Determination of Martensite Start Temperature of High-Speed Steels based on Thermodynamic

Calculations

J. Platl *,1, H. Leitner 2, C. Turk 2, R. Schnitzer1

1
Montanuniversität Leoben, Department of Materials Science, Franz-Josef-Straße 18, 8700
Leoben, Austria

2
Accepted Article
voestalpine Böhler Edelstahl GmbH & Co KG, Mariazellerstraße 25, 8605 Kapfenberg, Austria

E-mail:
mail: jan.platl@unileoben.ac.at

Abstract

High-speed
speed steels (HSS) exhibit a microstructure that comprises of a martensitic matrix with

carbides. Due to the generally high thermal stability of carbides, it is necessary to perform

hardening at very high austenitizing temperatures. Nevertheless, there are certain carbides in HSS

that are not dissolved. Therefore, the composition of the austenite, which can be tr
transformed into

martensite, is significantly different compared to the alloys’ nominal composition. Consequently,

commonly
ommonly applied formulae for calculation of martensite start temperature (𝑀� ) cannot be used

for HSS.

The
he present study demonstrates how empirical 𝑀� formulae, which are basically applied for low

alloyed steels,
steels, where no carbides are present at austenitizing temperature, can be modified for

high-alloyed HSS by applying thermodynamic calculations. Thermo-Calc© software was utilized to

calculate the composition of the austenite at two different austenitizing temperatures and with

these composition
compositionss 𝑀� were calculated subsequently. For experimental verification, 𝑀� of four

alloys,, which were quenched from these austenitizing temperatures were determined using

dilatometry. The experimental results showed good agreement with the corresponding

This article has been accepted for publication and undergone full peer review but has not
been through the copyediting, typesetting, pagination and proofreading process, which may
lead to differences between this version and the Version of Record. Please cite this article as
doi: 10.1002/srin.202000063.

This article is protected by copyright. All rights reserved

thermodynamic equilibrium calculations. Furthermore, the results revealed that Co does not raise

𝑀� as predicted by the commonly applied empirical formulae. Hence, adapted formulae for HSS in

a wide composition range are proposed.

Keywords: Martensite start temperature, High-speed steel, Dilatometry, Thermodynamic

Accepted Article
calculation, Thermo
Thermo-Calc
Thermo-
-Calc
Calc

1. Introduction

The microstructure of high

high-speed steels (HSS) in fully heat-treated condition consists of a

tempered martensitic matrix with embedded primary carbides. The heat treatment to obtain this

microstructure includes austenitization with subsequent quenching in order to achieve a

supersaturated martensite
martensite. The austenitizing is performed at high temperatures to guarantee

maximum dissolution of primary carbides. Nevertheless, certain primary carbides remain

undissolved
solved due to their high thermal stability and ensure the extraordinar
extraordinarily high wear resistance

of HSS. Furthermore, grain coarsening during austenitization is prevented by these carbides.

During subsequent tempering fin

fine secondary hardening carbides are formed. These finely

dispersed nm
nm-sized
--sized
sized carbides provide the exceptionally high hardness of HSS at elevated

temperatures.. To obtain the aforementioned martensitic matrix, the final temperature during
temperatures

quenching must fall below the martensite start temperature 𝑀� , which depends on the chemical

composition of the investigated steels. Additionally, 𝑀� can be used to estimate the amount of

microstructure.[1,2]
retained austenite in the resulting microstructure

empirical formulae were published in the past.[3–10] These studies

For calculation of 𝑀� several empiric

mainly focused on determination of 𝑀� for low and unalloyed steels. Within these investigations

𝑀� of steels was found to be decreased by most alloying elements except for Co and Al. The

largest decrease of 𝑀� was caused by C according to all publications. However, for low alloyed

steels 𝑀� was calculated by inserting the nominal composition of each alloy. A summary of these
This article is protected by copyright. All rights reserved
formulae [3–10] can be found in the publications of Kung et al.[11] and Ishida.[12] The latter calculated

changes in 𝑀� for dilute iron base alloys due to element additions based on thermodynamic

evaluations of Gibbs free energies. They concluded that besides the aforementioned elements Co

and Al, 𝑀� is also raised by additions of Ti and V. The same conclusions were drawn by Izumiyama
Accepted Article
et al.[13], who reported that 𝑀� of binary iron based alloys are increased by the addition of Al, Ti

and V (in minor additions up to approximately 1 wt%) as well as Co (up to 40 wt%). Kung et al.[11]

examined the validity of several 𝑀� formulae by comparing calculated to dilatometrically

measured values
values in a large range of steel compositions. They concluded that two of the previously

steels, namely the linear equation from Andrews[8] and the

cited formulae for low and unalloyed steels

formula proposed by Haynes et al.[9], also provide reasonable results for alloyed steels.

Within this study special interest was drawn to the following five formulae, which contain the

main alloying elements of HSS, such as C, Cr, W, Mo, V and Co.

𝑀� = 545 − 330 ⋅ C + 2 ⋅ Al + 7 ⋅ Co − 14 ⋅ Cr − 13 ⋅ Cu − 23 ⋅ Mn − 5 ⋅ Mo (1)

− 4 ⋅ Nb − 13 ⋅ Ni − 7 ⋅ Si + 3 ⋅ Ti + 4 ⋅ V

𝑀� = 539 − 423 ⋅ C − 30.4 ⋅ Mn − 12.1 ⋅ Cr − 17.7 ⋅ Ni − 7.5 ⋅ Mo + 10 ⋅ Co − 7.5 ⋅ Si (2)

𝑀� = 561 − 474 ⋅ C − 33 ⋅ Mn − 17 ⋅ Cr − 17 ⋅ Ni − 21 ⋅ Mo + 10 ⋅ Co − 7.5 ⋅ Si (3)

𝑀�� = 550 − 350 ⋅ C − 40 ⋅ Mn − 35 ⋅ V − 20 ⋅ Cr − 17 ⋅ Ni − 10 ⋅ Cu − 10 ⋅ Mo (4)

− 8 ⋅ W + 15 ⋅ Co + 30 ⋅ Al
� (5)
𝑀�� = 0
0.495
0..495
495 ∙ 𝑀���� + 0.00095 ∙ 𝑀�� + 40

The contents of the alloying elements for all formulae are given in wt% and the resulting 𝑀� in °C.

An adapted formula proposed in the aforementioned publication of Ishida[12] was specifically

designed for steels with minor alloying content and is given by equation (1). Andrews` formula[8] is

represented by equation (2) and the formula proposed by Haynes`[9] by equation (3). These three

formulae do not consider the carbide forming element W. Equation (1) predicts an increasing

effect of V to 𝑀� , as described above, and in the other two equations it is not considered at all.

This article is protected by copyright. All rights reserved

Additionally, equation (1) and (2) show lower factors for C, Cr and Mo compared to equation (3).

Another formula of this type, which additionally takes W and V into account, is represented by

Equation (4). It was published by Jaffe et al.[14] A further improved equation for 𝑀�� is given by

𝑀�� . This corrected formula for 𝑀�� is expressed by equation (5) and was proposed in [14][15]. The
Accepted Article
improvement with the quadratic approach was achieved after evaluating data summarized in

time-temperature-transition
temperature--transition
temperature transition diagrams of low alloyed steel. These collections of diagrams were

al.[16] and by Rose.[17]

published by Wever et al.

Although
ough many papers have been published in this field, tthere is a lack of a proper empirical 𝑀�

formula with experimental proof for high alloyed steels, such as HSS. Therefore, the aim of the

present study is the evaluation and subsequent adaption of such a formula for the investigated

composition range.
range. This was done by comparing calculated 𝑀� using different formulae with

dilatometrically measured values for four different HSS with varying alloying element contents. In

order to
o account for the difference between nominal alloy composition and actual austenite

composition caused by the formation of certain thermally stable carbides in HSS, 𝑀� calculations

were carried out with thermodynamically derived austenite compositions. To calculate these

compositionss at different austenitizing temperatures, Thermo-Calc© software was used.

phase composition

Two different Thermo

Thermo-Calc
Thermo-
-Calc
Calc databases for steels and iron base alloys, namely an older version

TCFE2 and a more recent version TCFE7, were compared with respect to their correctness in

predicting actual austenite compositions

compositions. The latter were derived at two different austenitizing

temperatures (1150 °C and 1200 °C) and thus different fractions of dissolved carbides and

subsequently the calculated 𝑀� were contrasted with dilatometrically measured values.

2. Experimental

Table 1 contains the nominal chemical compositions of the four investigated steels which were

produced via powder metallurgy (PM) by voestalpine Böhler Edelstahl GmbH & Co KG,
This article is protected by copyright. All rights reserved
Kapfenberg, Austria. These steels were chosen since they represent commonly used grades for

tooling applications in a wide range of chemical compositions (especially variations in C and Co).

Additionally, the PM processing route guarantees finely distributed carbides resulting in a more

homogenous material response in terms of austenite grain growth during austenitization in

Accepted Article
comparison to conventionally cast HSS, which contain coarse nonuniformly distributed eutectic

carbides.[1]

Table 1 Nominal chemical composition of the investigated HSS.

[wt%]
[wt %] C Si Mn Cr Mo Ni V W Cu Co

Alloy
A lloy 1 0.88 0.57 0.36 4.25 2.72 0.28 2.03 2.46 0.12 4.36

Alloy
A lloy 2 1.29 0.60 0.29 4.00 4.91 0.21 3.03 6.14 0.11 8.12

Alloy
A lloy 3 1.36 0.59 0.28 3.89 4.88 0.18 3.81 5.37 0.13 0.38

Alloy
A lloy 4 1.30 0.56 0.31 4.07 4.87 0.22 3.00 6.10 0.10 0.38

For dilatometer measurements samples were machined to dimensions of Ø5x10 mm³. All

experiments were performed on a TA Instruments® DIL 805A quenching dilatometer.

Temperature measurement was conducted using a type B thermocouple and elongation was

measured with a quartz (SiO2) connecting rod. In order to avoid reactions between sample and

thermocouple, thin Ta foils of approximately 3x3x0.127 mm3 were spot welded (Spot Welding

Fixture 9061, TA Instruments®) on the samples as diffusion barriers.

Samples
amples were heated to austenitizing temperature within two minutes. For both austenitization

conditions, which are listed in the following enumeration, three samples were tested.

1) T1 = 1150 °C, t1 = 10 min

2) T2 = 1200 °C, t2 = 3 min

This article is protected by copyright. All rights reserved

Rapid quenching was conducted with He gas. The cooling time from 800 °C to 500 °C was set to

t8/5 = 5 s (lambda parameter λ = t8/5/100 = 0.05 s). These conditions are commonly used for the

investigated high-alloyed steel grades.

Figure 1 (a) shows exemplary dilatometer measurement curves, where the elongation of the
Accepted Article
sample is plotted as a function of the temperature. Besides thermal expansion or contraction

during heating or cooling, two phase transformations can be observed. In addition to the

conversion from
from ferrite into austenite during heating at approximately 875 °C the martensitic

transformation is shown between 100 °C and 150 °C. Figure 1 (b) represents a closer view on this

transformation. The
The onset of this transformation is the 𝑀� , which is determined according to

SEP1681[18] as the temperature at which a volume expansion, expressed as an inc

increase in

elongation, occurs. It can be deduced that the determined 𝑀� of all three measured curves are in

good accordance with each other.

alloy 1

Fig. 1 Exemplary dilatometry evaluation of alloy 1 quenched from 1200 °C. Measurement of
elongation
longation above the full temperature range in (a). The detailed view in (b) shows good
accordance concerning 𝑀� between the three evaluated samples.

Thermo-Calc with the TCFE2 and TCFE7 Steels/Fe-alloys databases were used for the

thermodynamic calculations of the phase fractions and austenite compositions at 1150 °C and

1200 °C. For these calculations typical equilibrium phases in steels (melt, ferrite and austenite) and

This article is protected by copyright. All rights reserved

all available types of carbides in the databases (graphite, cementite, MC, M2C, M3C, M3C2, M6C,

M5C2, M7C3, M12C and M23C6) were taken into account. The final 𝑀� were calculated by inserting

the computed austenite compositions into equations (1) to (5).

3. Results
Accepted Article
For low alloyed steels, in which no carbides are present during austenitization, 𝑀� can be

calculated with the previously described formulae by inserting the alloys´ nominal compositions.

amounts of carbides in HSS[1,2] and thus significant differences between nominal and
Due to high amounts

actual austenite composition

compositions this approach does not lead to any agreement. Calculations of 𝑀�

for the alloys 2

2,, 3 and 4 using the nominal compositions, which are listed in table 1, resulted in

calculated values beneath room temperature (RT). This would yield a microstructure that does not

contain any martensite after quenching to RT. Dilatometrically determined 𝑀� of these alloys

however led to values above 100 °C. Therefore, the actual composition of the austenite which can

be transformed to martensite has to be inserted in these equations in order to determine proper

calculation results.
results. Assuming a martensitic transformation, which takes place without diffusional

processes and without a change in chemical concentration if quenching is sufficiently rapid [19], the

chemical composition could be determined by means of wavelength dispersive spectroscopy in an

electron microprobe or in a scanning electron microscope

microscope. Since the interaction volume of the

electron beam into the sample surface due to the necessarily high acceleration voltages in order

to determine heavy elements (e.g. Mo or W) with sufficient accuracy is larger than average

carbide interspacing[20], it is not possible to measure the accurate austenite composition. Another

possibility would be to conduct atom probe tomography measurements of rapidly quenched

specimens. However, such measurements are time consuming and the analyzed volume is very

small. A third opportunity, which is also viable for the high-temperature austenitizing regime, was

This article is protected by copyright. All rights reserved

chosen for this study. The actual austenite composition was derived by thermodynamic

calculations using different Thermo-Calc software databases.

Figure 2 exhibits a comparison between measured M� and calculated values with equation (1) in

(a) and (b) and with equation (2) in (c) and (d). Figure 2 (a) and (c) represent the results at an
Accepted Article
austenitizing
ng temperature of 1150 °C and (b) and (d) at 1200 °C. It can be deduced, that the

calculated M� with both databases are significantly higher than the measured values for all

investigated steels at both austenitizing temperatures

temperatures. Additionally, the element Co has been

neglected in the calculations

calculations,
cal culations see dashed bars. However, the obtained results still exceed the

measured values for all four investigated steels and for both databases and austenitizing

temperatures..
temperatures

Fig. 2 Comparison between measured and calculated 𝑀� with (a, b) equation (1) and (c, d)
equation (2) at austenitizing temperatures of (a, c) 1150 °C and (b, d) 1200 °C. For all investigated

This article is protected by copyright. All rights reserved

steels, the calculated 𝑀� with both equations, databases and austenitizing temperatures are
significantly higher than the measured values.

The measured 𝑀� in comparison to values calculated by equation (3) and (5) with

thermodynamically derived austenite compositions are shown in figures 3 and 4 for all four
Accepted Article
investigated steels. The
The results of the 𝑀� calculations with austenite compositions derived at

1150 °C are presented in figure 3 (a) and figure 4 (a). Figure 3 (b) and figure 4 (b) show calculated

values with derived austenite compositions at 1200 °C. Figure 3 exhibits the calculated 𝑀� using

equation (3) and in figure 4 the results of the 𝑀� calculations with equations (4) and (5) are

shown. For both figures,

figures, measured 𝑀� are indicated by the white columns with the error bars. The

grey backed columns exhibit the calculated 𝑀� using austenite compositions derived from TCFE2

database and the white dashed columns show the results for the 𝑀� calculations using the

compositions obtained by the more recent TCFE7 database. For the Co free alloys 3 and 4, good

agreement between measured and calculated values can be observed in both figures, whereas the

calculated values for the Co

Co-containing alloys 1 and 2 do not fit to the measured results.

Fig. 3 Comparison between measured and calculated 𝑀� with equation (3) at austenitizing
temperatures of (a) 1150 °C and (b) 1200 °C. Alloy 1 and 2 show no correlation between
measured and calculated values, irrespective of the used database.

This article is protected by copyright. All rights reserved

 Accepted Article
Fig. 4 Comparison between measured and calculated 𝑀� with equation (5) at austenitizing
temperatures of (a) 1150 °C and (b) 1200 °C. Alloy 1 and 2 show no correlation between
measured and calculated values, irrespective of the used database
database.

Since for the Co-

Co-containing
Co-containing
containing HSS (alloy 1 and 2) no correlation between measured and calculated

values was found,

found, the calculations of M� were repeated but the influence of Co was neglected.

The obtained calculation results with equation (3) and (5) in comparison to the experimentally

determined M� are depicted in figures 5 and 6, respectively. For all investigated alloys,

calculations in which Co was not taken into account lead to values that are almost consistent with

the conducted measurements. Con

Concerning the different databases, except for figure 5 (b), better

correlation can be achieved with TCFE7 database than with TCFE2. For this reason, the following

considerations are for the results obtained from the calculations with TCFE7-without Co.

This article is protected by copyright. All rights reserved

 Accepted Article
Fig. 5 Comparison between measured and without Co calculated 𝑀� with equation (3) at
austenitizing temperatures of (a) 1150 °C and (b) 1200 °C. In (a) the calculated 𝑀� (TCFE7 –
without Co) are in good agreement with the experim
experiments for the austenitization at 1150 °C.

Fig. 6 Comparison between measured and without Co calculated 𝑀� with equation (5) at
austenitizing temperatures of (a) 1150 °C and (b) 1200 °C. Best agreement between
measured and calculated values (TCFE7 – without Co) can be seen in (b) for the
austenitizing temperature of 1200 °C.

Figure 5 (a) shows that for austenitization at 1150 °C best agreement, which is within the

experimental tolerance, is achieved by applying equation (3). Measured 𝑀� for an austenitization

at 1200 °C, as shown in figure 5 (b), are significantly higher than the values calculated with TCFE7.

An examination of figure 6 (b) depicts contradictory results. Using formulae (4) and (5) reveals

calculated values for the higher austenitizing temperature of 1200 °C, which are within the

measurement tolerance of the dilatometer experiments. Except for alloy 4, the calculated values
This article is protected by copyright. All rights reserved
for the austenitization at 1150 °C, which are shown in figure 6 (a), also exhibit good agreement

with the measurements.

As expected, it can be observed in figures 5 and 6 that all calculated as well as the measured 𝑀�

are lower at an austenitizing temperature of 1200 °C compared to 1150 °C.

Accepted Article
All calculated austenite compositions and the amounts of stable phases according to both

Thermo-Calc
Calc databases are listed in table 2. It can be deduced that higher alloyed steels, which are

represented by the alloys 2, 3 and 4, contain large amounts of M6C carbides, whereas alloy 1 only

contains a little amount of this carbide type according to TCFE7 database. The amount of MC

carbides in the higher alloyed steels is also significantly higher than in alloy 1. Furthermore, higher

contents of C and V in the aaustenite can be deduced according to calculations with TCFE7 than

with TCFE2. The weight fractions of C as well as the carbide forming elements V, Mo and W are

heavily reduced in comparison to the nominal alloy composition shown in table 1. By comparing

thee amount of Cr to its nominal value, except for alloy 1, a slight increase can be observed. The

amount of Co in the austenite

austenite, especially for alloy 2 with the highest nominal Co content,

increases. The content of the other elements in the austenite is also increased in comparison to

the values in table 1

1..

Table 2 Calculated austenite compositions and amount of stable phases according to

different Thermo
Thermo-Calc
Thermo--Calc
Calc databases.
Chemical composition [wt%] Stable phases [%]
T [°C] Database
C Si Mn Cr Mo Ni V W Cu Co γ MC M6C
TCFE2 0.506 0.570 0.360 4.096 2.222 0.280 0.608 2.205 0.120 4.359 97.26 2.74 0.00
1150 °C
TCFE7 0.557 0.586 0.370 4.246 2.338 0.288 0.776 1.962 0.124 4.484 97.07 2.55 0.38
Alloy 1

TCFE2 0.571 0.570 0.360 4.113 2.357 0.280 0.833 2.266 0.120 4.359 97.76 2.24 0.00
1200 °C
TCFE7 0.634 0.582 0.367 4.246 2.505 0.286 1.039 2.179 0.122 4.448 98.01 1.99 0.00
TCFE2 0.517 0.687 0.332 4.015 2.522 0.240 0.444 2.676 0.126 9.287 87.27 5.12 7.61
1150 °C
TCFE7 0.553 0.669 0.338 4.095 2.181 0.245 0.665 2.012 0.129 9.118 85.46 4.63 9.91
Alloy 2

TCFE2 0.582 0.677 0.326 4.001 2.891 0.236 0.602 3.180 0.124 9.142 88.68 4.86 6.46
1200 °C
TCFE7 0.630 0.663 0.333 4.098 2.502 0.241 0.878 2.403 0.127 9.024 86.75 4.26 8.99
1150 °C TCFE2 0.508 0.679 0.322 3.916 2.377 0.207 0.751 2.234 0.150 0.437 86.88 5.51 7.61
Allo
y3

This article is protected by copyright. All rights reserved

 TCFE7 0.540 0.652 0.323 3.987 2.343 0.208 0.846 1.992 0.150 0.426 86.49 5.43 8.08
TCFE2 0.586 0.669 0.317 3.900 2.736 0.204 0.987 2.652 0.147 0.430 88.22 5.16 6.62
1200 °C
TCFE7 0.623 0.645 0.318 3.984 2.683 0.204 1.108 2.384 0.148 0.421 87.85 5.03 7.12
TCFE2 0.602 0.648 0.358 4.113 2.252 0.254 0.665 2.286 0.116 0.439 86.39 4.23 9.38
1150 °C
TCFE7 0.629 0.619 0.359 4.168 2.195 0.255 0.761 2.108 0.116 0.425 86.09 4.19 9.72
Alloy 4

TCFE2 0.679 0.638 0.353 4.105 2.602 0.250 0.890 2.723 0.114 0.432 87.73 3.85 8.42
1200 °C
TCFE7
Accepted Article0.713 0.612 0.354 4.171 2.523 0.251 1.012 2.527 0.114 0.420 87.43 3.75 8.82

4. Discussion

The aim of the present study was to find an empiric

empirical formula, that properly predicts 𝑀� of HSS in

the investigated composition range, see table 1. Besides several 𝑀� calculation formulae, which

were basically developed for low alloyed steels and therefore did not lead to sufficient results, five

formulae,, given by equations (1) to (5), were considered in detail. These five formulae contain

most of the main alloying elements of HSS. Calculations of 𝑀� with nominal compositions yielded

no agreement with measured values for all equations. The deviation between the nominal and the

austenite composition arises from the significant amount of carbides that are not dissolved in this

type of steels due to thei

their high thermal stability. Therefore, calculations were carried out using

compositions of the austenit

austenite according to thermodynamic data at certain austenitizing

temperatures. Application of equation (1) and (2) did not lead to any agreement with this

approach either,
either, see figure 2. For both equations this can be attributed to the factors of certain

elements (C, Cr and Mo), which are apparently too low for proper calculation of 𝑀� for HSS and

thus the calculations led to much higher values than those which were determined by dilatometer

measurements. An additional explanation for the significant deviations from the measured values

could be that equations (1) and (2) do not consider W. Regarding the element V, equation (2) does

not predict any influence and in equation (1) it is assumed to increase 𝑀� .

For the Co free HSS (alloy 3 and 4) at an austenitizing temperature of 1150 °C good accordance

(maximum deviation of 19 °C, see figure 3 (a)) between measured and with TCFE7 calculated

values was achieved with equation (3). As shown in figure 4 (b), applying equation (5) led to good
This article is protected by copyright. All rights reserved
agreement for the higher austenitizing temperature of 1200 °C (±5 °C deviation). For both, the

deviations between measured and calculated values are within the measurement tolerance of the

dilatometer experiments. For the lower austenitizing temperature equation (3) led to better

agreement because of the lower amount of carbides dissolved, which results in an austenite
Accepted Article
containing lower amounts of carbon and alloying elements, than at 1200 °C. Therefore, the

austenite composition is shifted towards lower alloyed steels, for which the formula was originally

used. Additionally, the formula does not include the elements V and W. However, this has less

influence because a high fraction of these elements are bonded into thermally stable carbides (VC

and high W-containing

containing M6C). This can be seen in the thermodynamic calculations of stable phases

and austenite compositions in table 2 and was also found by Boccalini et al.[21]

The better agreement with the quadratic approach for calculati

calculation of 𝑀� at higher austenitizing

temperatures can be related to the considered inf

influence of V and W in equation (4). At 1200 °C

more thermally stable carbides are dissolved leading to higher concentrations of carbide forming

elements in the austenite

austenite. This results in austenite compositions that are closer to the nominal

compositions of high-
high-alloyed
high -alloyed
alloyed steels than at 1150 °C. Therefore, the influence of especially V is

more pronounced because of its high factor (−35 ⋅ V) in equation (4).

No agreement between measured and calculated values was found ffor alloy 1 and 2, representing

HSS with high Co content

content. Figures 3 and 4 showed calculated 𝑀� that are significantly higher than

the experimental values. Therefore, the calculations were also performed without Co, as can be

seen in figures 5 and 6

6.. This yielded the desired agreement with measured values but contradicts

empirical formulae proposed in the past[3–10], in which is claimed that Co

findings summarized by empiric

increases 𝑀� for steels that form carbon martensite. In two publications published by Goldman et

al.[20, 21] martensitic transformations in high Ni-containing steels were investigated. These steels

form a soft nickel martensite representing a different type of martensite compared to the alloys

investigated in the present study, which form hard carbon martensite. Heat evolutions during
This article is protected by copyright. All rights reserved
these soft martensitic transformations were measured using differential thermal analysis.[22] These

measurements led to the conclusion that a ternary Co-containing alloy (Fe-31.3Ni-5.7Co) shows

lower heat evolution during martensite formation than the other investigated Co free alloys (Fe-

30Ni, Fe-25.1Ni-0.26C and Fe-11.7Ni-15.1Cr). Additionally, shear moduli of austenite and

Accepted Article alloys.[23] The shear
martensite were measured with resonance frequency techniques for the same alloys

module of the Fe-

Fe-30Ni
Fe -30Ni
30Ni austenite was determined to be 25% higher than that of the Co
Co-containing

austenite (Fe-
(Fe-31.3Ni-5.7Co).
(Fe-31.3Ni
31.3Ni--5.7
5.7Co).
Co). Therefore, it was stated that due to the lower modulus of the Co
Co-

containing austenite and the lower evolution of heat during martensitic transformation less

undercooling is necessary to guarantee the martensitic transformation and thus 𝑀� is increased by

Co..[22,23] These considerations apparently do not apply for the steels investigated
the addition of Co

within this study.

study. This can be mostly attributed to the different type of martensite in the

investigated high Ni-

Ni-containing
Ni -containing
containing steels compared to the carbon martensite, which is formed in HSS.

In order to analyze the influence of Co in HSS in detail, additional experiments, such as the

determination of heat evolution during martensitic transformations using differential scanning

calorimetry, would be necessary.

𝑀� calculations with austenite compositions obtained from TCFE7 Steels/Fe-alloys database

yielded better agreement with measured values than those derived from TCFE2. Table 2 shows the

higher contents of C and V according to TCFE7 in comparison to the older TCFE2 database yielding

lower 𝑀� due to the aforementioned high impact of V and the even higher impact of C in the used

equations and thus

thus better agreement with the experi
experiments. Besides C, strong carbide forming

elements [2], such as V, Mo and W

W, are incorporated in the formed carbides and therefore their

amount in the austenite is significantly reduced compared to the nominal alloy composition. The

formation of Cr-containing carbides is suppressed because of the aforementioned elements and

this leads to the increased Cr amounts in the austenite compared to the nominal composition of

the alloys 2, 3 and 4. The increased Co content of the austenite compared to the nominal
This article is protected by copyright. All rights reserved
compositions of the investigated alloys can be explained by the fact that Co does not form any

carbides and is therefore incorporated into the austenitic matrix. The same considerations can be

made for the remaining elements (Si, Mn, Ni and Cu). The more accurate calculation of austenite

composition and hence the better agreement with measured values of TCFE7 compared to TCFE2

database
Accepted Article
base can be assigned to several improvements concerning formation of stable carbide phases

implemented in the database updates[24]. These adaptions can be seen in the differing amounts of

carbides and austenite compositions according to the used databases, see table 2.

Regarding a possible influence of different austenite grain sizes aat austenitizing temperatures on

transition diagrams published by Wever et al.[16] showed, that for a typical

𝑀� , time-temperature-transition
temperature--transition
temperature

quenched and tempered steel the 𝑀� is not significantly influenced by different austenitizing

temperatures. They pointed out that an increase in austenitizing temperature from 850 °C to

1050 °C yielded a 𝑀� of approximately 360 °C for both conditions of the investigated steel

(42CrMo4). Since the latter does not form primary carbides during austenitization, nonuniform

grain growth may occur.

occur. Due to the homogenous primary carbide distribution after PM processing

yielding uniformly fine austenite grains[1] for the investigated steels, the influence of austenite

grain size on 𝑀� is not of major importance for tthe obtained results.

The findings of this study can be expressed by proposing new adapted formulae for calculation of

𝑀� for HSS. Accurate prediction of 𝑀� for HSS is enabled by applying the following formulae which

do no longer include the impact of th

the element Co. The alloying element contents of the austenite

according to Thermo
Thermo-Calc
Thermo-
-Calc
Calc Software TCFE7 Steels/Fe-alloys database at a certain austenitizing

temperature have to be inserted in wt%.

∗
𝑀�� = 550 − 350 ⋅ C − 40 ⋅ Mn − 35 ⋅ V − 20 ⋅ Cr − 17 ⋅ Ni − 10 ⋅ Cu (6)

− 10 ⋅ Mo − 8 ⋅ W

∗ ∗� (7)
𝑀�� = 0.495 ∙ 𝑀�� + 0.00095 ∙ 𝑀�� + 40

This article is protected by copyright. All rights reserved

 𝑀�∗ = 561 − 474 ⋅ C − 33 ⋅ Mn − 17 ⋅ Cr − 17 ⋅ Ni − 21 ⋅ Mo − 7.5 ⋅ Si (8)

A modification of equation (4) is given by equation (6). In addition to Co, also Al is excluded.

Equation (7) can be used to calculate 𝑀� of HSS in °C at very high austenitizing temperatures at

which many carbides are dissolved. For HSS, that are austenitized at lower temperatures, at which
Accepted Article
the austenite contains lower amounts of alloying elements and thus its composition is closer to

those of lower
lower alloyed steels, equation ((7) can also be applied. However, for lower austenitizing

temperatures, equation (8) leads to better results. This equation represents a modification of

equation (3), in which the influence of Co is neglected.

5. Conclusion
Conclusionss

The aim of the present study was to find a formula that properly predicts 𝑀� of HSS in the

investigated composition range

range. Already existing empirical formulae, that are commonly applied

for low alloyed steels, were used to calculate 𝑀� for HSS. For this purpose, austenite compositions

were derived by thermodynamic calculations using different Thermo

Thermo-Calc databases. These

calculations were performed at two different austenitizing temperatures (1150 °C and 1200 °C) for

four HSS alloys with differing chemical composi

compositions and then compared to dilatometrically

measured values. Following conclusions can be drawn:

 Calculation of 𝑀� with the nominal compositions of the investigated HSS did not lead to

any agreement with experimental results. This can be attributed to th

the difference between

nominal and actual austenite composition caused by the formation of thermally stable

carbides.

This article is protected by copyright. All rights reserved

  To account for the difference between nominal and actual austenite composition, 𝑀� were

derived using thermodynamically calculated austenite compositions. This led to good

agreement for the Co free alloys 3 and 4.

 Two databases provided by Thermo-Calc Software were compared for predicting the actual
Accepted Article
austenite compositions for subsequent calculation of 𝑀� with the suggested formulae. The

more recent version TCFE7 resulted in better agreement with experimental results

compared to an older version (TCFE2).

 Since no agreement for tthe Co-containing alloys 1 and 2 was found, calculations of 𝑀�

were also conducted without Co. TThis approach yielded the desired agreement between

measured and calculated values for all investigated alloys.

Based on the findings of this study new adapted formulae for HSS were proposed depending on

the amount of dissolved carbides during austenitization. These formulae are given by

equations (6) - (8).

Acknowledgements

Funding of the Austrian BMK (846933) in the framework of the program “Production of the

future” and the “BMK Professorship for Industry” is gratefully acknowledged.

References

[1] G. A. Roberts, R. Kennedy, G. Krauss, Tool steels, Vol. 5, ASM International, Materials Park, OH,
USA 1998
1998..
[2] H. Berns, W. Theisen, Eisenwerkstoffe - Stahl und Gusseisen, Vol. 4, Springer, Berlin,
Heidelberg, Germany 2008.
[3] P. Payson, Trans. Am. Soc. Met. 1944, 33, 261–280.
[4] R. A. Grange, Trans. Am. Inst. Min. Metall. Eng. 1946, 167, 467–490.
[5] A. E. Nehrenberg, Trans. AIME 167 1946, 494–501.
[6] E. S. Rowland, S. R. Lyle, Trans. Am. Soc. Met. 1946, 37, 27–47.

This article is protected by copyright. All rights reserved

[7] J. H. Hollomon, L. D. Jaffe, Ferrous metallurgical design: design principles for fully hardened
steel, Wiley, USA 1947.
[8] K. W. Andrews, J. Iron Steel Inst. 1965, 203, 721–727.
[9] A. G. Haynes, W. Steven, J. Iron Steel Inst. 1956, 183, 349–359.
[10]L. A. Carapella, Met. Prog. 1944, 46, 108.
[11]C. Y. Kung, J. J. Rayment, Metall. Trans. A. 1981, 13, 328–331.
Accepted Article
[12]K. Ishida, J. Alloys Compd. 1995, 220 (1-2), 126–131.
[13]M. Izumiyama, M. Tsuchiya, Y. Imai, Sci. Rep. Res. Inst. Tohoku Univ., Ser. A 1970, 22, 105–115.
[14]L. D. Jaffe, J. H. Hollomon, Trans. AIME 167 1946, 617–626.
[15]Verein deutscher Eisenhüttenleute, Werkstoffkunde Stahl: Band 1: Grundlagen, Vol.1 (Eds: W.
Jäniche, W. Dahl, H.-
H.-F.
H. -F.
F. Klärner, W. Pitsch, D. Schauwinhold, W. Schlüter, H. Schmitz), Springe
Springer,
Berlin, Heidelberg, Germany 1984.
[16]F. Wever, A. Rose, W. Peter, W. Strassburg, Atlas zur Wärmebehandlung der Stähle, Verlag
Stahleisen, Germany 1954.
[17]A. Rose, Atlas zur Wärmebehandlung der Stähle
Stähle, Verlag Stahleisen, Germany 1972.
en Prüfblatt (SEP) 1681, Guidelines for preparation, execution and evaluation of
[18]Stahl-Eisen
dilatometric transformation tests on iron alloys
alloys, Verein deutscher Eisenhüttenleute, 2nd
edition, Düsseldorf, Germany 1998.
[19]a) G. Gottstein, Physikalische Grundlagen der Materialkunde, Springer-Verlag, Berlin,
Heidelberg, Germany 2007; b) E. Hornbogen, H. Warlimont, B. Skrotzki, Metalle: Struktur und
Eigenschaften der Metalle und Legierungen
Legierungen, Springer, Berlin, Heidelberg, Germany 2019.
[20]J. I. Goldstein, D. E. Newbury, J. R. Michael, N. W. Ritchie, J. H. Scott, D. C. Joy, Scanning
electron microscopy and X X-ray microanalysis, Springer, New York, USA 2017.
[21]M. Boccalini, H. Goldenstein, Int. Mater. Rev 2001, 46(2), 92–115.
dman, W. D. Robertson, Acta Metall. 1965, 13(4), 391–401.
[22]A. J. Goldman,
[23]A. J. Goldman, W. D. Robertson, Acta Metall. 1964, 12(11), 1265–1275.
[24]Thermo-Calc
Calc Software Homepage, www.thermocalc.com/products
www.thermocalc.com/products-
services/databases/thermo -dynamic/tcfe-history/, accessed: 2019.

Martensite start temperatures for high-speed steels with varying alloy compositions are predicted by
applying commonly used calculation formulae for low alloyed steels. The actual austenite compositions
considering carbide formation during austenitizing are thermodynamically derived and compared to
experimental data obtained from dilatometer measurements. Adapted formulae for different austenitizing
temperatures are proposed.

This article is protected by copyright. All rights reserved

Accepted Article

This article is protected by copyright. All rights reserved