Professional Documents
Culture Documents
2.4. Intergranular stress corrosion The sum of the relative fragilities for all
In an aggressive environment the chemistry active impurities gives the fragility index for
of the grain boundary delineates a preferential the cast. In this context the fragility index is
fracture path through changes in the electro- an engineering parameter which depends on
chemical potential at the crack front. This the commercial material composition and, for
increases the rate of crack growth and reduces a fixed material specification and heat treat-
the service life. m e n t (a), will define its quality in relation to
the effect of impurities for the fracture process
a. Because the calculation of S~(a) usually
3. THE RELATIVE FRAGILITY AND THE FRA-
gives only relative effects with accuracy, the
GILITY INDEX
fragility index becomes an absolute parameter
From the point of view of the material's only by normalizing against the effect of one
integrity we require a useful quantitative element in the series. If we choose the most
engineering parameter that measures the like- deleterious element in the sum, the inaccuracy
lihood of the material failing through the in the largest term is removed. Thus the
presence of naturally occurring concentrations fragility index is defined by
of p o t e n t impurity elements. Thus we must E~(a)
distinguish clearly between the intrinsic em- fragility index - (2)
brittling power of an element and its ability ~m(a)S~n(a)
to embrittle in practice: the former is a funda- where the element m is the one which has the
mental property relating to the effect on highest relative fragility. The fragility index
atomic cohesion whereas the latter is a func- thus has the dimensions of weight concentra-
tion of the concentration of the element in tion and, as will be shown later, is an expres-
the locality of the fracture event and hence sion of phosphorus, tin or copper equivalent.
of the concentration of the element in the Thus the final representation of the fragility
bulk. In general we can define a parameter index is one which is readily understood in
¢~(a), the relative fragility of the alloy for the the fields of quality control and materials
element i: specification.
For certain engineering situations, general
~ ( a ) ~ Zi/~(a)S~(a ) (1)
overall parameters are available to specify the
where Xi is the bulk concentration of element suitability or otherwise of a batch of material.
i and ~ ( a ) is its enrichment factor, i.e. the For instance the conventional fracture tough-
ratio between the molar fractional concentra- ness parameter is used in this manner. The
tion at the interface and the bulk concentra- fragility index would normally be included in
tion by weight of the element i for the mate- a fracture toughness determination but in an
rial (a) and appropriate to the fracture process unspecified form. Here we show how one
a. S~(a), the embrittling sensitivity of element important c o m p o n e n t of such parameters,
i per unit area of interface, can be expressed that due to impurity concentrations, may be
most easily as a relative materials parameter. measured separately so that the overall para-
The relative fragility is n o t expected to be meter may be more effectively optimized.
linear with bulk concentration for long-term The introduction of the p a r a m e t e r / ~ ( a )
isothermal heat treatments [4] but will into eqn. (1} simplifies considerably the pre-
approximate to this for normal commercial dictive application of this equation because
heat treatments over a wide range of concen- it allows an estimate of the a m o u n t of segrega-
trations [ 5]. tion. The extensive studies on grain boundary
It is thus clear that a species to which the and surface segregation have led to a consider-
interfacial atomic cohesive force is highly able understanding of the segregation pheno-
sensitive will n o t necessarily be responsible menon and allow predictive correlations to be
for a high value of the relative fragility if its made so that, simply from the use of tabulated
bulk concentration is very low; conversely, a bulk parameters and a knowledge of the heat
relatively unaggressive element present in large treatment of the sample, fairly precise esti-
quantities may constitute the most dangerous mates may be made of grain boundary and
element in the system because of a high value free surface segregation levels throughout a
of ¢~(a). wide range of materials. The grain boundary
236
10 ¢ - - 10 4
. * Ni-B/
• Fe -AS
Cu-Sb ~Fe-Co •Fe~P
.4. •6Fe-P
• ~Fe PJI •
/•~Fe pl~~Fe'Sb
• ~eBFe-sn
@.Fe-Sn/~l" @.Fe-As
c 10 2 -- 10 2
-g O.Fe-Cu
~' FeSi • Pt-Au -
• YFe-Cr • / CuSb
• YFe"Mr~Fe S~/ *
E _ /
d
N • - P Ni-Cu
Cu-Au ~_
x _~Fe-SJ • / . -o
LU AueAg CuAu/~ PICu
ul
Au-Ni • •Fe-Cr /
A -Au /Au-Ag
I
I~IV@
/ Pt-Cr
• ~ /•••Pt - Ni
I •Cu-Pt p d / P d _ N i •
t/ •Au
Cu-Ni•
I /;u-N•• Au-Cu•
/ I J J J I i ~ J i
1 10-2 10-4 I 102 104
Atomic solid solubility Xc Predicted 13 (surface)
(a) (b)
Fig. 1. Correlations o f the equilibrium segregation enrichment ratio ~ ( a ) : (a) for grain boundaries after Seah and H o n d r o s [6] ; (b) for free surfaces after
Seah [ 8 ] . The solid lines s h o w the upper and lower quartiles o f the results, and data published since the original papers are included as stars. The ~ ( a ) here, in
line with the theory, are ratios o f a t o m concentrations and should be m u l t i p l i e d by the ratio o f atomic weights for use in eqns. (1) and (2).
t~
CO
238
TABLE 1
Temper brittleness of AISI 3340 (3.5Ni-1.7Cr) and 2 1 C r - l M o
AISI3340 a
P 0.017 1800 8 6 6
Sn 0.012 950 10 3 2~1
Sb 0.003 950 11 1 1•
As 0.016 300 8 1 12
214--Cr-1Mob
P 0.017 2000 4c 14 11
Sn 0.012 200 10 2~1 3
Sb 0.003 300 11 1 1
As 0.016 75 8 1 1
aThe temper brittleness fragility index of this cast is 0.031 wt.% Peq (theoretical value 0.030 wt.% Peq)-
bThe temper brittleness fragility index for this cast is 0.025 wt.% Peq (theoretical value 0.022 wt.% Peq ).
CThe embrittling sensitivity of phosphorus in molybdenum steels is approximately halved because of simultaneous
molybdenum segregation which improves the grain boundary cohesion [ 14 ].
239
TABLE 2
Stress relief cracking in 1M C r - M o - V
purity the relative detrimental importance is where Dv is the lattice diffusivity, d is the
as follows. interatomic spacing and 5 the width of the
grain boundary. This equation together with
Element Cu P Sn S
eqns. (4) and (5) allows the relative em-
~(a) 2~1 1~1 1 1
brittling sensitivities in Table 3 to be calculated
In order to compare these values with pre- as a change in tr or er concomitant with a
dicted performance, we adopt the model of change in segregation level, from which values
Fig. 3 and consider the effect of segregation of ¢~(a) can be computed.
on grain boundary transport. A comparison of the predicted relative
The appropriate enrichment factors for fragilities with the experimental values shows
grain boundaries in eqn. (1) may be assessed a clear agreement that phosphorus, copper
from solid solubility data in the manner dis- and tin are the dominant elements. The hier-
cussed earlier; here, however, they were mea- archies are different however. The measured
sured experimentally for improved accuracy. effect of sulphur, which was n o t predicted, is
At 550 °C equilibrium grain boundary seg- very probably due to sulphide precipitates
regation is attained after a very small fraction and not segregation, while the phosphorus
of creep life. In this alloy, while the grain dependence is somewhat lower than that
boundary segregation is sufficient to exacer- predicted by this model and may involve the
bate the creep problem, nevertheless it does contribution of phosphorus to alterations in
n o t provoke temper brittleness and so the microstructure [21]. Nevertheless it is clear
grain boundaries cannot be exposed easily by that copper and phosphorus are the most
low temperature fracture for AES examina- important elements, followed by tin.
tion. However, on the few samples where par- Both stress relief cracking and creep em-
tial failure on uncavitated boundaries occurred, brittlement are thus impurity sensitive, but
grain boundary enrichments were measured the vastly different strain rates cause differ-
and these are presented in Table 3. ent forms of segregation to be important and
Under the applied stress, cavities nucleate are thus dominated by different elements.
along the boundaries and grow by grain boun- Phosphorus is relatively much more active at
dary vacancy transport, leading inevitably to grain boundaries than surfaces and hence, in
intergranular creep rupture. this form of creep embrittlement, it is a major
The effect of segregation on the grain troublesome element along with copper and
boundary diffusivity D b may be measured tin. Here the fragility index, expressed in
through the attendant effect on the grain terms of the experimentally observed
boundary energy ~'b; these terms are linked dominant element copper, is 0.24 wt.% Cueq.
through the equation due to Borisov et al.
[20] in the form 4.4. Intergranular stress corrosion
The change in chemistry of the grain boun-
D b _ d exp/27ba2 ~ daries by impurity segregation, without any
| ] (6)
Dv 5 \ kT microstructural changes, can in some service
TABLE 3
Creep embrittlement at 550 °C in ½Cr-Mo-V
Cu 0.12 70 1.9 2½ 2½
P 0.017 900 1.9 4½ 1½
Sn 0.012 500 1.1 1 1
As 0.016 270 1.25 1 -
Sb 0.003 350 1.0 <½ -
The creep embrittlement fragility index for this cast is 0.24 wt.% Cueq(theoretical value 0.44 wt.% Cueq).
242
conditions affect markedly the propensity for ined. The grain boundary enrichments of the
cracking in the presence of a combination of observed elements are given in Table 4.
stress and an aggressive environment. The The importance of a corrosion process in
effect of impurities on stress corrosion crack- this environmentally assisted crack propaga-
ing is illustrated here by mild steel in a nitrate tion is shown by the fact that it can be con-
solution, where failure of commercial quality trolled electrochemically. This has led to the
material is invariably intergranular. association of cracks with pre-existing active
In extensive recent work at the National paths, defined by grain boundary segregants
Physical Laboratory [22] the relative harm- or precipitates, which are anodic to the main
fulness of tramp elements was analysed from volume of material. Either the active paths
mechanical test data on laboratory melts of formed by the presence of segregants at grain
high purity mild steel, each containing a single boundaries are intrinsically anodic, or the
impurity at a dilute level comparable with zones next to the grain boundaries, being
that found in commercial material. Following depleted of segregant, are anodic with respect
an equilibrium segregation heat treatment, to the grain boundaries. If the corrosion pro-
samples were tested at constant strain rate in ducts are soluble, intergranular corrosion
a hot nitrate solution and at the free corroding occurs in the unstressed state, but often a
potential. The time to failure of samples was stress is required to open cracks and to expose
used to derive a ratio R for each melt, this new surfaces to the action of the corrosive
being the ratio of the time to failure in the environment. For example, the surface of
corrosive nitrate solution to that in an inert mild steel in nitrate solution becomes covered
paraffin environment at the same temperature. with a magnetite layer which is protective and
It was found that after an equilibrium the function of the stress is to fracture this
segregation heat treatment the reduction in R magnetite layer to allow the reaction to
from its value for the high purity steel could proceed.
be expressed by [22] The capacity of segregants at the grain
b o u n d a r y - o x i d e interface to operate as a
--AR = 700 S + 1A1 + 0.4As + 27Ca + 1.9Cu +
local cell, enhancing the localized attack,
+ 0.2Ni + 20P + 0.95Sb + 1Sn + 0.3Zn
depends upon a very large number of factors,
where the element symbols are the bulk con- among them the local pH, the potential dif-
centrations and apply only to very dilute ference which exists within the crack, and the
levels. All these residual elements are detri- change in solution composition during disso-
mental, although several of the elements lution. Despite this, good theoretical pre-
become beneficial at levels only slightly higher dictions of stress corrosion crack growth rates
by acting through microstructural changes. In have been made from bulk electrochemical
commercial steels the sulphur, aluminium and properties by Doig and Flewitt [23].
calcium are stabilized as precipitates and The preferential dissolution of the grain
would not be detected as grain boundary boundary in an aqueous oxidizing environment
segregants. Thus at the typical impurity levels is shown schematically in Fig. 4. The em-
encountered in U.K. steels the relative harm- brittling role of hydrogen is not considered
fulness of the elements as derived from direct here because of the anodic nature of the crack
mechanical test data is as follows. tip in relation to its oxide-covered walls. To a
first approximation the rate of dissolution of
Element P Cu Ni Sn As Sb
the grain boundary is directly dependent on
¢~{a) 3 2 0.2 0.1 0.1 <0.1
the electrochemical potential set up between
As in the case of creep embrittlement, it is it and the bulk crystallites [24], and here we
normally difficult to expose the grain boun- assume that this reflects the term S~(a} in
daries for direct analysis. However, on stress- eqn. (1). In Table 4 the difference IAE0$
corroded samples where intergranular fracture between these two potentials is given, from
had been initiated following impact in the which ¢~(a) is calculated. In this case phos-
spectrometer, fresh grain boundaries were phorus and copper are important and to a
occasionally exposed just ahead of the crack lesser extent arsenic, tin and nickel. It should
front. In these instances the chemistry of the be noted that there is still uncertainty about
uncorroded grain boundaries could be exam- the appropriate dissolution reaction equation
243
TABLE 4
Stress corrosion cracking of mild steel in nitrate solution
The intergranular stress corrosion fragility index for this cast is 0.031 wt.% Peq (theoretical value 0.052 wt.% Peq).
REFERENCES
5. CONCLUSIONS
1 M.'P. Seah and C. Lea, Philos. Mag., 31 (1975)
In the more important categories of inter- 627.
2 C. Lea and M. P. Seah, Scr. Metall., 9 (1975) 583.
granular weakness considered here, two or
3 A. Joshi and D. F. Stein, A m . Soc. Test. Mater.,
three residual elements in the steel dominate Spec. Tech. Publ., 499, 1977, p. 439.
the fracture properties through equilibrium 4 M. Guttmann, Surf. Sci., 53 (1975) 583.
segregation. The order of p o t e n c y of the 5 G. C. Gould, A m . Soc. Test. Mater., Spec. Tech.
important elements and their relative effects Publ., 407, 1968, p. 59.
6 M. P. Seah and E. D. Hondros, Proc. R. Soc.
vary markedly from one metallurgical situa- London, Ser. A, 335 (1973) 191.
tion to another. In order to quantify this 7 S. Brunauer, P. H. Emmett and E. Teller, J. Am.
hierarchy of potency an engineering para- Chem. Soc., 60 (1938) 309.
244
8 M. P. Seah, J. Catal., 57 (1979) 450. 17 R. Bruscato, Weld. d., Res. Suppl., 49 (1970)
9 C. Lea and M. P. Seah, Surf. Sci., 53 (1975) 272. 148-s.
10 M.P. Seah, Acta Metall., 25 (1977) 345. 18 R. C. Miller and A. D. Batte, Met. Constr., 7
11 M.P. Seah, Philos. Trans. R. Soc. London, to be (1975) 550.
published. 19 R. D. Skelton, Met. Sci., 9 (1975) 192.
12 E. D. Hondros, M. P. Seah and C. Lea, Met. 20 V. T. Borisov, V. M. Golikov and G. V. Scher-
Mater., (January 1976) 26. bedinskiy, Phys. Met., Metallogr. (USSR), 17
13 M. P. Seah, P. J. Spencer and E. D. Hondros, Met. (1964) 80.
Sei., 13 (1979) 307. 21 P. Hayes and P. Grieveson, Met. Sci., 9 (1975)
14 M. P. Seah, Acta Metall., to be published. 332.
15 E. D. Hondros and M. P. Seah, Int. Metall. Rev., 22 C. Lea, Corros. Sci., to be published.
22 (1977) 262. 23 P. Doig and P. E. J. Flewitt, Proc. R. Soc. London,
16 J.P. Hirth and J. R. Rice, Metall. Trans., to be Set. A, 357 (1977) 439.
published. 24 C. Lea, Corros. Sci., to be published.