Professional Documents
Culture Documents
art ic l e i nf o a b s t r a c t
Article history: The present study investigates the effect of deformation temperature and strain on the formation of twin
Received 24 December 2013 boundaries in Inconel 600. With a constant strain rate of 0.003/s, deformation temperatures were varied
Received in revised form between 25 1C and 982 1C while strains varied between 11% and 80%. The resulting microstructures were
21 February 2014
characterized using electron back scatter diffraction both immediately following deformation and after
Accepted 24 February 2014
Available online 4 March 2014
an annealing cycle at 1025 1C. Comparison of the grain boundary characters associated with the samples
enabled correlation of the processing parameters to the formation of desirable coherent twin boundaries
Keywords: and the identification of a deformation mechanism map. Processing parameters associated with strain
Grain boundary engineering annealing and dynamic recovery were found to promote the formation and retention of twin boundaries,
Thermo-mechanical processing
while statically and dynamically recrystallized microstructures tend to contain large proportions of
Coincident site lattice
random grain boundaries that mitigate the effects of grain boundary engineering.
& 2014 Elsevier B.V. All rights reserved.
http://dx.doi.org/10.1016/j.msea.2014.02.078
0921-5093 & 2014 Elsevier B.V. All rights reserved.
B. Li, S. Tin / Materials Science & Engineering A 603 (2014) 104–113 105
growth are not compatible with the large thermal inertia inherent in specimens with a diameter of 10 mm and a height of 15 mm. Using
large complex structures. Moreover, the multiple deformation and an MTS servo-hydraulic testing system equipped with a resistance
thermal processing cycles would add manufacturing lead time and furnace, a total of 16 samples were compressed at a strain rate of
cost even if the aforementioned limitations could be overcome. As 0.003/s to true strains of 11%, 28%, 51% and 80% at temperatures of
such, the current approaches used for GBE are not ideally suited for 25 1C, 538 1C, 760 1C and 982 1C. Strains were measured using a
the fabrication of physically large, complex components for propulsion MTS high temperature extensometer with alumina extension arms
and power generation applications. while temperatures were measured by averaging the values
In polycrystalline materials, the character of the grain boundaries is obtained from three thermocouples attached directly to the sur-
extremely important and greatly influences both the physical and face of the sample. For tests conducted at elevated temperatures,
mechanical properties of the material, particularly at elevated tem- the samples were immediately quenched in water following
peratures [20–22]. In recent years, ultra-fine grain processing techni- deformation in order to preserve representative microstructures.
ques have been developed to produce bulk metallic materials with The compressed samples were sectioned longitudinally along their
sub-micron grain sizes [23–27]. At relatively low temperatures, the centers using a low speed diamond saw. For each deformation
high densities of grain boundaries serve as obstacles to restrict temperature and strain level, one half of the sample was subjected
dislocation activity and resist fatigue crack propagation. In these to an annealing treatment that consisted of heating the sample
instances, the Hall–Petch relationship is valid for numerous material from room temperature to 1025 1C at a rate of 23.8 1C/min and an
systems and relates changes in the yield strength to grain boundary 18 min isothermal hold at 1025 1C. To investigate the effect of
density or grain size. However, for structural materials intended for annealing time on the formation of Σ3 boundaries, multiple
use at elevated temperatures, fine grain sizes and high densities of samples were deformed to 25% strain at 25 1C and sectioned.
grain boundary area are not necessarily beneficial as grain boundary These samples were placed directly into a furnace set at 1025 1C
sliding and the accumulation of cavitation damage contribute to the and annealed for 5, 18, 38 and 60 min followed by a water quench.
degradation of creep properties. The random grain orientations that Following deformation and/or heat treatment, all of the sam-
exist in isotropic polycrystalline materials lead to the formation of a ples were prepared using standard metallographic procedures that
random grain boundary network consisting of a statistical distribution incorporated a final polish using 0.05 μm colloidal silica. A JEOL
of grain boundary orientations. Depending on the relative orientations, 5900 equipped with an Oxford Instruments Nordlys Nano EBSD
the coherency of the individual grains with their neighbors or CSL at system was used to measure the character of the grain boundaries
the boundary may vary significantly. CSL boundaries, with Σ429, and quantify the average intragranular misorientations. Multiple
tend to contain large concentrations of crystalline defects and vacan- EBSD scans were taken from each sample to produce statistically
cies that serve to both weaken the interface and promote diffusive significant results and provide quantitative information at differ-
mechanisms at elevated temperatures. Under these conditions, grain ent length scales. High magnification scans were taken over an
boundary diffusion and sliding are accelerated along these boundaries. area spanning 40 40 μm2 using a 0.1 μm step size. These EBSD
This leads to environmental degradation and poor resistance to creep scans were used to produce inverse pole figures showing the
deformation. On the other hand, adjacent or neighboring grains that individual grain orientations and the magnitude of the intragra-
exhibit CSL boundaries, with Σo29, have relatively coherent interfaces nular misorientations. Lower magnification scans over an area of
that contain few crystalline defects that would serve to weaken the 350 350 μm2 using 0.7 μm step size were used to generate
boundaries. In this regard, coherent twin boundaries with Σ¼ 3 are datasets that contained statistically significant numbers of grains
highly desirable features in grain boundary engineered materials since for determination of the grain boundary characters. Data from
they possess some of the lowest grain boundary energies [5–16,28,29]. samples in the as-deformed condition were compared to those
Moreover, since there are comparatively fewer vacancies and defects following annealing.
along low Σ interfaces, the mechanisms by which diffusion and mass
transport occurs along the interfaces are also more sluggish. Thus,
polycrystalline materials containing high proportions of special CSL 3. Results
boundaries are desirable for use at elevated temperatures. In order to
induce the formation of sufficiently large fractions of special grain In the as-received or hot rolled condition, the IN600 sample
boundaries, strain energy in the form of dislocations first needs to be contained equiaxed grains with an average grain size of approxi-
stored in the material. Upon subsequent annealing, the supplied mately 7.7 μm, Fig. 1a. Twin or Σ3 boundaries were observed in
thermal energy enables the system to effectively utilize the stored this initial condition and EBSD analyses revealed that their length
strain energy to form twins and other special CSL boundaries. It is still fraction (total Σ3 length/total length of all boundaries) was 36%.
not entirely clear whether the underlying mechanisms responsible can The grain orientation map for hot rolled IN600, shown in Fig. 1b,
be attributed to dislocation recovery or grain boundary migration also reveals that there were no measurable changes in the
induced by the differences in strain energy. In this paper, the effect of intragranular misorientations that correspond to plastic deforma-
various hot deformation processing parameters on the formation of Σ3 tion or the presence of pre-existing dislocation networks and is
boundaries in IN600 was investigated. consistent with a fully recrystallized microstructure.
Cylindrical compression samples were deformed to various
strain levels at different temperatures. Depending on the deforma-
2. Procedure tion temperature, the samples exhibited characteristic flow beha-
viors, Fig. 2. The peak flow stresses were observed to decrease
Fully recrystallized hot rolled bars of IN600 with a composition dramatically as a function of temperature and ranged from
listed in Table 1, were machined into cylindrical compression 143 MPa at 25 1C to 16 MPa at 982 1C. Additionally, the degree or
magnitude of strain hardening was also observed to decrease as
Table 1 deformation temperatures increased. Only a modest amount of
Nominal composition of Inconel 600 (wt%). strain hardening was evident at 760 1C and no strain hardening
was observed during deformation at 982 1C. As samples were
Ni Cr Fe C Mn Si Cu deformed to various levels of true strain at these temperatures, the
Bal. 15.5 8.0 0.15 1.0 0.50 0.50
corresponding changes in the microstructure were characterized.
In the as-deformed condition, the magnitude of the intragranular
106 B. Li, S. Tin / Materials Science & Engineering A 603 (2014) 104–113
Table 2
Average internal misorientation of as-deformed samples (deg).
25 1C 0.423 0.776 – –
538 1C 0.339 0.587 – –
760 1C 0.249 0.433 0.461 0.640
982 1C 0.168 0.207 0.265 0.397
Fig. 3. Orientation images from as deformed samples of IN600 at (a) 760 1C/11% strain, (b) 760 1C/28% strain, (c) 760 1C/51% strain, (d) 760 1C/80% strain, (e) 982 1C/11% strain,
(f) 982 1C/28% strain, (g) 982 1C/51% strain, (h) 982 1C/80% strain.
108 B. Li, S. Tin / Materials Science & Engineering A 603 (2014) 104–113
Fig. 4. Band contrast maps showing the grain structure and density of Σ3 boundaries following a one hour anneal at 1025 1C of the IN600 samples deformed at (a) 25 1C/11%
strain, (b) 25 1C/28% strain, (c) 25 1C/80% strain, (d) 538 1C/11% strain, (e) 538 1C/28% strain, (f) 538 1C/80% strain, (g) 760 1C/11% strain, (h) 760 1C/28% strain, (i) 760 1C/80%
strain, (j) 982 1C/11% strain, (k) 982 1C/28% strain and (l) 982 1C/80% strain.
Table 3
Grain boundary character of samples deformed at different temperatures and strain
followed by a one hour anneal at 1025 1C.
length in each of the samples was averaged and plotted in the form of
a surface plot as a function of both strain and deformation tempera-
ture, Fig. 5. Characterization of the as-deformed samples revealed that
in all instances, the overall length of the Σ3 boundaries decreased
modestly when compared to the total length of Σ3 boundaries in the
as-received IN600 material prior to deformation (13,211 μm). Follow-
ing annealing, however, Σ3 boundaries were created and the mea-
sured length of Σ3 boundaries in samples deformed at 25 1C/11%
strain, 538 1C/11% strain, 760 1C/11% strain, 25 1C/28% strain, 760 1C/
28% strain increased modestly to length values greater than 13,211 μm.
The Σ3 boundary lengths in all other samples were measured to be
less than 13,211 μm, with the sample deformed at 25 1C/80% strain to
have the lowest Σ3 boundary length of 9312 μm.
The length of the random boundaries for all of the annealed IN600
samples was also averaged and plotted in the form of surface plot,
Fig. 6. Surface plot showing the measured random boundary length following a Fig. 6. As the initial grain size of the as-received material was 7.7 μm,
one hour anneal at 1025 1C of the IN600 samples deformed to various strains at the overall grain length of random grain boundaries over a
different temperatures. 350 μm 350 μm scan area was 23,314 μm. Grain growth occurred
in all of the deformed IN600 samples following annealing and this led
to an overall reduction in the length of the random grain boundaries.
Since samples deformed to strains of 28% and 51% at deformation
temperatures of 25 1C, 538 1C and 760 1C were observed to recrystal-
lize either partially or fully, the resulting grain refinement served to
create random grain boundaries and counteract the reduction attrib-
uted to grain growth. This is reflected in by the bulge in the center of
the surface plot in Fig. 5. Overall, the average length of the random
grain boundaries was observed to be inversely proportional to the
average grain sizes measured for the specimens, Fig. 6. Large grain
sizes correspond to relatively low values of the random grain
boundary length while the fine grains were associated with the higher
values. The average grain size was observed to increase as a function of
temperature, but decrease as a function of increasing strain.
With quantification of the average lengths associated with
random and Σ3 boundaries in the various samples, the length
fraction ratio (total Σ3 boundary length/total length of all bound-
aries) was plotted, Fig. 7. By comparing the length fraction ratio,
samples containing large Σ3 boundary lengths and relatively small
random boundary lengths were readily revealed. Irrespective of
Fig. 7. Surface plot showing the variation in grain size following a one hour anneal
at 1025 1C of the IN600 samples deformed to various strains at different
the deformation temperature, the samples deformed to 11% all
temperatures. exhibited relatively large grain sizes and Σ3 lengths after anneal-
ing. This resulted in the Σ3 length fractions ranging between 64%
and 81% for samples deformed at 25 1C to 982 1C, respectively.
These values are significantly higher than the Σ3 length fraction of
36% measured in samples of IN600 in the as-received or hot rolled
condition. With recrystallization and grain refinement occurring
during deformation or following annealing in the samples
deformed to strains of 28% and 51% at temperatures of 25 1C,
538 1C and 760 1C, the resulting increase in the random grain
boundary length was responsible for a decrease in the Σ3 length
fractions. The microstructures of samples deformed to 80% strain
at 25 1C, 538 1C and 760 1C followed by the 1025 1C anneal were
fully recrystallized and experienced some grain growth. Under
these conditions, the Σ3 length fractions were observed to increase
modestly. As shown in Fig. 2e–h, no evidence of dynamic recrys-
tallization was observed following deformation to the various
Fig. 8. Surface plot showing the length fraction of Σ3 boundaries following a one
strain levels at 982 1C. The samples deformed at 982 1C exhibited
hour anneal at 1025 1C of the IN600 samples deformed to various strains at the largest average grain sizes, Fig. 6, and the highest Σ3 length
different temperatures. fractions, Fig. 7, following annealing. Although grain sizes of the
samples deformed at 982 1C are significantly larger than those
deformed at lower temperatures, they are still smaller when
twin boundaries. In the other samples, the majority of twin compared to the average grain size observed in undeformed
boundaries were observed to either bisect the grain or originate samples of IN600 that were subjected to the identical annealing
from grain boundaries. heat treatment at 1025 1C. Fig. 9 shows the electron band contrast
Results from the assessment of the character of the grain bound- image of an undeformed sample of IN600 annealed at 1025 1C for
aries in the various samples are summarized in Table 3 and Figs. 5–8. one hour. The average grain size was found to increase from 7.7 μm
Over a constant scan area of 350 μm 350 μm, the total Σ3 boundary to 69.9 μm as a result of the annealing treatment only.
110 B. Li, S. Tin / Materials Science & Engineering A 603 (2014) 104–113
4. Discussion
predominately confined to their slip planes. As temperatures dislocation densities following deformation. These residual dislocation
increase to 760 1C and 982 1C, thermal activation enables disloca- densities in dynamically recrystallized grain structures were found to
tion climb and the formation of low energy sub-grain networks. affect the resulting grain boundary character and formation of Σ3
Interestingly, the overall length of Σ3 boundaries in samples boundaries following annealing. As seen in Fig. 3a–d, deformation of
deformed to 11% at 25 1C, 538 1C and 760 1C were nearly equiva- IN600 at 760 1C resulted in dynamic recrystallization. The fraction of
lent, but their grain sizes and corresponding length of random the microstructure that had dynamically recrystallized was propor-
boundaries in these samples varied. Among these samples, the tional to the strain and the microstructure has completely recrystal-
observed increases in the Σ3 length fractions can be largely lized as the sample was being strained to 80%. Consistent with the
attributed to reductions in the length of the random boundaries intragranular misorientations measured in these samples, Table 2,
as opposed to increases in Σ3 length. Additionally high relative remnant plastic deformation was stored within the microstructure.
proportions of internal twins, or self-contained twins that are Annealing of these samples resulted in strain induced growth of the
present within the grains, were observed in samples deformed at dynamically recrystallized grains, Fig. 4h and i. Unlike the highly
760 1C and 982 1C. These observations were consistent with low strained samples deformed at 25 1C and 538 1C where static recrys-
mobilities associated with Σ3 boundaries during grain growth. tallization results in a strain or dislocation free microstructure where
As expected, strain annealing mechanisms were responsible for the reduction in surface or interfacial energies drive grain growth,
increasing the overall length and length fraction of Σ3 boundaries dynamic recrystallization produced a refined grain structure contain-
in the samples deformed to 11% strain, but increases in grain size ing a sufficiently high strain level or dislocation density that was able
due to elevated deformation temperatures also contributed to to enhance the formation of Σ3 boundaries during annealing.
increasing the Σ3 length fractions. Although the overall length of the Σ3 boundaries measured in the
sample deformed to 80% at 982 1C was nominally identical to that
4.2. Static recrystallization measured in the same sample following annealing, the average grain
size increased from 14.5 μm to 24.8 μm as a result of the annealing,
Recrystallization events are typically avoided during grain Fig. 7. The larger grain size also corresponds to a significant reduction
boundary engineering as the nucleation and growth of recrystal- in the overall length of random grain boundaries and results in a net
lized grains tends to populate the microstructure with high overall increase in the length ratio of Σ3 boundaries, Fig. 8. Hence,
fractions of high angle or random grain boundaries [6,8]. Σ3 strain-annealing and dislocation grain boundary interactions that
boundaries may form in recrystallized microstructures, but will occur within fully, dynamically recrystallized microstructures do
likely occur naturally as annealing twins induced during growth of appear to increase the overall length ratio of Σ3 boundaries at large
the recrystallized grains. The proportion of Σ3 annealing twin magnitudes of strain.
boundaries will scale proportionally as a function of grain size and
forming high proportions of them in fine grained microstructures 4.4. Dynamic recovery
will be limited as dislocation – grain boundary interaction events
are absent in recrystallized microstructures. A bi-modal grain size A deformation temperature of 982 1C corresponds to a homo-
distribution characteristic of partial recrystallization was observed logous temperature, Tm, of 0.77 for IN600. Considering that the
to occur during annealing of samples deformed to 28% and 51% microstructure of IN600 consists of a solid solution strengthened,
strain at 25 1C and 538 1C, Fig. 4b and e. Due to the relatively small single phase austenitic structure, dynamic recovery mechanisms
size and lack of Σ3 boundaries in the newly recrystallized grains, can dominate and accommodate high levels of strain during high
the fraction of random grain boundaries increased, Fig. 6, while temperature deformation [36,37]. Since there was no evidence of
the grain size decreased, Fig. 7. Since the recrystallized grains are grain refinement in the microstructures immediately following
also replacing the original grain structure that contained moderate deformation at 982 1C, Fig. 3e–h, dynamic recovery was likely the
fractions of Σ3 boundaries, the overall length of the Σ3 boundaries dominant mechanism responsible for accommodating the
in the partially recrystallized specimens also decreased, Fig. 5. imposed strain. The lack of hardening observed in the flow stress
Increasing the deformation strain to 80% produced fully recrystal- curves, Fig. 2, and the relatively low magnitude of the average
lized microstructures after annealing, Fig. 4c and f. Once fully intragranular misorientations measured in the as-deformed
recrystallized, the new strain free grains do grow during annealing microstructures, Table 2, were all consistent with dynamic recov-
and some Σ3 annealing twins form. Compared to the samples ery occurring during deformation. As the samples were being
deformed to 11% at 25 1C and 538 1C, grain refinement does occur deformed at 982 1C, the macroscopic strains were accommodated
following annealing of the 80% deformed samples, Fig. 7. As there by dislocations rearranging themselves into low energy configura-
was no residual strain or significant dislocation density within the tions or three dimensional subgrain networks. The enhanced
grains to stimulate the formation of Σ3 boundaries during anneal- mobility of dislocations at 982 1C enabled dislocation climb and
ing the average overall Σ3 length and Σ3 length fraction are annihilation events to occur within the subgrain walls and sup-
relatively low in these fully recrystallized samples, Figs. 5 and 8. press strain hardening and dynamic recrystallization. Interestingly,
deformation at 982 1C did not result in a significant overall
4.3. Dynamic recrystallization reduction in the measured length of the Σ3 boundaries, although
some grain growth did occur during deformation at 982 1C as the
During static recrystallization, dislocation recovery and recrystalli- average grain size increased from an initial size of 7.7 μm to
zation are limited to occurring following deformation and during the 14.7 μm immediately following deformation to a true strain of
subsequent heat treatment [30–35]. In these instances, the stored 80%. By preserving much of the original Σ3 boundary length while
strain energy serves as the driving force for the nucleation and growth simultaneously reducing the length of random grain boundaries
of strain free grains. Dynamic recrystallization, however, occurs when via grain growth during deformation, the length fraction of Σ3
the deformation temperatures and strain rates enable recovery and boundaries could be increased. The subsequent annealing of the
recrystallization events to occur while strain is being accumulated. In specimens deformed at 982 1C was found to induce some addi-
these instances, newly recrystallized grains form during deformation. tional grain growth and result in a corresponding reduction of the
As they form and grow, these initially strain free grains are also able to random grain boundary length, but did not serve to generate
accommodate additional straining. Unlike statically recrystallized grain additional Σ3 boundary length. Based on these observations, it
structures, dynamically recrystallized grains may contain significant appears that under deformation conditions where dynamic
112 B. Li, S. Tin / Materials Science & Engineering A 603 (2014) 104–113
[32] H.J. McQueen, J.J. Jonas, Recovery and Recrystallization during High Tempera- [36] H.J. McQueen, W. Blum, Mater. Sci. Eng. A290 (1) (2000) 95–107.
ture Deformation (Metal Hot Working). Plastic Deformation of Materials [37] E. Nes, K. Marthinsen, Y. Brechet, Scr. Mater. 47 (9) (2002) 607–611.
(A 75-46501 24-26), Academic Press, Inc., New York (1975) 393–493. [38] N. Souaï, N. Bozzolo, L. Nazé, Y. Chastel, R. Logé, About the possibility of grain
[33] R.L. Goetz, V. Seetharaman, Scr. Mater. 38 (3) (1998) 405–413. boundary engineering via hot-working in a nickel-base superalloy, Scr. Mater.
[34] W. Roberts, H. Boden, B. Ahlblom, Met. Sci.13 (3–4) (1979) 3–4. 62 (11) (2010) 851–854.
[35] S. Gourdet, F. Montheillet, Acta Mater.51 (9) (2003) 2685–2699.