is very fine, then it can easily form homogeneous solution. On the other hand, if the inoculant is very coarse, only a few regions have any chance of localized saturation. All the above arguments can also be put forward in favor of undissolved graphite particles acting as nuclei of eutectic freezing. It can be assumed that these graphite particles are present in the melt due to incomplete solution of the graphite added as an inoculant. It is also possible that the graphite added as an inocu— lant may first go into solution in the melt and during later cooling it may form colloidal dispersion of carbon or graphite. Also it is conceivable that the graphite forms colloidal dispersion in the melt as soon as it is added. This colloidal dispersion of carbon or graphite may coagulate to some extent as the temperature of the melt is lowered and finally at the eutectic temperature it may initiate the eutectic freezing. If this is the mechanism by which late addition of graphite nucleates the eutectic freezing, then it can be postulated that any late addition, which can induce carbon in the melt to be colloidally dispersed to some extent and later its agglomeration as graphite nuclei, will be able to nucleate the eutectic freezing. If the late addition of graphite nucleates the freezing of graphite—austenite eutectic by causing local saturation of the melt with respect to graphite, it would be instructive to know about the nucleation of eutectic freezing in melts containing increasing amounts of carbon. It is expected that as the carbon content of the melts is increased, due to statistical fluctuation of carbonjrithe melt, there is a greater probability of causing localized saturation, and so greater chance to form graphite nuclei and thus increased degree of nucleation of graphite-austenite eutectic. Also, as the carbon content of the melts is increased, there is a greater possibility of getting certain amount of carbon or graphite as colloidal dispersion in the melt at temperatures close to the eutectic, assuming that at least a part of carbon is so dispersed in the melt. This colloidal dispersion can then coagulate with greater ease in higher carbon melts and it can increase the degree of nucleation of eutectic. If any positive correlation can be found between the carbon content of the melt and the degree of nucleation of graphite-austenite eutectic, it may be possible in some future investigation to establish a unique mechanism for the nucleation of the eutectic freezing. lO 11 II SCOPE OF INVESTIGATION The present study was aimed at correlating the carbon content of the iron melts with the degree of nucleation of the eutectic freezing. It was decided that nothing would be added to the melt which might have any nucleating effect on the subsequent freezing of the eutectic. Since the degree of nucleation of the eutectic is thought to be related to the superheating and the pouring temperatures, it was decided to pour heats directly from the melting furnace at 2650°F after prior superheating of the melt to 2800°F. Only one heat was to be poured from 29OOOF to find out the dif- ferences caused by the increased superheat. Since the degree of nucleation of the eutectic is thought to be related to the undercooling in the eutectic freezing, it was decided to take time-temperature cooling curves for all the heats poured. These cooling curves were to be utilized to determine the proper moments for quenching the partly solidified wedge specimens. The technique of quenching wedges to suppress the progress of freezing of the eutectic was to be similar to that used by McGrady et al. (8). 12 III EXPERIMENTAL PROCEDURE All heats studied in this work were melted in a 20 Kw high frequency induction furnace. A magnesia crucible of 31 lbs capacity was used to melt the heats. The chemical analysis of the charge constituents is given in Table I. The actual amounts of the constituents Which comprised the charges of the various heats are shown in Table II. The procedure followed to prepare a heat is described below: The crucible of the induction furnace was first charged with a piece of cast iron slug, all the ferro-silicon was charged next, followed by some ingot iron, pieces of cast iron and the rest of the ingot iron. (It is to be noted that no ingot iron was used in heats D1 and D2.) Since all the materials could not be charged in cold, one piece of cast iron slug weighing about 6 lbs was withheld at the start. The cold charge was packed nicely in the crucible so that it could be covered with a refractory brick to minimize heat loss and oxidation of the charge during the melting. When a part of the charge was molten and there was some room in the crucible, the last piece of C. I. slug was charged. When the l3 materials in the crucible were completely molten, ferro- manganese, ferro-phosphorus and ferro—sulfur were added through a chute after parting the slag cover slightly with a steel rod. A few minutes after making these final addi- tions, the temperature of the melt was measured by dipping a platinum—platinorhodium thermocouple which was protected by a short fused silica tube. The temperature measurement was continued for some time. When the melt attained a tem- perature of 28000F (in case of heat D6, the highest bath temperature was 29000F), power to the induction furnace was cut off. Slag was then skimmed off the top of the melt. 'When the temperature dropped to 26500F, the melt was poured into the molds directly from the melting crucible. The pro- gress of a typical heat is given below: Heat No. D4 Date: 7th March, 1961 3:00 P.M. Power on. Power input: 20 KW. Fe-Si, ingot iron, all C.I. slugs except one piece charged. 3:35 Last piece of C.I. slug charged 4:00 32.6 gms Fe—Mn, 16.0 gms Fe—P and 5.0 gms Fe—S charged. 4:04 Temperature 26000F Power: 20 KW 4:08 Temperature 27000F 14 4:12 P.M. Temperature 27750F Power: 18 Kw 4:15 Temperature 28000F Power off. Slag skimmed. 4:20 Temperature 265OOF Poured into molds. Dry sand molds were made from a mixture of lake sand, cereal, linseed oil and water. The molds were baked for four hours at a temperature of about 4750F. Only the 1.2 inch diameter transverse test bar molds were then washed with a nonucarbonaceous silica slurry and the baking cycle was repeated on them. Rectangular chill blocks of approximately 2 1/4" x 3/4" x 3 7/8" were poured in molds prepared as above. For the quenching experiments with wedge-shaped castings, the molds were prepared in the same way as the chill block molds. The wedge-shaped castings were 2“ x 1" in cross—section at tOp, tapering to 2" x 1/4" at the reduced end. The length of the casting was about 5". For the quenching experiments, chromel—alumel thermocouples of 22 gage wire were used. The thermocouple wires were pro- tected by thin porcelain insulators and fused quartz tube. The tip of the couple projected from the quartz tube and was coated with a thin wash of alundum cement. The couple was located on the thinner side of the mold at a distance of one inch from the heavy end of the casting. The tip of the couple was placed 15 at a distance of about 3/16" from the inner mold wall. For automatic recording of the time—temperature curve, a high speed electronic temperature recorder was used. For heat D1, two of the molds for wedges were provided with thermocouples and the third mold had none. The melt was first poured into the two molds with thermocouples, then into the third mold without thermocouple, then into the chill block molds and finally into the vertically fixed 1.2 inch diameter transverse test bar molds. The first two wedge molds were poured over quenching water tanks. When the eutectic solidification was in progress, as indicated by the time- temperature curves, the molds were broken and the partially solidified castings were dropped into the quenching tanks. The third wedge was allowed to cool slowly to room temperature. On sectioning the quenched wedges, it was found that there were hollow regions in the central portion of the castings. These were probably formed due to ejection of molten metal through the thermocouple entry points. So, it was decided to pour only one wedge mold with thermocouple on and to allow it to cool slowly below the transformation temperatures and to determine from this time—temperature curve the moments at which the other two wedges were to be quenched. It was found in the heat D2 that this procedure worked well and so this was continued for the rest of the heats. The actual pouring se— quence in the three wedge molds was: 16 wedge mold No. 1 without thermocouple - poured first wedge mold No. 2 with thermocouple - poured next Wedge mold No. 3 without thermocouple ~ poured last After pouring in the three wedge molds, the two chill block molds and three 1.2 inch diameter transverse test bar molds were poured. The water quenched wedges were sectioned at a distance of about one inch from the heavy end by an abrasive wheel cutter. Precautions were taken during sectioning so that the martensite formed by the water quench is not tempered. The pieces were then prepared for macro— and micro-examinations. The slow cooled wedges were similarly prepared for microscopic examinations. The chilled blocks were fractured after making a little saw cut at a corner on the grey-iron side. All the pieces were fractured nearly at the mid-points. The depths of total chill and clear chill were measured. The results are tabu— lated in Table IV. To show the trend in variation of chill depth with carbon content, the fractured surfaces of three specimens were photographed. The transverse test bars were not tested in this study. They were stored for future references. 17 Samples for chemical analysis were drilled from the grey iron end of the chilled blocks. The chemical analyses were carried out by a commercial laboratory. The chemical analyses of the various heats are given in Table III. 18 IV DATA AND RESULTS Part A. Chemical Compositions Chemical compositions of charge constituents are given in Table I. Constituents of charge for various heats are given in Table II. Chemical compositions of experimental heats are given in Table III. Maximum ranges of variation for different elements were: Silicon - 0.40% Manganese - 0.19% Phosphorus - 0.014% Sulfur - 0.017% The mean values for the different elements other than carbon were: Silicon — 2.26% Manganese — 0.65% Phosphorus - 0.142% Sulfur - 0.061% Percentage carbon equivalent, as defined by the relation, %C equivalent = %C + 1/3 x %Si, for the various heats are also shown in Table III. On this basis, heats D1 and D2 are hyper-eutectic and the rest of the heats are hypo—eutectic. It is to be noted that heats D6 and D7 have very nearly the same carbon equivalent. Chemical Compositions of Charge Constituents TABLE I Constituent % C % Si % Mn % P ‘% S Cast Iron Slug No. 3.84 1.18 0.76 0.136 0.045 Cast Iron Slug No. 3.82 1.18 0.77 0.137 0.060 Ingot Iron 0.02 -- -- 0.02 0.02 Ferro—Silicon 0.44 27.40 0.84 0.033 0 018 Ferro-Manganese -- —- 80.00 -- -- Ferro-Phosphorus -- -— —- 25.00 -- —- -- -- -- 50.00 Ferro—Sulfur 19 TABLE II Constituents of Charge for Various Heats (In Pounds) 20 Heat No. C°nStltuent D1 & D2 D3 D4 D5 D6 & D7 Cast Iron Slug No. 1 15.1 14.0 12 9 ll 9 10.9 Cast Iron Slug No. 2 14.5 13.5 12.4 11.4 10.4 Ingot Iron — 2.0 3.93 5 88 7.73 Ferro- Silicon 1.32 1.41 1.65 1.67 1.78 Ferro— Manganese 0.0308 0.051 0.0715 0.091 0.114 Ferro— Phosphorus 0.0154 0.0268 0.0352 0.046 0.057 Ferro- Sulfur 0.007 0.0092 0.011 0.013 0.015 30.996 Total Charge 30.973 30.997 30.998 31.000 Chemical Composition of Experimental Heats TABLE III 21 Percent element %C equi- Heat No. C Si Mn ' P S valent Dl 3.79 2.21 0.75 0.145 .072 4.53 D2 3.75 2.02 0.71 0.134 .055 4.42 D3 3.56 2.11 0.70 0.147 .065 4.26 D4 3.29 2.35 0.65 0.148 .060 4.07 D5 2.91 2.38 0.62 ~0.142 .055 3.70 D6 2.68 2.42 0.56 :0.139 .055 3.49 2.75 2.33 0.57 0.138 .062 3.53 D7 22 Part B. Macro-structures of Quenched wedges Macro-structures of irons quenched at the start of the eutectic solidification are shown in Figs. 1, 2 and 3. These were prepared from the quenched wedges as described before under the heading "Experimental Procedure." After final metallographic polishing, the specimens were etched deeply with 2% Nital and only one set of specimens which were quenched at nearly an equivalent state of solidification (as was evident from near constancy of the thickness of the slow cooled rim in the macro—structure) were photographed at natural size. From left to right, Fig. 1 shows quenched specimens of heats D1, D2 and D3; Fig. 2 shows those of heats D4, D5 and D6; and Fig. 3 shows those of heats D6 and D7. Only the heat D6 was superheated to 29000F. The rest of the heats were superheated to 28000F. The round dark areas in the photo—macrographs were revealed by the microscope to contain graphite and to be centers of solidification and transformation of the eutectic in a back- ground of primary dendrites and quench ledeburite. The trans- formation cells appear dark in the macrographs because of the presence of graphite and of the absence of massive cementite. From the macrographs, it is observed that the decrease in the number of eutectic cells is not gradual as the carbon content is lowered from heat D1 to heat D7, taking into account 23 only those heats which were superheated to 28000 F (namely, heats D1 to D5 and D7). It is evident that there is a sharp reduction in the number of eutectic cells in heat D4 as com~ pared to those in heat D3. Also the average size of the cells increases as the number of the cells decreases. In Fig. 3 it is found that there are more cells in heat D6 than in heat D7, though the former was superheated to higher temperature. This is surprising, since it is common- ly believed that superheating destroys potential graphite nuclei. In Fig. 1, the macrostructure of heat D1 looks a little different than the others. It is not clear why it should be so. As the quenching technique used in heat D1 was slightly different and as the central part in the specimen is left hollow due to reasons previously mentioned, these may lead to the apparent difference in the macrostructure. J 24 Fig. 1. From left to right. Heats D1, D2 and D3 Fig. 2. From left to right. Heats D4, D5 and D6. Fig. 3. From left to right. Heats D6 and D7. Figs. 1-3. Macrostructures of irons quenched at the start of eutectic solidification. Nital etch. Natural size. 25 Part C. Microstructures of Quenched wedges After taking the macrophotographs, the same set of quenched wedges were re-polished and etched lightly with 1% Nital. Each specimen was examined carefully under the micro- scope and photomicrographs of typical eutectic cells (round, dark areas of the photomicrographs) were taken at 100x. Photomicrographs of the quenched specimens of heats D1, D2, D3, D4, D5 and D7 are shown in Figs. 4, 5, 6, 7, 8 and 9 respectively. All these heats were superheated to 28000F. Examination of the micrographs shows that in heats D1 and D2 (Figs. 4 and 5), the microstructures of the growing cells contain randomly oriented type A graphite. In heat D3 (Fig. 6), type A graphite is still seen, but here the graphite is much finer than those in heats D1 and D2. In heats D4, D5 and D7, the microstructures of the growing cells (Figs. 7, 8 and 9) contain interdendritic type D graphite. It is found that as the number of eutectic cells decreases, the average size of the individual cells increases. The dark, unresolved areas at the boundary of the eutectic cells were found to be very fine graphite at higher magnifi- This fine graphite is believed to be formed in the cations. initial stages of the quenching. The background of the micro- graphs consists of transformed austenite dendrites (dark) and quench ledeburite (lighter). At higher magnification, a 26 martensitic structure is found in all places where there was austenite at higher temperatures, namely primary austenite as dendrites and austenite in "quench” ledeburite. Most of the austenite thus present in the structure above the eutectoid temperature was transformed to martensite due to the water quench which was used to arrest the normal solidification of the irons. So, at room temperature the microstructure of the quenched wedges consists of graphite in a matrix of martensite, cementite and some retained austenite. In Fig. 4 (heat D1), there are some austenite dendrites visible in the microstructure, while in Fig. 5 (heat D2), there is no primary austenite visible. This is surprising, because heat D1 has higher carbon (as well as higher C—equivalent) than heat D2. The appearance of austenite dendrites in Fig. 4 might be due to non—equilibrium cooling; or, it may indicate that the commonly accepted C—equivalent value of 4.3% for the eutectic may not be correct. Heat D1. Quenched specimen. Nital etch. HEat D2. Quenched specimen. Nital etch. 100x. 100x. 27 28 i k5 “ .0} . “Tm _ i ~-v-s.‘.‘ V. . ' ‘ l ' ." ' . ‘- ‘. - _ ‘ -i,.;,"' ~ , ' . o " ‘. ‘1“ >‘. . ‘. "‘ ' on \- ‘3'.- Fig. 6. Heat D3. Quenched specimen. Nital etch. 100x. Fig. 7. Heat D4. Quenched specimen. Nital etch. lOOX. \ To .3. . .‘I c Fig. 0 Q " ‘ “_ ‘1. fit-"- :r\ $ ‘_ .‘l ’. A" -’ 3...“. ‘ww".'-wf,cx ‘-.4-'\ I— ‘ i. .5 l _ W... ' I o‘ Viki! Q w :4. i 'o .1. ‘. 8. Heat D5. Fig. 9. Heat D7. Quenched specimen. 29 ~ as“. h \ . V \1\..- ‘q - I ~‘ , ‘ . ‘Js‘ \‘ ‘ Nital etch. 100X. Nital etch. lOOX. Quenched specimen. 30 Part D. Microstructures of Slow Cooled wedges Since it is believed that there is a correlation between the degree of nucleation of eutectic freezing and the micro- structure at the surface of the casting [lesser degree of nucleation manifests itself as a tendency towards type D and E graphite at the surface (7)], it was decided to study the microstructures at the surface of the slow cooled wedges. Slow cooled wedges were prepared for microscopic examin— ation as described before. The specimens were etched with 1% Nital and were carefully examined both at 100x and 500x at the surface as well as in the interior. Photomicrographs of the representative struc— tures were taken mostly at 100X. Figs. 10, ll, 12, 13, 14 and 15 show the microstructures at the surface of the slow cooled wedges of heats D1, D2, D3, D4, D5 and D7 respectively, all at 100X. It was found that in those cases where there was a fine, interdendritic graphite (type D or E) at the surface, the microstructure changed gradually to coarse, random (in some cases, coarse, interdendritic) graphite as the interiors of the pieces were gradually approached. Figs. 16 and 17 show the microstructures at a distance of 1.5 mm below the surface in specimens of heats D3 and D4 respectively at lOOX. 31 Fig. 18 shows the interdendritic graphite in specimen of heat D7 at a magnification of 500x. Fine, interdendritic graphite which is not well-resolved at 100x in Fig. 15 can be clearly seen in Fig. 18. In Figs. 10, 11 and 12 (heats D1, D2 and D3 respectively), it is found that as the carbon content is reduced, the gra— phite flakes become shorter and finer. In all these three microstructures, graphite is mostly of random type A, though in Fig. 12, some amount of interdendritic disposition of graphite is visible. In Figs. 13, 14 and 15 (heats D4, D5 and D7 respectively), the graphite is clearly interdendritic and becomes finer and more interdendritic as the carbon in irons is lowered. Comparing Figs. 12 and 13, it is evident that there is a sharp transition in the mode of distribution of graphite. In Fig. 12, graphite is mostly random, while in Fig. 13, it is highly interdendritic. Comparing Figs. 12 and 16, it is found that the graphite distribution is more random in the interior, but the difference is not so marked. On the other hand, comparing Figs. 13 and 17, it is found that there is a marked difference in the graphite distribution between that at the surface and in the interior. This type of variation between the structures at the surface and in the interior was also noticed in the slow 32 cooled specimens of heats D5 and D7. So, from these microstructures, it is evident that the graphite distribution at the surface of the slow cooled specimens changes with the change in carbon content of the irons. But the change from random, coarse graphite to inter- dendritic, fine graphite seems to be quite abrupt as the carbon content is changed from 3.56% in heat D3 to 3.29% in heat D4. This abrupt change in the microstructures of the slow cooled specimens of heats D3 and D4 seems to be related to the sharp reduction in the number of eutectic cells in the quenched specimen of heat D4 over those of heat D3, as was pointed out before. Also, it is found that when the microstructure at the surface is fine, interdendritic graphite, there is a change from this type of distribution to more random and coarser graphite into the interior of the specimen. So, according to the criterion of McClure §£_al, (7), it can be concluded that heats D4, D5 and D7 have fewer nuclei for the freezing of the eutectic and the formation of random graphite. This is also evident from Figs. 7, 8 and 9 (microstructures of the quenched specimens of heats D4, D5 and D7) and Figs. 2 and 3 (macrostructures of quenched specimens). 33 ..,.,.:. \./ l \‘ ' “ L/\\ //A Fig. 10. Heat Dl..Slow cooled. Nital etch. 100x. At surface. . . x < K. ; ..\),v L . a. z ,. 1., W o I .1! L \dv. . . a 3.. \ s ._ . 1.1 \(KV\ .5 , Pfivfl \ Fig. 11. Heat D2. Slow cooled. Nital etch. lOOX. At surface. 34 I \ l. I I Q _ (ll. 7 L. hug...» At surface. lOOX 12. Heat D3. Slow cooled. Nital etch. Fig. Heat D4. Slow cooled. Nital etch. lOOX. At surface. 13. Fig. Fig. 14. Heat D5. Slow cooled. Nital etch. lOOX. At surface. .’ ’ , . _ 4 \3, . ‘1. - ’ .‘ _ 1‘ “‘6' y. l - f ‘1‘ .' .‘ \K‘ ' . 3 " —~ I. ‘ \t‘ _ 1~ . ' I) I ’y , I ‘I ‘\ J a. % r ”a, . . I ' . n. w \ H . i 1 a ___A t.) ‘./I-. '- {A 1 S--- ‘L 'A ¢ \ - Fig. 15. Heat D7. Slow cooled. Nital etch. lOOX. At surface. 35 Fig. 16. Heat D3. Slow cooled. 1.5 mm below surface. Nital etch. lOOX. Fig. 17. Heat D4. Slow cooled. 1.5 mm below surface. Nital etch. lOOX. 36 37 500x. Nital etch. Slow cooled. 18. Heat D7. In the interior. Fig. 38 Part E. Chilling Tendency The results of the chill depth measurements are given in Table IV. Fig. 19 is a photograph, at natural size, of the frac- tured surfaces of chill blocks for heats D1, D3 and D7 respec- tively, starting from the left. From Table IV, it is found that both the clear chill and the total chill increase as the carbon content is lowered. But neither the clear chill nor the total chill changes uni- formly as the carbon content is changed from 3.79% in heat D1 to 2.75% in heat D7. Comparing between heats D6 and D7, it is found that the clear chill in heat D6 (which was superheated to 29000F) is less than that in heat D7 (superheated to 28000F), though both the heats were nearly of the same composition. This is surprising, because as it is commonly believed, superheating destroys potential graphite nuclei, so on chilling it should produce a greater depth of chill. But the observation here made is contrary to this belief. Chil TABLE IV 1 Test Data 39 Chill depth in 32nds of an inch Superheating Heat no. temperature Clear chill Total chill oF D1 1 1 2800 D2 9 14 " D3 10 14 " D4 11 26 " D5 14 26 " D7 17 40 " D6 15 40 2900 Fig. 19. Chilling tendency from left to right. Heats D1, D3 and D7. Natural size photograph. 40 41 Part F. Cooling Curve Data The cooling curves are not reproduced here. Their general shape is similar to those of McGrady et a1. (8). The eutectic start temperature is taken as the lowest tem- perature recorded in the trough region of the time- temperature curve. The eutectic start temperatures for the different heats are shown in Table V together with %C and the superheating temperature for the various heats. For heat D1, the average readings of two thermocouples are recorded in the table. From Table V, it is indicated that when the superheating temperature is kept constant, the eutectic start temperature is lowered with lowering of carbon content in the melts. But this lowering of the eutectic start temperature does not seem to be gradual. Comparing between heats D6 and D7, it is found that super- heating the melt leads to a lowering in the eutectic start temperature. To interpret properly the data on the eutectic start tem- perature, it is necessary to know the equilibrium eutectic start temperature as a function of carbon and silicon content from the ternary Fe-C-Si diagram. For the moment, if it is assumed that silicon in the experimental heats is constant, the 42 effect of variation in carbon content of the melts on the eutectic start temperature can be found out from the constant silicon sections of the Fe-C—Si ternary diagram. Taking the 2% silicon section (the section nearest to the average sili- con content of the experimental heats) of the ternary diagram, it is found that there is very little lowering of the equili- brium eutectic start temperature when the carbon content of the melts is lowered from 3.79% in heat D1 to 2.75% in heat D7(LD. But from Table V, it is found that the eutectic start temperature is lowered with a lowering of carbon content in the melt. Obviously, all of this lowering in the eutectic start temperature can not be justified by pointing to the change in carbon content in the experimental heats. So, it seems fair to conclude that the undercooling in the eutectic reaction increases as the carbon content of the melt is lowered, when the superheating temperature is kept constant. TABLE V Cooling Curve Data Superheating Eutectic start Heat no. % C temperature temperature or 0F D1 3.79 2800 2026 D2 3.75 " 2027 D3 3.56 " 2012 D4 3.29 " 2022 D5 2.91 " 2015 D7 2.75 " 1990 D6 2.68 2900 1946 44 Part G. Effect of Superheating To study the effect of superheating on the degree of nucleation of the eutectic, two heats were made with identical charges. One of the heats (D6) was superheated to 29000F and the other (D7) to 28000F and both the heats were poured into molds under identical external conditions. Quenched and slow cooled wedges of heat D6 were examined under the microscope and photomicrographs were taken at 100X. Fig. 3 shows the macrographs of heat D6 to the left and of heat D7 to the right for comparison. From the macrographs, it is evident that the specimen of heat D6 contains larger number of eutectic cells than that of heat D7. Fig. 20 shows the microstructure of the specimen of heat D6 quenched at the start of the eutectic solidification. Fig. 21 shows the microstructure at the surface of the slow cooled wedge of heat D6. Comparing Fig. 20 with Fig. 9 (heat D7, quenched), it is found that the graphite in the eutectic cell of heat D6 (Fig. 20) is much coarser and less interdendritic than that in heat D7 (Fig. 9). Comparison of Fig. 21 with Fig. 15 (heat D7, slow cooled, at surface) does not show any marked difference in the graphite distribution near the surface, but it is still evident that graphite is coarser in Fig. 21 than in Fig. 15. 45 It was pointed out before that the depth of clear chill in heat D6 was less than in heat D7. According to popular belief, superheating the melt pro— motes the formation of undercooled graphite (type D or E) with lowering in the degree of nucleation of the eutectic freezing, due to loss of potential graphite nuclei on heating to higher temperature. Accordingly, superheating should also lead to a greater depth of chill. But in this study, it is found that superheating: a. produces a greater number of eutectic cells (Fig. 3), b. does not promote the formation of undercooled graphite, but to some extent it reduces such tendency (comparison between Figs. 20 and 9), c. reduces the depth of clear chill (Table IV). On the other hand, it is noted from Table V that super— heating reduces the eutectic start temperature, which might be explained on the basis of the loss of potential graphite nuclei. It seems then that the effects of superheating as observed in this study are contrary to the popular belief in some of the aspects of its effects. It is felt that no final con— clusion should be drawn from this study of one set of experi- mental heats. Fig. 20. Heat D6. Quenched. Nital etch. Fig. 21. Heat D6. Slow cooled. Nital etch. lOOX. lOOX. At surface. 46 47 V DISCUSSION In this investigation, it was decided to keep the silicon, manganese, phosphorus and sulfur contents of the experimental heats at constant level and to vary only the carbon content of the melts. Since variable amounts of charge constituents were used, it was found that the amounts of these elements picked up in the melt were not uniform and the pick up did not follow any consistent trend. This point will be made evident by noting that the same charge was used for heats D1 and D2, but the final composition in the two heats was not at all the same, though all external factors were apparently kept identical. It is felt that due to the small size of the heats (31 lbs), it was so difficult to obtain constant amounts of Si, Mn, P and S in the experimental heats while using different amounts of charge constituents. However, the ranges of variation of Si, Mn, P and S contents in the various heats do not seem to be of much significance. Since the findings of this study are of a qualitative nature, the assumption of constant Si, Mn, P and S contents in the experimental heats appears to be justified. From an examination of the macro- and micro-structures of the specimens quenched at the start of eutectic freezing, it is evident that: 48 a. there is a sharp decrease in the number of eutectic cells in heat D4 as compared to heat D3; b. in heats D1, D2 and D3, the eutectic cells contain randomly oriented type A graphite; c. in heats D4, D5 and D7, the cells contain interdendritic type D graphite. Examination of the microstructures of the slow cooled specimens also shows that the microstructure at the surface changes with the change in carbon content of the irons. But this change is very much marked between heats D3 and D4. In the former, the surface structure is of random type A graphite, while in the latter it is of interdendritic type D graphite. Thus it is found that the abruptness of the change in the degree of nucleation of eutectic between heats D3 and D4 is confirmed by the microstructures of the quenched and the slow cooled specimens. There appears to be a sharp transition in the mode of nucleation of eutectic as the carbon content in the melt is lowered from 3.56% in heat D3 to 3.29% in heat D4. The data on the chilling tendency (Table IV) show that both the clear chill and the total chill increase as the car- bon content of the iron is lowered. However, there are sudden increases in the total chill between heats D3 and D4 and between D5 and D7. The clear chill increases as the carbon content is lowered from 3.75% in heat D2 to 2.75% in heat D7, 49 at constant superheating temperature. It is not clear why there should be such a large difference in chilling tendencies between heats D1 and D2. The abruptness of the change in the degree of nucleation of eutectic between heats D3 and D4 is not reflected in the clear chill data. The total chill however shows a sudden increase in heat D4 over that in heat D3. But it also shows another sudden increase in heat D7 over that in heat D5 with- out any apparent reason. From the cooling curve data (Table V), it is evident that the higher the carbon content of the melt, the higher is the eutectic start temperature. Since the eutectic freezing in irons is initiated by graphite nuclei (12), a higher eutectic start temperature in higher carbon irons indicates that there might be more graphite nuclei present in them. An examination of the macrographs of specimens quenched at the start of eutectic freezing (Figs. 1-3) showed that higher carbon irons (heats D1, D2 and D3) gave rise to a larger number of eutectic cells as compared to lower carbon irons (heats D4, D5 and D7). It was pointed out before that there was a sharp reduction in the number of eutectic cells in heat D4 over those in heat D3. From Figs. 1-3, it is apparent that the nucleation characteristics of heats D1 to D3 are quite different from those of the other heats. In the 50 heats D1 to D3, the eutectic solidification seems to be easily nucleated, while in the other heats it is not so. In the first group, the eutectic freezes to form random graphite, while in the latter, it produces interdendritic graphite. Considering the general trend of the cooling curve data IA...»- h and the observations made from the macro- and micro-structures of the quenched specimens, it seems reasonable to conclude that up to a critical value, the higher the carbon content in the irons, the greater are the number of potential graphite , nuclei in the melt. During the freezing of the melt, more graphite nuclei will be available to higher carbon irons so that there will be less undercooling. The eutectic reaction will begin at comparatively higher temperatures at a large number of points all of which will grow slowly to produce random type A graphite. On the other hand, a melt containing less than the critical amount of carbon will lack graphite nuclei as it approaches the eutectic temperature. So, the melt will need greater undercooling to initiate the eutectic reaction. However, once the eutectic freezing starts at a few points, the cells will grow at a high rate producing type D graphite. The above sequences of freezing in the two types of irons are in accordance with the concept advanced by Boyles (11). It is apparent from this study that the critical carbon content, as defined above, lies somewhere between 3.56% 51 to 3.29% when the silicon content of the melt is of the order of 2.2%» When a melt contains carbon greater than this critical value, the eutectic freezing will resemble those of heats D1 to D3. But if the melt contains less carbon than the critical, the freezing sequence will correspond to those of heats D4 to D7. It is suggested here that inoculation by graphite produces localized regions in the melt containing carbon greater than the critical amount (as defined above). Eutectic freezing to produce random graphite starts in these regions. It is quite probable that in irons containing carbon greater than the critical amount, the distribution of carbon in the melt may be of a critical nature at temperatures close to the eutectic. It is conceivable that this critical distribution of carbon may be attained by the addition of other selected materials, some of which being presently known inoculating agents. This critical distribution of carbon may correspond to the colloid- ally dispersed carbon or graphite particles or to the localized high carbon regions in the melt formed due to segregation or chance fluctuation of carbon. The effect of superheating the melt on the degree of nucleation of eutectic and other related phenomena has already been discussed. It was pointed out before that some of the experimental observations of this study are contrary to the 52 commonly held belief that superheating destroys potential graphite nuclei in the melt. It was found that superheating actually increases the degree of nucleation of eutectic, produces more normal (less interdendritic) graphite structure and decreases the depth of clear chill. All these were unexpected. It was also found that superheating reduces the eutectic start temperature. Since there is only one set of data to show the effect of superheating, it was felt that no final conclusion should be drawn at this stage. The present study suggests the need to carry out investi- gations to confirm the existence of a critical carbon level at which there is a sharp transition in the mode of formation of graphite. It will also be of interest to study how this critical carbon level varies with variation in composition of the melt, the degree of superheating and the rate of cooling in the mold. 53 VI CONCLUSION The following conclusions can be drawn from the present study: 1. There is a sharp decrease in the number of eutectic cells as the carbon content cflfthe melt is lowered from 3.56% to 3.29% at an average silicon content of 2.2%. 2. The microstructures of the eutectic cells change sharply from randomly oriented type A graphite to interdendritic type D graphite when the carbon content is lowered from 3.56% to 3.29%. 3. The microstructures at the surface of the slow cooled specimens change with a change in carbon content of the irons, but the change in microstructure is very marked when the carbon is lowered from 3.56% to 3.29%. 4. The eutectic temperature is lowered with lowering of carbon content of the melt. 5. Chill depths increase with decrease in carbon content of the irons, but the sudden increase in the degree of nucleation when the carbon is lowered from 3.56% to 3.29% is not reflected in the chill data. 6. Superheating the melt tends to increase the degree of nucleation of eutectic, produces more normal (less interdendritic) 54 graphite structure and decreases the depth of clear chill. It also lowers the eutectic start temperature. The results of this study seem to lend support to the idea of the presence of carbon in inoculated melts, in a state of critical dispersion or distribution at temperatures close to the eutectic. 10. 55 VII BIBLIOGRAPHY W. Hume-Rothery and G. V. Raynor, "The Structure of Metals and Alloys,” Institute of Metals, London, 1956, p. 287. J. T. Eash, "Effect of Ladle Inoculation on the Solidi— fication of Gray Cast Iron," AFA Transactions, Vol. 49, 1942, p. 887. H. Morrogh and w. J. Williams, "Graphite Formation in Cast Irons and in Nickel—Carbon and Cobalt—Carbon Alloys," Journal of the Iron and Steel Institute, Vol. 155, 1947, p. 321. H. Morrogh and W. J. Williams, "Undercooled Graphite in Cast Irons and Related Alloys," Journal of the Iron and Steel Institute, Vol. 176, 1954, p. 375. H. Morrogh, Journal of Research and Development, British Cast Iron Research Association, Vol. 5, 1955, p. 665. A. Hultren, Y. Lindblom and E. Rudberg, "Eutectic Soli- dification in Grey, White and Mottled Hypo-Eutectic Cast Irons," Journal of the Iron and Steel Institute, Vol. 176, 1954, p. 365. N. C. McClure, A. U. Khan, D. D. McGrady and H. L. womochel, "Inoculation of Grey Cast Iron," AFS Trans- actions, Vol. 65, 1957, p. 340. D. D. McGrady, C. L. Langenberg, D. J. Harvey and H. L. Womochel, "Hypoeutectic Gray Cast Iron Ladle Additions," AFS Transactions, Vol. 68, 1960, p. 569. C. L. Langenberg, "The Effect of Late Addition of Graphite on the Properties and Microstructure of Induction Furnace, High Strength Cast Iron," Thesis, Michigan State Univer- sity, 1957. I. Igarashi, G. Ohira, K. Ikawa and T. Horigoma, "Forma— tion of Undercooled Graphite in Cast Iron," AFS Transactions, Vol. 66, 1958, p. 561. l. gfi‘v-x“ '3 mA_—.' . ll. 12. A. Boyles, "The Structure of Cast Iron,' Society for Metals, 1947. A. DeSy, Grey Iron, "Solidification Mechanisms of Eutectic and AFS Transactions, Vol. 67, 1959, p. American 486. 56 . A; Aj‘knr L I .1 f? are C’ . A. ‘ 1,0 Pg! " AUG: $5.22!, 1? "2; ‘