High Temperature Phase Chemistries and Solidification Mode

Prediction in Nitrogen-Strengthened Austenitic Stainless Steels

ANN M. RITTER, MICHAEL F. HENRY, and WARREN E SAVAGE

Nitronic 50 and Nitronic 50W, two nitrogen-strengthened stainless steels, were heat treated over a
wide range of temperatures, and the compositions of the ferrite and austenite at each temperature were
measured with analytical electron microscopy techniques. The compositional data were used to
generate the (y + 6) phase field on a 58 pct Fe vertical section. Volume fractions of ferrite and
austenite were calculated from phase chemistries and compared with volume fractions determined
from optical micrographs. Weld solidification modes were predicted by reference to the Cr and Ni
contents of each alloy, and the results were compared with predictions based on the ratios of calculated
Cr and Ni equivalents for the alloys. Nitronic 50, which contained ferrite and austenite at the solidus
temperature of 1370 ~ solidified through the eutectic triangle, and the weld microstructure was
similar to that of austenitic-ferritic solidification. Nitronic 50W was totally ferritic at 1340 ~ and
solidified as primary delta ferrite. During heat treatments, Nitronic 50 and Nitronic 50W precipitated
secondary phases, notably Z-phase (NbCrN), sigma phase, and stringered phases rich in Mn and Cr.

I. INTRODUCTION II. EXPERIMENTAL PROCEDURE

NITRONIC 50 is a nitrogen-strengthened austenitic stain- Two heats of material were used for this study, one of
less steel, developed as a substitute for 304 and 316-type Nitronic 50 (N50) and one of Nitronic 50W (N50W). The
stainless steels in applications where strength and corro- composition of the N50 alloy conformed to specifications
sion resistance were of primary importance.~ Mechanical for ASTM XM-19, which require a maximum carbon level
property data on Nitronic 50 (originally designated 22-13-5 of 0.045 pct, and the composition of the N50W alloy con-
and termed XM-19 by ASTM) have been compiled by sev- formed to a General Electric Company specification. The
eral authors, 2'3'4 and recent work by Espy 5 dealt with the weld filler alloy, N50W, has increased Cr and Mn and de-
weldability of Nitronic 50 and Nitronic 50W, the latter a creased Ni, relative to N50. The alloys were vacuum induc-
matching weld filler metal. tion melted, and poured under 1 atmosphere of nitrogen.
One measure of the weldability of a stainless steel is its The compositions of the heats are given in Table I. Differ-
tendency to hot crack, which is influenced by the amount 6 ential thermal analysis (DTA) curves obtained on both alloys
and distribution 7 of delta ferrite in the weld. These, in turn, were used to measure the liquidus and solidus temperatures.
are related to the solidification mode of the particular stain- The samples used for the phase study were 0.010 inch
less steel, and recent papers 8-H have sought to characterize thick and contained about 84 pct cold work. They were heat
the solidification of 300-series stainless steels. In general, it treated in a box furnace under an argon atmosphere, and
has been found that welds which solidify as primary delta- with the exception of a few low temperature ages, were heat
ferrite are more resistant to hot cracking than welds which treated in the cold-worked condition. The sample coupons
solidify as primary austenite. 8'~2'Z-~ were heat treated in a holder made from 0.25-inch thick
Few data exist on high temperature phases in Nitronic 50 plates of Nitronic 50, with the inside of the holder dusted
and Nitronic 50W although some characterization of phases with alumina powder to prevent sticking of the samples.
formed in the temperature range of 600 to 1150 ~ has been For quenching, the holder was removed from the furnace
done.14 The solidification modes of these alloys have not and tilted to let the sample coupons fall into an ice brine
been established, and predictions of solidification modes bath. The steel plates decreased radiation losses from the
using the Fe-Cr-Ni ternary diagrams, as has been done for coupons immediately before quenching, and the small
stainless steels of near-ternary compositions, ~5~J6 could sample thickness allowed a quench rapid enough to retain
be inaccurate because of the other alloying elements. The the high temperature phases. High temperature ages ranged
present investigation considers the chemistry and dis- from 1100 ~ to 1380 ~ generally at 10 to 20 ~ intervals
tribution of major and minor phases up to the solidus with times at temperatures from 1 hour to 10 minutes.
temperatures of both alloys, and relates these data to the Repeat runs at several temperatures were made. Speci-
solidification modes experienced during welding. mens aged at 800 ~ hr/WQ, 900 ~ hr/WQ, and
1000 ~ hr/WQ were vacuum encapsulated prior to heat
treatment, since the cooling rate from these temperatures
was not judged to be as critical. These low temperature heat
ANN M. RITTER and MICHAEL F. HENRY are Staff Metallurgists, treatments were also done on specimens which had been
General Electric Corporate Research and Development, Schenectady, NY solution-annealed at 1120 ~ for 30 minutes.
12301. WARREN F. SAVAGE is Professor and Director of Welding
Research, Department of Materials Engineering, Rensselaer Polytechnic
Optical microscopy was done on all samples. Samples
Institute, Troy, NY 12180. were etched in a solution of 10 ml acetic acid/10 ml nitric
Manuscript submitted December 21, 1982. acid/15 ml hydrochloric acid/2 drops of glycerine, and

METALLURGICAL TRANSACTIONS A VOLUME 15A, JULY 1984-- 1339

 Table I. Alloy Compositions
Alloy Cr Fe Ni Mo Mn C N Nb V Si S P
N50W 23.02 58.24 11.33 2.00 6.22 0.04 0.32 0.24 0.2 0.02 0.006 0.005
N50 21.32 59.51 12.42 2.05 5.18 0.009 0.30 0.23 0.2 0.08 0.006 0.005

etching was done by immersion. Samples for transmission

electron microscopy (TEM) were ground to 0.005 inch
thick and electropolished in a 20 pct perchloric acid/80 pct
methanol solution. The foils were examined in a JEOL
JSEM-200 scanning-transmission electron microscope
(STEM) equipped with an energy dispersive spectrometer
(EDS).
To calculate phase chemistry at any temperature, five
spectra were taken in the center of each grain analyzed, and
the mean composition (IX), standard deviation (s), and rela-
tive scatter (s/tx) determined. Three to five grains of both
ferrite and austenite were sampled in each foil. In specimens
containing sigma, individual readings were taken from 6 to
10 different sigma precipitates and in different austenite
grains, and average compositions obtained.
Integrated peak intensities for Cr, Ni, Fe, Mn, and Mo
were used to obtain elemental concentrations in weight per- (a)
cents by first normalizing the value of integrated intensity
for each element on the sum of integrated intensities of all
five peaks. Each normalized intensity was then multiplied
by an appropriate constant to obtain the concentration of
each element. The constants were determined from spectra
collected on a fully austenitic N50 sample and a fully ferritic
N50W sample. The compositions of these samples were
assumed to be the alloy compositions measured by X-ray
fluorescence. The details of the normalized intensity com-
putation method have been described elsewhere. 17
For samples heat treated at several temperatures, vol-
ume fractions of austenite and ferrite were determined
from light optical micrographs using both an areal sum-
mation (perimeter tracing) method and a systematic point
count method. 18

III. RESULTS
A. Microstructural Features of N50W and N50
Examination of optical micrographs from Nitronic 50W (b)
heat treated at 1100 ~ hr disclosed bands of fine ferrite Fig. 1 - - N 5 0 W , ll00 ~ h. (a) Bands of ferrite (8) and stringer phases.
particles (Figure l(a)). The banding occurred as a result of (b) Small intragranular Z-phase precipitates (arrows) in austenite matrix.
alignment of ferrite, present in the casting, with the rolling
direction. Growth of this original ferrite took place during
heat treatment. Stringers of large and small precipitates were Z-phase was present in this particular heat of N50W because
also observed. Energy dispersive spectra on these precipi- the specifications required Nb.
tates in thin foils found them to contain primarily Cr and After heat treating at 1150 ~ min the amount of fer-
Mn, with the ratio of these elements varying from phase to rite increased significantly, although it still tended to be
phase. Small amounts of V, and occasionally Ti, Nb, and concentrated in bands. Ferrite nucleated not only at austenite
Mo were also detected in these phases. A third type of grain boundaries, but both intragranularly and at twin
precipitate was observed in the thin foils. It had a rod-like boundaries. Annealing at 1180 ~ rain resulted in some
morphology when viewed in certain orientations, and a cir- coarsening of the ferrite. The rod-shaped precipitates were
cular cross-section in other orientations (Figure l(b)). These not observed, since the Z-phase coarsened at the higher
rod-shaped precipitates were identified as Z-phase, a com- temperatures and assumed a spherical morphology. At
plex nitride. 14.19,20Spectra showed this phase to be rich in Nb 1220 ~ min, the quantity of ferrite had increased, the
and Cr, with small amounts of V, Fe, and Mo. The stoichi- banding was no longer evident, and impingement of the
ometry of this phase has been approximated as NbCrN. 2~ growing ferrite was frequently observed.

1340-- VOLUME 15A, JULY 1984 METALLURGICAL TRANSACTIONS A

 ture, except that no ferrite was present in the lower-Cr,
higher-Ni Nitronic 50. Stringers of large and interme-
diate sized phases were seen (Figure 3(a)), and small rod-
shaped Z-phase precipitates were observed in the thin foils
(Figure 3(b)). Growth of the austenite grains occurred at
higher temperatures. Grains were 100 to 200/xm in size
after heat treating at 1250 ~ min. In contrast, in N50W
at 1260 ~ the largest grains were only 50/xm in diameter,
since growth of the austenite grains was inhibited by nucle-
ation and growth of the ferrite at the grain boundaries.
At 1260 ~ min, the first ferrite was observed in
Nitronic 50, primarily at austenite grain boundaries, and at
1300 ~ min, the ferrite covered most of the austenite
grain boundaries, with occasional intragranular ferrite
(Figure 4(a)). A large increase in the amount of ferrite oc-
curred on annealing at 1320 ~ min (Figure 4(b)), and
gradual coarsening and coalescence of the ferrite were
(a) observed at higher temperatures. At 1350 ~ rain, the
ferrite comprised half of the microstructure. Some increase

(a)
(b)
Fig. 2--N50W, 1280 ~ min. (a) Ferrite in dark-etching austenite.
(b) TEM micrograph showing straight-sided intragranular ferrite particles.

After heat treating at 1280 ~ min, the ferrite com-

pletely outlined the austenite grains (Figure 2(a)). It was
observed that, although the y/6 interface of grain boundary
ferrite was usually curved, intragranular ferrite was straight-
sided in some orientations (Figure 2(b)), suggesting a habit
relationship with the parent austenite. By 1280 ~ the fer-
rite was the continuous phase. At this temperature, the
stringer phases had decreased in quantity so that they were
rarely seen. A few of the intragranular phases persisted to
1320 ~ but after heat treating at 1330 ~ rain, they
were not observed. A small amount of austenite was present
in the 1330 ~ specimen, and at 1340 ~ the sample trans-
formed entirely to ferrite, with a microstructure consisting
of large (500 to 1000/xm) ferrite grains. This micro-
structural study has thus shown that the high-Cr, low-Ni
Nitronic 50W is fully ferritic at temperatures - 3 0 ~ below
the solidus, determined by DTA to be 1370 ~ (b)
The microstructure of Nitronic 50 annealed at 1100 ~ Fig. 3--N50, ll00~ h. (a) Austenite grains and stringer phases.
1 hr resembled that of Nitronic 50W at the same tempera- (b) Rod-shaped Z-phase (arrow) in austenite.

METALLURGICALTRANSACTIONSA VOLUME 15A, JULY 1984-- 1341

 (a) (b)
Fig. 4 - - N 5 0 . (a) 1300 ~ min. Ferrite at austenite/austenite grain boundaries. (b) 1320 ~ min. Ferrite in dark-etching austenite.

(a) (b)
Fig. 5 - - N 5 0 , 1380 ~ min. (a) Austenite in continuous ferrite matrix. Incipient melting at boundaries (arrows) appears also in (b) where liquid pool
has transformed.

in the amount of ferrite was observed up to the solidus 33 wt pct Cr, 5 wt pct Mn, 5 wt pct Ni, and 50 wt pct Fe.
temperature of - 1367 ~ together with evidence of incipi- In N50, the composition of the sigma was similar, with 6 to
ent melting at a few grain boundaries. A sample heat treated 7 wt pct Mo, 32 to 33 wt pct Cr, 4 wt pct Mn, 5 to 6 wt
at 1380 ~ rain exhibited large amounts of incipient pct Ni, and 52 wt pct Fe. More details on the formation of
melting at grain boundaries and especially near the areas of sigma in these alloys are given in Reference 17.
austenite (Figure 5(a)). Examination of what had been a The partitioning of elements to the ferrite and austenite
pool of liquid at 1380 ~ (Figure 5(b)) showed a lathy or was of interest, and the data from a range of samples are
Widmanst~itten-type structure. Dark-field images of this re- summarized in Table II. In both heats, ferrite which formed
gion (Figure 6) showed the distribution of lathy ferrite and at the lower temperatures, where the volume fraction of
austenite within an area of ferrite. This study thus has shown ferrite was low, was enriched in Cr and Mo and depleted
that the lower-Cr, higher-Ni Nitronic 50 does not transform in Ni relative to the alloy compositions. At higher tem-
completely to ferrite before melting begins. peratures, where the volume fraction of the austenite was
Sigma phase formed in both Nitronic 50 and Nitronic low, the austenite was depleted in Cr and Mo and enriched
50W in the temperature range investigated (800 to 1000 ~ in Ni, relative to the alloy compositions.
It was observed in samples aged in the cold-worked condi- Some partitioning of Mo to the ferrite was observed, and
tion and in samples annealed at 1120 ~ min prior to the austenite was most depleted in Mo at temperatures near
aging. Sigma in N50W contained 4 to 6 wt pct Mo, 32 to the solidus. The Mn content of the austenite tended to be

1342--VOLUME 15A, JULY 1984 METALLURGICAL TRANSACTIONS A

Fig. 6--N50, 1380 ~ min. (a) Bright-field TEM micrograph, lathy ferrite and austenite, (b) dark-field image using ferrite reflection, and (c) dark-field
image using austenite reflection.

Table II. Partitioning of Elements

Alloy Temperature Phase Cr Ni Fe Mo Mn

N50W 1100 ~ ferrite 28.7 wt pct 6.3 wt pct 55.7 wt pct 2.5 wt pct 5.9 wt pct
austenite 22.9 11.3 58.2 2.0 6.2
1180 ~ ferrite 27.2 7.6 56.0 2.4 6.0
austenite 23.3 11.4 58.3 1.5 6.8
1240 ~ ferrite 25.5 8.7 57.4 2.4 6.2
austenite 21.8 12.2 58.9 1.4 6.7
1280 ~ ferrite 24.8 9.4 57.6 2.2 6.2
austenite 21.8 12.2 59.0 1.3 6.9
1310 ~ ferrite 23.5 10.8 58.0 2.4 6.4
austenite 20.0 14.2 59.5 1.8 6.8
1320 ~ ferrite 22.9 11.2 58.5 2.1 6.2
austenite 20.1 13.6 59.8 1.3 6.5
1330 ~ ferrite 22.9 11.4 58.3 2.0 6.6
austenite 19.8 14.6 59.4 1.3 6.6
N50 1260 ~ ferrite 24.3 10.1 57.9 2.5 5.0
austenite 21.3 12.4 59.1 1.9 5.5
1300 ~ ferrite 23.7 10.5 58.3 2.3 5.0
austenite 21.1 12.5 59.7 1.8 5.3
1340 ~ ferrite 23.0 10.6 59.1 2.5 4.7
austenite 19.5 14.0 60.4 1.5 5.5
1370 ~ ferrite 21.8 11.9 59.4 2.1 5.1
austenite 19.3 14.3 60.4 1.4 5.5
1380 ~ ferrite 21.5 12.5 59.2 2.0 5.1
austenite 19.2 14.8 60.2 1.2 5.5

slightly higher than that of the ferrite. No strong influence performed on N50W and N50. Since N50W and N50 are
of temperature on the Mn content o f the phases was ob- complex Fe-Cr-Ni alloys with about 8 pct of other elements,
served. In N50 and N50W, ferrite at all temperatures con- the Cr and Ni contents of the coexistent phases cannot be
tained less Fe than did the austenite. represented on a single constant Fe section.
Figure 7 shows the Cr concentrations of the coexistent
ferrite and austenite in N 5 0 W as a function of temperature,
B. The (y + 6) Phase Field
together with the liquidus and solidus temperatures of the
Figures 7 to 10 summarize the results of the micro- alloy. A portion of the ( y + ~) phase field has also been
chemical analysis and differential thermal analysis studies determined, using compositional data from samples heat

METALLURGICALTRANSACTIONSA VOLUME 15A,JULY 1984--1343

 0 0
0 ED
! i I 1 i
Go

o
g g I 1.3 Ni
U3
=d23.0 Cr

L) U
OC~~1uldus_
iq __ g
,,r
_ li~uidus
*-4
solidus solidus
%. W

k
[]
V
g. "~ ,~ g
~J W
cr O~
g H-
g
o ~ !
/., ~'+o I O~
W I

'I:
&
L lw lz
O_
0o
Ld
1r GAMMA
li GAMMA
DELTA o
o ~ELTA
SIGMA SG
I MA
* I i 9 J I l
g I ~ I i 1

0 i0 20 30 40 40 30 20 10 0

CR CONTENT (WT/O) NI CONTENT (WT/O)

Fig. 7 - - C r concentrations measured by STEM/EDS on ferrite, austenite, Fig. 8 - - N i concentrations measured by STEM/EDS on ferrite, austenite,
and sigma in N50W specimens heat treated in furnace. Alloy Cr concen- and sigma in N50W specimens heat treated in furnace. Alloy Ni concen-
tration indicated. tration indicated.

o
treated in the cold-worked condition and from those an- o ! " I

nealed prior to low temperature aging. The (y + o-) phase

boundary determined from the cold-worked plus aged sam-

~
ples is indicated by the legend "CW". Similar (y + 6) and g 21.4 Cr
(7 + 03 phase fields appear in the plot of Ni concentrations (D
in these phases (Figure 8). In both Figures 7 and 8, only
U
the (y + o') field is labeled at low temperatures. Alloys
of a higher Cr concentration are needed to determine the 8 licluidus
o'/6 boundary. dZ solidus
Data from N50 furnace samples were used to generate Cr Iii
[]
and Ni compositions for the (3' + 6) phase field boundaries g
for that alloy (Figures 9 and 10). The (3/+ 6) phase field for N
N50, determined from Cr and Ni concentrations, is similar W
to that of N50W. Figure 11 presents all of the Cr concentra- c~
tions from both alloys for comparison. The (3' + o') phase 2>o
1

fields for the two alloys are not identical. However, this
study has not established the equilibrium volume fractions 1r '7+0
of austenite and sigma, and the differences in the (y + or) W
g
/
phase-fields may reflect non-equilibrium conditions. x

LLJ GAMMA
C. Volume Fraction Measurements p--
g DELTA
Volume fractions of austenite and ferrite were determined (D
from light optical micrographs using both an areal sum- SIGMA
mation method and a systematic point count method9 The
data were then compared with volume fractions calculated I I * , . I I I

from ferrite and austenite Cr and Ni concentrations. Results 0 i0 20 30 40

for N50W are given in Table II1, and those for N50 are CR CONTENT (WT/O)
summarized in Table IV. At a few temperatures, the chem- Fig. 9 - - Cr concentrations measured by STEM/EDS on ferrite, austenite,
istry of the major phase was measured to be identical to and sigma in N50 specimens heat treated in furnace. Alloy Cr concentra-
the alloy composition, and so no volume fractions could tion indicated.
1344--VOLUME 15A, JULY 1984 METALLURGICAL TRANSACTIONS A
 O
O
Q~ , i i
t-q

:t liquidus

solidus
. . . .
i,.,, 12.4 Ni

d
[]
,-I

o liquidus

so,,dus
21.4 c,-.~_I~23"~
,--?:,
W T W ~
t~

<C
rY
i "
& fy x=, x ")'+O
W
CO
~ x "y+O I "'~ m
/L i
W GAMMA
I---
DELTA o DELTA
0
r (D
SIGMA SIGMA
0
i I i I i I I I I i I ! I i
4O 30 20 10 O 0 10 20 30 40

NI CONTENT (WT/O) CR C O N T E N T (WT/O)

Fig. 1 0 - - N i concentrations measured by STEM/EDS on ferrite, aus- Fig. 11 - - C r concentrations measured by STEM/EDS in ferrite, austenite,
tenite, and sigma in N50 specimens heat treated in furnace. Alloy Ni and sigma in N50W and N50 specimens heat treated in furnace.
concentration indicated.

Table III. Volume Fractions for Nitronic 50W

Temperature Technique* Vol Pct Ferrite -+ 2s 2s/lz Vol Pct Austenite + 2s 2s/lz
1320 ~ STEM/EDS, Cr . . . .
STEM/EDS, Ni* 95.8 -+6.7 7.0 pct 4.2 +6.7 159 pet
perimeter 94.9 +2.7 2.8 pct 5.1 +2.7 53.3 pct
point count 95.0 -+2.8 2.9 pct 5.0 +-2.8 56.0 pet
1310 ~ STEM/EDS, Cr 85.8 +3.4 4.0 pct 14.2 +-3.4 23.9 pct
STEM/EDS, Ni 85.5 -+4.7 5.5 pct 14.5 -+4.7 32.4 pet
perimeter 76.0 -+2.0 2.6 pct 24.0 -+2.0 8.5 pct
point count 77.3 +-2.6 3.4 pct 22.7 -+2.6 11.4 pct
1280 ~ STEM/EDS, Cr 43.1 -+4.0 10.0 pct 56.9 +-4.0 7.0 pct
STEM/EDS, Ni 37.7 +-5.7 15.1 pct 62.3 -+5.7 9.1 pet
perimeter 48.7 -+6.4 13.2 pct 51.3 +-6.4 12.5 pct
point count 44.4 +6.8 15.3 pet 55.6 -+6.8 12.2 pet
1240 ~ STEM/EDS, Cr 34.7 -+2.5 7.2 pct 65.3 -+2.5 3.8 pct
STEM/EDS, Ni 29.5 -+4.6 15.6 pct 70.5 +-4.6 6.5 pct
perimeter 22.4 +-8.2 36.6 pct 77.6 -+8:2 10.6 pet
point count 19.2 -+7.8 40.6 pct 80.8 +7.8 9.6 pct
1180 ~ STEM/EDS, Cr . . . .
STEM/EDS, Ni 3.2 +-2.5 78.1 pct 96.8 +-2.5 2.6 pct
perimeter 3.7 +-0.8 22.1 pct 96.3 -+0.8 0.8 pct
point count 2.8 -+0.4 14.3 pct 97.2 +-0.4 0.4 pct
*All STEM/EDS calculations corrected for differences in Fe content, except where indicated by asterisk (*)

be calculated. Values for twice the standard deviation, set of data by the square root of the number of data sets
2s, are also reported in Tables III and IV. For the optical (micrographs). This defined the mean volume fraction, /z,
measurements made using each technique, these values with 95 pct confidence based on precision. Values for 2s/tz
were obtained by dividing the standard deviation for one were also tabulated.

METALLURGICAL TRANSACTIONS A VOLUME 15A, JULY 1984-- 1345

 Table IV. Volume Fractions for Nitronic 50W

Temperature Technique* Vol Pet Ferrite + 2s 2s/tz Vol Pet Austenite -+ 2s 2s/l~
1380 ~ STEM/EDS, Cr 86.0 -_.4.3 5.0 pet 14.0 -+4.3 30.7 pet
STEM/EDS, Ni . . . .
perimeter 84.4 -1.1 1.4 pet 15.8 -+l.1 13.5 pet
point count 87.0 -1.8 2.1 pet 13.0 -+1.8 13.8 pet
1370 ~ STEM/EDS, Cr 81.1 +4.8 5.9 pet 18.9 -+4.8 25.4 pet
STEM/EDS, Ni 80.5 ___6.7 8.3 pet 19.5 -+6.7 34.4 pet
perimeter 74.1 -+4.5 6.0 pet 25.9 +--4.5 17.4 pet
point count 74.9 -+3.0 4.0 pet 25.1 -+3.0 11.9 pet
1340 ~ STEM/EDS, Cr 53.5 ---3.5 6.5 pet 46.5 -+3.5 7.5 pet
STEM/EDS, Ni 49.9 ---4.7 9.4 pet 50.1 -+4.7 9.4 pet
perimeter 41.0 --+2.4 5.8 pet 59.0 -+2.4 4.1 pet
point count 39.3 __+4.6 11.7 pet 60.7 -+4.6 7.6 pet
1300 ~ STEM/EDS, Cr 9.5 -+4.5 47.4 pet 90.5 -+4.5 5.0 pet
STEM/EDS, Ni 9.0 -+4.0 44.4 pet 91.0 -+4.0 4.4 pet
perimeter 3.2 -+0.5 16.0 pet 96.8 -+0.5 0.5 pet
point count 3.1 ---0.7 22.6 pet 96.9 -+0.7 0.7 pet
*All STEM/EDS calculations corrected for differences in Fe content

For the volume fractions determined from phase chem- phases. Initial calculations assumed that the phase chem-
istries, the values of 2s were determined from statistics on istries could be represented on a constant Fe section. In
the STEM/EDS data. Spectra taken in any one ferrite or reality, the N50W (7 + 6) phase field shown in Figures 7
austenite grain yielded mean compositions with a standard and 8 is a projection onto a 58.2 wt pet Fe section. Fe
deviation of 0.22 wt pet Cr. Therefore, with 15 ferrite or concentrations range from 55.7 wt pet in ferrite at 1100 ~
austenite data points at a particular temperature, the 95 pet to 59.4 wt pet in austenite at 1330 ~ Similarly for N50,
confidence level of precision on a mean composition with a bulk concentration of 59.5 wt pet Fe, the ferrite at
would be 2s -- 0.12 wt/Cr. The value for Ni was 2s -- 1260 ~ contained 57.9 wt pet and the austenite at 1380 ~
0.16 wt pet. With this precision, and an average com- contained 60.1 wt pet Fe. The lever arms used to calcu-
position of 23.5 wt pet and 20.0 wt pet Cr in ferrite and late the volume fractions were corrected for the variation
austenite, respectively (N50W at 1310 ~ the minimum in Fe content of the phases, except for one case where
Cr in the austenite would be 19.88 and the maximum the Fe content of the major phase was measured to be
20.12 wt pet. The minimum Cr in ferrite would be 23.38 slightly greater than the alloy Fe level (N50W, 1320 ~
and the maximum 23.62 wt pet. Using these values, the Uncorrected and Fe-corrected volume fractions are shown
smallest calculated amount of austenite was 10.9 vol pet, in Table V. As can be seen, the effect of Fe-correction
and largest was 17.7 vol pet. The smallest calculated is to decrease the difference between volume fractions cal-
amount of ferrite was 82.3 vol pet and largest was 89.1 vol culated from Cr concentrations and those calculated from
pet. Therefore, the variation about the average calculated Ni concentrations.
volume fractions of 85.7 vol pet ferrite and 14.3 vol pet Consideration of the data in Tables III and IV showed that
austenite was 3.4 vol pet, on the basis of precision. Simi- the volume fractions obtained using the optical methods
lar calculations were done for the other Cr and Ni composi- were generally similar, and that the volume fractions
tional data. The resultant values for 2s and for 2s/iz are determined from the Cr concentrations were generally
shown in Tables III and IV. close to those obtained from the Ni concentrations. Some
The accuracy of the phase composition would be affected discrepancies were observed between volume fractions
by the accuracy of the X-ray fluorescence technique used to measured by optical techniques and those obtained from
determine alloy compositions, since these values were used phase chemistries. Possible reasons for these differences
to calculate elemental constants for STEM/EDS mea- and a discussion of the relative precision and accuracy of the
surements. X-ray fluorescence data from NBS standards techniques will be presented later.
were obtained when the N50 and N50W heats were ana-
lyzed. The calculated Cr and Ni concentrations of the NBS
D. Solidification Modes
standards were within about 0.2 wt pet of the certified val-
ues. This is a lower limit on the accuracy in determining the The microstructures of autogenous gas tungsten arc
elemental constants for STEM/EDS. Inaccuracy in these (GTA) welds on N50 and N50W were examined to deter-
constants would alter the measured ferrite and austenite mine the solidification mode of each heat and the amount
compositions. This would tend to translate the (3' + 6) and type of ferrite present. In a weld which solidified as
phase field but not change its width or the relative position primary delta-ferrite, most of the ferrite retained at room
of the alloy composition within the phase field. Thus, some temperature would be located at the dendrite cores ~~
inaccuracy in measuring the alloy Cr and Ni concentrations indicating that the first phase to solidify was delta-ferrite.
would not affect the volume fraction calculations. This solidification mode has been termed ferritic-austenitic
The volume fractions calculated from the STEM/EDS by Suutala et al. ,1o since delta-ferrite forms as the primary
data have been corrected for the differences in Fe in the or leading phase, with subsequent austenite formation

1346--VOLUME 15A, JULY 1984 METALLURGICAL TRANSACTIONS A

 Table V. Comparison of Volume Fractions
Uncorrected Fe-Corrected
Alloy Temperature Element Vol Pct Ferrite Vol Pct Austenite Vol Pct Ferrite Vol Pct Austenite
N50W 1320 ~ Cr . . . .
Ni 95.8 4.2 -- --
1310 ~ Cr 85.7 14.3 85.8 14.2
Ni 85.3 14.7 85.5 14.5
1280 ~ Cr 40.0 60.0 43.1 56.9
Ni 31.1 68.9 37.7 62.3
1240 ~ Cr 32.4 67.6 34.7 65.3
Ni 25.7 74.3 29.5 70.5
1180 ~ Cr . . . .
Ni 2.6 97.4 3.2 96.8
N50 1380 ~ Cr 91.3 8.7 86.0 14.0
Ni . . . .
1370 ~ Cr 80.0 20.0 81.1 18.9
Ni 78.3 21.7 80.5 19.5
1340 ~ Cr 51.4 48.6 53.5 46.5
Ni 46.5 53.5 49.9 50.1
1300 ~ Cr 8.5 91.5 9.5 90.5
Ni 4.0 96.0 9.0 91.0

through a liquid-ferrite-austenite three-phase reaction. By interfaces relative to the primary delta-ferrite/austenite

comparison, primary austenite solidification, also termed interfaces. ~7.2~,22
austenitic-ferritic solidification, ~~ occurs with austenite as These criteria were used to differentiate between the
the first phase to solidify from the liquid. In both modes two types of ferrite in the N50W and N50 welds. In N50W,
of solidification, the last liquid to solidify would be of a most of the ferrite was vermicular in morphology and located
eutectic composition, and ferrite would form from this as at the dendrite cores (Figure 12(a)), resulting in a micro-
a divorced eutectic phase. 2~ This eutectic ferrite would be structure typical of ferritic-austenitic solidification. The
located in interdendritic regions, since it forms during the total volume fraction of ferrite in this weld was measured at
final stages of solidification. The amount of eutectic ferrite about 6 pct, using the systematic point count method, and
present in welds tends to increase as the alloy composition most of this was primary delta-ferrite. A small amount of
approaches the eutectic composition. 21 eutectic ferrite was observed in the interdendritic regions.
Since it is possible for a weld which underwent ferritic- The eutectic ferrite/austenite interfaces were covered with
austenitic solidification to contain both primary delta-ferrite Nb(C, N) precipitates. As can be seen in Figure 12(b), the
and eutectic ferrite, these types of ferrite must be dis- density of coverage was generally much higher than on the
tinguished. This can be done on the basis of position in the primary delta-ferrite/austenite interfaces.
microstructure, solute distribution in austenite near the The microstructure of the GTA weld on Nitronic 50 was
ferrite, ferrite composition, and density of interfacial typical of solidification which had occurred with austenite
precipitation. As has been discussed, primary delta-ferrite as the leading phase. There was about 2 vol pct ferrite
is located at dendrite cores, while eutectic ferrite is found present in the weld, and all of the ferrite appeared to be of
in interdendritic regions. Solute profiles in austenite near the eutectic type, located in the interdendritic regions
eutectic ferrite have been shown by Cieslak et al. 21 and (Figure 13).
Ritter ~7 to be different from profiles adjacent to primary
delta-ferrite. Relative to austenite at the dendrite core, aus-
tenite near eutectic ferrite tends to be enriched in Cr, Ni, and IV. DISCUSSION
Mo, and depleted in Fe. By comparison, austenite adjacent
to primary delta-ferrite is generally somewhat depleted in In this section, the volume fraction measurements will
Ni and only slightly enriched in Cr. These profiles are in- first be considered. Following this, the question of so-
dicative of solute distribution occurring during solidifica- lidification mode prediction from the phase compositions
tion, modified by subsequent solid state diffusion during will be addressed. Predictions based on phase chemistries
cooling of the weld. 2L Because of the rejection of both will be compared with data from calculation of Cr and
ferritizers and austenitizers into the liquid during primary Ni equivalents.
austenite solidification, the eutectic ferrite has been found to
be somewhat enriched in Cr, Mo, and Ni and depleted in Fe
A. Volume F r a c t i o n D et erm i n a t i o n
relative to primary delta-ferrite formed in the same alloy. 22
The solute enrichment in the last regions to solidify also The data presented in Tables III and IV showed that there
results in denser precipitation at the eutectic ferrite/austenite were occasionally discrepancies between volume fractions

METALLURGICAL TRANSACTIONS A VOLUME 15A. JULY 1984-- 1347

 calculated from the Cr and Ni contents of the ferrite and
austenite. Similarly, the volume fractions determined from
the phase chemistries differed from those obtained by op-
tical methods.
The factors influencing the measurement of volume frac-
tions from phase chemistry will be discussed first. The
volume fractions were calculated from the Cr and Ni con-
tents, and the effects of other alloying elements, such as
Mn and Mo, were not taken into account. Since the peaks
from these elements were small, the statistical fluctuations
in measured compositions were large. It was felt that this
would introduce more error into the volume fraction calcu-
lations. Also, the amounts of strong austenitizers like C and
N could not be measured, and so their effect on the volume
fraction calculations could not be considered. It was seen
that the differences in Fe content of the phases influenced
the absolute value of the mean volume fractions. Therefore,
(a) the accuracy of the volume fraction measurements may also
be influenced to some extent by the secondary elements, like
Mn and Mo.
Alteration of phase chemistries by diffusion during cool-
ing of the samples would influence the calculated volume
fractions. In most of the heat-treated samples, the quench
rate was rapid enough to suppress diffusion. This was in-
ferred from the fact that the chemistries of both large and
small grains of ferrite were the same. If appreciable amounts
of diffusion had occurred, the ferrite chemistry would vary
with particle size. In the one specimen of N50 heat treated
at 1380 ~ some diffusion into the liquid pools may have
taken place, altering the chemistry of the adjacent austenite.
The chemistry of the intragranular austenite (reported in
Table II) was different from that of austenite located on
grain boundaries. The chemistry of the intragranular aus-
tenite was used to calculate the volume fractions of ferrite
and austenite at 1380 ~ but since it is uncertain what the
equilibrium austenite composition would be at this tem-
perature, the calculated volume fractions may be inaccurate.
(b) One final factor which would influence the calculation of
volume fractions from phase chemistries is the difficulty in
Fig. 12--(a) Optical micrograph showing primary delta-femte in N50W distinguishing the chemistry of the major phase from the
weld. (b) TEM micrograph of extraction replica from N50W weld. Eutectic
ferrite (arrows) is distinguished by heavy Nb(C, N) precipitation. alloy composition when the amount of the minor phase is
very small. This effect was seen at 1320 ~ and 1180 ~ for
N50W. At both temperatures, the Cr content of the major
phase was identical to the alloy Cr level, and so no volume
fractions could be calculated. The Ni contents of the major
phases were slightly different from the alloy Ni content, and
were used to calculate volume fractions. However, the dif-
ference between the alloy Ni content and the measured Ni in
the major phases was less than the 2s limit on precision for
Ni (0.16 wt pet). Therefore, the volume fractions calculated
from these measurements may be inaccurate.
The precision of the volume fraction measurements from
the optical micrographs can be affected by the size of
the second phase and by the morphology of this phase.
When the volume fractions of the second phase were small,
the phase tended to be small in size, and the interfaces
between the minor and major phases were difficult to define.
This influenced the precision with which the grid point
intersections could be determined in a systematic point
count. Precise tracing of the perimeter of small particles for
Fig. 13--Optical micrograph showing eutectic ferrite in N50 weld. areal summation was also found to be difficult. When the

1348--VOLUME 15A, JULY 1984 METALLURGICALTRANSACTIONSA

phases had grown, the interphase interfaces were more
easily resolvable, but the phase shape tended to be ir-
regular as growing phases coalesced. This deviation from a
~I 50~ FE SECTION
spherical morphology may have affected the volume frac- i~b,,.,2 3.0 Cr
tion measurements. 21,4-
It can be seen from Tables III and IV that the 2s values Cr
12.4 m~ I 1 " 3 Ni
for the optically determined volume fractions tended to be L) Ni '
slightly lower than for the volume fractions determined from . . . . . ~- -..l *d..
phase chemistries, when the volume fractions were high and dz
low. When the volume fractions were intermediate, the 2s
W
values for the phase chemistry data tended to be less than n
those for the optical data. The relative errors, as defined by II I \\
2s/l~, followed the same trends. The mean volume fractions
LJ \ ~5
determined from the two optical techniques were generally I
similar, and the volume fractions determined from the Cr l \ \
concentrations were generally very close to those calculated I
I \
from the Ni concentrations of the phases. The optically- <~
determined mean values were usually different from those
obtained from the phase chemistries, indicating that the W
accuracies of the techniques were different. As has been / 4- c
discussed, inaccuracy in measuring the Cr and Ni concen- L~ ! 3' I-- o ~+1~
trations would not affect the volume fraction calculations, F- /
suggesting that the chemically-determined volume fractions
may be more accurate than those obtained by the optical
methods. However, the effects of the minor elements, such
as Mn and Mo, on the volume fraction calculations are
n I I I | I
unknown. No assessment of the accuracy of the optical 1o 20 30 40
B
methods was made in this study. Comparison of results 4B 30 20 10 0
from measurements made on known volume fractions in a
ternary Fe-Cr-Ni system would be helpful in establishing
the relative accuracies of the optical and phase chem-
istry techniques. CR
NI
From the results in this study, it can be seen that, when
the chemistry of the major phase is not measurably different Fig. 14--The 60 pct Fe verticalsection, with nominal Cr and Ni contents
of N50 and N50W.
from the alloy composition, the STEM/EDS technique can-
not be used for volume fraction calculations. The optical
techniques could be used, although the accuracy of the tech- makes it impossible to show both the Cr and Ni contents of
niques should first be evaluated. In samples in which the these alloys on the 60 pct Fe section.
phases are not of equilibrium chemistries, as in GTA welds Comparison of the (3' + 8) fields for N50W and N50
where solid state diffusion2~ alters the chemistry of the fer- showed them to be essentially coincident within the scatter
rite and adjacent austenite, the volume fractions could not of the points (see Figure 11). The main difference between
be calculated from the phase chemistries, and the optical the solid state equilibria in these heats results from the
methods would again be necessary. relative positions of the bulk alloy compositions in the com-
When the phases are small, and the chemistry of the mon phase field. Nitronic 50W, with higher Cr and lower
major phase is distinguishable from the alloy composition, Ni, lies more toward the ferrite field. Heat treatment of this
the STEM/EDS technique would be preferred for calcu- alloy above 1330 ~ transformed the sample entirely to fer-
lating volume fractions. In the sigma-containing samples rite, predicting that this heat would solidify as primary delta
heat treated for this study, for example, the sigma phase was ferrite. The microstructure of the weld on N50W was con-
generally 0.5 to 1 /xm in size, precluding the use of the sistent with primary delta-ferrite solidification, in the
optical techniques. ferritic-austenitic mode discussed by Suutala et al. ~o
Reference to the position of the Nitronic 50 composition
within the (3' + 8) field revealed that this heat should not
B. Solidification Mode Prediction
transform completely to ferrite before incipient melting be-
The nominal Cr contents shown on the 60 pct Fe vertical gins. The equilibrium phase data from heat-treated samples
section from the Fe-Cr-Ni ternary data (Figure 14)23 indi- of N50 indicated that the microstructure at the solidus
cated that both alloys would solidify as primary austenite if consisted of 75 to 85 vol pct ferrite and 15 to 25 vol pct
their solidification mode were predictable from ternary dia- austenite. From this one infers that the line of bulk Cr con-
grams, and that both heats would consist of stable austenite centration passes through the eutectic triangle. The GTA
up to incipient melting. Predictions of solidification mode weld on N50 had a microstructure typical of solidification
on the basis of Ni content more closely approximated the occurring with austenite as the leading phase. The pres-
observed behavior of N50W but not the behavior of N50. As ence of eutectic ferrite in the interdendritic regions suggests
mentioned in the Results section, displacement arises as a that the solidification mode was austenitic-ferritic, as de-
result of the presence of other alloying elements, which scribed by Suutala et al.t~ However, since heat treatments

METALLURGICAL T R A N S A C T I O N S A V O L U M E 15A, JULY 1 9 8 4 - - 1349

were not done to establish the phase boundaries of the eu- the same manner as Ni, ~7 although not to as pronounced a
tectic triangle, it was not possible to predict the dominant degree, suggesting that, at Mn contents of 5 to 6 wt pet, it
solidification mode of the N50 heat from the constant-Fe was an austenitizer.
section. In addition, minor perturbations in alloy chemistry In Espy's paper~ the Cr and Ni equivalents were not
would influence the solidification modes, and thus other related to the solidification modes of the various welds, but
heats of the alloys might solidify by different modes. to the room temperature phase fractions. Therefore no pre-
Prediction of solidification modes in stainless steels can diction of solidification mode was possible from the data
be done by calculating the Cr and Ni equivalents, and relat- and equivalents presented by Espy. However, using Espy's
ing these numbers to the Schaeffler or DeLong diagrams. 24 formulas for Cr and Ni equivalents and the solidification
Several different calculations for Cr and Ni equivalents modes and compositions examined by Suutala, ~5 the transi-
have been developed for stainless steels of high alloy tion between austenitic-ferritic and ferritic-austenitic solidi-
content. These methods have recently been reviewed by fication appears to occur at a Creq/Nieq ratio of about 1.35.
Suutala, 25 who found that for nitrogen-containing stainless The ratios for N50W and N50 lie close to this value, with
steels with Mn contents less than 5 wt pet, the equivalents N50W with a ratio of 1.39, slightly higher, and N50 with a
proposed by Hammar and Svensson 26 best predicted the ratio of 1.31, slightly lower. This predicted that the N50W
solidification modes experienced by the welds. For stainless heat should solidify by the ferritic-austenitic mode, and the
steels containing 5 to 8 pet Mn, however, Suutala 25 found N50 heat by the austenitic-ferritic mode. It would be ex-
that the equivalents developed by Hull 27correlated best with pected that small heat-to-heat variations in compositions
observed solidification modes and ferrite contents. Brooks 28 could result in differences in solidification mode. Within the
investigated welds on heats of Nitronic 40, a high-N, high- specifications for N5OW, the Creq/Nieq ratio could range
Mn stainless steel, and established that Hull's equivalents from a minimum of 1.16 to a maximum of 2.44, and for
worked well in predicting ferrite contents. By comparison, N50, the range would be from 0.97 to 1.72. This indicates
Espy 5 proposed a different set of equivalents applicable to that for welds made with N50W filler wire, the N50W
the entire Nitronic alloy series. composition must be closely controlled to ensure the ferritic-
Calculation of equivalents using Hammar's formulations austenitic mode generally found to be necessary 8'12'13'2t for
gave a Creq/Nieq ratio of 1.41 for N50W, and a ratio of prevention of hot cracking.
1.33 for N50. With Hull's equivalents, the ratios were of
1.40 for N50W, and 1.31 for N50. Suutala 25 has found that
austenitic-ferritic solidification occurred in welds with
V. CONCLUSIONS
Creq/Nieq ratios of 1.52 or less, using Hammar's equiv-
alents, and that ferritic-austenitic solidification occurred for 1. The (y + 6) phase field generated from ferrite and aus-
welds with ratios between 1.52 and 1.95. Using these val- tenite compositional data successfully predicts the pri-
ues, welds on N50W and N50 should solidify in the mary delta-ferrite solidification behavior of GTA welds
austenitic-ferritic mode, which was not the case. With on Nitronic 50W.
Hull's equivalents, the demarcation between austenitic- 2. In welds of Nitronic 50, the microstructure appears to
ferritic and ferritic-austenitic solidification was a ratio of be that of a material which underwent austenitic-ferritic
approximately 1.43 to 1.45 according to Suutala's data. 25 solidification. The (y + 6) phase field indicates that
The ratio of 1.40 for N50W would then be very near the the heat should solidify through the eutectic triangle.
transition value between the two modes. The amount of Without establishing the boundaries of the eutectic tri-
ferrite expected in this weld would be about 3 vol pet, angle, it is not possible to predict the leading phase
using Hull's equivalents and Suutala's 2s data. This value is during solidification from the phase chemistry data.
considerably lower than the measured - 6 vol pct. Using 3. The Cr and Ni equivalents developed by Espy, in con-
Suutala's 25graphs and Hull's 27 equivalents, the weld on N50 junction with solidification microstructural data com-
was predicted to contain no ferrite, while the weld did con- piled by Suutala, were used to predict the solidification
tain --2 vol pct ferrite. Using the equivalents proposed by modes of the Nitronic 50W and Nitronic 50 heats exam-
Espy, 5 the N50W weld was predicted to contain slightly ined in this study. Correlation of solidification modes
more than 5 vol pet ferrite, and the N50 weld to contain with the equivalents proposed by Hull was good for the
1 to 2 vol pct ferrite, values closer to those measured on Nitronic 50 heat, but less conclusive for Nitronic 50W.
the welds. 4. Cr and Mo were observed to partition strongly to ferrite,
The major difference between the approaches of Espy 5 and Ni to the austenite. The austenite tended to be some-
and Hull 27 was the treatment of Mn and N. Espy 5 varied the what enriched in Mn relative to the ferrite, but no large
weighting factor for N in the Ni equivalent calculation, since influence of temperature on Mn content was seen. Ferrite
he observed that at N levels above 0.20 wt pet, the N was at all temperatures contained less Fe than the austenite.
not as strong as austenitizer as at lower levels. Hull, z7 as 5. Several secondary phases, notably sigma and Nb-rich
well as Hammar and Svensson, 26 used a constant multiplier Z-phase, precipitated in Nitronic 50W and Nitronic 50.
for nitrogen. With respect to Mn, Hull 2v found that at Mn The composition of the sigma phase remained nearly
contents about 4 pet, Mn was no longer an austenitizer but constant over a range of aging temperatures.
a ferritizer, while Espy 5 observed that Mn had little effect on 6. Phase chemistries determined by STEM/EDS techniques
the weld microstructures. Suutala 25 concluded that at high were used to calculate volume fractions of austenite and
levels of Mn (5 to 8 pet) this element acted as a ferritizer or ferrite. These volume fractions were compared with val-
a weak austenitizer, segregating as Ni segregated. In the ues determined by light optical micrographs. It was
welds on N50 and N50W, Mn was observed to segregate in found that the precision of the optical techniques tended

1350--VOLUME 15A, JULY 1984 METALLURGICAL TRANSACTIONS A

 to be somewhat better than that of the phase chemistry 3. E.E. Denhard and R.H. Espy: Met. Eng. Quart., 1972, vol. 12,
method when volume fractions were high or low, and pp. 18-20.
4. J.N. Kass, W. L. Bell, M. T. Wang, and J. Bobick: Metall. Trans. A,
that the precision of the phase chemistry data tended 1979, vol. 10A, pp. 715-25.
to be better when intermediate volume fractions were 5. R.H. Espy: Weld. J., 1982, vol. 61, pp. 149s-56s.
present. The accuracy of the phase chemistry method 6. E C. Hull: Weld. J., 1967, vol. 46, pp. 399s-409s.
appears to be different from the accuracy of the optical 7. J.C. Lippold and W. E Savage: Weld. J., 1979, vol. 58, pp. 362s-74s.
methods, although no absolute determination of accur- 8. T. Takalo, N. Suutala, and T. Moisio: Metall. Trans. A, 1979,
vol. 10A, pp. 1173-81.
acies was made. 9. N. Suutala, T. Takalo, and T. Moisio: Metall. Trans. A, 1979,
7. The optical methods are recommended for calculating vol. 10A, pp. 1183-90.
volume fractions when the chemistry of the major phase 10. N. Suutala, T. Takalo, and T. Moisio: Metall. Trans. A, 1980,
cannot be distinguished from the alloy composition. vol. llA, pp. 717-25.
11. S.A. David: Weld. J., 1981, vol. 60, pp. 63s-71s.
The phase chemistry method is superior to the optical 12. M.J. Cieslak and W. F. Savage: Weld. J., 1980, vol. 59, pp. 136s-46s.
methods when the phase size approaches the limit of 13. H. Thier: DVS-Berichte, DVS-GmbH, Dusseldorf, 1976, Band 41,
resolution of the optical microscope, as long as the phase pp. 101-06.
chemistry is measurably different from the alloy com- 14. K.K. Mallick: M.S. Thesis, University of Newcastle Upon Tyne,
position. A combination of optical and STEM/EDS Newcastle Upon Tyne, U.K., 1979.
15. G.L. Leone and H.W. Kerr: Weld. J., 1982, vol. 61, pp. 13s-21s.
techniques over a range of volume fractions would pro- 16. S.A. David, G.M. Goodwin, and D.N. Braski: Weld. J., 1979,
vide the best data base. vol. 58, pp.330s-36s.
17. A.M. Ritter: M.S. Thesis, Rensselaer Polytechnic Institute, Troy,
NY, 1982.
ACKNOWLEDGMENTS 18. J.E. Hilliard and J.W. Cahn: Trans. TMS-AIME, 1961, vol. 221,
The authors wish to acknowledge E.K. Weaver of pp. 344-52.
19. H. Hughes: JISI, 1967, vol. 205, pp. 775-78.
Wyman-Gordon Co. for the differential thermal analysis 20. D.H. Jack and K. H. Jack: JISI, 1972, vol. 210, pp. 790-92.
data. We wish to thank T. E. Douglas, R.T. Laing, W.F. 21. M.J. Cieslak, A. M. Ritter, and W. F. Savage: Weld. J., 1982, vol. 61,
Moore, E. C. Nagy, R. A. White, and R. H. Wiley of Gen- pp. ls-8s.
eral Electric for material processing. The contributions of 22. A.M. Ritter, M. J. Cieslak, and W. F. Savage: Metall. Trans. A, 1982,
vol. 14A, pp. 37-44.
M.F. Garbauskas and J.L. Methe', who performed the 23. ASM Bulletin of Alloy Phase Diagrams, ASM, Metals Park, OH,
X-ray fluorescence analysis, and C. R. Rodd, who assisted 1981, vol. 2, pp. 89-102.
in metallographic preparation, are also acknowledged. 24. W.T. DeLong: Weld. J., 1974, vol. 53, pp. 273s-86s.
25. N. Suutala: Metall. Trans. A, 1982, vol. 13A, pp. 2121-30.
26. O. Hammar and U. Svensson: Solidification and Casting of Metals,
REFERENCES The Metals Society, London, 1979, pp. 401-10.
1. E.F. Denhard: U.S. Patent No. 3,592, 634, 1971. 27. F.C. Hull: Weld. J., 1973, vol. 52, pp. 193s-203s.
2. R.R. Gaugh and D. C. Perry: Chem. Eng., 1972, vol. 79, pp. 84-90. 28. J.A. Brooks: Weld. J., 1975, vol. 54, pp. 189-95s.

METALLURGICALTRANSACTIONS A VOLUME 15A, JULY 1984--1351