Journal of Nuclear Materials 452 (2014) 607–613

Contents lists available at ScienceDirect

Journal of Nuclear Materials

journal homepage: www.elsevier.com/locate/jnucmat

Physical and mechanical modeling of the neutron irradiation effect on

ductile fracture. Part 2. Prediction of swelling effect on drastic decrease
in strength
Boris Margolin ⇑, Alexander Sorokin
Central Research Institute of Structural Materials ‘‘Prometey’’, Saint-Petersburg, Russia

a r t i c l e i n f o a b s t r a c t

Article history: A drastic decrease in the ultimate tensile strength of irradiated austenitic steels with high swelling values
Received 4 July 2013 is considered.
Accepted 20 May 2014 The physical–mechanical model proposed in Part 1 of the present paper is applied for the prediction of
Available online 2 June 2014
a drastic decrease in ultimate tensile strength. The mechanism called by the authors the ‘‘running col-
lapse mechanism’’ is used for modeling the material ductile fracture when stresses are less than the yield
strength. This ductile mechanism is similar to brittle fracture when crack propagates unstable manner.
Running collapse mechanism occurs due to evolution of vacancy voids resulting in irradiation swelling.
Nanoscale of vacancy voids (void sizes, distance between voids) results in the possibility of ductile frac-
ture in very small zones whose size is considerably smaller than the grain size.
Ó 2014 Elsevier B.V. All rights reserved.

1. Introduction It is known that at ductile fracture the ultimate tensile strength

There are a number of papers [1–3] devoted to experimental
investigation of the swelling effect on ultimate strength rul of irra-
diated austenitic steels. According to the paper [1], at a certain
swelling S a drastic decrease in ultimate tensile strength begins
at transcrystalline fracture by the mechanism of growth and coa-
lescence of vacancy voids. The decrease of rul may reach 5 times
over a short range of a swelling variation.
Fig. 1 shows the data obtained for 18Cr–10Ni–Ti steel with high
carbon content (0.11 wt.%) irradiated at the temperature Tirr = 400–
500 °C and tested at the temperature Ttest = Tirr [1]. According to the
results presented, rul decreases 5 times over the swelling range
from 17% to 27%, and the specimens fracture occurs at stresses
lower than the yield strength rY.
At present, there are scarcely any adequate physical models
explaining the sharp decrease in rul. Let us consider some of the
existing explanations.
The widespread explanation of such behavior of material
is reduction of specimen net cross-section due to vacancy voids
[1]. This explanation cannot be a satisfactory for the following
reason.
⇑ Corresponding author. Tel.: +7 812 710 25 38.
E-mail addresses: margolin@prometey2.spb.su, margolinbz@yandex.ru
(B. Margolin).
for a specimen with voids is higher than for a void-free specimen
when both the specimens have the same net cross-section area.
Such result is connected with a higher stress state triaxiality
(SST) in a specimen with voids.
Therefore the value rul of a specimen with the swelling S with
uniform distribution of vacancy voids over a specimen volume
[4] should be at least no lower than
 
rul ¼ 1  Av  rmul ; ð1Þ
where Av is the relative area of vacancy voids (the ratio of
voids cross-section area to the cross-section area of a specimen
with voids) resulting in irradiation swelling, rmul is the ultimate
tensile strength of a matrix material (a material without vacancy
voids).
The dependence of rmul on the neutron dose D is a monotonically
increasing one due to increasing the concentration of radiation
defects (dislocation loops and precipitates). With neutron doses
corresponding to a drastic decrease in rul (D > 40 dpa) rmul reaches
its maximum value and does not grow any more [4]. Thus a
decrease in ultimate tensile strength with swelling changing from
17% to 27% can be calculated by the formula:
 
rul jS¼27% 1  Av jS¼27%
¼ ; ð2Þ
rul jS¼17% 1  Av jS¼17%

http://dx.doi.org/10.1016/j.jnucmat.2014.05.050
0022-3115/Ó 2014 Elsevier B.V. All rights reserved.
608 B. Margolin, A. Sorokin / Journal of Nuclear Materials 452 (2014) 607–613

700

600

500
σul, MPa

400

300

200

100

0
0 5 10 15 20 25 30
S, %

Fig. 1. Dependence of ultimate strength on swelling for 18Cr–10Ni–Ti steel with

high carbon content (0.11 wt.%) for Ttest = Tirr = 400–500 °C: s, d are experimental
data [1] (black dot corresponds to specimen with flaw).

The swelling level of 17% corresponds to the upper limit of

swelling at which there is still no drastic decrease of ultimate ten-
sile strength according to [1]. The swelling level of 27% corre-
sponds to a decrease of ultimate tensile strength by 5 times.
To determine the relation between Av and S we will use the fol-
lowing simplified statements: all voids of a cubic form with the
Fig. 2. Voids coalescence area in the specimen of 16Cr–15Ni–2Mo–2Mn–Ti–V–B
same size are located in cube corners, i.e. they are spaced at equal
steel (ChS-68) irradiated by the neutron dose D = 67 dpa at Tirr = 475 °C; void
intervals. Then volume fraction is 9% [2].
3
dvac
S¼ 3 3
; ð3Þ
lvac  dvac Voids volume fraction f can be calculated by the formula

where dvac is the average size of a vacancy voids, lvac is the average  3
dvac
distance between centers of voids. f ¼ : ð5Þ
lvac
From Eq. (3) we have
2  2=3 According to Eqs. (3) and (5) S = f/(1  f). Then the value of
dvac S fc = 0.34 corresponds to Sc = 0.52.
Av ¼ 2
¼ ; ð4Þ
lvac 1þS Using the introduced parameter fc the dependence rul(f) is pre-
sented in paper [2] in the following form
Using Eqs. (2) and (4) let us calculate a possible decrease in rul
"  2=3 #
with swelling changing from 17% to 27%: f
m
rul ¼ r ul 1 : ð6Þ
rul jS¼27% fc
ffi 0:89:
rul jS¼17%
Based on Eqs. (3) and (5) the Eq. (6) can be represented in the
Thus a several times decrease in rul cannot be explained only by form
an increase in voids area when increasing swelling from 17% to 2 3
27%.
6 A
Considerable progress was made in the understanding of the rul ¼ rm
ul 41 
 v 7
5: ð7Þ
swelling effect on ultimate tensile strength of material in papers Av
c
[2,3]. The voids coalescence areas were revealed (see Fig. 2) when
investigating voids distribution by a transmission electron micro- As is seen, Eq. (7) describes a linear decrease in rul with a
scope. Such result was interpreted by the authors [2,3] as the pos- growth of Av beginning with Av = 0.
sibility of spontaneous voids coalescence in material under On the basis of the model analysis [2,3], the following conclu-
irradiation. sion may be made. The model does not allow one to describe the
In paper [2] on the basis of the so-called ‘‘sphere problem’’ [5] experimentally observed peculiarity of the dependence rul(S), i.e.,
the authors introduced the parameter X as a ratio of the sum of a drastic decrease in rul over a short range of a swelling increase.
volumes of all spheres to the volume of a material with spheres. According to the model, a decrease in rul proceeds monotonically
When X reaches some critical value Xc, the situation is possible beginning with S = 0.
when each sphere includes the center of a neighboring sphere. In Moreover, the model vulnerability consists in the assumption
this case the coalescence of spheres with the radius Rc can occur. that two events are identical: they are the coalescence of several
The numerical value of the parameter Xc is 2.7 [6]. vacancy voids and specimen fracture [2,3]. It is clear that the nucle-
Then it is assumed [2] that voids of the radius R = Rc/2 can also ation of a local flaw 1–3 lm in size does not necessarily mean spec-
coalesce spontaneously by contacting each other. Thus the critical imen fracture, especially if the fracture proceeds by the ductile
 3 mechanism due to the growth and coalescence of voids.
void volume fraction fc = Xc RRc = Xc/8 = 0.34 where the void vol- Thus the object of this paper is to investigate a cause of a
ume fraction is the ratio of the volume of voids to the volume of a drastic decrease in ultimate tensile strength rul and to apply the
material with voids. physical–mechanical model presented in the Part 1 of the present
 B. Margolin, A. Sorokin / Journal of Nuclear Materials 452 (2014) 607–613 609

paper [7] for prediction of the dependence of rul on irradiation As presented in the papers [11,12], at a certain swelling level
swelling. the specific voids area (the area of voids surfaces related to a mate-
rial volume unit) Asurf hardly changes with increasing swelling.
2. Model of the swelling effect on rul This means that from the moment when Asurf becomes a constant
the coalescence of voids begins because Asurf can be constant with
an increase in swelling in the case of void coalescence only (see
2.1. Setting up a problem Fig. 3).
Thus the introduction of Sinh as a swelling value corresponding
As was shown in the previous section, a drastic decrease in rul to beginning of formation of void distribution inhomogeneity is
with an increase in S cannot occur when the distribution of voids is experimentally justified [11,12].
uniform over the specimen volume. Hence, the necessary condition To describe the degree of inhomogeneity of void distribution let
max
for such a decrease is the formation of inhomogeneity in void dis- us introduce the ratio S where Smax is the maximum local (within
eS
tribution. Local fracture of a material is possible in the local region
the grain) swelling; eS is the average swelling by the specimen
with void volume fraction greater than average one. This fracture
volume.
can occur at stresses being considerably lower than the stresses
Let us take that this ratio is described by the dependence
corresponding to the fracture of a material with the average void 8
volume fraction. > 1 for e
S<S
As an example of a local fracture of material the investigation
Smax <  h inh i 
¼ 2 ð8Þ
results [2,3] discussed in the previous section of this paper (see e
S : exp a  e
> S  Sinh for e
S  Sinh
Fig. 2) can be presented. It seems the regions with coalesced voids
demonstrate the local fracture of a material with an increased void where a is the material constant.
volume fraction. This fracture may occur at the expense of the non-
uniform swelling of an irradiated material. 2.3. Specimen fracture conditions
The formation of inhomogeneity in void distribution is the nec-
essary condition of local fracture of a material at low stresses. The specimen fracture is analyzed on the basis of the following
However, is the local fracture followed automatically by the spec- considerations:
imen fracture, i.e., macrofracture? Generally, local fracture results
in macrofracture in case of brittle fracture only. Therefore, brittle 1. A polycrystalline material is presented as a conglomerate of
fracture is often described by the weakest link theory [8–10]. In cells with a linear size equal to the grain size. Under irradiation
most cases of ductile fracture, further loading of a specimen is the swelling of each cell is the same till e
S < Sinh. When eS P Sinh
required after its local fracture for the specimen macrofracture. the formation of void distribution inhomogeneity begins, and in
That’s why the ductile fracture of a material with a high void vol-
this case there are cells in which Smax P S > e
S. The dependence
ume fraction has a specific nature that results in the specimen
macrofracture immediately after the local fracture of a material of Smax on eS is described by Eq. (8).
with no increase in stresses. 2. In case of void distribution inhomogeneity, the fracture of cells
It may be assumed that ductile fracture at low stresses occurs if with S = Smax occurs first of all. With homogeneous void distri-
the following necessary and sufficient conditions are fulfilled. The bution the fracture of all the cells occurs simultaneously. The
necessary condition is the formation of inhomogeneity in void dis- fracture of a unit cell with voids proceeds by the mechanism
tribution that may result in the local fracture of a material at low of plastic collapse [7,13]. The condition of plastic collapse is for-
stresses; the sufficient condition is the extension of local fracture mulated as
zone (the propagation of a microcrack) till the specimen macro-
fracture with no increase in stresses.
dF eq
¼ 0; ð9Þ
In order to describe the fracture of a material with voids at low depeq
stresses it is necessary to understand the nature of inhomogeneity where Feq = req(1  AR ), req is the equivalent stresses acting in the
formation in the void distribution of a material and the nature of
void-free matrix material, AR is the relative void area, i.e. the ratio of
unstable ductile fracture as well as to formulate the necessary
voids cross-section area to the cross-section area of a unit cell with
and sufficient conditions quantitatively.
voids.; depeq is the equivalent plastic strain increment.

2.2. The condition of formation of inhomogeneous void distribution 160

First of all, to formulate such condition it is necessary to define

the homogeneous void distribution. The homogeneous void distri- 120
bution means the following situation: for a randomly chosen vol-
ume of a material with a linear size more than 10 times greater
80
than the average distance between voids the distribution of voids
sizes is the same.
During swelling the distribution of voids is homogeneous until 40
the intensive coalescence of voids begins. The process of void coa-
lescence results in an increase in void distribution inhomogeneity
since in some places the volume of voids will greatly increase. It is 0
assumed that void coalescence is the main process resulting in void
0 4 8 12 16
distribution inhomogeneity.
Void volume fraction f, %
A critical swelling level Sinh is proposed as a criterion of the
beginning of inhomogeneous void distribution formation. The Fig. 3. Dependence of the specific voids area on the value of void volume fraction at
investigation results [11,12] can serve as this criterion justification. different temperatures over the range from 410 to 560 °C [11].
610 B. Margolin, A. Sorokin / Journal of Nuclear Materials 452 (2014) 607–613

In Eq. (9) two populations of void are considered: vacancy voids where r eeq and eeeq are the average values of equivalent stresses and
resulting in swelling and deformation ones nucleating on inclu- strains under elastic deformation of a material, req and eeq are
sions and carbides under the material deformation. Value of AR equivalent stresses and strains under elastic–plastic deformation.
increases with increasing the Odqvist parameter (the strain path According to [18],
R
length) epeq ¼ depeq .
2  ð1 þ lÞ e
The constitutive equations for AR calculation (describing the eeeq ¼ r eq ; ð11Þ
3  Ev
nucleation of deformation voids as well as the growth of deforma-
tion and vacancy voids under deformation) are presented in Part 1 where l is the Poisson ratio under elastic deformation, Ev is the
of the present paper [7] and the dependences for calculation req Young modulus for a material with vacancy voids due to swelling.
are given in the papers [4,7]. It should be noted that the relative The value of Ev can be calculated by the following formula
volume and area of vacancy voids before deformation can be deter-
Ev ¼ E  ð1  f Þ ð12Þ
mined from radiation swelling.
where E is the Young modulus of a matrix material (the Young mod-
3. After fracture of the first cell a disk-shaped microcrack nucle- ulus of a material without swelling).
ates with a diameter equal to the grain size. Since voids do not grow under elastic deformation of a material,
4. At nominal stresses (related to the specimen cross-section) rN then f in Eq. (12) basing on Eqs. (3) and (5) may be calculated by
less than yield strength rY the specimen unstable ductile frac- the formula
ture proceeds by the mechanism referred to as a ‘‘running col-
lapse mechanism’’ (RCM). The RCM occurs if the following S
f ¼ : ð13Þ
conditions are fulfilled. Along the front of a microcrack nucleat- 1þS
ing due to the first cell fracture the condition (9) is met in the Taking into account the plane strain condition, r
 eeq is calculated
zone Z (so called ‘‘process zone’’) having the size rf at the same by the formula [18]
stresses as for the first cell fracture (see Fig. 4). It is taken that Z rf
swelling of the zone Z is equal to eS (see Fig. 4). If the condition 1 KI 2  ð1  2lÞ K I
r eeq ¼ ð1  2lÞ pffiffiffiffiffiffiffiffiffiffiffiffi ¼ pffiffiffiffiffiffiffi pffiffiffiffi ; ð14Þ
(9) is met, the fracture of the zone Z will occur, the microcrack rf 0 2p  r 2p rf
size and, correspondingly, the loading along the crack front will where KI for a disk-shaped microcrack is calculated by the formula
increase. Further microcrack growth will be unstable due to the [19]
following circumstances. rffiffiffiffiffiffiffiffiffi
2 dg
If the stress level is low and rp  rf, where rp is the size of a plas- K I ¼ rN p ð15Þ
 p 2
 2 
tic zone near a disk-shaped microcrack rp  61p rK0:2I ; then, where dg is the grain size, rN is nominal stresses acting in a
specimen.
according to [14], the ductile fracture of a material will be unstable
The value req is calculated for a material with voids by the
immediately after the microcrack start. If the stress level is high,
formula
rp > rf and KI > KC (KC is the critical value of stress intensity factor
h  i
meeting the condition (9) at the microcrack start), it is assumed req ¼ rmeq ðepeq Þ 1  AR epeq ð16Þ
that fracture can also be unstable since for highly irradiated mate-
dK R @K I dK R  p
rial is low [14] and the conditions KI = KR(Da) and
da
> [15]
@a da where rm eq eeq is the dependence describing a stress–strain curve

are fulfilled immediately after the nucleation of a disk-shaped for the matrix material.
microcrack, (KR(Da) is crack growth resistance in terms of SIF, The value of eeq is calculated by the formula
and a is the microcrack size; K R jDa¼0 ¼ K C and K R jDa>0 > K C .
2ð1 þ lÞ
eeq ¼ epeq þ req : ð17Þ
5. The RCM realization can be estimated by the following proce- 3Evd
dure. It is taken that in the zone Z the stress–strain state is In Eq. (17) Evd is the Young modulus for a material with vacancy
homogeneous. Strain in the zone Z is determined by the Nei- and deformation voids that depends on epeq . In this case the value of
ber’s equation applied to the stress–strain state calculation near Evd is calculated by Eq. (12) where Evd substitutes Ev and f is the
the crack tip [16,17] vacancy and deformation voids volume fraction taking into
account the growth of both void populations under deformation.
r eeq eeeq ¼ req eeq ð10Þ The dependence f(epeq ) can be calculated by the formulas pre-
sented in the paper [7].
Substituting Eqs. 11–17 in (10) we obtain

8ð1 þ lÞ ð1  2lÞ2 dg 2 2ð1 þ lÞ

v rN ¼ req epeq þ req : ð18Þ
3E p 2 rf 3E vd

At a given rN level, the plastic strain epeq in the zone Z is deter-

mined by Eq. (18).
The RCM is realized if the following condition is fulfilled:
 collapse
epeq  epeq : ð19Þ
Z
 collaps
In Eq. (19) epeq Z is the value of strain under which in a zone
Z condition (9) is met.
When calculating Eqs. (16), (17), and (19) the values of AR ; Evd
 collapse
Fig. 4. Zone Z (‘‘process zone’’) around the fractured first cell: is the zone Z;
and epeq Z are calculated under the stress state triaxiality
is the fractured grain schematized in calculations as a disk-shaped microcrack. qm  rreq
m
¼ 1:1 and q1  rreq1 ¼ 1:1 þ 23 where rm is the hydrostatic
 B. Margolin, A. Sorokin / Journal of Nuclear Materials 452 (2014) 607–613 611

component of stresses, r1 is the maximum principal stresses. Such Let us take the following parameters for modeling a sharp
values correspond to the average triaxiality near the tip of an inter- decrease in rul: Ttest = Tirr = 450 °C, E = 150 GPa, l = 0.3; Sinh = 12%;
nal disk-shaped microcrack with diameter equal to a grain size dg = 100 lm; a = 125.
[20]. The parameter Sinh is taken to be equal to swelling correspond-
ing to start of void coalescence. According to experimental data
2.4. Algorithm for rul calculation presented in Fig. 3 Sinh is assumed to be equal to 12%.
The parameter a is taken for best fit between available experi-
According to the conditions of specimen fracture formulated in mental data and predicted results.
the previous subsection rul can be calculated by the following As the zone Z in physical terms is area in which the elementary
algorithm. fracture event occurs, then the value of rf should exceed at least the
distance between voids. The value of rf in Eq. (18) may be deter-
1. When e S < Sinh the fracture of all unit cells occurs simulta- mined on the basis of dependence, proposed in Part 1 of the pres-
neously. In this case the value of rul is calculated on the basis ent paper [7].
of the following condition (
r def
f xe
S; when e
S<e
S
rf ¼ ð24Þ
h  i r vac when e
SeS
collapse m f ;
rul ¼ min r N ; r
ul  1  AR ; ð20Þ
where r def
f is the process zone size connected with the distance
where the nominal stress rcollapse
N is determined from the condition between deformation voids for material without swelling, rvac f is
 
(9) for unit cell with swelling e S. The dependences qm epeq and the process zone size connected with the distance between vacancy
 p voids, x is the coefficient of decreasing process zone size due to
q1 eeq for a unit cell are taken the same as for the center of a cylin-
drical specimen neck and calculated by the Bridgman equation [21]. swelling, e S is the some value of swelling for which the condition
  vac
rf = rf is met.
The condition rcollapse > rmul  1  AR corresponds to the
N Values of r vac
f and eS may be estimated by the following way.
situation when the load-carrying capacity loss of a specimen According to [23–25]
occurs before its fracture.
rvac
f ¼ k  lvac ð25Þ
2. When e S P Sinh the fracture of a unit cell with maximum swell- where k is the numerical coefficient; k = 2–8 [23–25], lvac is the
ing Smax occurs first. Value of Smax is determined by the Eq. (8). average distance between centers of vacancy voids.
In Eq. (25) value lvac is calculated by the formula
The value of rul in this case is calculated on the basis of the fol-
lowing condition 1
lvac ¼   ; ð26Þ
h  i q vac 1=3
v
f m
rul ¼ min r r  1  AR ;
N; ul ð21Þ
qvac
v is vacancy voids concentration.
where rfN is the nominal stress corresponding to specimen fracture According to [26], the value of qvac
v is swelling-independent over

by RCM. The value of rfN is calculated as maximum of two values the range from 1% to Sinh. For e
S > Sinh qvac
v shows a slight decrease:
h i a decrease in the vacancy voids concentration due to their coales-
rfN ¼ max rcollapse
N ; rRCM
min ; ð22Þ cence is compensated by the nucleation of new vacancy voids
under irradiation. According to the results obtained,
In Eq. (22) rcollapse
N is the nominal stress corresponding to the qvac  (2
4)  1012 1/mm3 [26] and qvac  4.5  1012 1/mm3 [11].
v v
fracture of the unit cell with maximum swelling, Smax; rRCM is
min Taking qvac
v = const  4.5  1012 1/mm3, k = 2 and substituting
the minimum nominal stress level at which the RCM is realized.
e these values in Eqs. (25) and (26) we obtain r vac
f  1.2  104 mm.
Value of rRCM
min can be determined for swelling S by the Eq. (18)
substituting plastic strain corresponding to fracture of zone Z with
 collaps
swelling e
S; epeq Z
, instead of epeq . 1000

2.5. Results of rul(e

S) calculation 800

In order to construct the dependence rul(e S) the weld metal of 600

σ, MPa

18Cr–10Ni–Ti steel made with 19Cr–11Ni–3Mo welding wire

without subsequent heat treatment is under study [4,7]. Depen- 400
dencies for determination of stress–strain curves for a matrix
material are presented in papers [4,7]. 200
The dependence of swelling on neutron dose for 18Cr–10Ni–Ti
grade steel and its weld metal is described by the following equa-
0
tion [22]
  0 5 10 15 20 25 30
S ¼ cD  Dn  exp r  ðT irr  T max Þ2 ð23Þ ~
S,%

where cD = 1.035  104; n = 1.88; r = 1.825  104 1/°C2; Fig. 5. Constructing the dependence of ultimate tensile strength rul (the line a–b–
Tmax = 470 °C. c–d) on swelling e
S: rm is the ultimate tensile strength of a matrix material (material
ul

Taking into account that Eq. (23) describes e S and specifying the without voids); rcollapse
N is the stresses corresponding to the fracture of first unit cell
value of Tirr, the neutron dose D can be calculated for each value of with swelling Smax; rRCMmin is the minimum level of stresses at which the RCM is
 
e
S. Then the parameters of stress–strain curves of matrix material realized; rm
ul  1  AR is a decrease in ultimate tensile strength due to the
are calculated for the obtained dose values. nucleation and growth of voids.
612 B. Margolin, A. Sorokin / Journal of Nuclear Materials 452 (2014) 607–613
The value of e
S is calculated from the condition
r def
f xe
S ¼ rvac
f : ð27Þ
Taking rdef
f= 0.1 mm and x = 1.884 [7], from Eq. (27) we have
e higher than some critical one show a brittle-to-ductile transition
S  5.2%.
Fig. 5 shows the calculated dependence rul(e
S) as the line a–b–c–d.
It also shows the dependence rm
ul (1  AR ) describing the loss of the
specimen load-carrying capacity. As is seen from the figure, for
e
S < 14% the ultimate tensile strength corresponds to loss of the
specimen load-carrying capacity; in this case fracture occurs at
higher stresses than the stresses corresponding to the necking.
For e
S > 14% the specimen fracture occurs before the loss of its
load-carrying capacity. For e S > 14%, i.e. for e S > Sinh, a drastic
decrease in rul occurs till S  24%. With e
e S > 24% rul decreases
insignificantly and remains at a very low level. This level coincides
with the minimum nominal stress (rRCM min ) level at which the RCM
is realized for a disk-shaped microcrack with size is equal to the
grain size.
The above dependence rul(e S) corresponds qualitatively to the
experimental data (see Fig. 1 as well as the Refs. [1–3]) including
a drastic decrease in rul over a short range of swelling change. It
should be noted that with e S > 14% the value of rul is controlled
by the fracture of a unit cell with maximum swelling Smax since
rcollapse > rRCM e fracture is the nucleation of a cleavage microcrack for which the
N min , but for S > 24% the value of rul is already con-
trolled by the RCM since rcollapse < rRCM
N min . The value of rul close to
zero (the black point in Fig. 1) seems to be connected with the
presence of a flow in the specimen.
3. Discussion of the results
According to the performed analysis, the fracture of a material
with e S > Sinh proceeds by the void ductile mechanism, however, it
has peculiarities typical for brittle fracture. Thus, unstable growth
of a microcrack with the size equal to the grain size proceeds
directly after its nucleation. As already noted, ductile fracture by
the nucleation of voids on inclusions, as well as by their growth
and coalescence does not have the described specific peculiarities.
Even for cracks with much larger sizes than the grain size its
growth is stable and occurs only with increasing loading. What is
the specific feature of ductile facture of an irradiated material with
voids? In our opinion, it is connected with a scale factor, namely,
with the size voids and distances between them. Let us consider
the size rf of the zone Z where an elementary event of fracture
can occur due to the coalescence of a microcrack nucleus with a
void. It was shown that rf is equal to rdef f for a material without
swelling and equal to r vac f when swelling is higher than 5% [7].
Thus, for an irradiated material with swelling rf = rvac f = 1.2  104
mm, while in the case of classical ductile fracture (the nucleation
of voids on inclusions) rf = r deff = 0.1 mm. As is seen, the zone size
rf differs
 approximately 1000 times. In addition, the critical strain
epeq collapse is considerably greater for the classical ductile mecha-
nism typical of an unirradiated material than for a material with
swelling and low strain hardening. Thus, rRCM min for the classical
ductile fracture is at least more than 100 times higher than for a
material with vacancy voids and makes up to over 20,000 MPa.
This value is vastly larger than the ultimate tensile strength of
austenitic steels in the initial condition; this value also exceeds
the ultimate tensile strength of steel irradiated at low tempera-
tures with no swelling. In these cases the RCM cannot be realized.
Thus, the ductile fracture of material occurs at low stresses due
to the nanoscale of voids nucleated under swelling. Such ductile
fracture at stresses lower than the yield strength shows features
typical for brittle fracture.
Another issue worthy of further discussion is connected with
the following circumstances. As was shown [27–29], swelling
results in a partial Fec ? Fea-transformation in 18Cr–9Ni and
18Cr–10Ni–Ti austenitic steels. Austenitic steels with the swelling
due to the partial Fec ? Fea-transformation [28,29]. For example,
cylindrical tensile specimen from 18Cr–10Ni–Ti steel with swelling
more than 7% tested at Ttest = 20 °C has the fracture strain close to
zero. In this case, fracture proceeds by the intercrystalline brittle or
transcrystalline quasi-brittle mechanism. So, the question arises
why fracture of a material with swelling S  7% (for example, for
S > 20%) occurs by the ductile mechanism and does not occur by
intercrystalline or quasi-brittle mechanisms?
Firstly, tests are normally carried out at Ttest = Tirr. In order to
obtain high swelling values Tirr should be close to a temperature
of 470 °C for 18Cr–9Ni and 18Cr–10Ni–Ti steels [22]. Therefore,
under the considered conditions Ttest will be higher than the brit-
tle-to-ductile transition temperature.
Secondly, at a high e S level the value of rul is lower than the
stresses required for the nucleation and growth of cleavage micro-
cracks. For example, for e S = 20–25% the value of rul = 100–400 MPa
(see Fig. 1), while the fracture stress rf at brittle frac-
ture  900 MPa. Therefore, ductile fracture will occur before stress
reaches the level sufficient for brittle fracture.
Thirdly, the necessary condition of transcrystalline brittle
Griffith condition is met. The minimum size of such a crack nucleus
is 100 nm. Relying on Eq. (3), for e S = 20% and the vacancy void
size dvac = 40 nm the distance between voids lvac  dvac  33 nm.
As is seen, the minimum size of a crack nucleus is larger than the
distance between voids. Therefore, a cleavage microcrack cannot
nucleate and result in brittle fracture.
4. Conclusions
1. It is shown that the experimentally observed effect of radiation
swelling on the ultimate tensile strength of austenitic steels
cannot be explained by the existing models.
2. There are two causes for ductile fracture of a material at low
stresses (lower than the yield strength) and a drastic decrease
in ultimate tensile strength:
– inhomogeneity of vacancy void distribution connected with
the process of void coalescence at a some swelling level
equal to Sinh;
– nanoscale of vacancy voids (void sizes, distance between
voids) resulting in the possibility of ductile fracture in very
small zones (100 nm) whose size is considerably smaller
than the grain size.
3. Ductile fracture with small process zone size makes an unstable
crack growth possible at low stresses by the mechanism called
‘‘running collapse mechanism’’.
4. The effect of radiation swelling on the ultimate tensile strength
of austenitic steel was modeled. The calculated results have
shown close agreement with the available experimental data.
References
[1] V.S. Neustroev, F.A. Garner, Very high swelling and embrittlement observed in
a Fe–18Cr–10Ni–Ti hexagonal fuel wrapper irradiated in the BOR-60 fast
reactor, Fusion Materials, vol. 43, Semiannual Progress Report for Period
Ending December 31, 2007, p. 109–122.
[2] A.V. Kozlov, I.A. Portnykh, S.V. Bryushkova, E.A. Kinev, J. Phys. Met. Metallogr.
95 (4) (2003) 379–389.
[3] O.V. Ershova, E.N. Shcherbakov, P.I. Yagovitin, et al., J. Phys. Met. Metallogr. 106
(6) (2008) 624–629.
[4] A.A. Sorokin, B.Z. Margolin, I.P. Kursevich, et al., J. Nucl. Mater. 444 (2014) 373–
384.
 B. Margolin, A. Sorokin / Journal of Nuclear Materials 452 (2014) 607–613 613

[5] H. Scher, R. Zallen, J. Chem. Phys. 53 (1970) 3759–3766. [20] K.F. Nilsson, N. Taylor, P. Minnebo, R.B. Bass, W. McAffee, Structural features
[6] J. Kurkijarvi, Phys. Rev. B9 (1974) 770. tests – embedded and surface flaws, in: Proceedings of International Seminar
[7] B. Margolin, A. Sorokin, J. Nucl. Mater., in press. ‘‘Transferability of Fracture Toughness Data for Integrity of Ferritic Steel
[8] W.A. Weibull, R. Swed. Inst. Eng. Res. 151 (1939) 5–45. Components’’, Petten, The Netherlands, 17–18 November, 2004, pp 108–124.
[9] F.M. Beremin, Met. Trans. 14A (1983) 2277–2287. [21] P.W. Bridgman, Studies in Large Plastic Flow and Fracture, McGraw-Hill, New
[10] B.Z. Margolin, V.A. Shvetsova, A.G. Gulenko, V.I. Kostylev, Fatigue Fract. Eng. York, 1952.
Mater. Struct. 29 (2006) 697–713. [22] N.K. Vasina, B.Z. Margolin, A.G. Gulenko, I.P. Kursevich, Probl. Mater. Sci.
[11] A.V. Kozlov, I.A. Portnykh, J. Phys. Met. Metallogr. 103 (1) (2007) 105– (Voprosy Materialovedenia) 48 (2006) 69–89 (in Russian).
109. [23] J.F. Knott. Fundamentals of Fracture Mechanics, Gruppo Italiano Frattura,
[12] A.V. Kozlov, I.A. Portnykh, J. Nucl. Mater. 386–388 (2009) 147–151. 1973.
[13] B.Z. Margolin, G.P. Karzov, V.I. Kostylev, V.A. Shvetsova, Fatigue Fract. Eng. [24] B.Z. Margolin, V.I. Kostylev, A.V. Ilyin, A.I. Minkin, Int. J. Pres. Ves. Pip. 78
Mater. Struct. 21 (1998) 123–137. (2001) 715–725.
[14] B.Z. Margolin, V.N. Fomenko, A.A. Sorokin, Strength Mater. 42 (3) (2010) 258– [25] J.R. Rice, M.A. Johnson, The Role of Large Crack Tip Geometry Changes in Plane
271. Strain Fracture, Inelastic Behavior of Solids, McGrow Hill, 1970, pp. 641–672.
[15] K. Hellan, Introduction to Fracture Mechanics, McGrow Hill Inc., 1984 [26] V.N. Voevodin, I.M. Neklyudov, Evolution of Structural-Phase State and
[16] G. Neuber, Stress Concentration (Russian Translation), Gostekhizdat, Moscow, Radiation Resistance of Structural Materials, Naukova Dumka, Kiev, 2006 (in
1947. Russian).
[17] B.Z. Margolin, A.G. Gulenko, V.A. Shvetsova, Int. J. Pres. Ves. Pip. 75 (1998) [27] D.L. Porter, J. Nucl. Mater. 79 (2) (1979) 406–411.
843–855. [28] B.Z. Margolin, I.P. Kursevich, A.A. Sorokin, et al., Strength Mater. 41 (6) (2009)
[18] S.P. Timoshenko, J. Gudier, Elastic Theory, Nauka, Moscow, 1978 (in Russian). 593–602.
[19] Y. Murakami (Editor-in-chief), Stress Intensity Factors Handbook, vol. 2. [29] B.Z. Margolin, I.P. Kursevich, A.A. Sorokin, et al., Strength Mater. 42 (2) (2010)
Pergamon Books LTD., 1987. 144–153.