Sei sulla pagina 1di 7

J Fail. Anal. and Preven.

(2010) 10:351357
DOI 10.1007/s11668-010-9374-3

FEATURE

Failures Induced by Abnormal Banding in Steels


F. DErrico

Submitted: 19 February 2010 / in revised form: 25 June 2010 / Published online: 16 July 2010
ASM International 2010

Abstract In industrial applications failures of mechanical


parts made of carbon and alloyed steels may develop either
during heat treatment steps or final finishing operations.
Such failures have high impact costs for manufacturers,
since heat treated steel products, in general, are high value
products which increase in value with each step in the
production process until the final life-cycle manufacturing
steps are completed. This work highlights the selection of
steels to avoid premature ruptures developing during either
the heat treatment steps or finishing operations with
emphasis on the role of banding in the failure process.
Failure does not have to involve fracture but may simply
imply a decrease in performance of surface treated components as consequence of surface properties, even in the
presence of correct heat treatment parameters. The root
causes for banding in steels, are described in literature, and
banding has major effects on final product properties (and
causal relationships). Therefore, the causes of banding are
studied and classified. Conclusions suggest that microstructural defects such as (micro)segregation bands and
other defects such as slag and oxides inclusions are
developed in the early fabricating cycle steps and can cause
premature failure of either semi-finished or finished
products.
Keywords Segregation  Heat treatments  Toughness 
Banding  Anisotropy  Martensite 
Transformation plasticity  Failure  Austenite

F. DErrico (&)
Department of Mechanical Engineering,
Politecnico di Milano, Milan, Italy
e-mail: fabrizio.derrico@polimi.it

Introduction
Basically the ultimate quality of steel products is determined from the steelmaking technological cycles and the
casting process technologies employed to fabricate raw
products. The blooms, billets, bars, and slabs that are used
as input to the forging, rolling, or machining operations are
used to finally shape the product and the post shaping heat
treatments and finishing all impact component quality.
While the chemical composition of a steel may be
defined, in practice a solid product may not reach uniform
chemistry over its entire section and the chemical composition varies, as does the final microstructure. Thus, the
mechanical properties, which are related to material
chemistry and microstructure, can vary greatly and product
quality is ultimately influenced by chemistry and microstructural variations.
A homogeneous distribution of alloying elements in
steels can be easily achieved by stirring while steel is in a
liquid state. At the end of steelmaking process, liquid phase
is essentially uniform in chemistry. On the other hand,
solidification processes generally develop macroscopic and
microscopic partitioning of chemical elements between
parent liquid and growing solid crystals. This fact causes a
non-uniform distribution of chemical elements to be
inherited by the as-cast material. For example, partitioning
(segregation) of chemical elements can occur abnormally
at the centerline of continuously cast steel products. This
chemical variation can be macroscopically observed by
naked eye inspection of microstructure simply comparing
the top and bottom, as well as the centerlines, of ingots.
On the other hand, a microscopic scale segregation (namely microsegregation) can occur when chemical partitioning develops along dendrites section. As
cast, semifinished products are hot deformed and any

123

352

J Fail. Anal. and Preven. (2010) 10:351357

Fig. 1 Banding on a UNI EN


18 CrNiMo 76 in normalized
condition: (a) acceptable
banding for post-heat treatments
and (b) worst case of banding
for further heat treating

micro-segregations in the casting are aligned to form


elongated bands. Since bands are formed by different
microstructural constituents, they are often identifiable by
metallographic analyses and clearly appear on normalized
or annealed hypoeutectoidic steels. These elongated
regions of varying chemical composition are termed bands
and the banding phenomena in steels produces alternating
layers of ferrite and pearlite [13].
Extensive literature on banding structure and metallurgical phenomena related to banding are available from the
early twentieth century [411]. According to these previous
studies, bands develop parallel to the rolling direction or
forging fibers especially in slow-cooled carbon and alloy
steels. The root cause of banding is ascribed to interdendritic microsegregation that is almost always present in cast
steel. The microstructural appearance of such banding
depends on both austenite grain size and cooling conditions that control austenite decomposition. Two possible
origins of microsegregation are recognizable, according to
Kirkaldy et al. [12]: (a) the pre-segregation and (b) the
transegregation. Pre-segregation refers to segregation
occurs during (dendritic) solidification and transegregation
refers to segregation occurring during the solid-state
transformations (austenite to ferrite-perlite structure) as
steels are heat treated and cooled.
Consider solidification of pure FeC alloy: in the austenite phase, carbon has a high diffusion coefficient,
namely, Dc and in fine dendritic structure, the microsegregation of C previously formed through the dendrite
solidification process should be homogenized by postprocess heating (i.e., hot-working steps, annealing heat
treatments, etc.). This implies that pre-segregation of C is
essentially nonexistent and banding should not common in
high purity FeC alloys.
Conversely, consider solidification of a generic FeCX
alloy, being X a solute element can be prone to presegregation. In contrast to carbon, the X element is generally characterized by very low diffusion coefficient in
austenite. This implies that any pre-existing (dendritic)
microsegregations of X should not be reduced during

123

heating and that homogenization of X is unlikely. Unfortunately, a pre-segregation of X may cause a presegregation of C, since X may influence the C activity.
The fact that banding is observed in carbon steels seems
to be in contrast with theory but in reality, a certain percentage of manganese and phosphorous (elements that
promote banding) remain from the steelmaking processes.
In these cases, ferriteperlite banding (see an example in
Fig. 1) is ascribable to the effects of manganese and
phosphorous on carbon activity. Manganese acts directly
on Ar3 temperature, stabilizing austenite by lowering Ar3.
Hot-working elongates the bands with varying Mn content
and the aligned bands are formed. By the premises that:
(a) Mn variations induce Ar3 variation and (b) proeutectoid
ferrite grains initiate to form (from austenite grains) as Ar3
temperature is reached, the first ferrite grains to form
provide low-content manganese areas. As the ferrite grains
are formed, carbon and manganese are progressively
rejected from the proeutectoid ferrite to form aligned bands
which are rich in carbon and manganese. Pearlite develops
in these areas [13]. One driving factor for the banding
phenomena is related to the dimension of prior austenitic
grains, in comparison with the inter-spacing between
aligned microsegregations of Mn (see scheme in Fig. 2).
When the austenitic grain size is smaller than the microsegregation interspacing, ferrite grains nucleate on
austenite grain boundaries and triple lines in the Mn-poor
regions (with high Ar3 temperatures); these grains grow in
alignment with those low-Mn content (i.e., the rolling
direction), and then proceed. In transverse direction carbon
is rejected and pearlite structures are formed. These highMn content regions that are adjacent, favorite growth of
pearlite bands, because of two contemporary effects:
(a) Ar3 is lowered for Mn high content, (b) increasing in
carbon content, up to eutectoidic concentration, in such
regions as it is rejected by proeutectoid ferrite grains
formed at higher Ar3 temperatures (i.e., lower Mn content
regions). A very large change in C concentration exists
between bands, as it must go from a value of around 0.02%
in a ferrite band to a value of around 0.77% in the pearlite

J Fail. Anal. and Preven. (2010) 10:351357

353

Fig. 2 Main driving factor of such phenomena is austenitic grain size vs. inter-spacing between aligned microsegregations of Mn. As for the
case I, if austenitic grain size (dotted line) is smaller than the microsegregation interspacing f, then ferrite grains (white grains) nucleate first on
austenite grain boundaries (and triple points) in Mn-poor regions (with high Ar3 temperatures). These grains grow in alignment with those lowMn content (i.e., the rolling direction), and then proceed. Conversely, in the case II, as austenitic grain size (dotted line) is greater than the
interspacing f of segregation, it could result in the disappearance of banding

bands. Increasing the austenitic grain size so that it is


greater than the interspacing of segregation, could result in
the disappearance of banding [13].
A quite different phenomenon, only in terms of initiating mechanisms, occurs as banding forms around elongated
MnS inclusions. In this case, previous interdendritic segregation concentrates both Mn and S and these elements
combine to form MnS inclusions. Around MnS inclusions,
Mn content is expected to be lower than in regions some
distance from the inclusion. The banding mechanism then
proceeds in similar way as previously described. In addition, phosphorus has a very strong tendency to segregate
during solidification and may be a factor in the banding
process. However, Mn is generally present in much higher
concentrations than P and plays a more important role in
segregation and banding than P.
The scale of chemical partitioning in the banded
structure is strongly dependent on the conditions of solidification that determine dendrite size. The smaller the
section size, the faster the solidification process and the
finer the scale of dendritic structure and associated segregation. Therefore, the continuous casting of smaller
sections is beneficial but may not eliminate banding,
because less homogenizing hot work is required to produce
finished sections. The potential for banding in continuously
cast billets, for example, may effect the suitability of hot
rolled to bars for forging applications.
High superheating in unstirred billets increases the size
of the columnar zone because the nucleation of equiaxed
dendrites is retarded. Electromagnetic stirring reduces the
effects of high superheats, but does not completely compensate for the increased size of columnar zones developed
by high superheat temperatures.
In summary, with a reasonable shape reduction realized
by hot-working steps, dendritic crystals are broken up and
then recrystallized into new austenite grains. Especially
during hot rolling steps, interdendritic segregation is broken but local variations of solute elements, mainly Mn,

develop in bands parallel to the rolling direction. Alternating regions of high and low concentrations of solute
elements have strong effect on Ar3 temperature and, consequently, inhibit the development of a homogeneous
carbon distribution inside products.

Integrity and Quality of Products


Several studies of the effect of banding on mechanical
properties exist in literature [1417]. In normalized or
annealed banded hypoeutectoidic steels, low variations can
be found on transversal and longitudinal tensile properties,
while significant variations of both reduction in area and
impact properties are found. Banded steels always contain
some degree of elongated inclusions (usually sulfides),
which can contribute to a further reduction of transverse
impact properties. Therefore, the anisotropy of area
reduction and impact properties may be due to the banding
or to the elongated inclusions and separation of these
effects is not so easy because heat treatments to removing
banding often lead to spheroidization of elongated sulfides.
Banded structure provide processing difficulties when
the heat treatments and post-finishing operations are
designed on the basis of the nominal chemical analysis of
steels. In Fig. 3 is shown the matrix of those critical steps
to control integrity and final mechanical properties of highvalue products.
The bulk chemical analysis of the steel is known
and hardening heat treatments are designed to fit time
temperature dependent transformation of austenite having
this chemical composition. This is done in order to optimized the mechanical properties in final products.
Unfortunately, the composition of neither band is equivalent to the bulk steel chemistry. Therefore, the bands may
respond to the heat treatments in an inappropriate manner.
In order to identify the root causes for industrial unconformities of final products that are associated with

123

354

J Fail. Anal. and Preven. (2010) 10:351357

Fig. 3 Matrix of critical steps to control microstructure through a


standard cycle for hot-working products

inhomogeneous in the pre-existing structures, a classification of possible problems is proposed: (a) intergranular
quench cracks induced by a high-C content that leads to
martensite bands after quenching; (b) high retained austenite levels in over-carburized layers that may result from
the high-C content bands; and (c) dimensional anisotropy
during phase transformation induced by banding.

Heat Treatment Failures Induced by Pre-Existing


Banded Microstructures
The use of incorrect cooling rates in steel processing is the
main cause of heat treatment failures. Banded regions are
characterized by either higher or lower carbon concentrations than nominal value and therefore should require a
different heat treatment schedule. This is clearly impossible and the schedule is typically set up on the basis of
nominal chemical composition. Therefore, a highly banded
steel will be subjected to heat treating cycles not properly
chosen for either portion of the band. This fact should be
taken into account and considered to be a basic condition
when developing the heat treatment cycle parameters for
banded steels.
For example, in quench hardening steels, highly banded
structure induces an Ar3 temperature that can significantly
vary in comparison to the standard composition. Therefore,
when the austenitizing temperature is set up for the standard steel grade of nominal composition the temperature
could be too high for high-C content areas (over-heating
can occur) and too low in low-C content bands (partial
austenitization can occur). As the steel is quenched to
obtain martensite structure, higher C content regions could
be subjected to (a) increasing crystal lattice stresses and

123

Fig. 4 A pinion gear shaft (UNI EN 18NiCrMo5) broken during


quenching step, after carburing: (a) the part fractured and (b) scheme
of crack path initiating from point A

(b) lowering Mf temperature. This can result in drastically


lowering toughness of the martensite after a direct
quenching step. In general, quenching cracks in such cases
are caused by a stress field that develops during the
transformation from austenite to martensite (because of
lattice volume changes). These stresses are often superimposed on thermal stresses that develop during the
cooldown. Depending on C content in bands the quenching
step may overstress the material and fast-rupture (quench
cracks) can occur. This is especially true when internal
defects such as inclusions are present.
For example, small inclusions that do not effect the
quenching process in a homogeneous medium-carbon steel
can become really unsafe in a highly banded microstructure
that is quench-hardened.
In the case shown in Fig. 4, a pinion gear was broken
during quenching after carburizing. The crack originated
from a fastening hole for machining operations. Such a
rupture could be ascribed to quenching step and one
could logically conclude that, as crack developed during

J Fail. Anal. and Preven. (2010) 10:351357

355

Fig. 5 Inner microstructure, as sectioned samples were polished: it is


visible the crack path starting from the origin point A (fastening hole
for machining). The crack originated from a quasi-continuous
network inclusions, basically produced in steel-making (continuous
casting) and so transferred in semifinishing hot-working steps

Table 1 Simulation data to easy evaluate minimum critical length


for a defect to cause fast brittle fracture at the end of quenching
Quenching residual
stress, MPa

Critical length
(ac), mm

25

300

1.535

15

300

0.553

300

0.061

KIC, MPa Hm

It has to be highlighted simulation considers the same value of


residual stress at the end of quenching, while for high-C content
regions it could be higher

quenching, incorrect heat treatments parameters should be


identified as root cause of failure. Unfortunately, such a
coarse deduction is based on a wrong premise. While it is
true that the crack started during quenching, but one must
consider that although quenching may induce the cracks,
but may not be the root cause of cracking. To make this
point clear, consider the simulated data reported in
Table 1: stresses lower than material strength could be

Fig. 6 Microstructure of a UNI EN 17 CrNiMo 67: (a) quenched


sample, after polishing, shows the presence of elongated MnS
inclusions and (b) the same after etching, showing a banded
microstructures with high-carbon martensite (light bands) and lowcarbon martensite (dark bands)

sufficient to make small defects to become unstable, and


brittle fast fracture might propagate during the quenching
step (that is, the final critical step for manufacturing
product cycle); the result of initiate point of fracturing after
quenching is visible in Fig. 5. As quenching step transforms crystal lattice from austenite to martensite, such a
microstructural change is critical for two concurrent
effects: (a) steel toughness decreasing and (b) thermal
stresses increasing. It is not infrequent that, even in absence
of a macroscopic notch, the presence of small defects (see
Table 1) can lead to fast fracture. In this case, a banded
microstructure (see an example in Fig. 6) is very sensitive
to such type of fast brittle rupture, since in region of highcarbon martensite develop a very low KIc value.

123

356

Fig. 7 Microstructure of a UNI EN 18 CrNiMo 76 after carburizing.


White areas with abnormal content of retained austenite are clearly
visible in elongated areas as they developed during quenching
because of presence of pre-existing high-carbon content bands

As high-carbon content increase, the Mf temperature


decreases. In carburizing low-alloyed steel, commonly
used in gear industry, other alloy elements (especially, Cr
and Mo) contribute to lower the Mf temperature. As consequence of such general premises, many alloyed steels can
develop non-conformities related to high retained austenite
content in most superficial carburized layers (Fig. 7).
Finally, dimensional anisotropy may develop due to an
abnormally (pre-existing) banded microstructure. Several
microscopic accommodations occur during heat treatments
and inhomogeneous plastic strains may influence macroscopic shape changes leading to part uncontrolled
distortions. As observed as in literature, a key in distortion
development is the dissimilar coefficient of thermal
expansion (CTE) and the relative yield strength of different
phases present. Kennedy et al. [18] noted the occurrence of
dimensional anisotropy for a variety of steels cooled at
high and low cooling rates. Goldberg [19] correlated
chemical segregation with variation of MS temperature
leading to early formation of martensite within definite
layers of austenite. Anisotropy was related to plastic
deformation and/or oriented transformation of the austenite, as the compressive stress caused by the preferential
growth developed.
The impact of banding has influences along entire thermal cycle between the austenite temperature and the
formation of martensite. During heating of a banded
microstructure, local austenitization temperatures (Ar1,
namely As, and Ar3, namely Af are defined, respectively, for
the start and finish of a ? c transformation) can be very
different; during heating, the result could be co-existence of
a sort of layer growth of austenite in those low-C content
bands (with lower As) alternating with unchanged ferrite

123

J Fail. Anal. and Preven. (2010) 10:351357

Fig. 8 Dilatometric heating and cooling curves for longitudinal


and transverse samples for a complete thermal cycle for a maraging
steel [20]

layers. From dilatometric heating curve (Fig. 8), austenite


formed layers exhibiting larger volumetric contraction
against those unchanged ferrite layers, in agreement with
higher density of crystal structure of austenite compared to
ferrite. Tensile stresses can develop inside austenite layers
because of the presence of ferrite layers that act as impediment to austenite volumetric contraction. Such tensile
stress can be sufficient to cause plastic deformation to
occur. As the austenitizing start temperature As is reached
in the low-C bands, ferrite start to transform into austenite,
but volumetric contraction cannot be accomplished because
of the presence of previously deformed austenite layers
(i.e., high-C content bands). Such anisotropy in strain
behavior between high-C content and low-C content bands
are key driving parameters to cause plastic deformations
along such bands and may result in residual compressive stresses. A net contraction in the direction of the
banded microstructure was observed for thermal cycles
a ? c ? a, while the length change during cycling was not
significant for the homogeneous structure [21]. Conversely,
a net expansion was observed in same experience for
sample bars taken in transverse direction with banding.
Therefore, one key result was that the direction of this net
deformation is dependent upon the orientation of the
chemical segregation pattern; contextually, final length
change is a net result of plastic deformation during heat up
and deformation in the opposite direction during cooling. In
fact the authors showed an artificially banded 5140 steel,
longitudinal banded specimens exhibited net length contraction, despite of transversal banded specimen that
resulted in axially elongation and quite elliptic section
shape. Also a result was that transformation-induced plasticity (TRIP) is a mechanism by which plastic deformation
occurs in the banded steel. This highlights the sensitivity of
the shape of the product phase to stress fields and the
potential for in-process failures in banded steels.

J Fail. Anal. and Preven. (2010) 10:351357

Conclusions
Many efforts have to be suggested in order to achieve high
chemically homogeneous steels to manufacture hardened
mechanical parts. Pre-existing local chemical composition
variations is common consequence of steelmaking process
and early post hot-working process. Different possibilities
exist to develop cost-effective improvements in precursor
products, before hot-working steps and heat treating postprocesses. In general, heat treatments are developed from
the nominal chemical composition and local segregation
can greatly influence the quality and conformity of final
product.
High banded structures are detrimental in all cases to
products subjected to thermal cycles and stress fields can
be generated because of strain anisotropy during heating
(a ? c transformation) and cooling (c ? a martensite
transformation). The banding phenomena in pre-cursor to
potential failure in semi-finished products and the large
distortions may induced by TRIP mechanisms.

References
1. Howe, H.M.: The Metallography of Steel and Cast Iron,
pp. 556565. McGraw-Hill, New York, NY (1916)
2. Cameron, A.E., Waterhouse, G.B.: The influence of arsenic on
steel. J. Iron Steel Inst. 113, 355374 (1926)
3. Whiteley, J.H.: Apparent relations between manganese and segregation in steel ingots. J. Iron Steel Inst. 144(2), 6377 (1941)
4. Bastien, P.G.: The mechanism of formation of banded structures.
J. Iron Steel Inst. 193, 281290 (1957)
5. Samuels, L.E.: Optical Microscopy of Carbon Steels, pp. 127135.
ASM, Metals Park, OH (1980)

357
6. Stead, J.E.: Some of the ternary alloys of iron carbon and phosphorus. J. Soc. Chem. Ind. 33(4), 173184 (1914)
7. Thompson, S.W., Howell, P.R.: Factors influencing ferrite/
pearlite banding and origin of large pearlite nodules in a
hypoeutectoid plate steel. Mater. Sci. Technol. 8, 777784
(1992)
8. Krauss, G.: Solidification, segregation, and banding in carbon and
alloy steels. Metall. Mater. Trans. B 34B, 781792 (2003)
9. Thompson, S.W., Howell, P.R.: Factors influencing ferrite/
pearlite banding and origin of large pearlite nodules in a hypoeutectoid plate steel. Mater. Sci. Technol. 8, 777784 (1992)
10. Stead, J.E.: Iron and phosphorus. J. Iron Steel Inst. II, 60-155
(1900)
11. Jatczak, C.F., Girardi, D.J., Rowland, E.S.: On banding in steel.
Trans. ASM 48, 279305 (1956)
12. Kirkaldy, J.S., von Destinon-Forstmann, J., Brigham, R.J.: Simulation of banding in steel. Can. Metall. Q. 1, 5981 (1962)
13. Eckert, J.A., Howell, P.R., Thompson, S.W.: Banding and the
nature of large, irregular pearlite nodules in a hot-rolled lowalloy plate steel: a second report. J. Mater. Sci 28, 44124420
(1993)
14. Grossterlinden, R., Kawalla, R., Lotter, U., Pircher, H.: Formation of pearlitic banded structures in ferritic-pearlitic steels. Steel
Res. 63, 331336 (1992)
15. Heiser, F.A., Hertzberg, R.W.: Structural control and racture
anisotropy of banded steel. J. Iron Steel Inst. 209, 975980 (1971)
16. Grange, R.A.: Effect of microstructural banding in steel. Metall.
Trans. 2, 417426 (1971)
17. Hellner, L., Norrman, T.O.: Banding in alloyed steels. Jernkont.
Ann. 152, 269286 (1968)
18. Kennedy, R., Grant A. I., Kinnear, A.J., Kirkpatrick, I.M.:
Dimensional changes in steels due to thermal cycling. J. Iron
Steel Inst. 601602 (1970)
19. Goldberg, A.: Trans. ASM 61, 2636 (1968)
20. Farooque, M., Qaisar, S., Haq, A.U.L., Khan, A.Q.: Dimensional
anisotropy in 18 Pct. Ni maraging steel. Met. Mater. Trans. 32A,
10571061 (2001)
21. Jaramillo, R.A., Lusk, M.T., Mataya, M.C.: Dimensional
anisotropy during phase transformations in a chemically banded
5140 steel; part II: modeling. Acta Mater. 52, 859867 (2004)

123

Potrebbero piacerti anche