Available online at www.sciencedirect.

com

ScienceDirect
Acta Materialia 77 (2014) 17–27
www.elsevier.com/locate/actamat

Atomic and nanoscale chemical and structural changes in quenched

and tempered 4340 steel
A.J. Clarke a,⇑, M.K. Miller b, R.D. Field a,c, D.R. Coughlin a, P.J. Gibbs a, K.D. Clarke a,
D.J. Alexander a, K.A. Powers b, P.A. Papin a, G. Krauss c
a
Los Alamos National Laboratory, Materials Science and Technology Division, PO Box 1663, Los Alamos, NM 87545, USA
b
Oak Ridge National Laboratory, Center for Nanophase Materials Science, PO Box 2008, Oak Ridge, TN 37831, USA
c
Colorado School of Mines, 1500 Illinois St., Golden, CO 80401, USA

Received 26 February 2014; received in revised form 14 May 2014; accepted 19 May 2014

Abstract

Atom probe tomography and transmission electron microscopy (TEM) have been used to determine the location and distribution of
carbon and alloying elements associated with the complex structural changes that occur at the atomic and nanoscale in 4340 steel after
quenching to martensite and tempering at 325, 450 or 575 °C. Tempering at 325 °C resulted in carbide formation without partitioning of
chromium, manganese, molybdenum, aluminum, nickel or phosphorus, but with early-stage silicon rejection from the carbide. TEM
verified the presence of cementite and the Bagaryatsky orientation relationship with the tempered martensite matrix and detected com-
plex precipitate structures. Tempering at 450 or 575 °C developed concentrations of all alloying elements at ferrite–cementite interfaces:
chromium, manganese and molybdenum partitioned into the cementite, and silicon, aluminum, nickel and phosphorus were clearly
rejected from the cementite. These results provide direct evidence for staged cementite growth, where early-stage growth likely occurs
under paraequilibrium conditions, followed by initial silicon redistribution and subsequent alloying element redistribution during
late-stage growth. Tempering at 575 °C induced spheroidization of the cementite, loss of the Bagaryatsky orientation relationship,
and phosphorus concentrations at Cottrell atmospheres within the cementite and at ferrite–cementite interfaces, correlating with early
observations of the retardation of spheroidization by phosphorus.
Ó 2014 Acta Materialia Inc. Published by Elsevier Ltd. All rights reserved.

Keywords: Steel; Tempering; Martensite; Atom probe tomography (APT); Transmission electron microscopy (TEM)

1. Introduction resistance to softening during tempering. Although some

Low-alloy, medium-carbon steels are austenitized,
quenched to martensite and tempered at various tempera-
tures to provide beneficial combinations of strength, tough-
ness and fracture resistance. These steels are alloyed with
substitutional alloying elements, such as chromium (Cr),
manganese (Mn), molybdenum (Mo) and nickel (Ni), to
provide good hardenability, i.e. depth of hardening, and
⇑ Corresponding author. Tel.: +1 505 665 3467.
E-mail address: aclarke@lanl.gov (A.J. Clarke).
alloy carbides may be retained, austenitizing is designed to
dissolve all carbon and other alloying elements into the
face-centered cubic crystal structure of the austenite. As a
result, the martensite, formed by the diffusionless shear trans-
formation of the austenite, is supersaturated with carbon and
other alloying elements compared to their equilibrium con-
centrations in ferrite-cementite microstructures. These super-
saturations are relieved by the precipitation of a variety of
carbide types and morphologies during tempering [1].
It is well known that the alloying elements in steels
retard softening during tempering, and Grange et al. [2]
have quantitatively documented the effects of various

http://dx.doi.org/10.1016/j.actamat.2014.05.032
1359-6454/Ó 2014 Acta Materialia Inc. Published by Elsevier Ltd. All rights reserved.
18 A.J. Clarke et al. / Acta Materialia 77 (2014) 17–27
elements in steels in retaining the hardness of as-quenched
structures as a function of element content and tempering
temperature. However, the mechanisms of alloying element
redistribution as the saturation in martensite is relieved are
only now beginning to be characterized by virtue of recent
developments in atom probe tomography (APT).
At low tempering temperatures, carbon is mobile, but
the diffusivity of the substitutional elements is too low to
significantly influence the microstructure and substructure
of the martensite [1,3]. Therefore, carbon plays a dominant
role in the chemical and structural changes that occur,
permitting the achievement of high strengths and moderate
toughness after tempering below 250 °C [1,3–5]. At
tempering temperatures above 250 °C, the replacement of
transition carbides and low-carbon martensite by cementite
and ferrite occurs [1,6–9]. Higher tempering temperatures
(>425 °C) provide sufficient thermal energy for substitu-
tional elements to be mobile and influence structural
coarsening, resulting in microstructures that exhibit
combinations of high toughness and moderate strengths
[1,4,5].
Babu et al. reported early-stage cementite growth under
paraequilibrium conditions (i.e. dependent only upon car-
bon diffusion) in an Fe–C–Si–Mn steel after tempering at
lower temperatures, whereas higher tempering tempera-
tures and longer times resulted in Si and Mn partitioning
[10]. Zhu et al. have shown no partitioning of Cr, Mo,
Mn or Ni with the initiation of Si partitioning after temper-
ing 4340 steel at 300, 350 or 400 °C for 1 h [11]. Tempering
at 400 °C for 10 h resulted in significant Cr and Mn parti-
tioning within the cementite and the rejection of Si and Ni
[11]. The purpose of this study was to systematically eval-
uate the redistribution of interstitial and substitutional ele-
ments and the subtle structural changes that occur in 4340
steel at the atomic and nanoscale after quenching and
tempering between 325 and 575 °C for 2 h using APT
and transmission electron microscopy (TEM).
2. Experimental procedure
Aircraft-quality 4340 steel was used for this study; the
composition of this steel in wt.% and at.% is provided in
Table 1. Nominally 6.35 mm thick plate was austenitized
under vacuum for 30 min at 845 °C, oil quenched, and tem-
pered in air at 325 or 450 °C or under vacuum at 575 °C for
2 h. A sample was also encapsulated in quartz under ultra-
high-purity argon, austenitized for 30 min at 845 °C, and
then ice-water quenched for microstructural comparison
with the oil-quenched condition.
Table 1
4340 steel composition.
C Mn Si Ni
wt.% 0.42 0.78 0.26 1.78
at.% 1.92 0.78 0.51 1.67
APT was performed using a local electrode atom probe
(Cameca Instruments Inc. LEAPÒ 4000X HR). Sections
were machined from quenched and tempered steel plate
and mechanically ground to a thickness of 0.25 mm.
Abrasive wire-cutting, followed by mechanical grinding,
was then used to fabricate specimen blanks with a square
cross-section, nominally 0.25  0.25 mm.
Specimen blanks were electropolished into sharp needles
by a two-stage double-layer and micropolishing technique
[12]. The electrolytes used were 25% perchloric acid
(70%) in glacial acetic acid for rough polishing and 2%
perchloric acid in 2-butoxyethanol for final polishing [12].
Focused ion beam (FIB) annular milling was then
performed in an FEI Nova 200 instrument to sharpen each
needle for APT. For the LEAP analyses, the specimen
temperature was 50 K and the specimens were run in
voltage mode. A pulse fraction of 0.2 and pulse repetition
rate of 200 kHz were used.
TEM specimens were prepared by two different meth-
ods: electropolishing and FIB milling. The electropolishing
solution used was 95% glacial acetic acid with 5% per-
chloric acid. Polishing was performed at room temperature
in a Fishione automatic polisher at a potential of 25–30 V.
The FIB foil preparation was performed in an FEI DB235
dual-beam FIB equipped with an Omniprobe nanomanip-
ulator for foil extraction. TEM analysis was performed in
an FEI TF30 Tecnai instrument operating at 300 kV.
3. Results
3.1. As-quenched martensitic microstructure
Oil quenching of the 4340 steel produced a lath martens-
itic microstructure. Martensite crystals contain dislocation
densities as high as 1012 cm2, resulting from the lattice
invariant deformation component of the diffusionless mar-
tensitic phase transformation [13]. Fine twins were also
observed in some laths, as shown in the bright-field TEM
images in Fig. 1.
Nanoscale, plate-like precipitates were also revealed by
TEM, as shown in Fig. 2a. Selected-area diffraction
(Fig. 2b) suggests that these precipitates are e0 (also referred
to as g) transition carbide [9,14,15]. Three reflections, labeled
as A, B and C, are highlighted in Fig. 2b. Position A corre-
sponds to a d-spacing of 0.20 nm, consistent with f0 2 1ge0
planes. Position B corresponds to superlattice reflections
from the ordered e0 carbide and position C results from dou-
ble diffraction between f0 2 1ge0 carbide and f1 1 0ga martens-
ite reflections. The pattern is consistent with two variants of
the following orientation relationship [15]:
Cr Mo Cu P S
0.83 0.24 0.03 0.010 0.003
0.88 0.14 0.03 0.018 0.005
 A.J. Clarke et al. / Acta Materialia 77 (2014) 17–27 19

Fig. 1. (a) Bright-field TEM image of lath martensite produced by oil quenching 4340 steel, containing twins highlighted by the arrow in (b). The inset
selected-area diffraction pattern in (b) is indexed as a h1 1 0i beam direction containing {1 1 2} twins and is properly rotated with respect to the image to
show the {1 1 2} habit plane. (c) Simulated electron diffraction pattern. The filled circles correspond to primary a reflections, the open circles correspond to
twin reflections of the filled circles, and the “x” symbols correspond to double-diffraction reflections.

Fig. 2. (a) Bright-field TEM image of nanoscale carbides (examples indicated by the arrows) in oil-quenched 4340 and (b) selected-area diffraction using
the ½0 0 1a zone axis. Data are consistent with that of Taylor et al. [15] for the e0 carbide with an orientation relationship of ½1 0 0e0 k½0 0 1a and
ð0 0 1Þe0 kð1 1 0Þa . Position A corresponds to f0 2 1ge0 reflections, position B corresponds tof0 1 1ge0 superlattice reflections, and position C corresponds to
double diffraction between f0 2 1ge0 and f1 1 0ga reflections. ð0 2 0Þe0 and ð0 0 2Þe0 reflections should be present near f1 1 0ga , but are presumed to be too weak
to be visible in close proximity to the much stronger matrix reflections.
20 A.J. Clarke et al. / Acta Materialia 77 (2014) 17–27
½1 0 0e0 k½0 0 1a
ð0 0 1Þe0 kð1 1 0Þa
From a tempering study of Fe–C–Ni steel, Taylor et al.
reported initial f0 2 3ga “spinodal” habit planes that
migrated toward f0 2 1ga for large precipitates [14,15].
These habit plane normals are indicated in Fig. 2a for ref-
erence. The large precipitates in the central region of
Fig. 2a have e0 carbide habit planes near ð0 2 0Þa , suggesting
that these precipitates are well beyond any early stage
“spinodal” clustering.
APT revealed ultra-fine-scale carbon clustering in the
oil-quenched martensitic microstructure, which is high-
lighted by the 5 at.% carbon isoconcentration surfaces in
Fig. 3a. Similar clustering, although qualitatively finer
in scale, produced after ice-water quenching is shown in
Fig. 3b. The features indicated by the arrows in Fig. 3a
and b display maximum carbon levels approaching 12–14 at.%,
as shown in the proximity histograms. Carbon atom
redistribution during the quench contributes to the high
strength of as-quenched 4340 steel [16]. Such rearrange-
ment has been attributed to carbon segregation to disloca-
tions and the formation of Cottrell atmospheres [17–19],
with carbon levels ranging from 6 to 8 at.% [18,19]. The
maximum carbon levels associated with the ultra-fine-scale
carbon clusters in Fig. 3 exceed the range typically
associated with Cottrell atmospheres at dislocations; these
levels are also lower than those anticipated for transition
carbides or cementite.
3.2. Martensitic microstructure tempered at 325 °C
APT revealed the formation of plate-like features with
carbon levels above the equilibrium amount of 25 at.%
anticipated for Fe3C-cementite after 325 °C tempering for
2 h, as shown in Fig. 4. There is no evidence of Cr, Mn
or Mo partitioning into the carbide (Fig. 4b), and the Ni,
Al and P levels are similar in the carbide and matrix
(Fig. 4c). There is apparent Si rejection from the carbide
(Fig. 4c). Similar rejection of Si from cementite at low
tempering temperatures, with no detectable diffusion of
Fig. 3. (a and b) Ultra-fine-scale carbon clusters in 4340 after oil or ice-water quenching, respectively, highlighted by 5 at.% C isoconcentration surfaces,
and proximity histograms showing the carbon levels associated with the features indicated by the arrows. APT results.
other substitutional elements, has been observed and
shown to correlate with the more rapid diffusion of Si com-
pared to that of Cr, Mn and Ni at temperatures between
300 and 400 °C [11].
TEM micrographs and selected-area diffraction and
microdiffraction results from the martensitic matrix and
carbide are shown in Fig. 5. The carbide contains internal
structure with laths ranging from 10 to 20 nm (Fig. 5a),
along with finer features ranging from 1 to 2 nm
(Fig. 5b). Diffraction data from these features are
consistent with the presence of Fe3C (cementite) with the
Bagaryatsky orientation relationship [20–22] (from
Fig. 5c–f) for the tempered martensite and cementite:
½0 1 1 k½0 0 1
a Fe3 C
ð2 1 1Þa kð0 1 0ÞFe3 C
Double diffraction accounts for the (weaker) reflections
in  16 f1 1 2ga positions. Complementary selected-area dif-
fraction and microdiffraction results from other precipi-
tates sometimes contained reflections only in  13 f1 1 2ga
positions (i.e. reflections at 16 f1 1 2ga positions are missing),
which may indicate cementite-like carbide and/or twin-
related structure. Simulated electron diffraction patterns
using the Isaichev orientation relationship [21,23] reveal
1
3
f1 1 2ga reflections, but additional reflections are also pre-
dicted that are not experimentally observed. Nagakura
et al. have reported that the transition from chi carbide
(Fe5C2) to Fe3C occurs by microsyntactic growth, or car-
bide intergrowth, requiring only iron atom displacements
and carbon diffusion [1,24,25].
3.3. Martensitic microstructure tempered at 450 °C
APT of a carbide after tempering at 450 °C for 2 h is
shown in Fig. 6a. A proximity histogram with a maximum
carbon level of 25 at.% is provided in Fig. 6b, suggesting
that the carbide is cementite. Strong evidence of Cr, Mn
and Mo partitioning to the carbide is also highlighted. Less
clear trends with respect to the carbide exist for Si, Al, Ni
and P in the proximity histogram in Fig. 6c, but based
 A.J. Clarke et al. / Acta Materialia 77 (2014) 17–27 21

Fig. 4. (a) C atom map and a 10 at.% C isoconcentration surface highlighting the carbide–matrix interface indicated by the arrow after oil quenching and
tempering 4340 steel at 325 °C for 2 h and (b and c) proximity histograms showing the C, Cr, Mn, Mo, Si, Al, Ni and P levels associated with the carbide
and martensitic matrix. Error bars are provided every fifth data point in the proximity histograms. APT results.

upon the rejection of Si from the carbide after tempering at
325 °C, it is speculated that Si is also rejected from the
carbide at 450 °C.
Carbides formed during high-temperature tempering are
intra- or inter-lath or at prior austenite grain boundaries.
Inter-lath carbides, typical of retained austenite decompo-
sition to ferrite and cementite, were observed after temper-
ing at 450 °C. Complex carbides, similar to those observed
after tempering at 325 °C (e.g. Fig. 5) with selected-area
diffraction and microdiffraction reflections at  16 f1 1 2ga
or  13 f1 1 2ga positions, were also observed after temper-
ing at 450 °C.
3.4. Martensitic microstructure tempered at 575 °C
APT results after tempering at 575 °C for 2 h are shown
in Fig. 7. A plate-like carbide is defined by 10 at.% isocon-
centration surfaces in Fig. 7a. These carbon isoconcentra-
tion surfaces are also shown in Fig. 7b–g to highlight the
location of the carbide within the analyzed volume and
the complex element partitioning associated with the car-
bide and ferrite (matrix). Atom maps are provided for
Cr, Mn and Mo in Fig. 7b–d, respectively. A complemen-
tary proximity histogram for these elements with respect to
the upper isoconcentration surface is provided in Fig. 7i.
The maximum carbon content is slightly higher than, but
close to, that anticipated for equilibrium cementite
(25 at.%). Cr, Mn and Mo clearly partition to the
cementite, with elevated levels within the cementite adja-
cent to the carbide–matrix interface. Equilibrium partition-
ing of these elements within the cementite has not yet been
achieved. Atom maps for Si, Al, Ni and P are shown in
Fig. 7e–h, respectively. A proximity histogram is shown
in Fig. 7j, highlighting the behaviors of these elements with
respect to the cementite and matrix. The Si map in Fig. 7e
and the proximity histogram in Fig. 7j clearly show that Si
is rejected from the cementite and partitions to the ferrite
matrix, but small Si clusters remain within the cementite,
perhaps trapped at internal structure or defects. The
behaviors of Al, Ni and P are less obvious, but closer exam-
ination of the proximity histogram and carbide–matrix
interface in Fig. 7j reveals that these elements behave
similarly to Si. Clustering of these elements within the
cementite, often at the same location(s) as the Si, is also
qualitatively observed in the element maps. P appears to
not only be segregated near the carbide–matrix interface
in Fig. 7h, but is also segregated within the cementite.
From three-dimensional visualization of this data set at dif-
ferent orientations, the lower arrowed P-enriched region is
linear in morphology, and therefore, most likely a disloca-
tion. The extent of P segregation around the line is consis-
tent with that of a dislocation. Such features have been
attributed to APT observations of Cottrell atmosphere for-
mation at dislocations [17], but in the case of the cementite
may also have been associated with an internal boundary in
the cementite, as shown in Fig. 5.
Bright-field TEM images of the ferrite (matrix) and
coarse carbides after tempering at 575 °C for 2 h are shown
in Fig. 8a and b. Diffraction from four different carbides in
three different ferrite grains matched orthorhombic cement-
ite, with lattice parameters of a = 5.090 Å, b = 6.748 Å and
c = 4.523 Å. A selected-area diffraction pattern for the
22 A.J. Clarke et al. / Acta Materialia 77 (2014) 17–27

Fig. 5. Cementite with internal structure (arrows) and selected-area diffraction for (a) the ½113 a zone axis and (b) the ½0 1 1a zone axis after oil quenching
and tempering 4340 steel at 325 °C for 2 h. (c) Selected-area diffraction for the ½0 1 1a zone axis and (d) a simulated electron diffraction pattern for the
matrix (filled circles) and cementite (open circles) and microdiffraction for (e) the matrix and (f) cementite used to determine the Bagaryatsky orientation
relationship. Bright-field TEM images and electron diffraction.

cementite and ferrite is shown in Fig. 8c, and a simulated
electron diffraction pattern is shown in Fig. 8d. The Bagar-
yatsky orientation relationship was determined. Many
cementite particles still have a f1 1 2ga habit plane, but
some particles appear to be spheroidizing, resulting in
slight rotation of the cementite with respect to the ferrite,
as shown in Fig. 7c. This rotation suggests a loss of precip-
itate coherency with the matrix and rotation away from the
Bagaryatsky orientation relationship. Simulated electron
diffraction patterns based on the Isaichev orientation rela-
tionship, which can be considered as a small rotation from
Bagaryatsky, did not produce the observed rotation in
the experimental pattern. Low dislocation density is also
observed in the ferrite, suggesting that 575 °C tempering
is sufficient to result in significant recovery of the
dislocation substructure produced by the martensitic
transformation.
4. Discussion
Medium-carbon, alloyed steels are typically tempered
after quenching to achieve the desired hardness, strength,
 A.J. Clarke et al. / Acta Materialia 77 (2014) 17–27 23

Fig. 6. (a) C atom map and a 5 at.% C isoconcentration surface highlighting the carbide-matrix interface indicated by the arrow after oil quenching and
tempering 4340 steel at 450 °C for 2 h and (b and c) proximity histograms showing the C, Cr, Mn, Mo, Si, Al, Ni and P levels associated with the cementite
and the matrix. Error bars are provided every fifth data point in the proximity histograms. APT results.

ductility and/or toughness. Along with sometimes dramatic
mechanical property changes, quenching and tempering
also results in subtle microstructural changes. Three dis-
tinct tempering stages are generally accepted for carbon
steels, including: stage I tempering from 100 to 250 °C
involving the formation of transition carbide and reduction
of the martensite carbon content; stage II tempering from
200 to 300 °C involving the decomposition of retained aus-
tenite into ferrite and cementite; and stage III tempering
from 250 to 350 °C involving the replacement of transition
carbide and low-carbon martensite by cementite and ferrite
[1,6,13,26,27]. Importantly, these tempering stages overlap
and the progression of microstructural changes that occur
are subtle and significantly influenced by alloying element
additions.
TEM revealed lath martensite, instances of twinning,
and indications of transition carbide that may have formed
close to the martensite start (Ms) temperature during oil
quenching in this 4340 steel. APT revealed ultra-fine-scale
carbon clustering after quenching with maximum carbon
levels ranging from 12 to 14 at.%, which are inconsistent
with those anticipated for transition carbides (e.g. e, e0 or
g, with anticipated carbon levels near 30 at.% or above).
The experimentally obtained carbon levels appear to be
influenced by quench rate, as shown qualitatively in
Fig. 3. Importantly, the carbon clustering produced during
the quench (to room temperature or perhaps an intermedi-
ate temperature below Ms for subsequent thermal process-
ing) will set the stage for subsequent tempering reactions,
such as carbide formation that occurs during the first stage
of tempering [1,28,29].
APT showed a carbon-enriched plate with a maximum
carbon level exceeding the 25 at.% anticipated for equilib-
rium cementite after tempering at 325 °C, traditionally a
stage III tempering temperature. Cr, Mn and Mo did not
partition within the carbide or tempered martensite. Simi-
larly, Al, Ni and P did not partition, but Si was rejected
from the carbide into the matrix. After tempering an Fe–
C–Mn–Si steel at 350 or 400 °C for 1.8  103 s (0.5 h),
Babu et al. reported no Mn or Si partitioning, suggesting
early stage paraequilibrium growth of cementite, whereas
tempering at 350 °C for 6.48  105 s (180 h) or 500 °C for
0.3  103 s (0.83 h) resulted in Si rejection from and Mn
enrichment into the cementite [10]. Zhu et al. reported that
quenching and tempering of 4340 steel at 300 or 350 °C for
1 h resulted in features containing 12–14 at.% carbon, with-
out Cr, Mn or Ni partitioning or substantial Si redistribu-
tion [11]. The carbon levels identified by Zhu et al. are
much lower than those obtained for the plate-like carbide
in Fig. 4 and agree better with those obtained after quench-
ing (Fig. 3). This discrepancy may be associated with
quench rate and/or sample size variations, or perhaps carbon
24 A.J. Clarke et al. / Acta Materialia 77 (2014) 17–27

Fig. 7. (a) C atom map and 10 at.% C isoconcentration surfaces (red) defining the carbide–matrix interface after oil quenching and tempering 4340 steel at
575 °C for 2 h and atom maps for (b) Cr, (c) Mn, (d) Mo, (e) Si, (f) Al and (g) Ni, including the 10 at.% C isoconcentration surfaces, and (h) P segregation.
(i and j) Proximity histograms from the upper isoconcentration surface showing complex element partitioning in the cementite and ferrite matrix. Error
bars are provided every fifth data point in the proximity histograms. APT results. (For interpretation of the references to color in this figure legend, the
reader is referred to the web version of this article.)
 A.J. Clarke et al. / Acta Materialia 77 (2014) 17–27 25

Fig. 8. (a and b) Ferrite (matrix) and cementite after oil quenching and tempering 4340 steel at 575 °C for 2 h. (c) Selected-area diffraction for the ½0 1 1a
zone axis and (d) a simulated electron diffraction pattern for the matrix (filled circles) and cementite (open circles). Some rotation of the cementite
diffraction data with respect to the matrix is observed in (c)—note splitting of reflections at 1 and 7 o’clock in (c), suggesting coherency loss during
spheroidization and rotation away from the Bagaryatsky orientation relationship. Bright-field TEM images and selected-area electron diffraction.

detection during APT [30–33]. Systematic studies on the
location of substitutional elements with respect to transition
carbide are limited, but Barnard et al. reported that after tem-
pering an Fe–C–Si–Mn–Cr steel for 24 h at 250 °C, epsilon
carbide was not enriched with Si [34].
Although the elevated carbon level in Fig. 4 suggests that
transition carbide is possible, APT alone cannot be used to
conclusively assign whether transition carbide or cementite
is present. Cementite was confirmed by TEM selected-area
diffraction and microdiffraction for the 325 °C condition
(Fig. 5), but more complex electron diffraction was also
observed, perhaps indicating cementite-like carbide and/or
twin-related structure. Based upon the available TEM and
APT shown here, the carbide in Fig. 4 is more likely
early-stage cementite with initial Si redistribution, which
was reported by Babu et al. [10]. Although Babu et al. sug-
gest the presence of cementite, their selected-area diffraction
reveals a f2 0 0ga carbide habit plane, whereas the selected-
area diffraction in Fig. 5 reveals a f1 1 2ga cementite habit
plane. Low-temperature tempering and more complemen-
tary APT/TEM microstructural studies may help to unam-
biguously clarify the details associated with the carbides
present after tempering at 325 °C.
After tempering at 450 or 575 °C, APT revealed plate-
shaped carbides containing 25 at.% C, the amount
expected in cementite. Cr, Mn and Mo clearly concentrated
in the cementite close to the ferrite–carbide interface, with
concentrations decreasing into the cementite (e.g. Fig. 7).
The Si, Al, Ni and P trends with respect to the cementite
and ferrite were less clear after tempering at 450 °C, but
Si appears to be rejected from the cementite. After
tempering at 575 °C, Si, Al, Ni and P were all noticeably
rejected from the cementite, persisting near the ferrite–
carbide interface. After tempering 4340 steel at 400 °C for
1 h, Zhu et al. reported carbide with carbon levels near
14 at.%, without significant Cr, Mn, Ni or Si partitioning
[11]. Tempering at 400 °C for 10 h resulted in a carbon level
exceeding 20 at.% and apparent Si and Ni rejection from
and Mn and Cr enrichment into the carbide [11]. The Cr,
Mn, Si and Ni trends observed in the current work agree
with the results of Zhu et al. [11]. The current results
provide direct evidence for staged cementite growth, with
early paraequilibrium growth based only on carbon
diffusion, followed by initial silicon redistribution and then
growth influenced by the diffusion of carbide-forming
elements to and within the cementite.
26 A.J. Clarke et al. / Acta Materialia 77 (2014) 17–27
While APT revealed complex substitutional element
partitioning and P segregation near the ferrite–carbide
interface, the TEM images from the steel tempered at
575 °C for 2 h showed coarsened cementite with indications
of spheroidization and loss of the Bagaryatsky orientation
relationship (Fig. 8) compared to more typical elongated
carbides that result from lower-temperature tempering
(Fig. 4), although it is important to note that complex car-
bides were still present in the 575 °C, 2 h temper condition.
P segregation at the carbide–matrix interface has been sug-
gested as a contributing mechanism for slowing cementite
growth and retarding the spheroidization of elongated car-
bide structures [35,36]. The subtle cementite spheroidiza-
tion at 575 °C suggests that this temperature is indeed at
the upper limit of the temperature range (generally 475–
575 °C) where solute mobility effectively retains elongated
carbide structures. Additionally, the enrichment of Si, Al
and Ni at the ferrite–cementite interface suggests that the
relative mobility of these elements may also contribute to
the retention of elongated carbides in alloyed steels such
as 4340. Alloy effects on carbide morphology have signifi-
cant ramifications for the mechanical performance of
quenched and tempered steels. For instance, James et al.
showed that elongated carbides retained during medium-
temperature tempering could effectively limit the decrease
in strength during tempering without significantly decreas-
ing ductility [37]. However, Materkowski and Krauss
reported a plateau in toughness during Stage III tempering
due to easy crack propagation along elongated brittle car-
bides [38]. Comparing high- and low-P 4340 steels, Mater-
kowski and Krauss also showed a decrease in toughness for
the high-P steel over the entire tempering range examined
[38].
Si and Al are commonly used to modify carbide forma-
tion temperatures and kinetics in low-carbon steels, thereby
promoting carbon availability for enriching and stabilizing
austenite (e.g. during austempering to produce transforma-
tion-induced plasticity (TRIP) steels with carbon-enriched
retained austenite). Direct experimental evidence that Al
and Si behave similarly with respect to partitioning from
cementite during martensite tempering is shown in Fig. 7,
making Al an effective alloying addition to help prevent
cementite formation during processing. Although Al is
often used to partially replace Si in alloying strategies for
promoting austenite retention, the results presented here
suggest that P plays a similar role in retarding cementite
development. Thus, alloying strategies including P, espe-
cially in conjunction with Si and/or Al, may present an
additional approach for controlling cementite development
and austenite retention, keeping in mind the possibility of
embrittlement associated with P segregation to austenite
grain boundaries [38–42].
5. Conclusions
Alloying of medium-carbon steels permits a range of
strength and toughness combinations to be achieved after
tempering. Attainment of desirable mechanical properties
in steels such as 4340 is related to the redistribution of sol-
ute and subtle microstructural changes that occur during
tempering. APT and TEM were used to systematically
examine the chemical and structural changes that occur
at the atomic and nanoscale in 4340 steel after quenching
and tempering at 325, 450 or 575 °C for 2 h, resulting in
the following conclusions:
(1) Lath martensite, instances of twinning and indica-
tions of e0 (g) transition carbide were identified in
4340 steel after oil quenching using TEM. Transition
carbides may have formed close to the martensite
start (Ms) temperature during quenching.
(2) APT revealed ultra-fine-scale carbon clustering in
4340 steel, with maximum carbon levels of 12–14 at.%.
Quench rate appears to influence the extent of carbon
clustering, setting the initial microstructural condi-
tion for subsequent tempering reactions.
(3) From APT, tempering 4340 steel at 325 °C resulted in
a carbon-enriched plate (>25 at.%) without Cr, Mn,
Mo, Al, Ni or P partitioning and the initiation of Si
rejection from the carbide.
(4) TEM electron diffraction was used to identify
cementite after tempering at 325 or 450 °C. The
Bagaryatsky orientation relationship for cementite
and the tempered martensite was determined, and
precipitates with internal structure after tempering
at 325 or 450 °C were observed. Complex electron
diffraction sometimes indicated cementite-like precip-
itates and/or twin-related structure.
(5) APT after tempering at 450 or 575 °C showed
cementite with carbon levels near 25 at.% and com-
plex element partitioning. Cr, Mn and Mo unambig-
uously partitioned to the cementite. Si appears to be
rejected from cementite at 450 °C, without significant
Al, Ni or P partitioning trends. Tempering at 575 °C
results in noticeable rejection of Si, Al, Ni and P from
the cementite.
(6) The APT and TEM results presented here provide
direct evidence for staged cementite growth, where
early-stage growth likely occurs under paraequilibri-
um conditions, followed by initial silicon redistribu-
tion and subsequent alloying element redistribution
during late-stage growth.
(7) APT clearly revealed P segregation at ferrite–cementite
interfaces and the formation of Cottrell atmospheres
within the cementite.
(8) From TEM, coarsened cementite existed after tem-
pering at 575 °C, with indications of spheroidization
and loss of the Bagaryatsky orientation relationship.
Acknowledgements
The authors gratefully acknowledge support from Los
Alamos National Security, LLC, operator of the Los
 A.J. Clarke et al. / Acta Materialia 77 (2014) 17–27
Alamos National Laboratory under contract number DE-
AC52-06NA25396 with the US Department of Energy and
the Innovative Manufacturing Initiative of the Office of
Advanced Manufacturing (AMO) at the US Department
of Energy (Award DE-EE0005765). Atom probe tomogra-
phy research (M.K.M. and K.A.P.) was supported through
a user project supported by Oak Ridge National Labora-
tory’s Center for Nanophase Materials Sciences (CNMS),
which is sponsored by the Scientific User Facilities
Division, Office of Basic Energy Sciences, US Department
of Energy. J.E. Scott, T.J. Tucker, T.V. Beard, and R.W.
Hudson are thanked for specimen heat-treatment and
machining assistance. Helpful discussions with Professor
S.W. Thompson regarding TEM electron diffraction data
and discussions with Professors G.D.W. Smith, J.G.
Speer, E. De Moor, B.C. De Cooman, D.K. Matlock,
and S.S. Babu and Dr. F.G. Caballero are also gratefully
acknowledged.
References
[1] Krauss G. STEELS: processing, structure, and performance. Materi-
als Park, OH: ASM International; 2005.
[2] Grange RA, Hribal CR, Porter LF. Metall Trans A 1977;8A:1775.
[3] Zhu C, Cerezo A, Smith GDW. Ultramicroscopy 2009;109:545.
[4] Grossman MA, Bain EC. Principles of heat treatment. 5th ed. Metals
Park, OH: American Society of Metals; 1964.
[5] Saeglitz M, Krauss G. Metall Mater Trans A 1997;28A:377.
[6] Williamson DL, Schupmann RG, Materkowski JP, Krauss G. Metall
Trans A 1979;10A:379.
[7] Lee H-C, Krauss G. Intralath carbide transition in martensitic
medium carbon steels tempered between 200 and 300 °C. In: Krauss
G, Repas PE, editors. Gilbert R. Speich symposium proceedings:
fundamentals of aging and tempering in bainitic and martensitic steel
products. Warrendale, PA: Iron and Steel Society; 1992. p. 39.
[8] Williamson DL, Nakazawa K, Krauss G. Metall Trans A 1979;
10A:1351.
[9] Hirotsu Y, Nagakura S. Trans Jpn Inst Metals 1974;15:1074.
[10] Babu SS, Hono K, Sakurai T. Metall Mater Trans A 1994;25A:
499.
[11] Zhu C, Xiong XY, Cerezo A, Hardwicke R, Krauss G, Smith GDW.
Ultramicroscopy 2007;107:808.
[12] Miller MK. Atom probe tomography. New York: Kluwer Academic/
Plenum Press; 2000.
[13] Speich GR. Trans Metall Soc AIME 1969;245:2553.
[14] Taylor KA, Chang L, Olson GB, Smith GDW, Cohen M, Vander
Sande JB. Metall Trans A 1989;20A:2717.
[15] Taylor KA, Olson GB, Cohen M, Vander Sande JB. Metall Trans A
1989;20A:2749.
[16] Leslie WC, Sober RJ. ASM Trans Q 1967;60:459.
27
[17] Chang L, Barnard SJ, Smith GDW. The segregation of carbon atoms
to dislocations in low-carbon martensites: studies by atom field ion
microcopy and atom probe microanalysis. In: Krauss G, Repas PE,
editors. Gilbert R. Speich symposium proceedings: fundamentals of
aging and tempering in bainitic and martensitic steel products. War-
rendale, PA: Iron and Steel Society; 1992. p. 19.
[18] Wilde J, Cerezo A, Smith GDW. Scr Mater 2000;43:39.
[19] Cochardt AW, Schoek G, Wiedersich H. Acta Metall 1955;3:533.
[20] Bhadeshia HKDH. Worked examples in the geometry of crystals. 2nd
ed. London: Institute of Materials; 2001.
[21] Zhou DS, Shiflet GJ. Metall Mater Trans A 1992;23A:1259.
[22] Bagaryatsky YA. Dokl Akad Nauk SSSR 1950;73:1161.
[23] Isaichev IV. Zh Tekh Fiz 1947;17:835.
[24] Nagakura S, Hirotsu Y, Kusunoki M, Suzuki T, Nakamura Y.
Metall Trans A 1983;14A:1025.
[25] Nagakura S, Suzuki T, Kusunoki M. Trans Jpn Inst Metals 1981;22:
699.
[26] Speich GR. Tempered ferrous martensitic structures. In: Metallogra-
phy, structures and phase diagrams. Metals handbook, Vol. 8. Mate-
rials Park, OH: American Society for Metals; 1973. p. 202.
[27] Speich GR, Leslie WC. Metall Trans 1972;3:1043.
[28] Genin JMR, Flinn PA. Trans Metall Soc AIME 1968;242:1419.
[29] Kurdjumov GV, Khachaturyan AG. Metall Trans 1972;3:1069.
[30] Marceau RKW, Choi P, Raabe D. Ultramicroscopy 2013;132:239.
[31] Miyamoto G, Shinbo K, Furuhara T. Scr Mater 2012;67:999.
[32] Takahashi J, Kawakami K, Kobayashi Y. Ultramicroscopy 2011;111:
1233.
[33] Yao L, Gault B, Cairney JM, Ringer SP. Philos Mag Lett 2010;90:
121.
[34] Barnard SJ, Smith GDW, Garratt-Reed AJ, Vander Sande J. Atom
probe studies: (1) the role of silicon in tempering of steel and (2) low
temperature chromium diffusivity in bainite. In: Aaronson HI,
Laughlin DE, Sekerka RF, Wayman CM, editors. Proceedings of
an international conference on solid-solid phase transforma-
tions. Warrendale, PA: Metallurgical Society of AIME; 1982. p. 881.
[35] Rellick JR, McMahon CJ. Metall Trans 1974;5:2439.
[36] Brown EL, Krauss G. Metall Trans A 1986;17A:31.
[37] James BA, Matlock DK, Krauss G. Interactive effects of phosphorus
and tin on carbide evolution and fatigue properties of 5160 steel. 38th
mechanical processing and steel working conference proceedings, Vol.
XXXIV. Warrendale, PA: Iron and Steel Society; 1997. p. 579.
[38] Materkowski JP, Krauss G. Metall Trans A 1979;10A:1643.
[39] Chen HC, Era H, Shimizu M. Metall Trans A 1989;20A:437.
[40] Traint S, Pichler A, Hauzenberger K, Stiaszny P, Werner E. Influence
of silicon, aluminum, phosphorus and copper on the phase transfor-
mations of low alloyed TRIP-steels. In: De Cooman BC, editor.
International conference on trip-aided high strength ferrous alloys—
proceedings. Aachen, Mainz: GRIPS; 2002. p. 121.
[41] Gallagher MF, Speer JG, Matlock DK. Microstructure development
in TRIP-sheet steels containing Si, Al, and P. 44th mechanical
processing and steel working conference proceeding, Vol. XL. War-
rendale, PA: Iron and Steel Society; 2002. p. 153.
[42] Zhao L, van Dijk NH, Peekstok ER, Tegus O, Brück E, Sietsma J.
Mater Sci Forum 2007;539–543:4321.