EFFECT OF LADLE ADDITIONS OF SOME
ALLOYS AND ACTIVE METALS ON THE
PROPERTIES AND
MECROSTRUC’TURE OF GRAY CAST IRON

Thai: for The Dogm of Ph. D.
MICHIGAN STATE COLLEGE
Howard Leroy Womochal
1954

Lt

0-169

This is to certify that the

thesis entitled

EFFECT OF LADLE ADDITIONS OF SOME ALLOYS
AND ACTIVE METALS ON THE PROPFRTIES AND
MICROSTRUCTURE 0F GRAY CAST IRON

presented by

HOWARD LEROY WOMOCE {EL

has been accepted towards fulfillment
of the requirements for

Ph.D. degree in Metallurgi cal
anineering

   

’ Major professor

Date 9 November lSSh

EFFECT OF LADLE ADDITIONS OF SOME ALLOYS
AND ACTIVE METALS ON THE PROPERTIES
AND MICROSTRUCTURE OF GRAY CAST

IRON

BY

HOWARD LEROY _W_0_MOCHEL

A THESIS

SUBMITTED To THE SCHOOL OF GRADUATE STUDIES OF ~~
MICHIGAN STATE COLLEGE OF AGRICULTURE AND
APPLIED SCIENCE IN PARTIAL FULFILLMENT
OF THE REQUIREMENTS FOR THE DEGREE
GP

DOCTOR OF PHILOSOPHY

DEPART ENT OF METALLURGICAL ENGINEERING

195“

q-:o-SL

ACKNOWLEDGEMENT

The author wishes to express
his appreciation for the aid and en-
couragement extended by the members
of the Metallurgical and Mechanical
Engineering Departments during the
course of this investigation. In
particular he acknowledges his in-
debtedness to Dr. A. J. Smith,

Mr. D. J. Harvey, Dr. C. C. DeWitt,
and Mr. D. G. Triponi.

I.

II.
III.

IV.
V.
VI.

TABLE OF CONTENTS

Introduction

Solidification of Gray Cast Iron
Factors Influencing Graphite Distribution
Inoculation of Gray Iron

Theories of Inoculation

Objective of Research

Experimental Proceedure
Experimental Results

Ferro-silicon and Calcium-silicon
Calcium Silicon and SMZ
Silico-manganese

High and Low Active Metal Ferro-silicons
Silicon as an Inoculant

Calcium as a Ladle Addition
Aluminum Additions

Effect of Acid and Basic Conditions
Magnesium as a Ladle Addition
Graphite Additions

Effect of Inoculants on Cell Size
Decarburization by Calcium
Discussion

Summary and Conclusions

Bibliography

Page

10
16
26
28

3'5

35
MI

'47

1+9
52
55
58
so
61+
66
7O
72
83
86

I. Introduction
Graphite Distribution in Cast Iron.

The properties of gray cast iron are determined by:

l) the character of the microstructural matrix and 2) the
amount and manner of distribution of the graphite flakes
in the matrix.

The matrix of an unalloyed high strength gray iron of
properly balanced silicon.and carbon contents may approach
one hundred percent pearlite. Under certain conditions un-
desireable ferrite appears in the microstructure. This
ferrite is generally associated with an unfavorable graphite
distribution, see Fig 16. The combination of ferrite and
graphite distribution of the type illustrated is detrimental
to all of the properties of cast iron.

Fig l is the Anerican.Foundrynen's Society- American
Society for Testing Materials Chart for graphite distribution
in gray iron. This chart illustrates the various possible
graphite distributions in cast iron of a wide range of carbon
contents. High strength hypoeutectic iron (carbon range 2.60-
3.10%) may exhibit types A, B, D or E of graphite distribution.

Type A with a random distribution of the graphite is
the preferred arrangement. The pattern arrangement of type
D is undersireable because of the resultant planes of weak-
ness in the iron. There is reason to believe that types
B and E are less serious manifestations of the conditions

causing the type D arrangement. Type D graphite is vari-

  
 

Fig. lZ—Repmduction of A.F.A.

Graphite Distribution Classifica-

tion Chart (Cast Metals Hand-
’ book").

 

 

Fig 1. American Foundry-en's Society Chart for Graphite
Distribution in Cast Iron. ’

eusly referred to in the literature as: abnormal, lacy,
dendritic, interdendritic, undercooled, reticular, eutecti-
forn, refined, and modified. Type B graphite distribution
is generally described as the rosette or whorl type. In
this report the undesireable kinds of graphite distribution
will be referred to as 'abnorml' with the understanding
that abnormal structures contain types B, D and 3. generally
with a preponderance of D. This terminology is Justified by
the fact that lost eo-erical irons of carbon of less than
3.10% my contain varying amounts of the four graphite types
in the same casting. The principle concern of the metal-
lurgist producing high. strength iron is to obtain a minus:
of normal or A graphite.

There are many factors which have an influence on the
graphite distribution in cast iron. As an introduction to
this paper it will be necessary to review the literature
concerning these factors, and also the subject of the solidi-
fication of gray iron. These matters have been discussed ex-
tensively in the literature and are generally understood by
metallurgists concerned with the production of cast iron.
However, the literature is so voluminous and in new cases
so conflicting that it appears advisable to review these
subJeots as briefly as possible for the convenience of the

reader and for the purpose of relating the data presented
in this paper.

Solidification of Gray Cast Iron.

As a simplification, cast iron is regarded as a
ternary alloy of Fe-C-Si, and is considered to solidify
and transform in accordance with the diagram of Fig 2
when the silicon content is in the vicinity Of 2%, as it
is in the case of the majority of commercially produced
irons. Fig 2 represents a vertical section through the
ternary solid at 2% silicon. Two forms of this diagram
can be recognized: the stable, and the metastable. In the
metastable diagram the carbon rich phase is cementite,
Fe3C, labeled in.Fig 2 "Ca". In the corresponding stable
diagram carbon is present as graphite. It is understood
that Fe3C has a higher free energy than the equivalent
Fe and C, and that the true equilibrium diagram is the
stable diagram.

White cast iron.solidifies and transforms in accordance
with the metastable diagram. Whether or not gray cast iron
solidifies in the stable or metastable manner was not deter-

l in

mined prior to the publication of a paper by A Boyles
1937.

In the experiments described in this paper Boyles
determined the manner of solidification of hypoeutectic irons
by arresting transformation in small melts by quenching
from temperatures between the liquidus and solidus. Photo-
micrographs from his paper are presented in Fig 3. These

structures make clear the solidification of a hypoeutectic

3A

¢r+€h

 

‘ Carbon,per
FIG. Pinon-Canon Dueun IN Pennies or 8.00 Pas Cam- Stuoon (GIIINII, Mum
we Summon).

Fig 2. Section through Iron-Carbon-Silicon Ternary Diagram
at 2.00% Silicon.

 

 

Fig 3. Microstructures Obtained by Quenching Partially
Solidified Hypoeutectic Iron. Heat Tinted. Upper Picture
x 20. Lower x 100. After Boyles.

iron. In the low ngnification micrograph at the top it
is apparent that the iron solidifies by the formation of
primary dendrites of austenite, and that the eutectic

solidifies in the remaining liquid (represented by the
light etching background) fro. centers of crystallization
I to form spherical masses'which upon completion of solidi-
fication constitute the 'cells' of the cast iron micro-
structure. Examination of these growing cells at higher
magnification (lower picture) reveals that the graphite
flakes form during the growth of the cells, and while the

iron is still partially liquid. It also appears that flake
growth takes place at the liquid-cell interface, a manner

of development which, by inhibiting sidewise growth, pro-
duces flakes rather than the thermodynamically more stable
nodular or temper carbon shape of graphite. All graphite
in the iron is seen to be located between the primary
dendrites.

This mechanism suggested an attractive explanation
for the variation-in size and distribution of flakes-
namely variations of~ in the relative rates of growth of
the austenite and graphite. For enmple a high rate of
austenite formation in the growing cell might produce
small (type D) flakes by ”pinching off" the developing
flake, . or a high availability of graphite for flake

growth and a high nucleation rate at the interface might
produce the same results.

This conception was modified for certain irons by
the publication of a paper by J. T. Eashz in which evidence
was presented in the form of quenching experiments to
show that irons highly abnormal in character solidified
in the metastable manner and that graphite of the D type
formed in the solid iron by the decomposition of cementite
at a temperature close to but below the solidus.

Papers confirming the findings of Eash have been pub-
lished by Morrogh and Williams3 and more recently by Owens
and Street.l+ It is believed by these investigators that
the metastable solidification leading to the formation of
type D graphite is associated with an "undercooling" of the
liquid iron. As a consequence type D graphite is frequent-
ly referred to as "undercooled graphite".

Although the possibility of the solidification of type
D irons in the metastable manner by undercooling cannot be
disputed, there are certain discrepancies in this theory
which require explanation and which will be discussed later

in this report in connection with other data.

Factors Influencing Graphite Distribution.

The following factors may be listed as having an im-
portant influence on graphite distribution:

1. Carbon content of the iron

2. Superheating temperature

3. Holding time at temperature

A. Method of melting

5. Pouring temperature

6. Sulphur and manganese/contents

'7. Gas content - hydrogen, oxygen, nitrogen
8. Cooling rate in the mold and extent of undercooling

9. Degree of inoculation

It is generally understood that high strength irons
with carbons less than 3.10% are much more susceptible to
the deveIOpment of type D graphite than are the higher
carbon irons. O

There is evidence to support the conclusion that
high superheating temperatures and long holding time at
temperature in the furnace promote the formation of ab-
normal structures in some irons. Sefing and Searlss,
Schneble and 0111333216 and Krynitshy and Saegcr" have pub-
lished papers leading to this conclusion. '

The opinion is frequently expressed that electric
furnace irons are more susceptible to D graphite fomtion
than are cupola irons. Direct experimental evidence to
support this contention is not to be found in the literature.

linens and Crosby8 have investigated the effect of
pouring temperature on the properties of high strength
irons and have concluded that pouring temperatures below
2650°1have an adverse effect on the properties of irons
of less than 3.00%b and that low pouring temperatures

'tend toward the dendritic, whorl, and eutectiform types
of distribution“.

Boylesl’9

has investigated the effect of sulphur
and manganese additions to iron-carbon-silicon alloys
very low in their sulphur and manganese contents. It
was found that in the absence of appreciable S that

the iron solidified in a highly abnormal condition with
pronounced type D graphite and a matrix almost entirely
ferritic. An addition of .018% of S or more altered the
structure to one containing a large amount of type A
graphite and pearlite. Boyles explained this effect as
an influence on carbide stability resulting in varia-
tions in the rate of nucleation and availability of
carbon during the simultaneous solidification of the
eutectic cell and the graphite flakes as postulated in
the above discussion of solidification. Manganese was
found to have an important but variable effect on gra-
phite distribution dependant on conditions.

The influence of gas content on the microstructure
of iron has been the subject of numerous investigations
and is a highly controversial matter. The subject is
rendered difficult for research by the problem of ana-
lytical determination of the gaseous elements. The
situation is particularly difficult in the case of
oxygen because of the reactions occuring between the
melt and the oxides of the furnace and ladle linings.

To review the literature dealing with the subject
of gases in ferrous alloys and of oxygen in particular
is beyond the scope of this paper. The bibliography

of the British Cast Iron Research Association lists more
than a thousand references. However, a few of the con-

clusions pertinent to this paper will be cited from the

recent literature.

Boylesl’9 has determined the influence of hydrogen
by melting in vacuo and in atmospheres of varying H con-
tent. He concluded that a certain H content is essential
to the formation of a normal structure in cast iron, and
that excessive amounts tend to promote unfavorable dis-
tributions. The influence of H was attributed to an
effect on the rate of graphitization.

The subject of oxygen in cast iron has been treated
recently in articles by Heinel'o’ll and Bach,12 and Loria
and Lewnie.13 From a study of these and other articles to
be discussed later the following conclusions may be ad-
vanced that are pertinent to this paper:

1) Above approximately 2600°F the 0 content of cast
iron is governed by the carbon content; below unis temper-
ature by the Si and Mn contents.10

2) The 0 content of cast iron is small being approxi-
mately .0025% in irons of about 2% silicon.13

3) Although the 0 content is small it is significant,
and changes of a temporary nature in the melt can be effect-
ed by the addition of active metals such as aluminum."

h) Thermodynamic equilibria are inapplicable because
cast iron.cannot be in equilibrium with iron oxide or

silicate slags except at high pressures.

Foundrymen have long believed that iron can be
"oxidized" and that oxygen increases chilling tendency.
Experiments described in the literature both confirm
and Oppose this conception. This matter will be dis-
cussed at greater length later in the report.

Nitrogen in cast iron has been discussed by

Dawson, Smith, and Bach.1h

It is concluded that N

has a strong stabilizing effect on massive cementite
and pearlite; that it affects the graphite flake size
and shape, the flakes becoming shorter and thicker with
increasing N content thus accomplishing an improvement
in mechanical properties; and that aluminum additions
reduce the effects of nitrogen presumably by the for-
mation of a nitride.

The situation regarding gases in cast iron can be
summarized by quoting Loria and Lownie.l3 "In spite
of this voluminous literature interpretation of the
data is still hazy even for steel......For cast iron
the situation is even.more nebulous."

D'Amico and Schneidewindls have determined the
effector cooling rate in the mold on solidification
temperature (extent of undercooling) and on micro-
structure by casting irons in a metal mold preheated
to various temperatures to produce various cooling
rates. It was concluded that a variation of the

solidification temperature approaching lOO°F could

10

be produced, and that as the degree of undercooling in-
creased the degree of abnormality increased so that the
severely undercooled irons exhibited type D graphite.

This conclusion is in accord with Eash2 and others3’
regarding the effect of undercooling on graphite distri-
bution.

It is evident from this discussion that the lower
carbon high strength irons are extremely sensitive alloys,
and that they are likely to develop type D graphite as
the result of the operation of a considerable number of
factors. In the ordinary course of events an electric
furnace cast iron of less than 3% carbon will show ab-
normal structures particularly near the surface of the
castings unless the iron is inoculated. The principle
means available to the metallurgist for controlling the
microstructure of these irons is the inoculation pro-
cess. The purpose of the research described in this

paper is to investigate the inoculation treatment.

The Inoculation of Gray Iron.

In the general sense in which the word is em-
ployed in physics, inoculation may be defined as the
addition of substances to melts for the purpose of
forming nuclei for crystallization.

Since it is by no means certain that the effect
of inoculation of iron is explainable solely on the

basis of nucleation effects, it has appeared advisa-

ble to use a broader definition for the term, and the
following has appeared in the literature.

Inoculation is the process in which an addition
is made late in the heat of molten cast iron or in the
ladle for the purpose of improving the microstructure
and properties of the iron to a degree not explainable
on the basis of the change in composition.

As an illustration, an iron of 2.80% and 2.25% Si
when melted to this composition with no late additions
will have markedly different properties from an iron
produced by withholding a portion of the silicon from
the cold charge and making up the deficiency with an
addition of silicon to the ladle in the form of one of
several commercially produced and complex alloys of
silicon.

The value of late treatments has been appreciated
and the process employed for a good many years, coming
into general use in the late nineteen-twenties. A
number of alloys for inoculation, notably calcium
silicide, have been patented. The use of calcium-
silicide constitutes the basis for the well known
Meehanite process, the basic patent for which has
expired.

18
Table I, taken from an article by Lownie, lists most

of the alloys employed as inoculants.
Table I

Classification of Typical Ladle Inoculants

Group I Group II
Ca-Metal Si C Cr-Si-Mn-Ti-Ca
Ca-Si Si-Mn Cr-Si-Mn-Zr
Ca-Si-Ti Si-Mn—Zr Mo-Si
Fe-Si Si-Ti Ni-Si
Graphite Si-Zr

Certain alloys containing relatively large amounts
of aluminum should be add ed to this list, such as an im-
portant prOprietary alloy of Si-Al-Ti-Fe. The above
listings are of the principal constituents of the alloys,
many of the alloys containing relatively small amounts of
active metals particularly Al, Ti and Ca.

A significant thing about this list is that with two
exceptions, all of the alloys contain silicon, and silicon
is commonly regarded as the essential constituents of ino-
culants. Most of the theories advanced to explain inocul-
ation effects are based on the action of silicon. A large
proportion of the papers dealing with inoculation have em-
ployed ferro-silicon as the addition agent, and in many of
these cases the composition of the alloy employed has not

been specified, indicating that the essential constituent

13

of the alloy is considered to be the silicon.

The literature concerning the use of these alloys
is voluminous, particularly that dealing with the practi-
cal aspects of inoculation. In the bibliography pre-
sented here no attempt has been made for a complete pre-
sentation. The articles selected for discussion are either
typical or are particularly pertinent to the experiments
described in this report.

The writer has been unable to find any discussion
of the use of calcium metal as an inoculant in the re-
cent literature.

Many references have been made to the use of graphite
as a ladle addition. The consensus of Opinion is that the
addition of graphite to the ladle decreases the chilling
tendency and improves the properties somewhat. This con-
clusion is supported by very little experimental data.
Dahlberg16 has recommended the use of 8 mesh graphite as
being most suited for ladle additions. The Opinion is
prevalent that the effects of graphite additions are soon
lost by holding in the ladle, presumably by complete so-
lution of dispersed graphite, and the suggestion is made
that the addition be made immediately before pouring into
the mold.

Ferro-silicon has been used in many of the investi-
gations of inoculation reported in the literature. The
alloy is generally employed in the grades ranging from

75% Si to 95% Si. Ferro-silicon is available in low

and high active metal contents (Al and Ca in particular)
but very few if any of the writers have specified the
exact composition or even the grade employed, the feel-
ing evidently being that silicon is the essential consti-
tuent in the inoculating alloy.

Crosbyl7 has experimented with inoculation with in-
creasing amounts of silicon, presumably in the form of
ferro-silicon,as the alloy used is not specified. He
reports an improvement in properties up to an addition
of approximately 60% of the total silicon content of
the iron (3.05% C, 2.38% Si). Beyond this proportion
of silicon as a late addition the properties fell off
rapidly. Graphical representation of the data from this
paper summarizes the effect of inoculants on the mechani-
cal properties in general and Fig.k presents this in,
formation. It is observed that the strength properties
are improved while there is some reduction in hardness.
The effect of inoculation on.chill depth is not represent-
ed here, but it is generally understood that the use of
inoculants effects a marked reduction in chilling tenden-
cy, and this reduction is one of the important justifications
for ladle treatments. See Fig 1%. In addition it will be
understood by the reader that the improvement in properties
resulting from increasing the inoculation results from a
change in the microstructure from type D to E and finally
to A.

__._ _— _ __.IL._ J—A h u. .. -—..
> k_‘,_.. —

 
  
   

 

 

      
 

 

  
 
 

  

      

   

 
  
 

   
      
    
 
     

 

 

 

4m 3200 0.360
'3".
. 3” 0.330
E
gm i 2000 i. 0.300
t; g I"
3 mo 2 0.270
i E a
24m to 0.240 °
3 a
a a 0.2m
;
5 3. also
0 25 50 75
PER CENT SILICON Am LATE
5'10. l7—Erner or Len Anemone or Smoou ON PHYSICAL Paonmas or CAST home Snow in
Tam 12.
Table I2
Eraser or LATE SILICON Anni-nous 'ro CAST IRON
Per Cent Silicon
Added as Late
Addition 0 25 50 75 100
Total Carbon,
per cent ...... 3.07-3.09 3.10-3.11 3.04-3.04 3.04-3.06 3.08-3.10
CombinedCar-
bon, per cent. . 0.58 0.61 0.63 . 0.67 0.64
Graphitic Car-
bon, per cent. . 2.50 2.50 2.41 2.38 2.45
Manganese, .
per cent ...... 1 0.88 0.90 0.89 0.89 0.88
Silicon, per cent,
Original ...... 2.17 1.50 1.00 0.50 . . .
Late ......... . . . 0.58 1.15 1.57 2.13
Total ....... 2.17 2.08 2.15 2.07 2.13
Transverse
Strength, lb . ‘ 2266-2332 2420-2530 2840-2895 2990-2860 2442-2510
Dcflection,in..... 0.187-0.l94 0.233-0.250 0.314—0.335 0.350-0.320 0.245-0r250
Brinell Hardness.. 228-228 223-223 217-217 217-217 217-217
Tensile Strength, 36,650- 42,000- 47,600- 48,200- 39,400-
pei ........... 38,650 43,120 42,500 48,600 39,550

 

 

 

 

 

 

 

 

 

Fig 1+. Effect of late Silicon Additions. After Crosby.

17

lhA

It should be mentioned in connection with chilling
tendency that inoculants are of two types; the graphiti-
sing type, represented by Group I of Table I; and the
”alloying element“ containing type, typified by the
members of Group II. Alloys of this second type may or
may not be graphitising depending on the nature of the
elements contained. lost of the inoculating agents are
designed to reduce chill or to permit the addition of
carbide stabilizing elements such as chromium, either
with a reduction in chilling tendency or with no in-
crease in chilling tendency. The experiments reported
in this paper are concerned only with agents of the
first group.

Another aspect of inoculation that deserves em-
phasis is that the effect of all inoculants is tempo-
rary, and will disappear gradually with holding time in
ladle or furnace. It is variously stated that the
duration of the m... is from s to so minutes. it no
doubt varies with the alloy used and with other cir-
cumstances. As mentioned above the effect of graphite
additions is particularly short- lived.

In reviewing the literature on inoculation the
reader is impressed by the lack of information dealing

with the relative effectiveness of the various ino-
culating agents. This is surprising in view of the

16

wide range of compositions of alloys recommended in the
literature. Flinn and Reese” have determined the re-
lative effectiveness of several inoculants containing
silicon and found them to be roughly equivalent except
for one alloy which was particularly low in silicon and
gave inferior strength. They state that all successful
inoculants contain silicon.

Theories of Inoculation .
In view of the importance of inoculation in a

practical way and of its theoretical implications con-
cerning the fundamental nature of cast iron, many attempts
have been made to develop a satisfactory explanation for
the effectiveness of late additions. Lewniela. Iorrogh
and Williams:5 and Creme19 have presented sumries and
discussions of the various theories. A brief review of
some of the theories will be presented here.

lost of the ideas have been based on a nucleation
mechanism as the name “inoculation" implies. It is
argued that the molten iron requires a certain nucleation
if the freezing is to occur in the stable manner producing
type A graphite. This conception was strengthened by the
publication of cooling curves indicating that irons with
type D graphite solidify at a lower temperature than irons
with normal structures. Curves showing this have been pub-

lished by D'Amico am Schneidewindls, lashz, and Boyles

17

20 as well as by others. Fig 5. presents curves

and Lorig
from the latter reference. The group of curves to the
left represent an iron inoculated with calcium silicide;
those to the right the same iron held at temperature for
a period Of time sufficient for the inoculation effect to
disappear. The fact that the inoculated iron solidifies
over a range of temperature, which is higher than that
of the other iron, is interpreted by Eashf Morrogh.and
Williams? and by others as indicating that the iron
which is not inoculated undercools, while the inoculated
iron is prevented from undercooling by nuclei provided
by the ladle treatment. These investigators have quench-
ed the "undercooled" irons from just below the supposed
solidification.temperature and found that the iron is
white. This is interpreted as evidence that the undercool-
ed iron solidifies in the metastable manner as previously
mentioned in this report. This conception is widely held
at present.

Having accepted the nucleation hypothesis, it has
been.necessary to designate the nucleating substance.
Two principle ideas have been advanced as to the nature
of the nuclei 1) the silicate slime theory originally
proposed by VonKeil21 and coworkers, and 2) the graphite

theory of Piw'owarsky.22’23

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

am ' I
[I'm No. I”

6‘ 5.06

2200 Si 2.29
Held Held
go elMimtes 52 Min/tea
\\ L- \\

2000

 

Minutes

Fig. 48—Cooling Curves Of Castings Made From Iron No. 133.
‘ Curves on left front castings poured 2 mmutcs after inoculation. Curves p
on right from castings poured 3.2 minutes after inoculation.

Fig 5. Time-Temperature Curves for Solidification of

Hypoeutectic Cast Iron. After Boyles and Lorig.20

17A

The silicate slime theory proposes that the addition
Of silicon alloys produces a colloidal dispersion of
ferrous or other silicates which nucleate solidification
and graphite formation thus promoting type A graphite.
This theory is supported by experiments of Diepschlagzn
who conducted melts under aluminous and silicate slags
and found that the former, supposedly by absorption of
silicate nuclei, produced finer graphite. Experiments
reported by Morrogh3, and one reported in this paper.
fail to confirm this finding. Inability to demonstrate
the presence of silicate inclusion in solid cast iron
is another serious Objection to this hypothesis.

One recognized tendency of abnormal iron is to
show a pronounced type D structure at the surface of the
casting or at any other point of rapid cooling in thin
sections. The 1.2" diameter transverse bar will show
type D at the surface and considerably type A graphite
at the center with the poorest structure at the immediate
surface if the iron is not completely inoculated and is
hypoeutectic. (See Fig 11 and Fig 12). Since the mold
wall is made up of silica and silicates and since its
rough surface should be ideal for nucleation, the
present writer feels that this is a powerful argument
Opposing both the silica slime theory, and the under-
cooling conception Of the formation of type D

19

structures.

The graphite nucleus theory is supported by the
argument that late graphite additions roduce chill
and promote coarse graphite flakes. However, there
is very little actual data in the literature to
support this contention; in fact some of the ex-
periments described in this report fail to confirm
this point. It has been argued that the addition
of silicon containing alloys promotes precipitation
of graphite in highly localized regions of the melt
by reducing solubility for carbon, and that this
solid graphite nucleates the solidification of the
melt and the formation of coarse random flakes. How-
ever, all available evidence indicates that the limit
of solubility 01 the melt for graphite is not approached.
This is shown by the fact that graphite additions dis-
solve rapidly in melts of low carbon irons. The persis-
tance of the effects of inoculation by silicon alloys
over periods of 30 minutes would indicate the unlikeli-
hood of the local concentration mechanism.

It has been pointed out in the literature that any
nucleatiom.hypothesis for cast iron must include three
distinct nucleation reactions 1) the nucleation of
primary dendrites 2) nucleationrof eutectic cells, and
3) nucleation of the graphite flakes. .Anyone or all
of these can cOnceivably influence graphite distri-
bution.and any complete nucleation theory should in-

volve all three processes.

It is the opinion of the writer that too little
attention is given to the previously mentioned mechanism

of formation proposed by Boyles9

who has argued that
small and large flakes can so formed as a result of
variations in the relative rates of fomtion of susten-
ite and graphite. This matter is deemed of sufficient
importance to justify an extended quotation from this
paper: ”an the other hand, the availability of carbon
may be limited by a slow rate or dissociation. The
growth Oi the flakes then lags behind the advancing
front of solidification. The flakes are pinched off,
so to speak, by the freezing of the metal and as a re-
sult small flakes are formed. Such cases are approach-
ing graphitization in the solid state and the fine
graphite so produced differs in appearance from that
formed by rapid graphitization. Between the two ex-
tremes of very rapid and very slow carbide dissociation,
a great range Of flake sizes may be produced witn the
same rate of cooling. Somewhere in this series, the
flakes will have a minim size.“

For the past several years it has seen customry
to minimise this conception of pattern flake formation

and to accept without reservation the undercooling,

meta-stable mechanism cf type 1) formation as proposed

by Bash? and previously discussed in this paper. A1-

21

though it is very difficult to disprove this under-
cooling theory, there are certain discrepancies and
unanswered questions in the mind of the writer con-
cerning this mechanism. Five of these are discussed
here:

1) The time-temperature curves presented by
Boyles (see Fig 5) are typical of many Of those found
elsewhere in the literature. Both the irons show a
pronounced dip just prior to the maximum.’ Elementary
text books Of physics refer to this dip as a result
of undercooling and state that the dip is followed by
a recalesence to the correct solidification temperature.
This being true it is hardly logical to refer to the
iron solidifying to give type D graphite as undercool-
ed as distinct from the other iron. Both irons ex-
hibit almost the same amount of undercooling.

2) Not all of the curves are similar to those of

Fig 5. D'Amico and Schneidewind15

have published the
curves of Fig 6, and Bash? those of Fig 7. It appears
from Fig 7 that no undercooling is taking place in
some Of the irons, in the usual sense of the word
"undercooling". The curve of Fig 6 most closely re-
presents solidification in accordance with the diagram
for the ternary alloy (Fig 2). Many investigators

appear to have overlooked the fact that ternary alloys

 

rEiEsairuas-wv '

_

I ' ‘7
.— ‘8 W
.

)-

' ORIGINAL MOLD TEME-2053°E_‘

 

_ieist'D-bis‘qbo
l

b
9
O

 

 

 

l l l l l i l l l

l l
mmmnw|maxmw2
nus- 5m_ __ ,, _, g . _

 

4 5.9

Fig 6. Time-Temperature Curve for Solidification of a 6
‘ 1
Hypoeutectic Cast Iron. After D'Amico and Schneidewind.

21B

 

Fig 7. Time-Temperature Curves for Solidification of
2
Hypoeutectic Irons. After Eash.

have complex time-temperature curves, and that in such alloys
liquid and two solid phases are in equilibrium over a range
of temperature supposedly represented by the three phased
field, Gamma + Liquid f Ca Of Fig 2. One is compelled to
wonder if the horizontals or maxima of some of these curves
do not represent an equilibrium between recalesence and heat
loss by the system rather than a non-variant equilibrium.
The diverse nature Of curves presented by the severalin-
vestigators raises some question about the extent Of the
range Of solid and liquid equilibrium.

This point appears to be particularly significant
to the writer in View of the fact that the principle support
for the undercooling theory rests on the microstructure Of
alloys supposedly quenched just below the "eutectic temperature",
evidently regarded as the solidus temperature by these investi-
gators. For example Morrogh and Williams3 in presenting evi-
dence to confirm Eashs' finding regarding metastable solidi-
fication of type D irons stated that the eutectic arrest for
a particular sample occured at 1133°C and that the sample was
quenched after solidification at ll32°c, a difficult achieve-
ment. Eash also refers to the "eutectic temperature" and
speaks Of quenching at a temperature 16° below the "eutectic

temperature" and finding the alloy completely white.

3) The photomicrographs presented by these in-
vestigators are particularly confusing. As an example
Fig 8 taken from the article by Morrogh and Williams
represents the microstructure of a sample quenched
just bglg! the eutectic temperature. This structure
is almost identical with structures presented by
Boyles as representing samples quenched just gbgzg
the eutectic temperature. ‘

Owens and Streeth in presenting evidence to
support the contention Of Eash, and Morrogh and
Williams, have published the photomicrograph of
Fig 8A. This structure was obtained by quenching
"partway through the eutectic arrest" and shows a
considerable area of type D graphite in the bottom
portion Of the picture immediately adjacent to the
inter dendritic accicular structure which Boyles
has taken in his work as representing liquid metal
at the moment of quenching. It appears to the
writer that this is evidence contrary to the thesis
of Eash, and of Owens and Morrogh, who have con-
tended that type D graphite develops after, the
alloy is solid and in a manner distinctfrom that
of type A graphite.

#) The most pronouncedly abnormal structures
with type D graphite occur at the extreme surface
of castings. This, in the light Of our under-

 

FIG. lI.—STRUCTURE 0F SLOWLY COOLEZ) INGOT

UENCHBD IMMEDIATELY ma SOLIDIFICATION.

; SHOWINO Tmmssomo Ausrsnma Dsmmas

. mu Wnrrs IRON Eumcnc or ACICULAa PAT-
rsaN. Brenna [N PICRlC ACID. x 60.

Fig 8. Microstructure of Iron Quenched from Below the

Solidus Temperature. According to Morrogh and Williams.3

 

. l-—Acicular carbide and undercooled
graphite. Etched in picrul ~-300

 

Fig 8A. Microstructure of Hypoeutectic Iron "Quenched Part-
way Through the Eutectic Arrest." According to Owens and
Street.

21+

standing of undercooling, should be the region most
difficult to undercool because of the Opportunity for
nucleation on the mold surface, a surface ideally con-
stituted for the purpose. It would appear logical tO
attribute the abnormal structure to the rate Of cool-
ing in the manner proposed by Boyles rather than di-
rectly to undercooling and metastable solidification.

5) Meny'microstructures Of low carbon irons show
type A and type D graphite structures immediately ad-
jacent to one another. It is very difficult to re-
concile this occurance with the undercooling theory.
Since the fine graphite adjacent to type A flakes
appears to originate at the center of the cells, it
should be possible to find cells with a white interior
and a gray periphery in the microstructure Of irons
quenched during solidification. TO our knowledge no
such peculiar distribution of the cementite and
graphite has appeared or been described in the litera-
ture.

The writer has embarked on this somewhat lengthy
discussion Of the shortcomings of the undercooling-
metastable theory because some of the observations
made during the experimental work Of this thesis
suggest another explanation for the displacement of
the time temperature curves Observed by these in-

vestigators. This matter will be discussed later.

Many attempts have been made to formulate an
acceptable "gas theory" Of inoculation. That gases
have an important effect on microstructure and pro-
perties has been discussed in this report. Hydrogen,
nitrogen, and oxygen appear to be effective in chang-
ing properties and microstructure of cast irons. It
is argued that the effects Of superheating are a re-
sult Of increased solubility for these gases with in-
crease in temperature. The effects Of inoculants can
be attributed to deoxidation, inasmuch as the commer-
cial alloys all contain elements commonly regarded as
reducing agents or deoxidizers in the metallurgy Of
steel. The undue chilling Of abnormal iron in parti-
cular has been attributed tO oxidation, and the re-
duction of chill by inoculants as a matter Of de-
oxidation. It is very doubtful that cast iron con-
taining large amounts of carbon, silicon, and manganese
could long contain an excess above the normal amount Of
oxygen. Attention has already been called to this
matter by citing Loria and Lowniel3 who claim an
equilibrium value for oxygen in the presence Of 2%
silicon to be .0025%.

The problem Of formulating a theory for the
mechanism Of inoculation and its effect on graphite
distribution has been greatly complicated by the dis-

covery Of nodular iron, which is produced by a double

26

ladle treatment - generally an initial addition Of

an alloy containing magnesium or cerium followed by
treatment with ferrO-silicon. Morrogh and Williams26
have discussed the theory of this process at great
length. Many Of the conclusions in their paper which
relate to grey cast iron have been discussed in COD!
nection with the previous reference to another paper

by these authors.3

Objective Of Research

Two facts become apparent from the literature
survey presented here 1) There is a lack of informa-
tion concerning the relative effectiveness Of the
alloys recommended in the journals as inoculants
2) A great deal Of uncertainty exists as to the mecha-
nism or theory Of inoculation.

Morrogh and Williams26 have commented on this
lack of information and have listed forty-one "graphite
formation anomalies". This list includes the effect of
practically every element introduced to iron melts even
including the graphitizing action of silicon, a most
fundamental principle of ferrous metallurgy. Also
included in this list are all of the practical and
theoretical aspects Of inoculation, and of the solidi-

fication Of cast iron.

The writer has long been of the opinion that much
of this uncertainty has resulted from failure to determine
the relative effectiveness of inoculating substances in
a careful and systematic manner. It is believed that
such a study should provide information of practical
value by showing the best alloys for inoculation, and
that the performance of the various alloys, should
they vary in the degree of effectiveness, should pro-
vide valuable clues as to the mechanism of inocula-
tion and of the solidification of various types of
gray iron. Accordingly research of this nature was
undertaken by the writer and his associates in the
engineering school of Michigan State College in l9M6.
It is the purpose of this report to summarize the re-
sults of the investigations previously reported by
McClure27 and Khan28, and to present new data dealing
with the inoculating substances, and with the melt-
ing conditions listed in the table of contents under
the heading of Experimental Results. This data was
acquired by comparison of inoculated heats with blank
heats with no ladle treatment, by comparing iron in-
oculated with various substances, and by comparing

irons melted under varying conditions.

28

II Experimental Proceedure

All of the heats described in this report were
melted in a 250 lb capacity, indirect arc, rocking
furnace unless otherwise noted. A typical charge is:
100.00 lbs pig Lot 1 (cu.15, Si 1.18, Mn .72, P.128,S.039%)

38.00 lbs pig Lot 2 (Ch.l2,Si 1.23, Mn.88, P.2h, 8.033%)
56.00 lbs of structural steel scrap
7.20 lbs of 27% ferro-silicon
.60 lbs of 80% ferro-manganese
.15 lbs of 50% iron sulphide

In.a few cases ingot iron punchings were substituted
for some of the steel. All of the constituents of the
melt were charged in the cold furnace except the ferro-
manganese and the iron sulphide which were added as soon
as there was a melt down.

All melts were brought to the temperature range of
2850-2900°F and tapped into 50 lb ladles. The nominal
tapping temperature being 2875°F as determined by
optical pyrometer readings through the spout. This
degree of superheat was selected as being representative
of general practice, and as providing sufficient tempera-
ture latitude to transport the metal to the molds, skim
the ladle, read the temperature with an optical pyro-
meter, and pour at the temperature selected for these

experiments, 2650°F to 2675°F.

All ladle additions, unless otherwise indicated,
were made by tapping a small amount of metal into the
ladle and then adding the inoculating alloy cont inu-
ously as the balance of the iron entered the ladle.
During this Operation an effort was made to carry the
inoculant under the surface with the stream of molten
iron.

In the case of ladle treatments with active metals
special proceedures were required. These will be des-
scribed for the individual cases. ’

The time interval between tapping and pouring
varied in general from 2 to 3 minutes. This time was
required for the metal to drop from the tapping tempera-
ture, and for transportation, skimming, and determination
of temperature.

lists]. from each ladle was cast into vertical, dry
sand core, 1.2" standard test bar molds, washed with a
non carbonaeious silica wash. All transverse test bar
results are the average of three or more here unless
otherwise indicated. Bars showing defects in the
fracture and abnormally low results for breaking load
and deflection were not included in the average. The
number of tests bars poured varied from three to five.

Bars were cleaned with a wire brush.

Small wedge chill test specimens were used in some
experiments, in others a chill block having the dimensions
3/h" x 2i-x 3-7/8" was employed. Two or more chill speci-
mens were poured from the ladle, and the results for chill
are the average of the two Specimens. In general the
agreement between chill specimens from the same ladle
was very good. In the case of two ladles an apprecia-
ble difference was found between the check specimens.

In these two cases the maximum chill is reported.

Tensile test results and hardness tests are not
reported in all cases. As the work continued it be-
came evident that the transverse breaking load, the
deflection, and in particular the triangular resilience
provided a more sensitive index of variations than did
the tensile strength. (Triangular resilience =
Trans. Breaking Load 1 Deflection). The most potent
influence of inoculants on.microstructure is at the sur-
face of the casting. This renders the transverse bar,
tested in the unmachined condition, a very suitable test
specimen for the purpose of demonstrating inoculation
action.

The subject of reproduceability of results is im-
portant in considering both chill tests and transverse
tests. Table II is presented to show the degree of re-
produceability of properties. ’This table gives test
results along with chemical analysis data on several

31

of the blank heats - that is heats which were melted
and poured without ladle treatment for the purpose of
comparison with various other irons of similar analy-

sis but ladle treated.

Table 2.
Iron C Si Trans. Deflec. Res. Chill.
T32-2 2.81 2.28 25M3 .219 278 16-30
T25-2 2.88 2.28 2575 .230 296 12-18
$25-1 2.86 2.22 2663 .2uo 320 1h-18
T33-2 2.84 2.28 2680. .233 30“ 16-31
T27-l 2.8h 2.39 2M90 .232 289 12-20
T27-2 2.85 2.H3 25Ml .2Hl 306 12-22
T30-B 2.86 2.60 2521 .222 280 12-20

In this table and subsequent tables "Trans" is trans-
verse breaking load for the 1.2 inch bar, "Deflec" is de-
flection in inches, "Res" is the triangular resilience in
inch pounds, and "Chill" the chill depth.in 1/32 of an
inch units. The figures before the dash give the depth of
clear chill, and those after the dash, the total chill.

If all of the results in this table are considered
it will be concluded that results for transverse break-
ing load are not significant unless the difference be-
tween two irons with different treatments exceeds 200 lbs.
However, in all cases the difference in silicon contents
between irons for comparison will be much less than the

range included here. The maximum variation.in deflection

is .022 inch; and in resilience, 42 inch pounds. The
variation in chill is large. It would appear that clear
chill is somewhat more consistent than total chill, and
that conclusions regarding chill depth should be confined
to those cases where the difference in chill is very mark-
ed. Variations in properties cannot be attributed to
variations in sulphur or manganese as these are controll-
ed within narrow limits for the several irons, and the
slight variation cannot be correlated with preperties.
It should be noted that there is a general impression
that uninoculated irons are erratic in their behavior
particularly with respect to chilling tendency and that
uniformity of behavior, particularly with respect to
chill, is one of the justifications for inoculation.

In making comparisons between irons in this paper
an effort has been made to keep the carbon equivalents
(C 4-§i, ) within one tenth of one per cent. This is a
degree of control which compares very favorably with
that of similar experiments reported in the literature.

Carbon and silicon contents were determined on
each ladle. Sulphur, manganese, and phosphorous con-
tents were determined only on one ladle from each heat
unless these elements were involved in the additions,
or when there was reason to believe that their contents

were altered by the ladle addition.

Photomicrographs were not made of all of the irons.
Typical structures are presented for the purpose of illus-
tration and discussion. Since the microstructures of any
one sample, unless it is fully inoculated, are quite vari-
able it becomes a very difficult matter to select an area
on the polished surface which is representative. The
microstructures of all significant irons were examined,
and it was concluded that the results of the transverse
tests were well correlated with the variations in micro-
structure, except in one instance which will be discuss-
ed later. The presence of a large percentage of type A
graphite, particularly at the surface of the casting,
seems to be very effective in producing a high trans-
verse breaking load and a high deflection. This ob-
servation is in accord with the experience of others.

Some of the conclusions reached in this investi-
gation are based on additions of active metals or of
alloys containing active metals. In this phase of the
research no attempt has been made to determine the amount
of the active metal retained in the solid iron. This
is justified by the fact that the retention of sub-
stantial amounts of the metals is not necessary in order
that profound changes in the microstructure and pro-
perties be produced. For example in producing nodular
iron the amount of residual magnesium has been reported

29
as varying from .020 to .085. McElwee and Barlow

have discussed this matter and have concluded that the
effects of inoculants cannot be measured by analysis of
the iron for residuals - the chemical changes being in-
significant or imperceptable.

In view of this feeling and in view of the time
consuming nature and of the uncertainty involved in the
determination of small quantities of active metals, it
was decided to defer analytical studies until their value
was indicated by the present research for continuing in-
vestigations.

In.making additions of the various alloys an.attempt
was made to keep the crushed sizes of the alloys comp
parable in each case. In general the commercial size,

8 mesh and down, was employed.

35

III Experimental Results

Since the amount of data in this report is con-
siderable, the results are presented as a number of
separate experiments for the convenience of the reader.

Each comparison of irons is given in two tables,
the first tabulating the available analytical data, and

the second the mechanical properties.
Comparison of Ferro-silicon and Calcium-silicon.

The two most widely employed inoculants have been
ferro-silicon and calcium-silicon. Very little of data
is to be found in the literature regarding the relative
effectiveness of these two alloys. Lorig30 has compared
irons treated with .2% additions of calcium-silicon with
irons treated with .6% additions of ferro-silicon. In
his experiments the irons were superheated over a range
of temperature up to 3000°F. The specific grade or
analysis of the ferro-silicon was not stated. Lorig
concluded that .2% additions of the calcium alloy were
equivalent to .6% additions of the ferro-silicon for
superheating temperatures up to 2900°F. Beyond this
temperature the irons treated with the ferro-silicon
were somewhat superior in mechanical properties.

In view of the industrial importance of these two
alloys and of the lack of comparative information, ferro-

silicon and calcium-silicon were selected by McClure27

36

and the writer for the initial experiments of this in-
vestigation.

Some difficulty is experienced in selecting the re-
lative amounts to be used in making a comparison between
alloys as diverse in chemical analysis as are these two.
The situation is rendered more difficult by the fact that
the silicon pick up is more complete when ferro-silicon
is employed than is the case with calcium silicon. It
was decided that the most satisfactory basis for comparison
would be afforded by use of amounts of the alloys such that
the silicon pick up would be the same in each case. This
would involve using somewhat larger amounts of calcium-
silicon. This basis of comparison has been proposed by
Morrogh and Williams3 for various alloys. In the experi-
ments performed by the author this required using amounts
of the two alloys in the proportion of .30% addition of
Si as ferro-silicon to .35% addition of Si as calcium-
silicon in the first comparison, and .60% to .70% in
the second.

The following proceedure was used. A 200 pound heat
was melted and brought to 2875°F as described above under
experimental proceedure. Two pairs of 50 pound ladles
were tapped from the heat. One member of the first
pair (iron no. T2-Fl) was inoculated with .30% Si addition
as ferro-silicon and the second with .35% Si addition as

calcium -silicon (T2-Cl). This proceedure was repeated for

the second pair of the ladles using .60% Si as ferro-silicon
(T2-F2) and .70% Si as calcium silicon (T2-C2).

Chemical analysis data are presented in Table 3,
and mechanical property data in Table h. As previously
mentioned manganese, phosphorous and sulphur were not
determined on all ladles. In these tables the irons to

be compared are grouped as pairs.

Data on Ferro-silicon and Calcium Silicon as Inoculants.
Table 3

Iron Addition C Si Mn P S

T2-Fl .30% Si as FeSi 2.95 2.00

T2-C1 .35% Si as CaSi 2.9# 1.95

T2-F2 .60% Si as FeSi 2.90 2.17 .056

r2-02 .70% 81 as CaSi 2.8% 2.21 .99 .102 .0h6
Table #

Iron. Inoc. Trans. Deflec. Res. Tensile Chill

T2-Fl Fe-Si 2665 208 277 “6,910 30-h5

TZ-Cl Ca-Si 3597 361 651 58,750 13-18

T2-F2 FeSi 2905 261 379 51,h70 12-20

' r2-02 CaSi 3728 39k 7u5 59,200 h-8

38

Analysis of Fe-Si

S1 92.73
A1 1.68
Ca .25

Table h shows the marked superiority of calcium-
silicon over ferro-silicon. All mechanical properties
are superior in the case of the irons treated with
the high calcium alloy. The increase in triangular
resilience is particularly noteworthy being 100% or
more. It is also evident that the high calcium alloy
is much.more effective than ferro-silicon in reducing
chilling tendency.

Examination of the microstructure of the cross
sections of the transverse bars revealed an entirely
normal (type A) graphite distribution in the interior
of the bars treated with the heavy addition of calcium—
silicon (T2-C2). The corresponding bars (T2-F2)
treated with ferro-silicon showed a tendency towards
type D and type E graphite distribution, see Fig 9.

At the immediate surface the calcium-silicon iron was
found to have a microstructure consisting mainly of
type A graphite with only a slight tendency towards D,
see Fig 10. The structure at the surface of the ferro-
silicon iron was highly abnormal with considerable
ferrite, Fig 11.

 

Fig 9. Microstructure of Interior of Iron T2-12,

Treated with .6% Ferrc-Silicon. Nital, 100x.

38B

 

Fig 10. Microstructure at Surface of Iron T2-C2, Treated
with .7% Calcium Silicon. Nital, 100x.

 

Structure at the Surface of Iron T2-F2, Treated

Fig 11.

with

6% Ferro-silicon. Nital 100x

39

The above pictures are from experiments conducted
independently by the writer, and are in good agree-
ment with microstructures presented previously by
McClure.

In general the microstructures of Figures 9,

10 and 11 are typical. Incomplete inoculation
manifests itself as a tendency towards D and E
graphite distribution with associated ferrite, and
this tendency is shown to be particularly marked at
the surface. In the succeeding experiments micro-
structures will not be presented in all cases, and
reference can be made to these photomicrographs as
being representative. As previously mentioned a

good correlation exists between the transverse
breaking loads, deflections, and their products, the
triangular resilience, and the microstructure. These
values will be taken as measuring the effectiveness
of the various ladle treatments in subsequent experi-
ments.

The results of Table 9 are in good agreement
with those of McClure who performed the initial
experiments of this gray iron research program. His
results showed the superiority of CaSi over FeSi for
a wide range of percentages of additions.

It is remarkable that this superiority of calci-

um silicon over ferro-silicon.has received little

attention in the literature. This difference appears
to be very significant both in a practical and a theo-
retical way. Since the difference in price is not
appreciable (20 to 21 cents per pound) it would appear
advisable to use calcium.silicon more generally than
has been the case subsequent to the expiration of the
patent covering its use. An additional conclusion of
practical importance which can be drawn from these re-
sults is that somewhat larger additions of inoculants
are required than is usually supposed to be the case.
Additions of .20% total inoculant are frequently re-
commended in‘the literature, and to the writers know-
ledge some foundries are using even less. However, it
should be mentioned that the employment of heavy treat-
ments involves some practical difficulties in addition
to the increased cost. The use of large amounts of
calcium silicon forms a dry layer of oxides on the
surface which is difficult to remove, and which, if
carried into the mold, may contribute to casting dirty-
ness and surface imperfections.

These results appeared to the writer to have con-
siderable theoretical significance. The superiority
of the alloy containing the larger amount of active
metal was indicative of an important role of the
active metal in inoculation. Accordingly a series of

experiments have been carried out in which the effect

of active metals have been determined. It is remarkable
that in.most of the theorizing which appears in the
literature, that the agency of active metals has been
almost entirely neglected.

The analysis of the ferro-silicon used in this
experiment was: 81 92.73%, A1 1.68%, Ca .25%. The
calcium silicon was: 63.97% 51, 31.72% Ca, 1.70% Fe,
.oss Zr, .16% Ti, 1.09% A1.

Calcium Silicon and SMZ

SMZ (Silicon-Manganese-Zirconium) is widely re-
commended and employed in industry as an inoculant.
Iownie31 states that suz is somewhat more effective
than ferro-silicon but fails to give the analysis of
the latter alley or supporting data. Morrogh and
Williams3 make similar statements but also neglect
to give data or analysis of the ferro-silicon. Burgess
and Bishop32 present data on irons inoculated with SMZ
but give no information as to its effectiveness re-
lative to other inoculants.

In view of the importance of SMZ as a commercial
inoculant experiments were undertaken.to compare the
alloy with calcium-silicon. The proceedure followed
was that of the previous experiment. A 200 lb heat
was divided into two pairs of ladles - the members of
the first pair were inoculated with .30% Si as SMZ and

.35% as Ca-Si. The members of the second pair of
ladles were inoculated with .60% Si as SMZ and .70%
as Ca-Si respectively. The somewhat greater addition
of ca-Si was necessitated by the lower silicon pick
up from the latter alloy.
Chemical analysis of the irons is presented in
Table 5, and mechanical property data in Table 6.
The analysis of the alloys employed is tabulated be-
low Table 6. In this and the other cases of this re-
port the alloy analysis given is that supplied by the
producer of the alloy.
Calcium Silicon and SMZ
.. _ Table 5 . v...
Iron. Addition 0 Si Mn P
T8-Sl .30% Si as SMZ 2.80 1.97 .9h
T8 -Cl .35% Si as CaSi 2.86 1.96 .92

T8 -82 .60% Si as SMZ 2.86 2.30 .96

18 ~02 .70% Si as CaSi 2.82 2.22 .91 .083
Table 6

Iron. Inoc. Trans. Deflec. Res Tensile

T8-Sl SMZ 3023 283 M30 k8,100

T8-Cl CaSi 3327 322 536 50,390

T8-82 SMZ 3138 326 511 h9,010

$8-02 CaSi 3516 365 6th 57.590

.056
.0h6

Chill
12-22
lh-22

10-15
12-17

Ca Si SMZ

63.92 81 62.79 Si
31.72 Ca 6.95 Mn
1.70 Fe 5.95 Zr
.08 Zr 2.19 Ca
.16 Ti 1.50 Al
1.09 Al .19 T1

These results indicate that SMZ is somewhat in-
ferior t0 Ca-Si as far as the mechanical properties
are concerned when the additions are made to give
approximately the same recovery of silicon. The
difference in chilling tendency is not sufficient to
permit a final conclusion, but it is indicated that
SMZ is slightly more effective than Ca-Si in reducing
chill.

A rough comparison of SMZ and FeSi is afforded
by comparing Iron T2-F2 of table 8 with Iron T8-S2 of
Table 6. It would appear that SMZ is somewhat superior
to FeSi of the grade employed particularly with re-
spect to deflection.

Figures 12 and 13 show the surface microstructures
of the .30% SMZ iron (18-31) and the .35% Ca—Si iron
(T8-Cl) respectively. It is evident that these amounts
of the alloys are not sufficient to produce normal struc-
tures at the surface of castings of 1.2 inch section.

The microstructure of the Ca-Si treated iron is somewhat

li-3A

~

" \s I. ‘_ w
\‘2-'[““.‘
5‘" v‘ .50._
l a. r i

 

Fig 12. Microstructure at Surface of Iron T8-Sl. Treated
with .3% SMZ. Nital 100x.

43B

 

Fig 13. Microstructure at Surface of Iron T8-Cl. Treated
with .35% Calcium-Silicon. Nital 100x.

superior to that of the SMZ iron. These results in,
dicate that the present practice of recommending re-
latively small amounts of inoculants is in error if
maximum properties are desired, and in particular if
wear resistance is a consideration.
Silica-manganese as an Inoculant.

Silica-manganese is not commonly used as an
inoculant. However, it is listed in Table 1 by
Lownie18 as being an inoculant, and is also mentioned
as an inoculating substance in the article dealing
with gray cast iron.in the American Society for
Metals Handbook, page 515, 1998 edition. The fact
that SMZ is an effective inoculant may be taken as
an indication of some value in a late addition of
manganese.

Boyles.9 has discussed the effect of sulphur
and manganese on.graphite distribution and has sug-
gested that the precipitation and distribution of
manganese sulphide inclusions is related to graphite
distribution. These relationships are indicative of
the possibility of some effect of a late addition of
manganese.

The following experiment was undertaken to

determine the effect of silica-manganese additions.

Heat T3 was melted and ladles l and 2 were treated

with a 1.0% addition of silica-manganese (.5 lb of

alloy per 50 lb ladle).

Blank heat T“ was melted

with the composition of the cold charge adjusted with

respect to silicon and manganese to give approximately

the same final analysis as for heat T3.

Results for these heats are given in Tables 7
and 8.

The composition of the silica-manganese em-

ployed is given below Table 8.
Data on.Silico Manganese and Blank Heat;

Iron
13-1
T3-2
Th-l
Th-Z

Iron
T3-1
T3-2
Th-l
Th-2

ilTable 7
Addition C . 81
1.0% Si Mn 2.88 2.31
1.01 Si an 2.93 2.3%
Blank 2.9% 2.3%
Blank 3.00 2.33
Table 8
Inoc. Trans Deflec
Si-Mn 236k 18%
Si-Mn 2375 188
Blank 2580 218
Blank 2631 223

Mn
1.06
1.07

.98
1.00

Res
217
223
282
29%

P s
.075

.060
.077 .062

Tensile Chill
- 32-52
- 2M-MH
- 16-35
- 18-3h

#6

Analysis of Si Mn

Mn 66.5
Si 18.5
C 1.25
Fe 13.00
Ti .17

Ca and Al present only as traces.

The results would be interpreted as indicating
that silica-manganese 0f the composition employed is
not of value as an inoculant and that the with-holding
of silicon and manganese from the charge for the purpose
of making late additions is detrimental to the pro-
perties of the iron including chilling tendency.

As a check on these results the first two
ladles of heat T6 were treated with 1.0% additions
of silico-manganese, using the first two ladles of
blank heat T5 for comparison. Results are presented
in Tables 9 and 10.

Data on Silica-Manganese and Blank Beats
.0. . , Table 9 ..... ._ . ..
Iron. Addition C Si Mn P S
T6-l 1.0% Si Mn 3.0M 2.26 .95 .075 .073
T6-2 1.0% Si Mn 3.0% 2.30 1.02
T5-l Blank 3.05 2.21 .99
T5-2 Blank 3.07 2.20 .97 .075 .072

Table 10

Iron Inoc. Trans Deflec Res Chill

T6-1 Si-Mn 2885 218 272 18-38
T6-2 Si-Mn 2506 2H0 301 16-30
T5-l Blank 2586 215 279 20-92
T5-2 Blank 2588 22% 290 » 20-h2

These results are not in complete agreement with
those of the first silico-manganese series. The silico-
manganese heats of this second series are slightly
superior in deflections and slightly lower in chill-
ing tendency than the correSponding blanks. These
small differences must be attributed to individual
variations in heats as discussed above in connection
with the subject of reproduceability of results.

The conclusion from these results is that silico-
manganese of this grade is without value as an ino-
culant. It is noteworthy that the silica-manganese

employed is free of calcium and aluminum.

High and Low Active Metal Ferro-silicons.

The superiority of alloys containing calcium,
aluminum, and zirconium as demonstrated by the above
data indicated the desirability of determining the
relative effectiveness of low and high active metal
content ferro-silicons.

High and low aluminum grades

were obtained from a major producer with the analysis

given just below Table 12. Irons of the compositions

presented in Table 11 were produced by inoculating with-
.50% Si additions of the two alloys for comparison with
one another, and for comparison with a blank heat. Re-

sults are given in Table 12.
Qgta on Grades of Ferro-silicon
. . ...... Table 11' ...,....-.
Iron Addition C Si Mn P S

T26-3 .51 Si,Low Al alloy 2.85 2.3» .85 .115 .066
r26-h .55 Si,High A1 alloy 2.90 2.3M

T27-1 Blank 2.88 2.39 .83 .122 .066
T26-2 .5% Si, High A1 alloy2.86 2.29 .85 .115 .066
Table 12
Iron. Inoc. Trans. Deflec. Res. Chill
T26-3 Low A1,FeSi 2591 .230 298 13-29
T26-k. High Al,FeSi 2920 .280 #09 8-13
T27-1 Blank 2890 .232 289 12-20
T26-2 High li,res1 2919 .288 520 8-1»
Low Aluminum FeSi High A1uminum.FeSi
93.00 81 90.57% 81
.MS Al 2.20% A1

e10 Ca 346% C‘

l+9

Comparison of Iron T26-3, inoculated with the low
active metal content alloy, with T27-l, the uninoculated
blank, shows that the low aluminum and calcium alloy
has a negligible value as an inoculant. Consideration
of the data for the irons inoculated with the high
active metal content shows the pronounced superiority
of the high active metal alloy.

These data afford strong evidence for a theory of
inoculation based on the action of active metals, parti-
cularly aluminum and calcium. They also indicate the
serious nature of failure to state the analysis of
inoculants with respect to these elements when used in
research. Such failure is usual in work reported in
the literat ure. Many of the discrepancies noted in the
literature may be attributed to variations in the active
metal contents of the alloys employed.

Although the ferro-silicons used here are desig-
nated as "high" and "low aluminum" alloys, it should
be noted that calcium as well as aluminum is varied in
the composition.

Silicon as an Inoculant

Consideration of the above data pointed to an im-
portant role of the active metals in inoculation, and
indicated the desirability of determining the effect
of pure silicon as a ladle addition. The following

experiments were performedz'l) Two ladles were tapped
from heat T2h and one ladle was treated with a .50%
addition of high aluminum ferro-silicon and the other
with a .50% addition of purified silicon. 2) A ladle
from heat T26 was treated with .50% purified silicon
for comparison with two blank or uninoculated irons.
Results are presented in Table 13 and 1h. The high
aluminum ferro-silicon is the one employed in previous
experiments with the analysis given on.page #8. The
"purified" silicon employed was the highest grade

commercially obtainable (99.8% $1, balance iron).

Data for Purified Silicon

‘ Table 13
Iron. Addition C Si Mn P
rzh-s .5% Si as pure Si 2.78 2.61 .91 .130 .059
T2k-F .5% Si as H1 Al FeSi 2.75 2.62 .91 .130 .059
T26-1 .5% Si as pure Si 2.87 2.32 .85 .115 .066
125-2 Blank 2.8% 2.2M .80 .122 .062
T27-1 Blank 2.8M 2.39 .83 .122 .066

Table 1*

Iron. Inoc. Trans Deflec. Res Chill
TZM-S .55 Si 2601 .218 28k 8-21
T2h-F .sx Si as Hi A1 FeSi 3059 .312 #76 3-3
T26-l .5% Si 2533 .210 266 13-36
T25-2 Blank 2575 .230 296 12-18
127-1 Blank 2390 .232 289 12-20

Comparison of the results for irons T2k-S and
T2k-F shows that pure silicon is much inferior in
all respects to the ferro-silicon.

Comparison of the values obtained for the blank
heats T25-2 and T27-l with T26-l, treated with .50%
of silicon, indicates that silicon has no value as
an inoculant, and suggests that with-holding a portion
of the silicon and adding it late may be slightly
detrimental rather than beneficial to properties.

These data render it difficult to support the
silica slime theory of inoculation particularly if
the nucleating agent is considered to be a ferrous
silicate particle. It also tends to discredit a
theory based on precipitation of graphite particles
resulting from local concentrations of silicon. It
would appear that the silicon in commercial inoculants
is simply a convenient carrier for active metals or
that the simultaneous addition of silicon with alumi-
num and calcium is in some way capable of improving
graphite distribution and reducing chill. ‘

Fig. 1% shows chill block fractures of TZH-S,
treated with purified silicon and T2k-F, treated with
"high aluminum" ferro-silicon, and shows the powerful
chill reducing tendency of the active metals contain-
ed in the ferro-silicon.

51A

 

Fig 1h. Fractures of Chill Blocks. Pair to left are of
Irons TZH-S, treated with .S% purified silicon and of T2h-F,
treated with .5% of high aluminum ferro-silicon, and shows
the pronounced effect of aluminum on chilling tendency.
Right hand pair are of Iron T30-A, treated with 1.1% alum-
inum and of T30-B the corresponding blank.

52

Calcium as a Ladle Addition.

In view of the evident importance of the active
metals in the inoculation mechanism Khan28 and the
writer undertook experiments to determine the effect
of calcium and aluminum additions as a contribution
to the general project. Data from Khan's report are
presented here along with confirming data acquired in-
dependently by the writer.

Khan inoculated ladles with .21% Ca, .H%% Ca,
and .68% Ca for comparison with blank ladles from
the same and other heats. The calcium additions were
accomplished by wiring the metallic calcium to an iron
rod and plunging the metal beneath the molten iron in
the ladle immediately after tapping. When solution
and reaction were complete, the ladle was skimmed,
the temperature allowed to dr0p to 26SO°F and the iron
poured in the usual manner.

The chemical analysis of the irons from Khan's re-
port are recorded in Table 15 and the mechanical pro-
perties in Table 16. It is evident from these data
that calcium is very effective as a ladle addition.
The pr0perties of iron No. T9CS-9 are superior to the
properties of all other irons of this report except-
ing those inoculated with calcium silicon. The

calcium silicon treated irons of Tables h and 6

show superior properties but have somewhat lower carbon

equivalents.
Data on Calcium.

Table 15' ‘
Iron Addition C Si Mn P S
T9Bl-h Blank 3.07 2.30 .95 .l .063
T901-h .21% Ca 3.0# 2.29 .062
T9B5-9 Blank 3.09 2.30 .066
T9C5-9 .Hhfi Ca 3.00 2.30 .057
T12B1-5 Blank 2.90 2.23 .05?
11101-5 e68% Ca 2e92 2e31 0°52

Table 16
Iron Inoc. Trans. Deflec. Bea. Chill
T9Bl-M Blank 2679 .251 33“ 7-22
r9Cl-k .211 Ca 2988 .283 #25 5-17
T9B5-9 Blank 2656 .261 3M5 1-17
T12Bl-5 Blank 2202 .196 216 12-30
21101-5 .68; Ca 3013 .361 5h5 1-8

The writer has checked the data reported by Khan
by treating a ladle with .50% Ca and comparing the re-
sulting data with two blank irons of approximately the

same analysis. Results are presented in Tables 17 and

18 and are in agreement with those of the proceeding

report.
Table 17
Iron Addition. C Si Mn P S
11-2 95 srns Ca .5% 3.02 2.ho .86 .098 .055
11-82 Blank 3.03 2.38 .86 .102 .058
Th-Z Blank 3.00 2.33 1.00 .077 .062
Table 18
Iron Inoc. Trans. Deflec. Res. Chill
11-2 Ca 302“ .385 58% 6-8
Tl-BZ Blank 2521 .262 330 12-20
Tk-z Blank 2631 .223 295 18-3k

All of this data is significant and is strongly
indicative of the major role played by Ca additions
in determining the graphite distribution.and pro-
perties of cast iron. Consideration of the data for
the other experiments shows that those ladle additions
which are free or relatively free of calcium (silico-
manganese, purified silicon, and "low aluminum" ferro-
silicon) are without value as inoculants, while there
is a correlation between the calcium contents and the
effectiveness ofthe other inoculants employed.

These data show that silicon additions are not
necessary to the production of excellent properties

in high strength cast iron.

55

Aluminum,Additions.

Those inoculants which have been effective in
the research covered by this report contain sub-
stantial amounts of aluminum along with the calcium.
The high aluminum ferro-silicon.contains 2.20% A1,
.% Ca; and the suz, 1.50% A1 and 2.19% Ca. The
effect of aluminum additions were determined by
Khan?8 and the writer, and were checked by the
writer.

Aluminum additions were made in the same
manner as the calcium additions - by wiring the
aluminum to an iron rod and by plunging beneath
the surface of the metal in the ladle. The ad-
dition of the aluminum gave the surface of the
metal a peculiar ropy appearance somewhat like
that of molten brass in the ladle. The aluminum
addition was approximately 1.1% in both cases.

The data reported by Khan28 are presented in
Tables 19 and 20. The effect on.mechanical pro-
perties is small but the effect on chill tendency

is remarkable.

Data on.Aluminum.Additions.

n.b1.‘19
Iron Addition C Si Mn P S
T13A6-10 1.1x Al 2.90 2.23 .067
Tl3B6-10 Blank 2.95 2.26 .068
Table 20
Iron Inoc.“ Trans. Deflec. Res. Chill
Tl3A6-10 A1 2881 .213 266 0-1

Tl3B6-10 Blank 2280 .18“ 210 7-2H

The properties for the irons of Table 19 are
lower than those obtained for other irons of about
the same carbon equivalent in this report. The low
values may be a result of prolonged superheating.

Check data on this aluminum experiment is given
in Tables 21 and 22. It would appear that aluminum
additions of 1% have no effect, or a negligible
effect on properties with the exception of the effect
on chill, which is pronounced.

Table 21
Iron. Addition C Si Mn P S
T30-A 1.15 A1 2.86 2.58 .87 .122 .06h
T30-B Blank 2.86 2.60 .87 .122 .062

Table 22
Iron Inoc. Trans. Deflec. Res. Chill
T30A A1 2588 .216 280 3-5

T30B Blank 2521 .222 280 12-20

57

More data on aluminum additions are needed with a
range of size of additions before a final conclusion is
Justified. However, it would appear that the effect of
aluminum is distinct from that of calcium. Aluminum is
seen to be the most effective element for reduction of
chill. This ability to reduce chill was noted in the
case of the irons inoculated with high aluminum ferro-
silicon, see Fig. 1%.

No adverse effect of the aluminum additions were
observed on the castings. Foundrymen have long be-
lieved that large aluminum additions are detrimental
to casting properties and tend to promote pin hole de-
fects in the casting surface, particularly when the
casting is poured in green sand molds. The absence of
such defects in the castings of these experiments may
be due to the fact that all of the castings were made
in dry sand cores.

Aluminum is capable of reducing chilling tendency
without improving graphite distribution ormechanical
properties, thus indicating that high chilling tenden-
cy is not a necessary concomitant of abnormal graphite
distribution. This is surprising in view of the fact
that abnormal irons as usually produced exhibit a high
chilling tendency. This observation renders it diffi-
cult to base a theory of inoculation on a reduction of

carbide stability.

58

Effect of Acid and Basic Melting Conditions.

It has long been understood that at higher melting
temperature there is a reduction of silicon from furnace
linings by carbon in the melt with a resulting loss in

10’11 has dis-

carbon and an increase in silicon. Heine
cussed this matter and has placed the temperature at
which this reaction starts at 2671°F. This reaction is
accompanied by a boiling of the melt presumably from
the liberation of carbon monoxide from the reduction.
The resulting changes in the carbon and silicon cons
tents contribute to the difficulty of controlling the
carbon and silicon within narrow limits in commercial
melting equipment.

Diepschlag2h

has claimed that melting in furnaces
with aluminous linings produces irons with different pro-
perties than those melted in silica linings. As previous-
ly mentioned Morrogh and Williams3 have failed to confirm
this finding.

In view of the possible effect of the silica re-
duction and gas evolution, and in view of the contra-
dictory nature of the reports on the effect of furnace
linings, experiments were undertaken.tc determine the
effect of strongly acid and basic melting on the macro-
structure and properties of cast iron.

Melts for this purpose were made in a 30 lb, high

frequency induction furnace using a magnesia crucible

for the basic melts and a silica crucible for the acid
melts. During and after the melt down small additions
of calcium oxide were made to the basic heats, and
small additions of silica to the acid heats for the
purpose of assuring basic and acid melting conditions.
Charges consisted of electric furnace remelt, steel
and ferro alloys. No inoculants were added. Since
some desulphurizing occured in the basic heats the
work was repeated using sulphur additions (slightly
larger in the case of the basic melts) for the purpose
of adjusting sulphur contents. Results are presented
in Tables 23 and 2%.

Data on Acid and Basic Melting.

59

Table 23"
lron. Addition C Si Mn P S
IA-l Acid, 90 grls 8102 2.83 2.38 .79 .111 .039
TA-Z Acid, 90 grms 3102 92.82 2.3% .80 .111 .058
TA-5 Basic,30 grms CaO 2.82 2.30 .81 .118 .029
TA-6 Basic,30 grms Ca0 2.82 2.k3 .070
Table 28
Iron Trans. Deflec. Res. Chill
TA-l acid 2855 .193 236 2h-52
TA-2 acid 2311 .182 210 20-82
11-5 Basic 2495 .185 232 30-52
TA-6 Basic 2AA? .178 218 22-h8

60

From these data it is concluded that acid and basic
melts are equivalent - the variation in the pr0perties
being negligible. Since the tapping temperature 2875OF
was well above the temperature set by Heine for the be-
ginning of reduction of silica, and since boiling of the
melt occured before tapping, it is concluded that the
reduction of silica from the furnace lining is without

effect on structure and properties.
Magnesium as a Ladle Addition.

Magnesium additions to molten iron are of interest
in connection with the productions of nodular or ductile
iron, which is formed by an addition of magnesium,
usually as a nickel magnesium alloy, followed by an
inoculation with ferro-silicon. No satisfactory theory
has been advanced to explain the effectiveness of this
double treatment in producing an iron with the graphite
entirely in the form of spheres or "nodules". However,
it is understood that magnesium is an active desulphur-
izer and that low sulphur content is one of the require-
ments of ductile iron production.

Khan?8 and the writer have attempted to determine
the effect of magnesium additions to molten iron with-
out the subsequent addition of ferro-silicon. The

technique employed for the addition of calcium and
aluminum, whereby the metal is plunged and held be-

neath the surface of the molten bath, cannot be employed
in the case of magnesium because of the low boiling point
(llOOOC) of this element and the consequent violence of
the reaction. In the experiments described here the
magnesium, in the form of 3" rods of commercially pure
metal, were dropped through a flanged tube with flange
covering the ladle onto the molten metal. Even under
these conditions the reaction is extremely violent, and
the pick up of magnesium somewhat uncertain. For these
reasons the use of pure magnesium has been generally
abandoned in industry, and the magnesium is introduced
in the form of an alloy. In these experiments magnesium
was used instead of the alloy in order to avoid the com-
plicating presence of the alloying element.

Results from the experiments reported by Khan are
presented in Tables 25 and 26. It is apparent that the
magnesium treatment has effected a small improvement in
properties, a marked decrease in chilling tendency, and
a considerable desulphurization.

Data on Magnesium.

Table 25
Iron Addition C Si Mn P S
T13 Bl-5 Blank 2.93 2.26 .057

62

Table 26
Iron. Inoc. Trans. Deflec. Res. Chill
T13 B1-5 Blank 2188 .173 188 21-35
T13 Ml-5 1.101 Mg 2333 .199 233 8-18

Microstructures of T13-B, the untreated blank and
of T13-M, the magnesium treated iron are presented in
Figs 15 and 16. The magnesium addition has an adverse
effect on graphite distribution and has promoted the
formation of associated ferrite. The extent to which
the microstructural changes are due to the lowering of
the sulphur content is problematical. The sulphur re-
maining in T13-M should be sufficient to permit a normal
structure in the iron according to Boyles.1’9

These results as far as microstructure is concerned

are in.agreement with the Opinion of Morrogh and Williams26

who have regarded the formation of highly abnormal type
D graphite and associated ferrite as a step in the di-
rection of the formation of nodular graphite structures
by magnesium.and cerium additions.

In.attempting to check the results reported by Khan
the writer performed the following experiments. Three
ladles were tapped from heat T32. The first ladle was
treated with .50% Mg, the second was untreated to serve
as a blank, and the third ladle was treated with 1.05 Mg

plus an amount of S calculated to replace the sulphur re-

62A

 

Fig 15. Microstructure of Blank Heat T13-B at the Surface.
Nital 100x.

 

Fig 16. Microstructure of Iron T13-M, Treated with Magnesium.
Nital 100x.

moved by the magnesium. The sulphur was added in the
form of iron sulphide after the magnesium reaction had
ceased, and amounted to an addition of .OH%S.

Results are reported below in tables 27 and 28
and are seen to be negative as far as properties and
chill tendency are concerned. Examination of the irons
under the microscOpe revealed no influence on structure.
Figures from the sulphur determinations show only a
minor desulphurization.by the magnesium addition. It
would appear that either the addition of magnesium is
without effect, or that the recovery of the magnesium
was too small to be effective. The latter conclusion
would appear to be Justified by the small amount of de-
sulphurization produced. Magnesium additions as pure
metal have been found to be uncertain in the industrial
production of modular iron, and have been abandoned in
favor of nickel-magnesium alloys. The writer regards
these results as inconclusive, and is of the Opinion
that further work with magnesium is Justified in view
of the importance of developing a theory for the mecha-
nism of module formation.

Table 27
Iron. Addition C Si Mn P S
T32-1 .501 Ma 2.86 2.28 .061
T32-2 Blank 2.81 2.28 .87 .105 .068

232-3 1.00% Hz - -OH% 3 2.83 2.21 .85 .083

Table 28

Iron Inoc. Trans. Deflec. Res. Chill
r32-1 .50% Mg 2530 208 268 16-32
T32-2 Blank 25h3 219 280 16-30

r32-3 1.00% Mg - .ohs 2537 220 279 16-32

The photomicrograph of Fig 16 shows the micro-
structure of iron T13-M, treated with magnesium, to be
more abnormal than that of the corresponding untreated
iron T13-B of Fig 15. However, the preperties of the
magnesium treated iron are somewhat superior to those
of T13-B, and the chill tendency is markedly less. This
is an instance of failure of microstructure and proper-
ties to correlate, and a further example of lack of
correlation between degree of abnormality and chilling
tendency. *

Graphite Additions.

The idea is widespread in the foundry industry
that benefits are to be derived from late additions of
graphite to molten iron, particularly with respect to
a reduction in chilling tendency. This conception has
been fostered by advertising claims made in the trade
Journals, but it is not made clear as to whether the
chill reduction is due to an inoculation effect or to

an increase in the carbon equivalent. The only ex-

perimental evidence in the literature is found in an
article by Dahlbergl6 who recommended the use of eight
mesh graphite, and reported an increase in mechanical
preperties and a decrease in chilling tendency with

very small additions of graphite.

The following experiment was performed to determine
the effect of appreciable graphite additions. Heat T29
was melted down to a low carbon content and a ladle treat-
ed with a .khfl addition of graphite for comparison with
blank heats of similar analysis. Results are presented
in Tables 29 and 30 and would indicate that the effect
of graphite additions are negligible if they exist at
all. The idea that the effect of graphite additions is
of very short duration, and that the graphite addition
should be made as the metal enters the mold has been
suggested to the writer. This possibility should be
checked before a final conclusion is reached, as the
matter of graphite in a colloidal dispersion serving
as a nucleating substance is of theoretical interest.

In the above experiment the graphite was added at the
furnace spout.

A Data on Carbon Additiong.

' “an.ag

Iron {Addition, C Si Mn p s
T29-2 .hhfi C 2.82 2.30 .89 .110 .06h
T32-2 Blank 2.81 2.28 .87 .105 .06M
133-2 Blank 2.81» 2.28 .86 .122 .062

66

Table 30
Iron Inoc. Trans. Deflec. Res. Chill.
T29-2 Graphite 2699 .242 326 12-18
T32-2 Blank 2583 .219 280 16-30
T33-2 Blank 2680 .233 311 16-31

The Effect of Various Inoculants on Cell Size.

The cell size of cast iron refers to the size of the
eutectic units in the solid iron. Boylesl has described
the manner of solidification of a hypoeutectic iron as:

l) the formation of primary dendrites of austenite in the
liquid 2) theneucleation of the eutectic which solidifies
as spherical "cells" between and around the primary den-
drites 3) the development of graphite flakes in the cells
simultaneously with the growth of the cells, with the
flakes growing at the liquid solid interface, and k) the
segregation of some elements in the remaining liquid
surrounding the growing cells - notably phosphorous,
sulphur, and hydrogen.

This conception of the solidification is based on ex-
periments in which transformation is arrested by quenching
at temperatures between the liquidus and solidus to pro-
duce microstructures like that of Fig 3, in which the
light etching background represents liquid at the moment
of quenching.

As previously discussed, Boyles has proposed a theory

to explain graphite size variations on the basis of varia-

67

tions in the rate of formation of the austenite and graphite
flakes of the eutectic cells. It is apparent that the number
and size of the cells in the solidified iron is an important
theoretical consideration, inasmuch as it indicates the possi-
bility of a cell nucleation by inoculants.

The cell size in irons is not made evident by ordinary
polishing and etching proceedures. Two methods have been
employed for the delineation of cell boundaries: 1) Heat
tinting, and 2) overetching with nital to bring out the cell
size by the position of steadite, which is located in the
cell boundaries as a result of the segregation of phosphorous,
and which is light etching.

Boyles1 has successfully employed heat tinting in certain
irons, but does not give specific instructions for applying
this method, and such directions are not to be found else-
where in the literature beyond the statement that temperatures
up to 900°F may be employed for heat tinting the polished sur-
face. The writer has been unable to apply the heat tinting
method successfully for measuring the cell size of the irons
of this report. This may be a result of the hypoeutectic
character of these irons with the consequent limitation on
the interdendritic space. Heat tintmng at a temperature of
800°F was found to bring out the primary dendrites but not
the cell boundaries.

The second method, which determines the cell boundaries
by the distribution of steadite, is dependent on the phosphor-

68

ous content of the iron. Steadite does not appear in the
microstructure unless the phosphorous content is approxi-
mately .10% and it increases in amount with increasing
phosphorous. High strength irons of the type investigated
in this report are generally produced with phosphorous con-
tents of less than .15% and the irons of this investigation
have been melted in conformation with this practice. This
has made it somewhat difficult to apply this steadite dis-
tribution method, but a number of irons were examined this
way for cell size.

Adams38

has discussed the steadite method, which as
previously mentioned employs deep etching with nital to
create a contrast between the light etching steadite in the
cell boundaries and the matrix. He has presented a chart
with numbers ranging from 1 to 7 where the low numbers in-
dicate large cell size. On this chart, size no. 2 represents
an average cell diameter of approximately one inch at a mag-
nification of 25; no. 3, a cell diameter of three quarters
of an inch; and no. u, of one half inch. Using the deep
etching method and this chart, the writer has determined the
cell size of a number of the significant irons of this re-
port. Data is presented in Table 31 where the treated irons
are compared with the corresponding blanks in the second

column or with other irons.

69

Data from Cell Size Determinations.

t‘Table 31..
Iron Addition Cell Iron Addition Cell
Tll-C2 .68% Ca h Tll-B3 Blank 3
T1-2 .50% Ca 3 Tl-B2 Blank 2
r2-02 .70% CaSi 3 T2-F2 .6o% FeSi 2
rzh-F .50% F051 3 r2u-s .50% Pure Sin 2
T29-2 .MA% 0 2 T25-2 Blank, 2
T30-A 1.10% Al 2 T30-B Blank 2
T13-M3 1.10% Mg n T13-B3 Blank 3

The ferro-silicon used to treat iron T2-F2 was an inter-
mediate active metal content alloy containing 1.68% Al and
.25% Ca (See Table 3 and 4 Page 37, and Page 38). The ferro-
silicon addition to iron T2h-F was the high active metal alloy
containing 2.20% Al and .M6% Ca (See page 50). With this in
mind consideration of the data on the first four pairs of
iron of the table would indicate that the addition of calcium
or of alloys containing appreciable calcium has the effect
of reducing the cell size.

Comparison of graphite treated iron T29-2 with blank
heat T25-2 of very nearly the same analysis indicates that
the graphite addition has no influence on cell size.

The aluminum treated iron T30-A has the same cell size
as T30-B, the corresponding blank.

An addition of 1.10% magnesium to iron T13-M3 appears

to have reduced the cell size from 3 to b when comparison

is made with the blank T13-B3.

0n the basis of the above data it would be concluded
that the addition of calcium to gray iron effects a reduction
in the cell size, and that the possibility of a similar re-
duction by magnesium exists. It would be advisable to cone
tinue this study using a series of higher phosphorous irons,
to give a better delineation of cell size, before a final

conclusion is reached.
Decarburization by Calcium.

During the experiments in which pure calcium.was added
to the ladle, Khan?8 and the author observed that the calci-
um addition tended to reduce the carbon content of the iron.
This conclusion was based on comparison of the analysis of
blank ladles tapped at the same time with the analysis of
the treated ladles. Table 32 gives data from.Khan's report
and includes data on the calcium treated ladle (Heat Tl)
from the work performed independently by the writer. Data

on ladles treated with large amounts of Ca Si are also in-

cluded (Heats T2 and T8).

71

Data on Decarburizing Effect of Ca.

Table‘32
% Ca Blank % C Treated % C Loss
.21 T9 Bl-5 3.07 T9 01-5 3.0% .03
.MB T9 B6-10 3.09 T9 06-10 3.00 .09
.63 T10 B1-5 2.96 T10 01-5 2.80 .16
.68 T11 B1-5 3.10 T11 01-5 2.92 .18
.87 T10 Bo-lo 2.99 Tlo 06-10 2.83 .16
1.11 T11 B6-lo 3.08 T11 06-10 2.8M .2h
.50 T1 - B2 3.03 Tl - 2 3.02 .01
.22 T2 - F2 2.90 T2 -02 2.8M .06
.22 T8 - 82 2.86 T8 -02 2.82 .03

The last column under "Loss" gives the reduction of
carbon content brought about by the addition of calcium.
The last two pairs of irons at the bottom of the table
give results for irons treated with calcium-silicon (31.7%Ca)
when compared with accompanying ladles treated with ferro-
silicon.

These results are fairly consistent with the exception
of the data from heat T1, and furnish strong evidence for
a decarburizing action of calcium. It is difficult to attri-
bute this effect to any mechanism other than one involving
the formation of a calcium carbide of high stability and low
solubility. This possibility is discussed below.

72

IV Discussion

The following observations can be made from the data of
the experiments of this report:

1. The element silicon is not effective as an inocula-
ting agent.

2. Silico manganese, which is relatively free of calcium
and aluminum, is not effective as an inoculant.

3. The inoculating ability of various commercial grades
of ferro-silicon increases with the calcium and aluminum con-
tents. Low calcium grades have very little effect.

H. Silicon-manganese-zirconium alloys containing some-
what more calcium than is found in commercial ferro-silicon
are more effective than ferro-silicon.

5. The most effective inoculant employed in these ex-
periments is the calcium-silicon alloy with approximately
30% of calcium.

6. Although aluminum is a very powerful chill reducer,
its addition to cast iron does not promote type A graphite
distribution.

7. Metallic calcium additions to the ladle bring about
marked improvements in properties and graphite distribution
of cast iron. Calcium is superior to all other inoculating
agents tested in these experiments, with the exception of
calcium-silicon, when added in the amount of approximately
.h%.

8. Metallic calcium is effective in reducing the chill-

ing tendency of cast iron.

9. Metallic calcium is effective in reducing the cell
size of cast iron microstructures.

10. Calcium additions effect a reduction of the carbon
content of molten iron. ‘

A survey of the literature and the data of this re-
port indicates that zirconium, titanium, aluminum, lithium,
magnesium and possibly cerium are either not effective in
changing graphite distribution or tend to produce type D
graphite in cast iron. Thus it appears that calcium is
unique in its effect of promoting large randomly distributed
graphite flakes, and that calcium is an active agent in com-
merical inoculants.

This conclusion and the above observations regarding
silicon render it difficult to support the silica slime or
‘ the graphite nucleus theory of inoculation, and suggests the
need for further examination of the mechanism of inoculation
in the light of the agency of calcium.

What are the ways in which calcium.might influence
graphite formation? The following are among the possibilities:

1. By deoxidation.

2. By nucleation.

3. By carbide formation.

4. By adsorption of calcium or a calcium compound on an
interface existing during solidification and graphite forma-
tion with a resultant influence on the energy distribution
in the system.

78

0f the several elements ordinarily occuring in ferrous '
materials calcium ranks high in its affinity for oxygen. This
affinity for oxygen can be expressed in terms of the free
energy of oxide formation. Data on these free energies have
been plotted by Richardson and Jeffes33 and their information
is presented graphically in Fig 17. These curves show the
variation with temperature of the free energy change accom-
panying the conversion of one gram molecule of oxygen at one
atmosphere pressure into the equivalent quantity of oxide.
These data are indicative of the high deoxidizing ability of
calcium., However, such data must be interpreted with caution
since it cannot be applied to alloys without reservation.
Bach12 has pointed out that the activities of any particular
element are influenced by variations in the concentrations of
the elements in the system, and that our information regard-
ing thermodynamic relations even for the simple Fe-Si-O
system is far from reliable or complete. ~Nevertheless, it
does appear probable that a reduction in oxygen content would
be accomplished by the addition of calcium.

The failure of aluminum and some other active metals to
influence graphite formation in a favorable manner in gray
iron can be advanced as an argument against the deoxidation

hypothesis. Heinelo’ll

has presented evidence to show that
aluminum effects a deoxidation of cast iron, and the failure
of aluminum to influence graphite formation renders it diffi-

cult to support the deoxidation mechanism as the principle

TEMPERATURE,

  
  

RT log‘m‘,

. pr [09‘ ,Lb) (KILOCALORIES)

}'—Absolutc zero

Fig 17. Relation between Free Energy and Temperature. After

Richardson and Jeffes.33

factor in determining graphite distribution unless it is
argued that the oxygen content of the iron is very critical.
Chilling tendency has been attributed to oxidation and both
calcium and aluminum are chill reducers.

The possibility of calcium additions producing a nucle-
ation is indicated by the decrease in cell size accompanying

1’9’20 has discussed

an addition of this element. Boyles
the relationship between cell size, rate of development of
the cells, and graphite distribution, and it is possible
that the sudden formation of a large number of centers of
crystallization of the eutectic by nucleation would in influ-
ence the rate of solidification and thereby the manner of for-
mation and distribution of graphite.

. It is conceivable that colloidally dispersed calcium
oxide would serve as a nucleus for initiating cell growth or
flake development. Another possibility, which deserves con-
sideration, is the existence of calcium.carbide or a complex
carbide containing calcium in molten iron treated with calcium
or calcium alloys. Calcium carbide, because of certain
structural considerations, would be ideally constituted for
nucleation of graphite flakes, and possibly for nucleation of
eutectic cells.

As previously mentioned the writer and his associates

have observed a decarburizing action of heavy calcium additions.
This can be taken as an indication that a carbide of low solu-

bility is produced in molten iron. Examination of the ladles

in which calcium additions or heavy calcium silicon additions

have been made discloses a gray compound on the top of the
ninterior surface of the ladle which resembles the familiar
"calcium carbide" and which evolves acetylene gas to such
an extent that the odor is readily detected and recognized
in the foundry where calcium-silicon additions are being
made. Finally, a freshly fractured gray iron casting, which
has been inoculated with calcium silicon, gives off a faint
but unmistakeable odor of acetylene, the gas evolved when
calcium carbide is moistened. These observations suggest
the formation of calcium carbide in the iron.

Wells3n has discussed the structure of metallic car-
bides. Three types are recognized:

l. Interstitial - Ti, Zr, V, Mo, Ta, W

2. Transition - Cr, Mn, Fe, Co, Ni

3. Electropositive metal carbides, similar to ionic
crystals, and including carbides of Mg, A1, and Ca.

It is significant that magnesium, aluminum, and calcium,
three metals known to be effective in influencing gray iron
behavior, are grouped in one class. The ionic crystal nature
of the carbides of these elements suggests a high stability,
and a low solubility in ferrous melts.

wells has pictured the structure of calcium carbide,
CaCz, as shown in Fig 18. The unbalanced structure of a grow-
ing crystallite of calcium carbide might be ideally suited for
adsorption at interfaces between graphite and solid metal, and
between graphite and liquid iron with the non-metallic or

76A

 

Fig 18. Representation of Structure of Calcium.Carbide.
After Hells.3l+

77

graphite rich face on the graphite phase, and the pronounced-

ly metallic face of the crystallite oriented towards the metal
phase. Such absorption would promote formation and growth of

the graphite flakes by lowering the surface energy. Wells

has suggested that the structure of calcium carbide is unique

as it is the only carbide which evolves acetylene.

Hughes35 has suggested that adsorption occurs in the
formation of nodules in.magnesium treated irons but does not
identify the adsorbed substance.

Buttner, Taylor and W01f36 have observed a difference in
the tendency of magnesium treated nodular irons in their a-
bility to wet graphite ladles, as opposed to untreated iron
which do not wet such ladles, and have suggested an absorp-
tion phenomena.

The existence of a new carbide phase, such as calcium
carbide, in the alloy system would be expected to alter the
phase diagram and the time-temperature curves for the alloy
system. Such changes in the time-temperature curves accom-
panying inoculation have been observed as a shifting of the
eutectic arrest to a higher temperature, but this change has
been attributed to a nucleation effect which prevents under-
cooling. The writer has discussed discrepancies in this con-
ception at some length (refer to page 21) and has pointed out
that the displacement of the time-temperature curve is not
necessarily a matter of undercooling but may be due to other

causes e

The existence of a new carbide either simple calcium
carbide or a more complex carbide may well account for the
observed displacement of the time-temperature curves to a
higher temperature. Hume-Rothsry37 has discussed the ele-
vation of the eutectic temperature by the formation of inter-
mediate phases of increasing stability in a series of alloys.
The possibility of the active metals influencing the phase
diagram.and the time-temperature curves deserves considera-
tion and further investigation. However, one serious ob-
Jection may be advanced to this hypothesis - the extremely
small amount of calcium and aluminum contained in the vari-
ous grades of ferro-silicon which have some value as inocu-
lants. The data of page #8 on the effectiveness of low and
high alloy ferro-silicons shows that the high active metal
ferro-silicon (2.20 Al and .#6% Ca) has appreciable value as
an inoculant. With an addition of the inoculating alloy of
.557, Which contains a total active metal addition of only
.0138% to the iron, it is apparent that the amounts of a1-
uminum and particularly of calcium involved are extremely
small, and certainly this raises some question about the
ability of such small amounts to produce a perceptible
change in the time-temperature curves. The alterations in
the time-temperature curves reported by Eash2 (see Fig 7)
were produced by ferro-silicon of unstated composition.
Nevertheless, we are compelled to attribute the effectiveness

of ferro-silicon to the contained active metals in the ab-

79

rsence of any effect of additions of purified silicon or of
a silicon alloy relatively low in active metal content.

In completing the discussion of these experiments
attention should be called to the low values obtained for
the transverse strength of certain of the blank irons of the
data reported by Khan?8 Referring to page 53, Tables 15 and
16, it is found that iron T12 Bl-5 shows a transverse
strength of 2202 pounds, which is far below the average for
similar untreated irons of this report. Other irons of
Khan's report show similar low values and render some of
the data of his report questionable, particularly when a
direct comparison of calcium treated ladles with ladles from
the same heat was not possible because of the decarburizing
action of the larger calcium additions. Irregularities of
properties of these irons can be attributed to variations
in temperature and time at temperature, and to the possi-
bility of a critical temperature at which there is a sharp
drop in properties and a sudden increase in the degree of
abnormality of the irons.

That uninoculated irons are likely to show variations
in properties which cannot be explained on the basis of com-
position is evident from other reports in the literature.
Adams38 has made a survey of the correlation of micro-
structure, analysis, and properties of commercially pro-
duced irons and has reported a wide variation in properties

for irons of inappreciable variation in microstructure and

80

composition. As an example data from his report is given
for two irons, each of which show D graphite distribution
at the surface, in Table 32 below:

Table 32
Iron C Si Trans
100 2.83 2.13 2710
210 ‘ 2.86 2.15 2215

Other data in the Adams report show similar wide varia-
tions. In the case of the irons described above the differ-
ence in properties was ascribed to a small variation in flake
size at the immediate surface of the casting.

That superheating is detrimental to the properties of
gray iron has long been understood. This matter has been in-
vestigated by Seffing and Surlss, Schneble and Chipman,6 and
Lorig.3o Although the conclusion that superheating is detri-
mental is accepted by gray iron metallurgists generally,
exact information regarding this phenomena is lacking because
of the failure of the several investigators to control the
composition changes in carbon and silicon which accompany
superheating. In the report by Lorig the variations in car-
bon content exceed .20% and in that by Schneble and Chipman
the variation is as much as .37% C. These variations in
compositions make it difficult to interpret the data, and
these papers have been critisized extensively in the liter-

81

ature. More exact data is required on this important sub-
Ject, and better means of control of composition and tempera-
ture in commercial melting units are necessary for facili-
tating gray iron research.

In the work conducted independently by the writer very
careful attention to time and temperature control was given
with the degree of success evident in the data of Table 2,
page 31. As previously noted, the uninoculated irons are
particularly eratic in their chilling tendency.

The irons melted in the experiments reported by Khan
were produced with different pig irons than those employed
for the irons of this report. This may account in part for
the tendency towards low values for the blank irons of the
previous report. Morrogh and Williams3 have ascribed such
variations in tendency to produce D graphite in pig irons to
variations in titanium content.

In this work the effect of the inoculating substance
could have made more spectacular by reducing the properties
of the blank irons by higher superheating temperatures and
longer times at temperature. However, the writer deemed it
advisable to conform more closely to commercial practice.
High heating temperatures promote composition changes, ir-
regularities in pr0perties, and undue lining wear, and the
advantages of the high temperatures have yet to be demon-

strated.

82

The adverse effects of superheating have been attributed
to l) the solution or distruction of nuclei, and 2) to an in-
crease in the amount of dissolved oxygen, hydrogen, and nitro-
gen. Since inoculants are capable of correcting the bad in-
fluence of superheating on.microstructure and properties, any
successful theory of inoculation must be related to the effects
of superheating. Boylesl has claimed that a certain Optimum
amount of hydrogen is essential to the formation of normal
structures in iron, and many others have discussed the possi-
ble effects of oxidation. The formation of oxides,nitrides,
and hydrides of calcium and other metals, and their possible
influence on behavior provide interesting material for Specu-

lation.

V Summary and Conclusions

A survey of the literature has revealed that silicon
is regarded as the essential constituent of gray iron inocu-
lants, that no satisfactory theory for the inoculation mechan-
ism has been prOposed, and that comparitive data on the re-
lative effectiveness of the various commercial inoculants are
lacking. Accordingly a general investigation has been under-
taken in which commercial inoculants and the elements occuring
in these alloys have been compared as to their effectiveness
as a step in the direction of formulating a theory and of pro-
viding data of practical value regarding the most effective in-
oculating proceedure.

This investigation has indicated that the element silicon
is not effective as an inoculant. Data has been acquired which
shows that the inoculating power of alloys increases with the
aluminum and calcium contents. It has been shown that alum-
inum is particularly effective in reducing chilling tendency,
and that calcium is effective both in decreasing chilling ten-
dency and improving graphite distribution and the mechanical
properties of high strength gray iron.

The most effective of the commercial inoculants has been
established as calcium-silicon. The advisability of using
larger quantities of the inoculating alloys than is generally
recommended is demonstrated.

Observations have been made which indicate that calcium

forms a carbide in molten iron. The literature has disclosed

that calcium is unique among the elements found in inoculat-
ing alloys in two reapects: 1) its free energy of oxide for-
mation under standard conditions is the lowest of the several
elements and 2) its carbide has a structure believed by
chemists to be different from that of other active metal car-
bides. The possibility of formulating theories on the basis
of deoxidation and of carbide formation is discussed.

Specific conclusions and observations from this report
are:

l. The element silicon is not effective as an inocula-
ting agent.

2. Silico manganese, which is relatively free of calcium
and aluminum, is not effective as an inoculant.

3. The inoculating ability of various commercial grades
of ferro-silicon increases with the calcium and aluminum con-
tents. Low calcium grades have very little effect.

#. Silicon-manganese-zirconium alloys containing some-
what more calcium than is found in commercial ferro-silicon
are more effective than ferro-silicon.

5. The most effective inoculant employed in these ex-
periments is the calcium-silicon alloy with approximately
30% of calcium.

6. Although aluminum is a very powerful chill reducer,
its addition to cast iron does not promote type A graphite
distribution.

7. Metallic calcium additions to the ladle bring about

marked improvements in properties and graphite distribution

of cast iron. Calcium is superior to all other inoculating
agents tested in these experiments, with the exception of
calcium-silicon, when added in the amount of approximately
.H%.

8. Metallic calcium is effective in reducing the chill-
ing tendency of cast iron.

9. Metallic calcium is effective in reducing the cell
size of cast iron microstructures.

10. Calcium additions effect a reduction of the carbon

content of molten iron.

11. Acid and basic melting conditions are equivalent in

their effects on gray iron melts.

1.

2.

3.

4.

5.

Bibliograpm

A. Boyles "The Freezing of Cast Iron” Transactions
American Institute of Mining and-Metallurgical Engi-
neers Vol 125, 1937, PP 141-203

J. T. Eash “Effect of Ladle Inoculation on the
Solidification of Gray-Cast Iron. Transactions
American Foundrymen's Association Vol 49, 1942
PP 887-910 ,

H. Morrogh and W. J. Williams “Graphite Formation
in Gray Cast Iron" ' Foundry Trade Journal V 82
Aug 21, 1947 '

W. 8. Owen and B. G. Street "A Note on the Formation
Of Undercooled Graphite in High Purity Iron-Carbon-
Silicon.Alloys' Journal of the Iron and Steel
Institute Vol 171 July 1952 pp 237-238

11. Surls and F. Sefing "The Properties of Gray
Iron Castings as Effected by Superheating Temper-
atures" Bulletin No. 65 Michigan Engineering Experi-
ment Station, Michigan State College, East Lansing,
Michigan No. 1935 .

6.

8.

II..-14-.

A. W. Schneble and J. Chipman "Factors Involved in
Superheating Gray Cast Iron and Their Effects on Its
Structure and Properties“ Transactions of the ‘
American Foundrymen's Association Vol 52 Sept 1944
PP 113-172 ”

A. I. Krynitsky and C. M. Saeger "Effect of Super-
heating, Pouring Temperature and Microstructure on
the Elastic PrOperties of Some Plain and Alloy Cast
Irons” Transactions of the American Foundrymen's

Association V01 50 Dec 1942 P]? 451-489

C. A. Timmons and V. A. Crosby ”Effect of Pouring
Temperature on the Strength andMicrOstructure of
Gray Cast Iron“ Transactions Of the American
Foundrymen” Association V01 49 Dec 1941

PP 225 - 254 '

A. Boyles ”Formation of Graphite in Gray Iron"
Transact ions of the American Foundrymen's‘AssO-

ciation V 46 Dec 1938 PP 297-340

10. R. W. Heine "Some Effects of Deoxidising Additions

on Foundry Malleable Irons" Transactions of
American Foundrymen'sSociety V 58, 1950
PP 277-297 '

11.

12.

13.

14.

15.

16.

17.

R. W. Heine "Oxidation-Reduction Principles
Controlling the Composition of Molten Cast Irons“
Transactions of the American Foundrymen's Society

V 59, 1951 PP 121-138

B. B. Bach 'Theoretical Aspects of Oxygen in
Cast Iron' American Poundrynen's Society
Preprint No. 54 - e, 1954 ' *

E. A. Loris and H. W. Lonnie I'The Residual
Oxygen Content of Cupola llelted Gray Cast Iron" _
Transactions of American Society for Metals

V 46, 1954 PP 409-417

J. W. Dawson, Le V. Smith and Be Be BSOh
”Sons Effects of Nitrogen in Cast Iron"
American Foundrymen v 26 July 1954 P 60.

C. D. D'Anico and R. Schneidewind “The

Solidification and Graphitisat ion of Gray
Iron" Transactions of the American Foundry-
men's Association. v 48 June 1941 913 775-803

H. 9R. Dahlberg "Graphite Particle Size in
Inoculation" Foundry V 71 Aug 1943 P 125

V. Crosby, Climax Molybdenum Co. Alloy Cast
Irons Handbook 1944 Edition, Chap III Page 65

18.

19.

20.

21.

22.

23.

24.

25.

E. W. Lownie “Theories of Gray Cast Iron Inocula-
tion'I Transactions of the American Foundrymen's

Association V 54, 1946 PP 837-844

L. Creme "Opinions on Graphitization' Foundry
V01 75, Dec 1945 p 102 '

A. Boyles and C. H. Lorig “Notes on the Under-
cooling of Gray Cast.Ir0n' Transactions of the
American Foundrymen's'Association V 49, 1941
p 769 -

0. Von Keil Archiv f’u'r Das Eisenhuttensesen
v 7, 1954 P 579 ‘

E. Piwovarski “Progress in production of High
Test Cast Iron" Transactions of the American
Foundrymen's Association v 54, 1926 P 914

F. Pirovarski, Archiv fur das Eisenhuttenwesen
V '7 1953 P 431

E. Diepschlag, Die Giesserei V 16, 1929

p 822, v 18, 1951 P 705; v 24, 1954 p 495

R. Flinn and D. Reese " The Development and
Control of EngineeringGray Cast Irons"
Transactions of the American Foundrymen's

Association v 49 March 1942 pp 559-607 ’

In

26. 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 v 155, 1947 p 521 '

27. N. McClure "Ferro-silicon, Calcium-silicon, and
Metallic Calcium as Ladle Inoculants for Gray
Iron' Thesis, Michigan State College, 1949
Department of Mechanical Engineering

28. A. U. Khan ”The Effects of Ladle Additions of
Active Metals on the Physical Properties, Micro-
structure, and Chemical Composition of Cast Iron".
Thesis, Michigan State College, Department of
Metallurgical Engineering. ‘

26.

27.

28.

29.

300

31.

32.

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 V 155, 1947 P 321.
N. McClure "Ferro-silicon, Calcium-silicon, and Metallic
Calcium as Ladle Inoculants for Gray Iron" Thesis,
Michigan State College, 1999 Department of Mechanical
Engineering.

A. U. Khan "The Effects of Ladle Additions of Active Metals
on the Physical Properties, Microstructure, and Chemical
Composition of Cast Iron." Thesis, Michigan State College,
Department of Metallurgical Engineering.

B. McElwee and T. Barlow "Structure Control of Gray Cast
Iron" Foundry V 72, Feb. 1999, P 113.

C. Lorig. "Gray Iron Metallurgical Practice." Foundry

v 67, March 1939, P 26. ’

H. w. Lonnie. "Ladle Inoculation Improves Gray Iron Pro-
perties and Structure." Foundry V 71, Nov 1993 and Dec
1993.

C. Burgess and R. Bishop "Alterations in Cast Iron Pro-
perties Accompanying Use of a Strong Inoculant of the
Silicon-Manganese-Zirconium Type." Transactions of the
American Foundrymen's Association V 52, 199%, P 671.

F. Richardson and J. Jeffes "The Thermodynamics of Sub-
stances of Interest in Iron and Steel Making from 0°C to
2900°c." Journal of the Iron and Steel Institute 7 16,
Nov. 1998 P 261.

3".“

35-

36s

37-

38-

A. F. Wells "Structural Inorganic Chemistry." 1950,
Oxford.

I.C.H. Hughes "Some Effects of Magnesium on the Forma-
tion of Graphite in a Solidifying Cast Iron." Foundry
Trade Journal v 93, Sept 25, 1952 P 399.

F.Buttner, H. Taylor, J. welf "Graphite Nodules; a Note
on Spherulization of Graphite in Cast Iron." American
Foundryman V 20, Oct. 1951, P #9.

w;HMme-Rothery "The Structure of Metals and Alloys,"
Published by The Institute of Metals, London, 1936.

R. R. Adams "Cast Iron Strength Versus Structure"
Transactions of the American Foundrymen's Association

v 50, June 1943, P 1063.

4337

"II
II"
"I
S"
"
llI
"
”I
u
H

3 1293 03178