ScriptaMetallurgicaet M&ii&, Vol. 32, No. 10,pp.
1595-1600,1995
Copyright0 1995Else&r Science.Ltd
PrintedintheUSAAllrighW-ed
0956-716X/95$9.50+ .OO
0956-716X(95)00242-1
EFFECT OF COMPOSITION ON THE TRANSFORMATION OF S-FERRITE TO TVIN
TYPE 316 STAINLESS STEEL WELD METALS
T.P.S. Gill, V. Shankar, M.G. Pujar and P. Rodriguez
Indira Gandhi Centre for Atomic Research
Kalpakkam 603 102, TN, India
(Received October 13, 1994)
(Revised December 13, 1994)
htroduction
One of the most important factors which determine the upper limit of service temperature of components
made of austenitic stainless steels is the inferior microstructural stability of the weld metal at elevated temperatures
as compared to that of the base metal. This difference in behavior is because of the presence of metastable h-ferrite
phase in the austenite matrix, which is intentionally retained to avoid the occurrence of solidification cracking in
the weld metal. During high temperature exposure as encountered in service, the ferrite transforms to a variety of
secondary phases such as M,C, carbide, cr, x, ~1’etc (1). These phases are known to degrade the mechanical and
corrosion properties of the weld metal (1,2). However, the formation of these phases is influenced by a number
of factofi such as weld metal composition, ferrite content and its morphology, in addition to temperature and time
of transformation. Thus far, it has not been possible to accurately predict the ferrite transformation kinetics and
hence the precipitation of secondary phases, though a few studies have been published which have advanced the
understandmg of the mechanisms of transformation of ferrite to various phases (3-6). It has been well recognised
that carbon plays a crucial role in retarding the ferrite to sigma transformation. Gill and co-workers (6) reported
that in high carbon (>o.O3%C) type 3 16 stainless steel welds, the ferrite decomposed by a two-stage transformation
process. In the first stage, at shorter aging times, ferrite to austenite transformation was assisted by the rapid
precipitation of M,,C, carbides at the 6/y interface. Thereafter, in the second stage, the transformation of the
remaining S-ferrite to either o or y (or both) was sluggish. On the other hand, in low carbon welds (<O.O3%C),
h-ferrite transformed rapidly to o and y.
In this study, the influence of Cr and ferrite content on the transformation kinetics of g-ferrite to carbides,
austenite and cr has been investigated in type 3 16L stainless steel weld metals. A novel technique was employed
to increase the chromium content in the weld metal while keeping the concentration of the remaining elements
essentially constam thus bringing out the effect of Cr alone on the stability of g-ferrite. An equation was derived
to predict the ferrite transformation kinetics. Further data analysis yielded a nomogram to estimate the sigma
content in type 316 stainless steel weld metals. The nomogram can be employed to optimize the weld metal
composition and to minimise sigma formation from h-ferrite at high temperatures.
An AISI type 3 16L stainless steel plate of 7mm thickness was employed for this investigation. The samples
for ferrite transformation studies were prepared by keeping the welding arc stationary at one spot and adding
controlled amounts of Cr in the molten puddle. The process of Cr addition and its homogenization in the molten
alloy was carried out at two current levels, In the Cr addition stage, the arc (current-SOA, voltage-W) was kept
stationary for 60s. During this period a known quantity of Cr chips (purity-99.99 wt.%) was manually added to
the molten puddle. The current was then increased to 80A and the molten pool allowed to homogenize for another
1595
1596 STAINLESS STEEL WELDS Vol. 32, No. 10
90s. After a cumulative period of 150s of arc-on time, the current was switched off and the molten puddle was
allowed to cool in flowing argon (flow rate - 12 Vmin.). A sample, obtained by using the above conditions but
without intentional addition of Cr, was also prepared. The actual composition of the solidified puddle without Cr
addition is given in Table 1. Four more samples with increasing Cr contents were produced for this investigation.
Table 2 gives the amount of Cr determined in the solidified samples, designated as Alloys 1 to 5.
TABLE 1: Chemical Composition of Alloy 1 (wt%)
C Si Mn s P Cr Ni MO N
0.027 0.5 1.79 0.011 0.02 16.3 10.1 2.1 0.029
The amount of b-ferrite in each sample was measured TABLE 2:Chromium and S-ferrite
using a calibrated instrument (Magne Gage) based on the Contents of Modified Weld Puddles
magnetic permeability principle. A large number of readings (>
/^““I
30) were taken on each sample to arrive at an average value. The
amount of g-ferrite present in stainless steel welds is expressed as
Ferrite Number (F$ and corresponds to volume percent ferrite
up to about 8FN (7). The average b-ferrite content measured in
each of the samples is given in Table 2. Aging was carried out at
65O’C for various durations until almost full transformation of
ferrite was achieved. The g-ferrite transformation kinetics were
studied by measuring the ferrite content after interrupting the
aging process at intervals.
An electrochemical extraction procedure (electrolytic dissolution in 1O%HCI-CH,OH solution at 1.5 volts)
was employed to estimate the total amount of precipitate in the samples after about 0.90 fraction of ferrite had
transformed. Under the above conditions, the austenite and b-ferrite dissolve in preference to Cr-rich phases like
M,,C$, o and x. At the end of the extraction process, the residue consists of a mixture of secondary phase
precipitates. The precipitates present in the residue were identified by using the X-ray powder diffraction method
employing a Cu-K, radiation source. The constituent phases, which were identified as Ma&, and o, were then
quantitatively estimated by treating the residue with a solution of 5%Br, in CH,OH. In this solution, cr phase
dissolved rapidly, while M&, remained unattacked, and was obtained as residue (8,9). By subtracting the weight
of the carbide obtained from the total weight of the residue taken, the amount of cr phase present was estimated.
This quantity was designated as (T,,~,the amount of (I phase formed after 90% ferrite transformation,
The S-ferrite transformation kinetics in the type 3 16L weld metals were studied by measuring the variation
of fraction of ferrite transformed with aging time during isothermal aging at 65O’C. The ferrite transformation data
were then fitted to the Johnson-Mehl equation which on simplification has the form
In[-ln( l-X)] = In(b) + nln(t) PI
where X is the fraction of ferrite transformed in time TABLE 3: Kinetic Parameters for S-ferrite
t (hours), b is the rate constant and n is a constant Transformation at 650’
which depends on the reaction mechanism. Figure
Alloy n, n2 b, b, chJ.9 X
1 shows the variation of ln[-ln( l-X)] with in(t). It is
observed that the S-ferrite in all the samples showed (wt”/s)
a distinct two-stage transformation behaviour. 1 0.13 0.49 0.50 0.47 nd. 0.98
Therefore, two different sets of Johnson-Mehl 2 0.17 0.53 0.26 0.23 0.9 0.94
parameters are required to define the complete 3 0.16 0.56 0.2 1 0.16 0.8 0.96
transformation of b-ferrite. These are designated as 4 0.17 0.67 0.25 0.10 1.2 0.98
n, and b, for the first stage, and n, and bz for the 5 0.18 0.96 0.14 0.05 1.6 0.98
second stage. These parameters are given in Table 3.
i
n.d. - not determined
Vol. 32, No. 10 STAINLESS STEEL WELDS 1597
Figures 2 and 3 show the variation of n, and n, with Cr content and b, and b, with FN, respectively. The
best-fit equations for these parameters are given below:
n, = O.Ol(Cr) - 0.08; (r = -0.71, s = +0.02) PI
n2= 0_19(Cr) - 2.69; (r = -0.99, s = kO.02) [31
bi = 0.45$N)a-46; (r = -0.89, s = aO.07) [41
b,= 0.53(FN)a-94; (r = -0.99, s = *0.03) PI
where Cr is expressed in wt.% and r and s are the correlation coefficient and standard error of determination,
respectively. It is observed that n, is essentially independent of Cr content, whereas n2 shows a strong linear
dependence on Cr. The Johnson-Mehl intercepts b, and b, exhibited a better con-relation with FN, than with the
amount of Cr. By substituting Eqs. [2] to [S] in [ 11, it is possible to represent ferrite transformation kinetics as
a function of Cr content and FN. A comparison between the calculated and experimental values of X is given in
Fig. 4 which shows an excellent correlation between the two quantities.
The above method to predict the ferrite transformation will be of limited use in designing weld metal
compositions, unless a method to correlate the kinetics parameter for the 6 to o transformation with the weld metal
composition is established. An attempt is made to evolve relationships for n2 with composition and Q content
which can be combined to give an empirical equation to estimate 00,9content from weld chemistry.
The kinetics parameter n, has been shown to represent the transformation of S-ferrite to o and y (6). From
data provided in Table 2, it is observed that with an increase in n, there is a corresponding increase in the CT~,~
content Linear regression analysis relating n2 and o~,~content gives:
n2= 0.53(%00.,J + 0.09; (r = 0.97, s = f 0.06) if51
Since in this study ‘synthetic’ weld metals were employed to investigate the influence of Cr on ferrite
transformation kinetics, while keeping the concentrations of the other elements constant, it will be necessary to
establish the validity of the expression [6] to actual welds. Data from four weld metals which were deposited using
commercial consumables, were employed for this purpose. The chemical compositions, n2 and 00,9contents for
these welds are given in Table 4 and a plot of nz vs 00,9for the actual and ‘synthetic’ weld metals is shown in Fig.
5.
TABLE 4: Chemical Compositions (w-t%), nz and cr0.9Content (wt%) in Different Commercial Weld Metals
Aged at 650°C
Alloy C Si Mn S P Cr Ni MO N n, oo.9 X
6 0.022 0.25 0.35 0.006 0.020 17.8 12.4 1.93 0.068 0.82 1.32 0.90
7 0.029 0.27 0.28 0.006 0.025 18.0 12.6 2.23 0.060 1.00 2.28 0.92
8 0.055 0.45 1.14 0.013 0.030 18.55 11.85 2.21 0.064 0.30 0.25 0.94
9 0.081 0.50 1.26 0.008 0.010 18.0 11.7 2.7 0.031 0.25 0.35 0.95
The following best-fit relationship was obtained:
n2= 0.4(%0,,) + 0.21; (r = 0.96, s = + 0.09) ]71
The predictive capability of Eq. [7] can be further improved if the o content can be determined exactly after 0.90
fraction of ferrite transformation. However, the data for (T~.~content employed in this study to develop Eq. [7]
pertained to X values ranging from 0.90 to 0.98 (Tables 3 and 4) as it was difficult to predetermine the exact aging
time required to transform 0.90 fraction of b-ferrite.
The variation of n2 with C content is given as (6):
n2= -11,65(%C) + 0.99; (r = -0.99, s = *0.04) PI
1598 STAINLESS STEEL WELDS Vol. 32, No. 10
From Eqs. [3] and [S], n, can be expressed as a function of C and Cr contents. Figure 6 shows a linear
relationship between nZ and the quantity (-11.65C+O.l9Cr). This relation can be expressed as
nZ= -12.O(%C) + 0.2(%Cr) - 2.45; (r = 0.88, s =h 0.14) 191
The change in the coetfficient for C in Eq. [8] is marginal tier the inclusion of the Cr term in Eq. [9], signifying
that the effects of C and Cr on nZare not interactive. Equation [9] was further modified by taking into account the
influence of other alloying elements on n,. It was found that only MO, apart from C and Cr, influenced n2
significantly, as shown in Fig. 6. The coefftcients for C and Cr also changed when the MO contribution was
considered. The following equation gave a good correlation:
n2 = -21.7(%C) + 0,25(%Cr) + 0.93(%Mo) - 5.09; (r = 0.96, s = f 0.08) [lOI
While the inclusion of the MO contribution increases the coefficient for C in Eq. [lo] significantly, the change in
the coefficient of Cr is not appreciable. This indicates a strong interactive effect between C and MO, but an
insignificant interaction between Cr and MO. The amount of cr precipitated, after about 0.90 fraction of b-ferrite
has transformed at 65O’C, can then be calculated from the following expression obtained by combining Eqs. [7]
and [lo]:
cr0,9(wt.%) = -55(%C) + 0.63(%Cr) + 2,35(%Mo) - 13.42; (r = -0.87, s = hO.35) Ull
It is observed that carbon plays a dominant role in retarding cr formation from S-ferrite. While both Cr and MO
aid the formation of cr, MO is observed to be nearly four times as effective as Cr. Equation [ 1 I] can also be used
to optimize the weld metal chemistry with respect to cr formation from b-ferrite. Since in deriving Eq. [ 111, the
efffects of other alloying elements and interactive terms have been neglected, this equation is essentially valid only
in the range of compositions studied. A nomogram has been constructed from Eq. [ 1 I] to estimate the (T~.~ content
from the chemical composition of a weld metal, and is presented in Fig. 7. The nomogram divides the regions of
composition (as a function of Cr, MO and C content) in terms of a sigma forming tendency. A series of parallel
lines corresponding to various wt.% C and a constant Cr/Mo ratio for which the S-ferrite to o transformation will
be negligible are identified as 0% cr lines. The 0% cr line for each carbon level divides the nomogram into no-o
and o zones. The amount of o,,~ (in wt.%) likely to form can be read directly from this diagram, for a given
combination of C, Cr and MO, from the scale drawn perpendicular to the 0% cr lines.
The nomogram is applicable for type 3 16 stainless steel weld metals and for compositions employed to
develop this nomogram. It has been documented that n2 remains constant for a given composition in the
temperature range 600 to 700°C thus implying that the transformation kinetics of S-ferrite also remains the same
(6). The accuracy of the nomogram was checked for weld metals not used in this study and also at various aging
temperatures in the range of 600 to 7OO’C (10-13). Figure 8 shows a comparison between the calculated and
experimental values for the amount of Q,,~. It is observed that experimental values fall close to slope 1, thus
establishing the accuracy of the nomogram in predicting the (T,,~content.
It is recognised that the proposed nomogram, in its present form, cannot be used to estimate the maximum
amount of cr that can form in a weld metal. Further work is required to correlate o content after almost complete
ferrite transformation and the maximum amount of o that can form. Several investigators have reported
transformation of S-ferrite to x on elevated temperature aging (14- 16). In the present study, x was not observed
to any significant extent and the main decomposition product of S-ferrite was o. Hence, the applicability of the
proposed nomogram to weld metal compositions in which S-ferrite transforms to any significant degree to x is not
established.
A study of the effect of Cr on S-ferrite transformation kinetics has yielded a relationship between the
kinetics parameter n, and weld metal chromium content. The amount of 00.9 formed from g-ferrite has been
correlated with weld metal composition. A nomogram is proposed, to predict the amount of (J~.~formed after 90%
Vol. 32, No. 10 STAINLESS STEEL WELDS 1599
ferrite transformation in a given weld metal, as a Rmction of C, Cr and MO contents. The nomogram can be
employed to optimize the composition for reducing high temperature embrittlement resulting from o precipitation
in type 3 16 stainless steel weld metals. The nomogram has been developed from data obtained at 650°C but has
been shown to be valid in the temperature range 600-700°C.
Acknowlew
The authors wish to thank Dr. Baldev Raj, Director, Metallurgy and Materials Group and Dr. S.L. Mannan,
Head, Materials Development Division for their continuous support during the course of this investigation. The
authors also thank Mr A.R.Hasan Shaikh of Metallurgy Division for his useful comments. Technical assistance
by Mr. C. Ramakrishnan, former student of the Kamataka Regional Engineering College, is gratefully
acknowledged.
References
1. J.J.Smith andR.A.Farrar, Int. Mat. Rev., 38,25 (1993).
2. T.P.S.GiII, J.B.Gnanamoorthy and K.A.Padmanabhan, Corrosion, 43,208 (1987).
3. R.G.Thomas and D.Yapp, Weld. J. Res. Suppl., 57, 361s (1978).
4 R.A.Farrar, J.Mater. Sci., 20,4215 (1985).
5. J.M.Vitek and S. A.David, Weld. J. Res. Suppl., 63, 246s (1984)
6. T.P.S.Gill, V.Shankar, M.Vijayalakshmi and P.Rodriguez, Scripta Metall. Mater., 27, 3 13 (1992).
7 EStalmasek, WRC Bulletin No.3 18, (1986).
8. T.P.S.Gill and J.B.Gnanamoorthy, J. Mater. Sci., 17, 1513 (1982).
9 T.P.S.GiIl and J.B.Gnanamoorthy, Trans. Ind. Inst. ofMetals, 36,269 (1983).
10. M.D.Mathew, G.Sasikala, T.P.S.GiII, S.L.Mannan and P.Rodriguez, Mater. Sci. Tech. (in press).
11. Hasan Shaikh, M.G.Pujar, N.Sivai Barasi, P.V.Siva Prasad and H.S.Khatak, Mater. Sci. Tech. (in press).
12. V.Shankar, A.Selvam, T.P.S.GiIl and S.L.Mannan, Joining of Materials for 2000 AD, ~411, The Indian
Institute of Welding, Tiruchirappalli (1991).
13 T.P.S.Gill, Ph.D. Thesis, Indian Institute of Technology, Madras, 1984.
14. J.M.Leitnaker, Weld. J. Res. Suppl., 61, 9s (1982).
15. R.A.Farrar, CHuelin and R.G.Thomas, J.Mater. Sci., 20, 2828 (1985).
16 G.F.Slattery, S.R.Keown and M.E.Lambert, Metals Technology, 10, 373 (1983).
-2.5
;:j A1
-3 -2 -1 0 1 2 3 4 5 6 7 16.0 17.0 18.0 19.0
In(t) CHROMIUM, wtw
FIG. 1. Variation of In[-ln( l-X)] with In(t) FIG. 2. Variation of n, and n, with chromium
content.
1600 STAINLESS STEEL WELDS Vol. 32, No. 10
0.6 -
0.5 -
0.8
0.4 -
p.6
so.3 -
. go.4
so.2 -
0.1 -
0.0 I . I * I . I ’ 0.0 ‘. * - - -. . - - . . . . . c
0.0 2.0 4.0 6.0 8.0 10.0 oh 0.120.’ 0.6 oh 1 I
FERRITE NUMBER, FN X 4est)
FIG. 3. Variation of b, and b, with ferrite number. FIG. 4. Comparison between the observed and
estimated values of fraction of ferrite transformed, X.
1.2 , I
-11.65C+O.l94Cr
4 -12.80C+O.l5Cr+OS5Mo
d
0.8
r”
0.4 x 0
x
FIG. 5. The relationship between n2 and q9 content. FIG. 6. The variation of n, as a function of weld
metal composition, f(comp.)
FIG. 8. Comparison between
experimental and estimated values of
o0.9.
FIG. 7. Nomogram for the estimation of q9 content in Type 316
stainless steel weld metals.
