Very strong low temperature bainite

F. G. Caballero, H. K. D. H. Bhadeshia, K. J. A. Mawella, and D. G. Jones,

and P. Brown

Bainite has been obtained by heat treatment at temperatures as low as 125°C in a high carbon, high silicon steel.
This has had the effect of greatly re® ning the microstructure, which is found to have a strength in excess of 2.5 GPa
together with an ability to ¯ ow plastically before fracture. Such properties have never before been achieved with
bainite. In this paper metallographi c details are reported of the very ® ne bainitic microstructure associated with the
incredibly low transformatio n temperature, where during the time scale of the experiments, an iron atom cannot
diffuse over a distance greater than ~ 10 ­ 17 m. Yet, the microstructure has a scale in the micrometre range,
consistent only with a displacive mechanism of transformation . MST/5034

Dr Caballero ([email protected]) is in the Department of Physical Metallurgy, Centro Nacional de Investigaciones

Metalu rgicas, CSIC, Avda. Gregorio del Amo 8, 28040 Madrid, Spain, Professor Bhadeshia is in the Department of Materials
Science and Metallurgy, University of Cambridge, Pembroke Street, Cambridge CB2 3QZ, UK, and Dr Mawella, Mr Jones,
and Dr Brown are in the Structural Materials Centre, Defence Evaluation Research Agency, R1079, Building A7, Farnborough
GU14 0LX, UK. Manuscript received 20 March 2001; accepted 22 May 2001.
# 2002 IoM Communication s Ltd.

bainite can be obtained by transformation at temperatures,

Introduction as low as 125 °C during heat treatment for many days. To
put this into context, an iron atom diffuses a distance of
The addition of ~ 2 wt-% of silicon to steel enables the just 10 ­ 17 m during the course of the transformation, even
production of a distinctive microstructure consisting of though the resulting bainite has a scale of many micro-
a mixture of bainitic ferrite, carbon enriched retained metres. The aim of this work presented here was to charac-
austenite, and some martensite. 1­ 5 The silicon suppresses terise and understand this ® ne bainitic microstructure in the
the precipitation of brittle cementite, and hence should lead context of the mechanism of transformation.
to an improvement in toughness. However, the full bene® t
of this carbide free bainitic microstructure has frequently
not been realised. This is because the bainite reaction stops
well before equilibrium is reached, i.e. when the carbon Experimental procedures
concentration of the residual austenite reaches a point
given by the To curve, beyond which diffusionless growth is The chemical composition of the steel studied was Fe ±
prevented. This leaves large regions of untransformed auste- 1 .59Si ± 1 .94Mn ± 1 .33Cr ± 0.30Mo ± 0.02Ni ± 0.11V (wt-%).
nite that under stress decompose to hard, brittle marten- The alloy was supplied as a cast ingot, specimens of which
site.6 were then homogenised at 1200°C for 2 days while sealed in
There is, however, a design procedure6,7 that avoids this partially evacuated quartz capsules ¯ ushed with argon. The
dif® culty in three ways: by adjusting the To curve to greater sealed specimens were cooled in air following the homo-
carbon concentrations using substitutional solutes, by con- genisation heat treatment.
trolling the mean carbon concentration, and by minimis- The homogenised specimens were austenitised for 15 min
ing the transformation temperature. Steels were designed at 1000 °C, and then isothermally transformed at tempera-
on this basis and when tested, revealed a reduction in the tures ranging from 125 to 500 °C for different times before
impact transition temperature of more than 100 K.6,7 Frac- quenching into water.
ture toughness measurements con® rmed the remarkable Optical microscopy and SEM were used to examine
levels of toughness that could be achieved, in some cases the etched microstructures. Specimens were ground and
matching the much more expensive maraging steels.8 ± 10 polished using standard techniques and etched in 2 vol.-%
Other aspects of the theory have been veri® ed using a nital solution and examined using a Jeol JXA 820 SEM
variety of advanced research techniques.11 operated at 10 ± 15 kV. The volume fraction of bainite
The ® rst major commercial exploitation came with the Vb was estimated by a systematic manual point counting
development of a kinetic theory 12 to complement the three procedure on SEMs.15 A grid superimposed on the micro-
thermodynamic criteria, so that continuous cooling trans- structure provides, after a suitable number of placements,
formation, which is typical in industry, could be dealt with. an unbiased statistical estimate of Vb. Stereological errors
More recently, it has been demonstrated experimentally for Vb values correspond to the standard deviation of the
that models based on phase transformation theory can be measurement.
applied successfully to the design of novel, strong, tough Quantitative X-ray analysis was used to determine the
steels.13,14 These alloys were designed to ensure that the volume fraction of retained austenite. After grinding and
hardenability is consistent with speci® c industrial speci® ca- ® nal polishing using 0.25 mm diamond paste, the specimens
tions and have achieved the highest ever combinations of were etched to obtain an undeformed surface. They were
strength and toughness for bainitic steels. Toughness values then step scanned in a Philips PW 1730 X-ray diffracto-
of nearly 130 MPa m 1/2 have been obtained for strength in meter using un® ltered Cu Ka radiation The 2h scan rate was
the range of 1600 ± 1700 MPa. This compares well with 1 ° min­ 1. The machine was operated at 40 kV and 40 mA.
maraging steels, which are at least ninety times more The retained austenite content was calculated from the
expensive. integrated intensities of (200), (220), and (311) austenite
During the course of this research on the design of peaks, and those of (002), (112), and (022) planes of
bainitic steels, 13,14 a discovery was made, that very ® ne ferrite.16 Using three peaks from each phase avoids biasing

DOI 10.1179/026708301225000725 Materials Science and Technology March 2002 Vol. 18 279
280 Caballero et al. Very strong low temperature bainite

a b

c d

a 125°C for 29 days; b 150°C for 29 days; c 190°C for 2 weeks; d 300°C for 2 weeks
1 Optical micrographs of isothermal decomposition of austenite: Hv Vickers hardness (30 kg load)

the results owing to any crystallographic texture in the could not be conducted because of extensive casting defects
specimens.17 The carbon concentration in the austenite and in the form of porosity.
ferrite was estimated using the measured lattice para-
meters.18
Specimens for TEM were machined down to 3 mm
diameter rods. The rods were sliced into 100 mm thick discs
Results and discussion
and subsequently ground down to foils of 50 mm thickness
on wet 800 grit silicon carbide paper. These foils were MICROSTRUCTURE OF ISOTHERMALLY
® nally electropolished at room temperature until perfora- TRANSFORMED SPECIMENS
tion occurred, using a twinjet electropolisher set at a voltage Isothermal transformation in the temperature range
of 40 V. The electrolyte consisted of 5% perchloric acid, 125 ± 300°C led to the formation of bainite, as illustrated
15% glycerol, and 80% methanol (vol.-%). Carbon replicas in the optical micrographs presented in Fig. 1a ± d. The
were extracted in 5 vol.-% hydrochloric acid and 95 vol.-% early stages of transformation are shown in Fig. 1 a and b
methanol. Foils and replicas were examined in a Jeol JEM for specimens isothermally heat treated for 29 days at 125
200 CX TEM at an operating voltage of 200 kV. and 150°C respectively. Transformation remains incomplete
The martensite start temperature of the steel was mea- even after 60 days of holding time at 125°C. Transformation
sured using a Thermecmaster Z thermomechanical simu- at temperatures much in excess of 300°C did not lead to
lator. Cylindrical specimens 12 mm in height and 8 mm in the formation of bainite, suggesting that the bainite start
diameter were heated at a rate of 5 K s ­ 1 to 1000 °C and temperature is between 300 ± 350 °C; experiments were
then cooled at two different rates (40 and 100 K s ­ 1). conducted up to 500 °C, which revealed only eutectoid
The formation of martensite during cooling was detected decomposition.
by monitoring the fractional change in dilatation with Figure 2 shows the evolution of hardness and the fraction
temperature. of bainite as a function of the isothermal transformation
Austenite formation begins during heating at the Ac 1 temperature and time. The rate of transformation is fastest
temperature and is completed when the Ac 3 temperature is when the temperature exceeds 150 °C. For example, it is
reached. Tempering must be carried out below the Ac 1 estimated that the fraction of bainite is 0 .8 for transforma-
temperature to avoid the accidental formation of austenite. tion at 190 and 250 °C for 9 and 2 days, respectively.
The austenite formation temperatures were determined However, a very long heat treatment is required (more than
using a Thermecmastor Z simulator. Dilatometric speci- 2 months) to obtain substantial transformation when the
mens were heated at a rate of 5 K s ­ 1 to 1000 °C and then temperature is as low as 125°C. At 300 °C, the maximum
cooled at 40 K s­ 1. The formation of austenite during volume fraction of bainite formed was 0 .6 and 4 days are
heating was detected by monitoring the fractional change in needed to complete the transformation.
dilatation with temperature. An Ac1 temperature of 805°C X-ray analysis was used to estimate the quantities of
and an Ac 3 temperature of 840 °C were determined from the retained austenite present in the microstructure, and its
heating experiment. Specimens were tempered at tempera- carbon concentration, following two weeks at different
tures ranging from 400 to 700 °C for 30 min and 1 h. temperatures. The experimental results are presented in
Compression tests were carried out at room temperature Table 1 and Figs. 3 and 4. The retained austenite fraction
on specimens transformed isothermally at 190 °C for 2 is expected to increase for the higher transformation
weeks. The compression specimens of 5 mm in diameter temperature. This is because less bainite forms at 300°C,
and 15 mm in length were tested in accordance with and in contrast to low carbon bainitic steels, the residual
ASTM E9 ± 89a (Ref. 19) on a 100 kN Mayes servohy- austenite is stable to martensitic transformation on cool-
draulic machine at a crosshead speed of 1 mm min ­ 1. Two ing to ambient temperature even at the higher bainite
specimens were tested in compression. Reliable tensile tests transformation temperatures.

Materials Science and Technology March 2002 Vol. 18

 Caballero et al. Very strong low temperature bainite 281

a 125°C; b 150°C; c 190°C; d 250°C; e 300°C

2 Plot of hardness (30 kg load) values and bainite volume fraction as function of time at different transformation temperatures

The bainite transformation progresses by the diffusion- where xa is the amount of carbon trapped in the bainitic
less growth of tiny platelets known as `subunits’ . The excess ferrite and x¯ is the overall amount of carbon in the steel.
carbon in these platelets partitions into the residual auste- The fact that the measured carbon concentrations of the
nite soon after the growth event. Diffusionless growth of retained austenite in Fig. 4 lie between the To and the
this kind can only occur if the carbon concentration of the paraequilibrium Ae 3 phase boundaries is consistent with
residual austenite is below that given by the T’o curve. The a mechanism in which the bainite grows without any
To curve is the locus of all points, on a temperature versus diffusion, but with excess carbon partitioning into the auste-
carbon concentration plot, where austenite and ferrite of nite soon after transformation. It is for this reason that the
the same chemical composition have the same free energy.
The T’o curve is de® ned similarly but taking into account
the stored energy of the ferrite owing to the displacive
mechanism of transformation. It follows that the maximum
amount of bainite that can be obtained at any temperature
is limited by the fact that the carbon content of the residual
austenite must not exceed the T’o curve on the phase
diagram.20 ± 23
The variation of the carbon content of austenite xc as a
function of the average carbon content of the alloy xo and
the volume fraction of bainitic ferrite Vb is given by24

Å + Vb xo {xa :
( )
xc ~x : : : : : : : : : : (1)
(1{Vb )

Table 1 Experimental quantitative data on microstructure*

Bainite formation
temperature, °C Vc Vb xc , wt-% xa , wt-%

300 0.35+ 0. 01 0.65+ 0. 01 1.69+ 0. 10 0.20+ 0. 07

250 0.16+ 0. 01 0.84+ 0. 01 1.76+ 0. 10 0.32+ 0. 07
190 0.13+ 0. 01 0.87+ 0. 01 1.72+ 0. 10 0.32+ 0. 07
3 Plot of volume fraction of retained austenite as func-
* Vc volume fraction of austenite. tion of isothermal transformation temperature

Materials Science and Technology March 2002 Vol. 18

282 Caballero et al. Very strong low temperature bainite

4 Calculated T ’o, To, and paraequilibrium Ae3 curves for

studied steel and experimental data for carbon compo- a
sition of retained austenite in specimens heat treated
for 2 weeks at different temperatures: x¯ represents
overall carbon content of steel; phase boundaries were
calculated as in Ref. 25

bainite reaction remains incomplete, with the maximum

fraction limited by a boundary well below the paraequili-
brium Ae3 phase boundary.25 The fact that the measured
carbon concentrations of the austenite exceed the To con-
centration is a consequence of the fact that the isolated ® lms
of austenite between the bainite plates can accumulate
carbon, beyond To (Ref. 26). They cannot, of course,
transform to bainite once the To curve is exceeded.
According to the experimental data in Table 1, despite
the quite large fraction of bainite, austenite is not greatly
enriched in carbon. This can only be possible if a substan-
b
tial quantity of carbon is trapped in the bainitic ferrite.
Experimental results in Table 1 reveal this carbon in solid a thin foil; b carbon extraction replica
solution in the ferrite and Fig. 5 con® rms the presence of
5 Micrographs of some carbides precipitated in bainitic
some cementite in the bainitic ferrite, in the specimen
ferrite in microstructure formed at 190°C for 2 weeks
isothermally transformed at 190 °C for 2 weeks. The car- (TEM)
bides are extremely ® ne, approximately 20 nm wide and
175 nm long. These observations explain the small extent of
carbon enrichment detected in the residual austenite. microstructure formed at 190°C for 2 weeks. Tempering
Using all these experimental data, the time ± temperature ± that microstructure at temperatures lower than 500°C for
transformation (TTT) diagram of the steel was calculated25 30 min did not result in any signi® cant loss of hardness.
and compared against the experimental data (Fig. 6). The
curves represent the initiation of transformation whereas
the points all represent a ® nite amount of bainite (a volume
fraction of ~ 0 .05); in the circumstances, the agreement Conclusions
between the measured data and the calculated TTT curve is
considered reasonable. Bainite has been obtained by transformation at tempera-
tures as low as 125°C in a high carbon, high silicon steel.

MECHANICAL PROPERTIES
As mentioned above, the bainite that forms has a very ® ne
microstructure because of the low transformation tempera-
tures. This is illustrated in the TEMs shown in Fig. 7. Quite
remarkably, the bainite plates formed at 190 °C have a width
that is less than 50 nm, with each plate separated by an even
® ner ® lm of retained austenite. Not surprisingly, the ® ne
microstructure leads to a very high strength.
Figure 8 shows the compression curve of the micro-
structure formed by transformation at 190°C for 2 weeks; a
view of the specimen after testing is shown in Fig. 9. The
specimen clearly has considerable plasticity for deformation
at room temperature even though the yield strength in
compression is ~ 2000 MPa.

TEMPERING EXPERIMENTS
Experiments were carried out to study the stability of the
microstructure to tempering heat treatments. Figure 10 6 Calculated time ± temperature ± transformation diagram
shows the hardness as a function of the tempering tem- of steel: points in graphs represent start and ® nish
perature. The experiments were conducted on the bainitic time for bainite transformation

Materials Science and Technology March 2002 Vol. 18

 Caballero et al. Very strong low temperature bainite 283

9 Compression specimen after testing

7 Thin foil TEMs of microstructure obtained at 190°C for

2 weeks

The bainite plates thus obtained are incredibly ® ne, with a

width of some 50 nm, giving a very high strength in excess
of 2 .5 GPa and perceptible ductility. Such properties have
never been achieved with bainite. It is the intention now to 10 Plot of hardness values (30 kg load) as function of
design new alloys using the theory of the bainite reaction, tempering: horizontal line represents hardness of
for creating a new class of ultrahigh strength steels and then microstructure formed at 190°C for 2 weeks
to conduct comprehensive mechanical and microstructural
tests.

Acknowledgements

This work was carried out as part of Technology Group 4

(Materials and Structures) of the MoD Corporate Research
Programme. The authors would like to thank Professor
Alan Windle for the provision of laboratory facilities at
the University of Cambridge.

References

1. s. j. matas and r. f. hehemann: Trans. AIME, 1961, 221 , 179 ±

185.
2. r. entin: in `Decomposition of austenite by diffusional
processes’ , (ed. V. F. Zackay and H. I. Aaronson), 295 ± 311;
1962, New York, NY, Interscience.
3. r. f. hehemann: in `Phase transformations’ , 397 ± 432; 1970,
Metals Park, OH, American Society for Metals.
4. t. lyman and a. r. troiano: Trans. AIME , 1945, 162, 196.
8 Compression curve for microstructure formed at 190°C 5. r. lehouillier, g. begin, and a. dube: Metall. Trans. A, 1971,
for 2 weeks: YS yield stress 2A, 2645.

Materials Science and Technology March 2002 Vol. 18

284 Caballero et al. Very strong low temperature bainite

6. h. k. d. h. bhadeshia and d. v. edmonds: Met. Sci., 1983, 17 , 15. g. f. vander voort: `Metallography: principles and practice’ ,
411 ± 419. 427; 1984, New York, NY, McGraw ± Hill.
7. h. k. d. h. bhadeshia and d. v. edmonds: Met. Sci., 1983, 17 , 16. j. durnin and k. a. ridal: J. Iron Steel Inst., 1968, 206 , 60.
420 ± 425. 17. m. j. dickson: J. Appl. Cryst., 1969, 2, 176 ± 180.
8. v. t. t. miihkinen and d. v. edmonds: Mater. Sci. Technol., 18. d. j. dyson and b. holmes: J. Iron Steel Inst., 1970, 208, 469.
1987, 3, 422 ± 431. 19. Standard E9 ± 89a, ASTM, Philadelphia, PA, 1995.
9. v. t. t. miihkinen and d. v. edmonds: Mater. Sci. Technol., 20. h. k. d. h. bhadeshia: Acta Metall., 1981, 29, 1117 ± 1130.
1987, 3, 432 ± 440. 21. h. k. d. h. bhadeshia and a. r. waugh: Acta Metall., 1982, 30 ,
10. v. t. t. miihkinen and d. v. edmonds: Mater. Sci. Technol., 775 ± 784.
1987, 3, 441 ± 449. 22. l. c. chang and h. k. d. h. bhadeshia: Mater. Sci. Eng. A,
11. h. k. d. h. bhadeshia: `Bainite in steels: transformations, 1994, A184 , L17 ± 20.
microstructure, and properties’ , 2nd edn; 2001, London, The 23. i. stark, g. d. w. smith, and h. k. d. h. bhadeshia: in
Institute of Materials. `Solid® solid phase transformations’ , (ed. G. W. Lorimer),
12. h. k. d. h. bhadeshia: in `Modelling and simulation for 211 ± 215; 1988, London, The Institute of Metals.
materials design’ , (ed. S. Nishijima and H. Onodera), 227 ± 232; 24. h. k. d. h. bhadeshia and d. v. edmonds: Acta Metall., 1980,
1996, Japan, National Research Institute for Metals. 28, 1265 ± 1273.
13. f. g. caballero, h. k. d. h. bhadeshia, k. j. a. mawella, d. g. 25. h. k. d. h. bhadeshia: in `Materials algorithms project’ ,
jones, and p. brown: Mater. Sci. Technol., 2001, 17, 512 ± 516. www.msm.cam.ac.uk/map/steel/programs/mucg46-b.html
14. f. g. caballero, h. k. d. h. bhadeshia, k. j. a. mawella, d. g. 26. h. k. d. h. bhadeshia and a. r. waugh: Acta Metall., 1982, 30 ,
jones, and p. brown: Mater. Sci. Technol., 2001, 17, 517 ± 522. 775 ± 784.

Materials Science and Technology March 2002 Vol. 18