Author's personal copy
Atom probe tomography characterization of heavily cold drawn pearlitic
Y.J. Lia,b,?, P. Choib, C. Borchersa, Y.Z. Chena, S. Gotoc, D. Raabeb, R. Kirchheima,b
aInstitut f¨ ur Materialphysik, Georg-August-Universit¨ at G¨ ottingen, Friedrich-Hund-Platz 1, D-37077 G¨ ottingen, Germany
bMax-Planck Institut f¨ ur Eisenforschung, Max-Planck-Str. 1, D-40237 D¨ usseldorf, Germany
cDepartment of Materials Science and Engineering, Faculty of Engineering and Resource Science, Akita University, Tegata Gakuencho, Akita 010-0852, Japan
a r t i c l e i n f o
Available online 24 November 2010
Atom probe tomography
Pearlitic steel wire
Severe plastic deformation
a b s t r a c t
Atom Probe Tomography (APT) was used to analyze the carbon distribution in a heavily cold drawn
pearlitic steel wire with a true strain of 6.02. The carbon concentrations in cementite and ferrite were
separately measured by a sub-volume method and compared with the literature data. It is found that the
carbon concentration in ferrite saturates with strain. The carbon concentration in cementite decreases
with the lamellar thickness, while the carbon atoms segregate at dislocations or cell/grain boundaries in
ferrite. The mechanism of cementite decomposition is discussed in terms of the evolution of dislocation
structure during severe plastic deformation.
& 2010 Elsevier B.V. All rights reserved.
It is well known that severe plastic deformation, e.g. heavy cold
drawing, can strengthen pearlitic steel wires. Depending on the
drawing strain the interlamellar spacing can be refined to 5–20 nm
[1–3] and the ultimate tensile strength reaches values between 3.2
and 5.0 GPa [1,4–8]. In addition, many efforts have been made to
and/or adding Cr to the material [9–12]. In such cases, the tensile
strength can be as high as 5.7 GPa for wires with a diameter of
40 mm . The addition of Cr not only improves the strength of as-
patented wires by refining their lamellar structure, but also
increases the work hardening rates during cold drawing. The latter
effect can also be obtained by increasing the carbon content .
Accompanied by the formation of a nano-scaled lamellar structure
and the enhancement of wire strength, cementite is found to be
partially amorphized [3,8,13] and to decompose upon plastic
deformation. This makes an understanding of the strengthening
mechanisms more complicated. The decomposition of cementite
has been extensively investigated by means of M¨ ossbauer spectro-
scopy , one-dimensional atomprobe(1DAP) [4,6,11,13,15–17],
and atom probe tomography (APT) [3,8,15,18–20]. Up to now, the
mechanisms of cementite decomposition and the positions of
carbon atoms in ferrite still remain unclear [19,21–24].
To better understand the mechanism of cementite decomposi-
tion, it is important to accurately quantify the carbondistribution in
the material. The reliability of the quantification depends on the
technique. For example, 1DAP has been frequently used to analyze
the cementite decomposition. However, 1DAP is limited with
respect to the analyzed volume and field of view . Regarding
the limitation of specimen preparation methods, the needle-shaped
specimens for atom probe measurements were previously prepared
a lamellar structure, the electro-polishing method restricts the
accuracy of quantification because the specimen tips are parallel
to the pearlitic lamellae. Thus, only very few (one or two) cementite
lamellae can be detected and the spatial resolution in the direction
Takahashi et al. .
In the present work, we study the cementite decomposition of a
heavily cold drawn (true strain e ¼ 6:02) Cr-containing pearlitic steel
of cementite decomposition with a large field of view and high mass
resolution. The tip-shaped specimens for LEAP measurements are
prepared by the FIB based lift-out method with tips perpendicular to
the number of detected cementite lamellae and also enables analyses
cementite decomposition is analyzed by comparing our results with
literature data for similar materials taken at different wire strains .
of carbon atoms in ferrite are discussed.
Contents lists available at ScienceDirect
journal homepage: www.elsevier.com/locate/ultramic
0304-3991/$-see front matter & 2010 Elsevier B.V. All rights reserved.
?Corresponding author at: Max-Planck Institut f¨ ur Eisenforschung, Max-Planck-
Str. 1, D-40237 D¨ usseldorf, Germany.
Tel.: +49 211 6792853; fax: +49 211 6792333.
E-mail address: firstname.lastname@example.org (Y.J. Li).
Ultramicroscopy 111 (2011) 628–632
Author's personal copy
The material used in this work is a commercial pearlitic steel
0.20Cr–0.01Cu–0.006P–0.007S wt% or Fe–4.40C–0.30Mn–0.39Si–
0.21Cr–0.003Cu–0.01P–0.01S at.%) provided by Suzuki Metal
Industry Co., LTD. The wire with an initial diameter of 0.18 mm
was patented. This process includes austenitization at 1223 K for
80 s followed by pearlitic transformation in a lead bath at 853 K for
20 s. The patented wire was then deformed by lubricated drawing
e ¼ lnðd2
diameters of the wire, respectively.
0=d2Þ ¼ 6:02, where d0 and d are the initial and final
2.2. Local electrode atom probe and measurement conditions
A LEAP 3000X HRTM, Cameca Instruments was employed to
analyze the carbon distribution in three-dimensions. This equip-
ment enables the analysis of volumes up to 100?100?1000 nm3
(Dm=m ¼ 1/1100, FWHM at 27 Da).
LEAP measurements were performed in voltage mode at 70 K
under an ultra-high vacuum of 6?10?11Torr. The total voltage
during probing was in the range of 6.2–7.2 kV. The pulse fraction
was 15%. The pulse repetition rate was 200 kHz at detection rate of
0.005 atom per pulse.
Samples for APT analyses were prepared with the tips perpen-
dicular to the wire axis using a dual beam focused ion-beam (FIB)
(FEI Helios NanoLab 600TM) according to the procedure described
3.1. Mass spectrum analyses
6.5, 12, 13, 18, 18.5, 24, 24.5 and 36 Da are all due to carbon (m:
ionic mass; n: ionization state). Since carbon atoms evaporate as
4), the peaks at m=n ¼ 12 and 24 Da are ambiguous. While
the peak at 12 Da can be assigned as either12C+or
mixture of both, the ambiguity for the peak at 24 Da lies between
. A peak decomposition algorithm (supplied by the
software IVAS, Cameca Instruments), which is similar to the
method used by Sha et al.  and takes into account the isotope
ratios, was applied to these two peaks. The results show that both
signals can be ascribed to a mixture of ions. For the peak at 12 Da
the contributions to this peak are 81% from12C+and 19% from
. Therefore, this peak is assigned as
measurement. For the peak at 24 Da the percentages are 48% from
, as shown in Table 1. Compared to the
nominal carbon concentration (see Table 2), the assignments for
concentration by 16% or an overestimation by 4%, respectively.
, or a
12C+for the present
2and 52% from12C2þ
to the peak into account, the calculated overall carbon concentra-
tion is 4.4 at.%, which virtually equals the nominal one. This means
that the LEAP has accurately detected the carbon element in
pearlitic steel wires. For the 3D elemental map, the assignment
of this peak to12C2þ
Note that the assignment of the peak at 24 Da may differ between
cementite and ferrite if their mass spectra are analyzed separately.
100% from Cþ
2in ferrite, while 85–90% from C2þ
3.2. Carbon concentration in cementite and ferrite
3.2.1. 1D carbon concentration profiles
Fig. 1(a) shows the 3D atom map for carbon. To visualize the
morphology and size of the cementite lamella, a carbon isoconcen-
tration surface drawn at 7 at% is shown. Due to a relatively sharp
carbon concentration profile at the interface-near region, the
selection of iso-value is not critical for the present case. The iso-
value of 7 at.% C is selected using the criterion that a phase with a
defined as cementite. With this definition the morphology and size
of the cementite lamella are visualized.
Results from peak decomposition performed for the peak at 24 Da.
Ion type% of rangeDecomposition abundance
assignments of the peak at 24 Da, and the nominal values. Unit: at.%.
Fig. 1. (a) 3D carbon atom map (containing over 12?106ions within a volume of
55 ? 55 ? 230 nm3) with selected boxes for the regions of interest (ROIs). The
carbon isoconcentration surface was drawn at 7 at.%. (b) 1D carbon concentration
profiles (fixed bin width of 0.2 nm) for the selected boxes along the main axes of
ROIs. The ROIs and the corresponding concentration profiles are related to each
other by the same colors.
Y.J. Li et al. / Ultramicroscopy 111 (2011) 628–632
Author's personal copy
Frequently, cementite decomposition has been quantitatively
analyzed by plotting an 1D carbon concentration profile for a
selected box containing many lamellae [15,29]. In Fig. 1(a) four
regions of interest (ROIs) of different colors are selected for
analysis. The advantage of this method is that the extent of both
the cementite decomposition and the amount of carbon in the
ferrite can be immediately taken from the concentration profiles.
As shown in Fig. 1(b), the peak values of carbon concentration lie
between 7 and 18 at.% for carbon-enriched lamellae and between
1 and 3 at.% for carbon-depleted regions.
However, this method can only provide a rough quantification
of the carbon concentration. The reasons are as follows. First, since
layer contains two different phases with different evaporation
pearlite cementite is known to have a lower evaporation field than
ferrite [20,31]. This leads to a widened apparent carbon concen-
tration gradient at the interfaces between ferrite and cementite.
Thus, the carbon concentration from 1D concentration profiles
strongly depends on the dimensions of the analysis box selected. If
we take a look at the spatial orientations and the morphologies of
cementite lamellae from the 3D atom map shown in Fig. 1(a), the
cementite lamellae have decomposed and/or fragmented into
various forms with various sizes regarding the cross-sections of
the cementite lamellae. Therefore, the local magnification effect
cannot be avoided. Secondly, since the assignment of the peak at
24 Da differs between cementite and ferrite, as will be further
phases cannot give an accurate quantification of the carbon
concentration in each phase. If the peak at 24 Da is assigned to
for such a ROI, the carbon concentration in ferrite is over-
estimated. Otherwise, the carbon concentration in cementite will
3.2.2. Quantification of carbon concentration by the sub-volume
In the following, the sub-volume method is used to treat
different phases separately and thus obtaining higher composi-
Fig. 2 displays the analysisproceduresof the sub-volume method.
For cementite, the ROIs (green) are selected and positioned indivi-
dually for each lamella in such a way that they contain the largest
possible volume without any matrix included. Here, the matrix is
defined as a phase with carbon concentration lower than 7 at.%. For
ferrite, the ROIs (pink) are selected from the matrix, which is,
however, defined as a phase with carbon concentration lower than
away from the interface-near zones, where the local magnification
effectissignificant. All the selected ROIsare cutaway fromthe whole
volume and their mass spectra are analyzed separately. Peak decom-
position is performed for the peak at 24 Da for each selected sub-
volume. Figs. 2(b) and (d) show ROIs representing cementite and
ferrite, respectively. The results from peak decomposition show that
86% is due to12C2þ
for cementite. Based on these assignments, the
average carbon concentration in ferrite is determined to be 0.60 at.%
concentration measured in the middle of the cementite lamella is
17.5 at.%(seeFig.2c) andthisvalueistakenasthecompositionofthe
Fig. 3 shows the average carbon concentrations in ferrite for all
the ROIs as a function of wire strain. The data for wire strains eo5
from  are also shown for comparison. It is seen that the obtained
values are different even for the same strain. Since no detail for the
determination of carbon concentrations in  is given, it is hard to
local magnification effect, the carbon concentration in ferrite
strongly depends on the location of the selected ROI. For the data
at e ¼ 6:02, obtained from the present work, such an influence can
indicates that the distribution of the carbon atoms is inhomoge-
neous in ferrite.
To clarify the dependence of carbon concentration on the wire
strain, the final average values for each strain are plotted as the red
Fig. 2. Example showing the analysis procedures of sub-volume method. (a) The
wholereconstructured 3D image with theselected sub-volumes to be cut. Green for
cementite and pink forferrite. (b) Atom map fora selected sub-volume of cementite
for Cþand C2þ, pink for Cþ
concentration profile for the ROI of (b). As shown by the thin horizontal line, the
value at the center of the profile is taken for the lamella. (d) As (b) for ferrite. (For
the web version of this article.)
2, green for C2þ
and red for C2þ
4 . (c) 1D carbon
Fig. 3. Carbon concentrations in ferrite as a function of wire strain for hypereu-
tectoid pearlitic steel wires. Literature data (Cr free) from  are displayed as small
black open circles. The large red symbols represent the final average values
calculated from all the data shown for each strain. (For interpretation of the
references to color in this figure legend, the reader is referred to the web version of
Y.J. Li et al. / Ultramicroscopy 111 (2011) 628–632
Author's personal copy
circles shown in Fig. 3. For eo5, the carbon concentration increases
with increasing wire strain. This implies that plastic deformation
promotes the cementite decomposition up to e? 5. For e45, the
carbon concentration in the ferrite appears to reach the saturated
state.Theslightlyhighervaluemeasuredate¼ 4:7 fromthanthat
fore ¼ 6:02 mayresultfromthelocalmagnificationeffect,whichwill
get more pronounced, if the probing direction is parallel to the
lamellar interfaces. This is exactly the case in . It may also result
work. It is known that Cr has the effect of strengthening the bonds of
the cementite lattice , and thus suppresses cementite decom-
position. In addition, the data evaluation method is also crucial to the
obtained concentration values, as stated in Section 3.2.1. No detail on
the analysis of the quantification is given in .
Fig. 4 displays the carbon concentrations measured in the
middle of the analyzed cementitelamellae asa functionof lamellar
thickness. The use of concentration values at the center of the
lamella may be a pertinent approach to quantify the extent of
cementite decomposition, since they remain unaffected by the
local magnification effect. For each cementite lamella, its thickness
is determined based on the isoconcentration surfaces drawn at
7 at.% C. The results show that no cementite with a stoichiometric
carbon concentration remains after heavy cold drawing. Also, we
observe that thinner lamellae are more prone to undergo deforma-
tion-induced chemical decomposition than thicker lamellae. This
observation was not reported before in the literature.
Although it is well accepted that cementite decomposes during
used to explain this phenomenon. One is based on thermodynamic
considerations and the other one on carbon–dislocation interac-
tions in ferrite. The first one attributes the decomposition to a
reduction in thermodynamic stability of cementite caused by
refinement of the lamellae structure and the increase of steps on
is driven by the difference between the high bonding energy
between dislocations and carbon atoms in ferrite (0.75 eV )
and the low binding energy between the atoms of iron and carbon
in cementite (0:4020:42 eV [34,35]). Thus, dislocations drag car-
bon atoms out of cementite during deformation [21,22]. Based on
the present APT results, we consider that the second approach
might be the more likely mechanism of atomic-scale mixing across
the hetero-interfaces. Beyond the elementary interaction between
carbon atoms and lattice dislocations, which can carry the solutes
a dislocation-shuffle mechanism to contribute to the deformation-
enhanced mixing process. Dislocation-shuffling refers to a phe-
nomenon of trans-phase plastic dislocation slip (from ferrite into
the cementite), potentially on more than one slip system. Such
shearing of atomic planes through dislocation glide on more than
one slip system across the interface of abutting layers can create
small embedded particles consisting of atoms from one phase in
the other. Such tiny material portions can be further cut by
through the Gibbs–Thomson effect so that they finally dissolve. In
mass transport among the two phases against the thermodynamic
equilibrium conditions. This and related possible mechanisms that
may lead to deformation-induced chemical non-equilibrium mix-
ing were recently discussed in .
According to the carbon–dislocation interaction mechanism,
dislocations mainly get stored at the phase boundary between
ferriteand cementiteduring severe plastic deformation.The ferrite
matrix is relatively free of dislocations. This is reasonable if the
mean free path of dislocations is larger than the interlamellar
density of r ¼ 1016=m2in the material after a drawing strain of
6.02. The mean free path of dislocations may be estimated with a
small prefactor 10 to be 10r?0:5¼ 100 nm, which is 10 times larger
than the measured interlamellar spacing of 12 nm. Clearly, the
motion of dislocations must be hindered by the phase boundaries.
Indeed, in many TEM studies a large density of dislocations are
observed at phase boundaries [23,37]. On the other hand, one
should not totally rule out the possibility of formation of cell
boundaries and/or grain boundaries inside ferrite lamellae during
severe plastic deformation. In fact, Tarui et al.  observed cell
structures in the size of 10 nm in the ferrite of a pearlitic steel wire
with e¼ 4:2 by TEM. The material used in the present work has
been further strained by a strain of approximately 2, the cell
structure is expected to evolve towards (sub)grain structures with
few dislocations inside the cells/grains.
Fig. 4. Carbon concentrations in the cementite as a function of lamellar thickness.
Fig. 5. (a) Carbon atom map for a selected ROI (20 ? 3 ? 4 nm3) from matrix
showing inhomogeneous distribution of atoms in ferrite. (b) Distribution of the first
nearest neighbor distance between carbon atoms in (a). Solid and dashed plots
represent the observed data and the randomly reordered data, respectively. (c)
Carbon contour plot with 0.3 nm distances between both samples and sampling
planes. The numbers mark the local carbon concentrations in atomic percentage.
Y.J. Li et al. / Ultramicroscopy 111 (2011) 628–632
Author's personal copy Download full-text
inhomogeneity. The 2D concentration profile shown in Fig. 5(c) marks
the local concentrations. It is likely that the areas with high carbon
concentrations mark the cell/grain boundaries or individual disloca-
tions, where the carbon concentration is five times higher than that in
carried here by dislocations leaving from the phase boundaries. They
may also diffuse from the phase boundaries through dislocation cores.
in cementite with decreasing cementite thickness (Fig. 4). From this
we can infer that a decrease of the cementite thickness in the as-
patented material promotes cementite decomposition during defor-
mation. This is understandable because for a given volume of total
cementite, the thinner the cementite lamella, the higher the fraction
of the phase boundary, and also the lower the interlamellar spacing.
This would enhance the dislocation density at the phase boundaries
upon deformation and therefore improve cementite decomposition.
a heavily dissolved originally coarse cementite, then the cementite
must have undergone a strong local deformation which involves a
large amount of dislocations. Thus, this would be a strong hint
towards the importance of dislocations on cementite decomposition.
Cementite decomposition in a Cr-containing pearlitic steel wire
atom probe tomography. It is found that the decomposition of
cementite reduces their carbon concentration to 12–18 at.% and
enhances the average carbon concentration in ferrite to 0.63 at.%.
The comparison of the measured carbon concentrations in ferrite
with literature data shows that plastic deformation promotes
the cementite decomposition up to the wire strain of 5. For e45
the decomposition process seems to reach its saturated stage. The
carbon concentration in cementite varies proportionally with the
thickness of the cementite lamella. These quantitative analyses
together with the inhomogeneous distribution of carbon atoms in
the ferrite allow us to conclude that the decomposition of
cementite is presumably controlled by carbon–dislocation inter-
effects (dislocation shuffling). We assume that the carbon atoms
are preferentially segregated at cell/grain boundaries or individual
dislocations in ferrite.
The authors thank Dr. H. Yarita from Suzuki Metal Industry Co.,
also thank U. Tezins and A. Sturm for their help with LEAP
measurements and FIB-based specimen preparation. We are grate-
ful to the Deutsche Forschungsgemeinschaft for funding this
 A. Taniyama, T. Takayama, M. Arai, T. Hamada, Scr. Mater. 51 (2004) 53.
 G. Landford, Metall. Trans. A 8 (1977) 861.
 C. Borchers, T. Al-Kassab, S. Goto, R. Kirchheim, Mater. Sci. Eng. A 502 (2009)
 H.G. Read, W.T. Reynolds Jr., K. Hono, T. Tarui, Scr. Mater. 37 (1997) 1221.
 H. Tashiro, Nippon Steel Tech. Rep. 80 (1999).
 N. Maruyama, T. Tarui, H. Tashiro, Scr. Mater. 46 (2002) 599.
 M. Zelin, Acta Mater. 50 (2002) 4431.
 S. Goto, R. Kirchheim, T. Al-Kassab, C. Borchers, Trans. Nonferrous Met. Soc.
China 17 (2007) 1129.
 T. Takahashi, I. Ochiai, H. Tashiro, S. Ohashi, S. Nishida, T. Tarui, Nippon Steel
Tech. Rep. 64 (1995) 45.
 H. Tashiro, S. Nishida, T. Tarui, S. Ohashi, S. Sasaki, K. Nakamura, A. Yoshie,
H. Demachi, Nippon Steel Tech. Rep. 80 (1999) 38.
 T. Tarui, N. Maruyama, J. Takahashi, S. Nishida, H. Tashiro, Nippon Steel Tech.
Rep. 91 (2005) 56.
 T. Ochi, S. Nishida, M. Sugiyama, J. Takahashi, T. Tarui, Nippon Steel Tech. Rep.
96 (2007) 5.
 M.H. Hong, W.T. Reynolds Jr., T. Tarui, K. Hono, Metall. Mater. Trans. A 30
 V.N. Gridnev, V.G. Gavrilyuk, Y.D. Ekhtyayru, Y.A.M. Eshkov, P.S. Nizina,
V.G. Prokopenko, Phys. Status Solidi A 11 (1972) 689.
 K.Hono, M. Ohnuma, M.Murayama, S. Nishida, A. Yoshie,Scr. Mater. 44(2001)
 T. Tarui, T. Takahashi, S. Ohashi, R. Uemori, Ironmaking Steelmaking 21 (1994)
 T. Tarui, T. Takahashi, H. Tashiro, S. Nishida, in: H.G. Paris, D.K. Kim (Eds.),
Processing and Applications of Metal Wires, TMS, Warrendale, PA, 1996, p. 87.
 F. Danoix, D. Julien, X. Sauvage, J. Copreaux, Mater. Sci. Eng. A 250 (1998) 88.
 X. Sauvage, J. Copreaux, F. Danoix, D. Blavette, Philos. Mag. A 80 (2000) 781.
 X. Sauvage, W. Lefebvre, C. Genevois, S.OhsakiK. Hono, Scr. Mater. 60 (2009)
 V.G. Gavriljuk, Scr. Mater. 45 (2001) 1469.
 V.N.Gridnev,V.V. Nemoshkalenko,
G. Prokopenko, O.N. Razumov, Phys. Status Solidi A 31 (1975) 201.
 J. Languillaume, G. Kapelski, B. Baudelet, Acta Mater. 45 (1997) 1201.
 X. Sauvage, Y. Ivanisenko, J. Mater Sci. 42 (2007) 1615.
 T.F. Kelly, M.K. Miller, Rev. Sci. Instrum. 78 (2007) 031101.
 M.K. Miller, K.F. Russell, G.B. Thompson, Ultramicroscopy 102 (2005) 287.
 M.K. Miller, K.F. Russell, K. Thompson, R. Alvis, D.J. Larson, Microsc. Microanal.
13 (2007) 428.
 K. Thompson, D. Lawrence, D.J. Larson, J.D. Olson, T.F. Kelly, B. Gorman,
Ultramicroscopy 107 (2007) 131.
 J. Takahashi, T. Tarui, K. Kawakami, Ultramicroscopy 109 (2009) 193.
 W. Sha, L. Chang, G.D.W. Smith, L. Cheng, E.J. Mittemeijer, Surf. Sci. 266 (1992)
 X. Sauvage, L. Thilly, D. Blavette, J. Phys. IV Fr. 11 (2001) 27.
 A. Inoue, T. Masumoto, Trans. ISIJ 17 (1977) 143.
 A.W. Cochardt, G. Schoek, H. Wiedersich, Acta Metall. 3 (1955) 533.
 R.A. Johnson, G.I. Diens, A.C. Damask, Acta Metall. 12 (1964) 125.
 R.A. Johnson, Acta Metall. 15 (1967) 513.
 D. Raabe, S. Ohsaki, K. Hono, Acta Mater. 57 (2009) 5254.
 M. Janecek, F. Louchet, B. Doisneau-Cottignies, Y. Br? echet, N. Guelton, Philos.
Mag. A 80 (2000) 1605.
Y.J. Li et al. / Ultramicroscopy 111 (2011) 628–632