J Mater Sci (2023) 58:1905–1924

Metals &Ccorrosion
METALS & ORROSION

Effects of tempering temperature on temperature-

dependent thermal properties of 1045 steel
Nı́colas Pinheiro Ramos1 , Mariana de Melo Antunes1 , Antonio Augusto Araújo Pinto da Silva1 , and
Sandro Metrevelle Marcondes de Lima e Silva1,*

1
Institute of Mechanical Engineering, Federal University of Itajubá, Itajubá, MG, Brazil

Received: 28 October 2022 ABSTRACT

Accepted: 25 December 2022 The 1045 steel is extensively heat treated because its mechanical properties can
Published online: be easily improved by thermal processes. Reliable knowledge of the relationship
9 January 2023 between microscopic structure and thermal properties is essential for the ther-
mal design of parts and tools. This paper reports on coupling microstructural
 The Author(s), under aspects with the effects of tempering temperature on the temperature-depen-
exclusive licence to Springer dent thermal properties of 1045 steel. Thermal conductivity (k) and specific heat
Science+Business Media, LLC, (cp) were simultaneously estimated using inverse heat conduction analysis. The
part of Springer Nature 2023 thermal characterization technique handled a simple and low-cost experimental
setup, based on transient one-dimensional heat transfer on samples subjected to
different heat-treatment conditions. The results, which were discussed in con-
nection with the resulting microstructures, demonstrated that all quenching and
tempering (Q&T) conditions led to lower values of both thermal properties
throughout the temperature domain. The tempering temperature was found to
impact k more substantially. The robustness of the results was ascertained by
recovering the applied heat flux based on linear regression equations for k and cp
and temperature measurements. Accuracy analysis indicated that the inverse
approach provides estimates with an uncertainty close to 4%.

Handling Editor: Catalin Croitoru.

Address correspondence to E-mail: metrevel@unifei.edu.br

https://doi.org/10.1007/s10853-022-08137-0

Content courtesy of Springer Nature, terms of use apply. Rights reserved.

1906
GRAPHICAL ABSTRACT
Quenching
30 minutes
Temperature (°C)
Tempering
60 minutes
Time (min)
Introduction
A fundamental aspect of materials science is the
characterization of engineering materials to obtain
theoretical and practical information on their struc-
ture and properties. The materials characterization
framework has many broad facets since it combines
knowledge from multidisciplinary backgrounds,
such as chemistry, metallurgy, physics, and even
biology and environment [1, 2]. Materials character-
ization techniques have evolved over time, from
primitive tools to highly advanced devices. Materials
testing can cover a wide range of properties (electri-
cal, mechanical, thermal, etc.), can evaluate all classes
of materials (metals, ceramics, polymers, composites,
etc.), can be applied to scales from atomic to macro-
scopic, and can handle ultra-short experiments or
tests lasting for years [3, 4]. Materials analysis has
contexts that vary from fundamental science, which
often lacks a prompt and direct understanding, to
real-world applications, where the behavior of an
engineering component is predicted to improve per-
formance and prevent failure [5]. Developing novel
approaches to explain the behavior of materials
enables us to expand and improve the scope of the
materials characterization universe, which is crucial
for scientific advancement.
The capacity of a material to transfer and store
thermal energy is governed by its thermal properties.
These properties play a crucial role in mechanical
Content courtesy of Springer Nature, terms of use apply. Rights reserved.
J Mater Sci (2023) 58:1905–1924
Microstructural Characterization
900°C
550°C
400°C
Microstructure
250°C
Thermal Analysis
Temperature-Dependent
Thermal Properties
design, as they enable the optimization of several
processes by predicting the thermal response of
engineering devices under working conditions [6].
Standardly, there are three different approaches for
investigating the thermal properties of materials. The
first one consists of steady-state measurements of
heat flux and temperature gradients, as in the guar-
ded hot plate method [6]. Despite the high accuracy,
this group of techniques is very time-consuming and
can hardly ever be applied to high-conducting
materials, such as metals. Also, they require special-
ized equipment and can furnish only thermal con-
ductivity. The calorimetric approach, on the other
hand, is very well suitable when the goal is to obtain
specific heat. Differential scanning calorimetry (DSC),
for example, has been extensively applied for this
purpose. Nevertheless, these methods do not furnish
thermal conductivity and manage to provide thermal
properties only for high-temperature levels. More-
over, they also require specific and expensive
equipment, and difficulties arise when dealing with
inhomogeneous materials, due to the mandatory
small sample size [7]. Lastly, the third approach
introduces transient techniques, such as the transient
hot wire method [8] and the laser flash method [9].
For both techniques, thermal conductivity and ther-
mal diffusivity can be obtained within a much
smaller period than that needed in steady-state
experiments. Anyhow, the experimental assemblies
also demand considerable financial and time resour-
ces for obtaining each thermal property. In recent
J Mater Sci (2023) 58:1905–1924
years, all these classical techniques have already been
improved, yet, none of them is unrestricted, which
means that methods for obtaining thermal properties
are usually defined for specific materials and tem-
perature ranges [6]. Experimental methods for esti-
mating the thermal properties of high-conducting
materials, for example, are scarce as a consequence of
the lower thermal sensitivity and higher contact
resistance presented by these materials [10]. Addi-
tionally, there is a lack of methods capable of
including the influence of temperature changes on
thermal properties. Accounting for temperature
dependence is essential when dealing with consid-
erable variations in the working temperature, since
regarding k and cp as constants can lead to unrealistic
simulation results [11].
The 1045 steel is employed in a large range of
engineering applications and is frequently quenched
and tempered to improve its mechanical properties.
This material can be used at high levels of pressure
and friction and combines strength, hardness, and
ductility. The properties of steel, including thermal
properties, are strongly influenced by microstructural
aspects [12–16]. Microstructure, in turn, depends on
the material’s thermal history and can therefore be
modified by heat treatments [12, 13]. Steels can
hardly ever be applied right after quenching, due to
the high brittleness provided by the process. Tem-
pering seeks to relieve internal stresses, reducing
some excessive hardness and increasing ductility.
Quenching and tempering are combined to enable
the heat-treated material to obtain a high degree of
toughness and tensile strength [12]. Q&T heat treat-
ment provides a wide range of properties without
changing composition [15]. Thus, the study and
knowledge of the microscopic structure of materials
are crucial in materials research and engineering.
Much has been discussed about the effects of heat
treatments on the microstructure of steels [12–16] and
also about their influence on mechanical [12, 15, 16]
and thermal properties [14, 17, 18]. However, even
though many researchers have studied the relation-
ship between microstructural aspects and mechanical
properties [19–21], little has been addressed about the
connection between microstructure and k and cp of
1045 carbon steel [22–24]. A proper investigation of
these thermal properties, as performed in this study,
can lead to improvements in the understanding of
several mechanical processes such as welding and
Content courtesy of Springer Nature, terms of use apply. Rights reserved.
1907
machining, as well as in the design of devices and
structures subjected to thermal solicitations.
Many studies have been developed to study the
impacts of heat treatments on the microstructural
characteristics and thermophysical properties of
steels. Klein et al. [14] and Kuepferle [18] used laser
flash analysis, DSC measurements, and CALPHAD
calculations to investigate copper-infiltrated steel and
boron alloyed steel, respectively. Both studies con-
cluded that microstructure and thermal properties
are significantly influenced by heat-treatment condi-
tions. Raju et al. [17] investigated the temperature-
dependent specific heat of reduced activation ferritic-
martensitic steel by DSC measurements in normal-
ized and tempered samples. Their study analyzed the
influence of different heat treatments on the material
microstructure and their relation to specific heat.
They indicated that tempered martensite has a higher
cp than untempered martensite. Esfahani et al. [21]
developed a numerical method for coupling the
mechanical properties and microstructural changes
during the quenching of 1045 steel. Volume fractions
of resulting microstructural phases were predicted by
finite element simulation along with internal stresses
caused by inhomogeneous temperature gradients.
Although thermal analyses were performed, no effect
on the thermal properties of the material was con-
sidered. Klein et al. [24] evaluated the impact of heat
treatment on the thermal conductivity of 1045 steel
using laser flash measurements and CALPHAD
simulations. The thermal property was studied dur-
ing the cementite precipitation and then correlated
with aging time, establishing its behavior during the
tempering process. Carbide precipitation kinetics was
correlated to constant thermal properties, neglecting
temperature dependence. Differently, Wilzer et al.
[22] investigated the behavior of the thermal con-
ductivity and specific heat of 1045 steel, associating
these with the final material microstructure after heat
treatment. The microstructural investigations were
performed for different tempering temperatures, but
thermal properties were regarded as constants and
determined at room temperature via two distinct
methods: DSC and laser flash. Since each thermal
characterization technique requires a different sam-
ple preparation, using distinct methods for inferring
thermal conductivity and specific heat consumes
more time and financial resources. Wilzer et al. [23]
accounted for the temperature dependence of the
thermal properties of quenched and tempered 1045
1908 J Mater Sci (2023) 58:1905–1924

steel. Their study, which was also based on DSC and data and their temperature dependence is established
laser flash measurements, only considered tempering via linear regression equations. Scanning electron
at 600 C, which can be considered severe and microscopy, X-ray diffraction, and microhardness
excessively high in many cases [12, 16]. testing are performed to analyze the relationship
Tempering temperature varies according to the between microscopic structure and thermal
mechanical property requirements of a particular properties.
application, where the hardness/toughness ratio
must be adapted. Temperatures from 150 to 550 C
are normally used for tempering low-alloy steels Quenching and tempering steel heat
[12, 16]. There is a lack of information about the treatment
behavior of k and cp of 1045 steel when quenched and
tempered within this range. In this context, this paper Quenching and tempering is a combined heat-treat-
investigates the influence of tempering temperatures ment technique to obtain steels with high strength
of 250, 400, and 550 C on the temperature-dependent and hardness along with gain in ductility and
thermal properties of 1045 steel, seeking to provide toughness. Cooling rate and chemical composition
suitable practical information within the temperature are mainly responsible for the microstructural trans-
range most applied to non-alloyed carbon steels. The formations that evolve during Q&T, and the final
results are particularly useful for performing reliable properties of a quenched and tempered material
simulations and accurately predicting the thermal depend on the microstructure [19]. In practice, both
behavior of quenched and tempered mechanical ele- quenching and tempering processes involve the
ments. Additionally, since most studies focus on continuous cooling of a steel part to room tempera-
investigating mechanical properties, the research lit- ture. Thus, the microstructure that results from
erature concerning the effects of heat treatment, quenching and tempering can be reasonably pre-
microstructural aspects, and temperature on result- dicted by using the continuous-cooling-transforma-
ing thermal properties of steels is scarce [24]. Thus, tion (CCT) phase diagram of the steel being heat
the present study is also meaningful in providing treated [26, 27]. CCT curves show the microcon-
knowledge to better understand metallo-thermo-me- stituents that should develop from austenite for each
chanical processes involving heat-treated steels. cooling rate, indicating the percent complete for a
In this study, an inverse methodology simultane- specific phase transformation [26, 27].
ously estimates k and cp as functions of temperature
via transient, low-cost, and single-shot experiments
conducted at room temperature. The thermal model
must not be subjected to different initial tempera-
tures, avoiding using an oven and performing several
tests. As thermal conductivity and specific heat are
determined at once, two or more different procedures
are not required for obtaining both thermal proper-
ties, unlike standard tests, such as the guarded hot
plate method, the laser flash analysis, and the dif-
ferential scanning calorimetry technique. Inverse
techniques have already proven their effectiveness in
heat transfer applications. In inverse heat conduction
problems (IHCPs), thermal properties, heat fluxes,
and even geometrical features can be indirectly
obtained by comparing experimental temperature
data with analytical or numerical temperatures [25].
The IHCP of interest is solved using a modified
version of the Levenberg–Marquardt (L-M) algo-
rithm. Thermal properties of different heat-treated Figure 1 CCT diagram for the studied 1045 steel with an
samples are determined using single thermocouple austenitizing temperature of 900 C.

Content courtesy of Springer Nature, terms of use apply. Rights reserved.

J Mater Sci (2023) 58:1905–1924 1909

JMatPro simulation software was used to construct

a CCT diagram for 1045 steel austenitized at 900 C,
as shown in Fig. 1. This diagram was constructed
based on characteristics of the studied 1045 steel, as
per chemical composition, obtained by X-ray fluo-
rescence (XRF) spectrometry, and grain size, calcu-
lated from the austenitizing/soaking time. If the
temperature decreases greatly enough to avoid
touching any knee of the CCT diagram, atomic dif-
fusion has no time to take place, so the carbon atoms
are enclosed within the lattice structure, forming
martensite [27]. Ferrite, pearlite, and bainite appear if
the cooling rate cannot drop the temperature below
the ‘‘martensite start’’ (Ms) temperature quickly
enough. Information about the kinetic and thermo-
dynamic mechanisms involved in the Q&T process
can be found in Nunura et al. [19], Esfahani et al. [21],
and Deng and Ju [26].
Tempering hardened steels decreases hardness and Figure 2 Schematic representation of the thermal model (a),
increases toughness and ductility. It also improves some experimental aspects (b), and heater-sample interface in
detail (c). 1: heater; 2: metallic sample; 3: insulation blocks; and 4:
dimensional stability, since as-quenched martensite
temperature sensor.
has high internal stress, and is very strong, brittle,
and hard. This isothermal heat treatment adjusts the mathematical formulation for the heat conduction
microstructure and physical properties of martensite problem of interest here can be expressed as follows
to desired levels [26]. Tempered martensite, which is [10, 29]:
generally composed of precipitates (iron carbides)
ox ½kðTÞox T ¼ qcp ðT Þot T; 0  x  L; t[0 ð1aÞ
dispersed into a ferritic matrix, is obtained from as-
quenched martensite due to solid-state reactions [28]. T ¼ T0 ; 0  x  L; t ¼ 0 ð1bÞ
Tempering effects are determined by the extent of the  
kðT Þox T ¼ hc T rh  T mp ; x ¼ 0; t[0 ð1cÞ
carbide precipitation, which in turn is defined by the
temperature and time of the heat-treatment process kðT Þox T ¼ 0; x ¼ L; t[0 ð1dÞ
[12, 16].
where x is the heat flux direction; t is the time; q is the
density, considered constant; L is the metallic plate
thickness; T0 is the initial temperature; hc is the con-
Inverse heat conduction approach
tact conductance, i.e., the reciprocal of the contact
for thermal characterization resistance, Rc; Trh and Tmp correspond, respectively,
to the temperatures of the resistive heater and the
Thermal modeling
metallic plate at the contact interface between these.
Consider the transient nonlinear one-dimensional Temperature is a function of x and t, i.e., T = T(x, t).
heat conduction throughout a finite metallic plate,
which is subjected to a constant heat flux on the top Sensitivity analysis
surface and an insulation condition on all other
Sensitivity can be understood as a ratio denoting how
boundary surfaces. Figure 2 shows a schematic rep-
well the output/measured quantity varies with a
resentation of the thermal model under study. Ther-
change in the parameter, which is the input/esti-
mal conductivity, k, and specific heat, cp, are assumed
mated quantity. For IHCPs, the sensitivity coefficient
to vary only with temperature, T. Radiation, phase
of a thermal parameter quantitatively evaluates the
change, heat generation, and convection phenomena
relationship between the temperature and the
are neglected in thermal modeling. Thus, the
parameter analyzed. Mathematically, the sensitivity

Content courtesy of Springer Nature, terms of use apply. Rights reserved.

1910
coefficient of a parameter is defined as the first
derivative of the temperature (independent variable)
in relation to the parameter (dependent variable). To
facilitate the comparison of different plots, the sen-
sitivity coefficient is frequently multiplied by the
parameter, as shown in Eq. (2) for parameter P. This
new quantity is called the scaled sensitivity coeffi-
cient (J) and has units of temperature [25].
oT
JP ¼ P ð2Þ
oP
The higher the sensitivity, the greater the possi-
bility of reliably estimating the parameter under
consideration because high sensitivity magnitudes
represent that the temperature data are sufficiently
informative [25]. When two or more parameters are
to be estimated simultaneously, it is also fundamental
to examine if there is a linear dependence relation-
ship between their sensitivity coefficients, i.e., if their
sensitivities are uncorrelated. Sensitivity coefficients
that differ greatly in magnitude and/or are correlated
can yield inaccuracies and difficulties in the estima-
tion procedure.
Inverse problem
Normally, the thermal properties of conducting
materials can only be assessed indirectly. As a result,
temperature data can be obtained experimentally to
estimate these unknown quantities through inverse
analysis. Inverse solutions are difficult to evaluate
because inverse problems are usually ill-posed. This
condition implies intrinsic instability to input data and
measurement errors [25]. For the heat conduction
problem described by Eqs. (1a, 1b, 1c, 1d), one can
construct an inverse problem in which the tempera-
ture-dependent thermal properties are initially
unknown. To solve the IHCP and then simultaneously
identify the thermal conductivity and specific heat, an
objective function is built by fitting numerically com-
puted temperatures to corresponding experimentally
measured temperatures. Thus, the inverse heat con-
duction approach for thermal characterization is con-
structed based on a minimization problem.
Thermal conductivity and specific heat are con-
tained in the vector of unknown parameters P = [k,
cp]. These are simultaneously estimated by minimiz-
ing an objective function based on the sum of squared
temperature residuals, as follows:
Content courtesy of Springer Nature, terms of use apply. Rights reserved.
J Mater Sci (2023) 58:1905–1924
OFðPÞ ¼ ½Y  TðPÞT ½Y  TðPÞ ð3Þ
where Y = [Yi,Yi?2,…,Yi?r] is the measured temper-
ature obtained by experiment; T = [Ti,Ti?2,…,Ti?r] is
the numerical temperature obtained from the direct
solution of the heat conduction problem; i is the time
index; and r is the number of future time steps along
N total number of measurements.
In this study, the commercial software COMSOL
Multiphysics was used to solve the direct problem at
different P and the L-M algorithm was implemented in
MATLAB to evaluate which combination of k and cp
best matches T with Y. A modified version of the
Levenberg–Marquardt method was used [29]. This
adapted version used the ‘‘discrepancy principle’’ to
establish convergence and incorporated the concep-
tion of future times to improve regularization [25].
Point estimates for k and cp were determined at tem-
perature intervals of about 8 C. This implies that the
temperature domain was sectioned into intervals. Both
thermal properties were assumed to remain constant
throughout the metallic plate for a period of time
within these intervals. Constant k and cp values were
iteratively computed at each temperature interval
using data from time ti to time ti?r. The average tem-
perature Ta = 0.5(Ti ? Ti?r) was used as the indepen-
dent variable to assign these point estimates, i.e., k(Ta)
and cp(Ta). Although the effects of temperature varia-
tions of up to 10 C on k and cp are not statistically
significant [11], considering constant thermal proper-
ties over the metallic plate undoubtedly introduces
some bias into the simultaneous estimation procedure.
Accuracy analysis
The temperature measurements used to solve the
IHCP being studied are obtained using real measur-
ing devices, which can exhibit variability in their
responses. As a result, the measured data should be
treated as an estimate of the measurand. In this same
context, the numerical methods used to perform the
simultaneous estimation also do not produce exact
responses, since convergence tolerance conception is
inherent to these. Accuracy analysis on experimental
and computational error sources is an adequate
approach to quantify the estimation uncertainties,
showing how incomplete the estimates are [30].
The total uncertainty associated with the estima-
tion results (Utotal) can be obtained by the quadratic
J Mater Sci (2023) 58:1905–1924 1911

sum of the independent errors from numerical tem-
perature (UT), measured temperature (UY), and L-M
tolerance (UL-M). Equation (4) gives all error sources
considered when quantifying the total uncertainty
related to the estimation of P [30].
 Scanning electron microscopy, X-ray
2 2
Utotal ¼ Uelectrical resistance þ Uelectrical current þ Uroughness þ Uweight þ Udimensions
2 2
þUhardness þ UPARDISO
2 2
þUtemperature reading þ Udata acquisition þ Uthermalinsulation þ ULM
The terms contained in the expression of Utotal are
relative uncertainties. To obtain these, the numerical
methods’ tolerances and the measuring devices’ res-
olutions are taken as individual uncertainties. Then,
one divides each individual uncertainty by the mean
value of the analyzed parameter; except for temper-
ature, for which its maximum rise is taken.
Experimentation
Table 1 shows the chemical composition of the 1045
steel sheet taken as the object of study. The chemical
composition was determined by the material supplier
using a Niton XL3t-800 XRF spectrometer. Four
specimens were machined from the as-received
material into the following dimensions (± 0.10 mm):
50.00 mm length, 50.00 mm width, and 10.00 mm
thickness, measured by a Mitutoyo 530104BR analog
caliper. To determine the density, specimen weights
were evaluated using a Bel S2202H analytical scale
and divided by the volume, resulting in an average
value of 7796 kg m-3.
Heat-treatment processes
Table 2 lists the investigated heat-treatment condi-
tions. One of the specimens was analyzed as
received, i.e., normalized, which is the typical con-
dition given by manufacturers. The other three
specimens were quenched into room-temperature
water from the same austenitizing temperature
(900 C for a soaking time of 30 min) and subse-
quently tempered for a soaking time of 1 h at
different temperatures: 250, 400, or 550 C, followed
by air cooling. Austenitization and tempering pro-
cedures were performed in a Brasimet K-250 electri-
cal resistance muffle furnace.
2 2 2
2 2
diffraction, and microhardness testing
þ UBDF þ Uthermocouple calibration
12
2 2
To qualify the microstructural and crystallographic
ð4Þ features of 1045 steel and validate the different ana-
lyzed conditions, cubic samples (edge length 10 mm)
were heat treated along with the specimens, for
metallographic analysis purposes. All heat-treated
pieces were first cleaned and visually checked for
distortion and cracks due to thermal stress. The
metallographic samples were prepared according to
the following route: (1) abrasive cutting (sectioning);
(2) hot mounting in epoxy resin; (3) wet grinding on
silicon carbide papers with different grit sizes using
an Arotec Aropol 2 V-PU manual grinder; (4) final
polishing on felt cloth with colloidal silica suspension
by using an Allied MetPrep 3/PH-3 semiautomatic
polisher; (5) ultrasonic cleaning; (6) chemical etching
in 3% nital solution (3v/v% nitric acid–ethanol). It is
important to mention that the epoxy resin used is
both electrically non-conductive and thermally non-
conductive, which helps to avoid tempering effects
when embedding the specimens.
Microstructural investigation and phase identifi-
cation were performed by microscopy. These analy-
ses were conducted using a ZEISS EVO MA 15
scanning electron microscope (SEM) in backscattered
electron mode (BSE) at an accelerating voltage of
15 kV and a working distance of 19.5 mm. Bulk
samples were also submitted to X-ray diffraction
(XRD) analysis in a Panalytical X’Pert PRO XRD
diffractometer, with Cu-Ka radiation, scan angle
from 30 to 90 with 0.02 step, and time per step of
1 s. Diffractograms were submitted to a smoothing
function to reduce XRD variation.
Vickers microhardness for all heat-treatment con-
ditions was measured using a TIME TH712 digital
microhardness tester. Indentations were made with a
load of 0.2 kgf and a dwell time of 15 s. Since the
specimens are macroscopically homogeneous,

Table 1 Chemical
composition of the studied Element C Si Mn Cr Ni Mo P S Al Cu Fe
1045 carbon steel (mass
Content [wt%] 0.45 0.15 0.63 0.21 0.19 0.16 0.03 0.03 0.25 0.23 Bal.
percentage)

Content courtesy of Springer Nature, terms of use apply. Rights reserved.

1912
Table 2 Heat-treatment
conditions for the studied 1045 Specimen
steel specimens
N As-received—normalized
Q&T250
Q&T400
Q&T550
measurements were taken at 15 spots randomly on
each sample, and the averages were evaluated.
Heat conduction apparatus
The solution to the IHCP at hand is based on com-
paring temperature measurements with numerical
results obtained by solving the direct problem at
different thermal property values. This way, the goal
of the experiment described next is to provide
experimental data Y to be inserted into the method-
ology described in Section ‘‘Accuracy analysis’’ and
then infer thermal conductivity and specific heat. The
parameter estimation problem is solved when a set of
k and cp that is likely to be responsible for the mea-
sured thermal response is found.
Figure 3 shows a scheme of the experimental heat
conduction apparatus used under controlled labora-
tory conditions to determine the temperature-de-
pendent thermal properties of 1045 steel. This
apparatus basically consists of a metal sample placed
inside a pocket milled in a refractory ceramic fiber
block and then covered by another insulation block.
At room temperature, the sample was heated at its
top surface (x = 0) by an Omega SRFGA20210 square-
section resistive heater. While the top surface was
subjected to a heat flux of about 15,400 W m-2 during
300 s, all other surfaces were kept thermally insu-
lated. An ITech IT6953A digital DC power source
was used to regulate and supply the voltage input
required by the resistive heater. The experimentally
applied heat flux density was determined by
Figure 3 Scheme of the
experimental heat conduction
apparatus used to estimate the
temperature-dependent
thermal properties of 1045
steel.
Content courtesy of Springer Nature, terms of use apply. Rights reserved.
J Mater Sci (2023) 58:1905–1924
Heat-treatment condition
Austenitized at 900 C, quenched in water, and tempered at 250 C
Austenitized at 900 C, quenched in water, and tempered at 400 C
Austenitized at 900 C, quenched in water, and tempered at 550 C
dividing the electric power supplied by the heating
area. The electric power was obtained using the
resistance and the current of the electric circuit,
which were measured with Instrutherm MD380 dig-
ital multimeters.
The average contact conductance, which was
1388.2 W m-2 K-1, was determined by measuring
some physical quantities related to the experimental
setup and then settling these in COMSOL Multi-
physics. A Mitutoyo SJ210 digital roughness meter
measured average surface roughness of 0.83 lm for
the resistive heater and 0.18 lm for the metal plates.
The contact pressure, which was 1521.4 N m-2, was
evaluated by dividing the weight of the experimental
apparatus, which was assessed by the analytical
scale, by its cross-sectional area, which was deter-
mined using the dimensions assessed by the analog
caliper.
The location of temperature sensors is essential to
evaluate an inverse thermal problem since it is nec-
essary to convey a sufficiently complete knowledge
of the transient behavior of the thermal model. Fur-
thermore, in data-driven inverse problems, it is
important from a practical sense that a minimum
number of sensors be used. For metals in 1D formu-
lations under ambient conditions, taking temperature
measurements using thermocouples is not feasible at
any surface other than at the bottom surface, i.e., at
x = L [29]. This is because the presence of the ther-
mocouple wire compromises the accommodation of
the heater on the top surface (x = 0), strongly affect-
ing the heat exchange. Researchers have already
J Mater Sci (2023) 58:1905–1924
shown that this location can transmit information
that is sensitive enough to carry out the estimation
process, given appropriate initial values [10, 31]. In
this context, sensitivity coefficients were evaluated to
ensure that a single thermocouple at the bottom
surface can transmit temperature information that is
sensitive enough to enable an accurate inverse
solution.
Since sensitivity analysis is a prior study toward
the simultaneous estimation feasibility, it was per-
formed only for the normalized condition. The sen-
sitivity-related calculations were based on k and cp
values obtained from introducing the measured
temperatures into thermal property equations found
in Grzesik and Nieslony [32]. Figure 4 shows the
modified sensitivity coefficients of k and cp at the
insulated surface over the experimental time. Exam-
ining this figure, one can observe that the modified
sensitivity coefficient of k (J k ) responds rapidly to
heat flux load and increases in the initial moments
until reaching a constant magnitude. J k is propor-
tional to variations in the heat flux intensity. The
modified sensitivity coefficient of cp (J cp ) increases
uniformly with temperature. As the temperature
raises throughout the duration of the experiment, the
magnitude of J cp raises continuously. Both thermal
properties have sensitivity coefficients with good
magnitude during practically the entire experiment
length. The effects of low sensitivity at the initial
stages of the experiment are overcome at later times
during the estimation process [25]. Additionally, one
Figure 4 Modified sensitivity coefficients at x = L for normalized
condition.
Content courtesy of Springer Nature, terms of use apply. Rights reserved.
1913
can also note that J k and J cp are linearly independent,
i.e., they have uncorrelated (different) shapes. Given
suitable starting values and enough sensitivity, one
can anticipate that there will not be major difficulties
encountered when computing the sensitivity and
information matrices using single sensor temperature
data collected at the bottom surface. Therefore, one
can ensure that x = L is a feasible point to be explored
to take temperature measurements to solve the IHCP
under study. Transient temperature information was
measured by a 30 AWG T-type thermocouple, which
was welded at the bottom surface using a capacitive
discharge. A Marconi MA184 thermostatic bath was
used to calibrate the thermocouple against the read-
ings of a PT-100 sensor. The temperature history was
collected at a sampling rate of 10 Hz by a Keysight
34980A data acquisition. After being collected by the
data acquisition system, the measurements Y were
recorded in a computer, in which the inverse analy-
ses were performed jointly using COMSOL and
MATLAB. Measurements were repeated at least five
times for each specimen.
Results and discussion
Microstructural characterization
Knowing that the ‘‘martensite final’’ (Mf) temperature
for 1045 steel is above environmental temperature
[33], retained austenite is not expected in the
microstructure of quenched alloys when austenitiza-
tion is properly carried out, i.e., an appropriate tem-
perature and soaking time to reach a homogeneous C
distribution. Moreover, water quenching should be
effective to avoid bainite formation in samples with
the present dimensions [27]. Figure 5 shows the
micrographs (SEM/BSE) of the studied 1045 steel, in
the conditions evaluated in the present work, i.e.,
(a) normalized, and austenitized ? water
quenched ? tempered at: (b) 250 C, (c) 400 C and
(d) 550 C for 1 h (details in subsection ‘‘Heat-treat-
ment processes’’). The normalized microstructure
(a) shows allotriomorphic ferrite along with perlite,
while the austenitized/quenched/tempered condi-
tions (b–d) essentially show complete lath martensite
structures. One can observe that the complete
martensitic microstructures in (b–d) practically do
not differ in morphology due to the same austeni-
tizing temperature applied [19]. Some parallel
1914
Figure 5 Micrographs (SEM/
BSE) of 1045 steel after
normalizing (a), and after
water quenching and
tempering at 250 C (b),
400 C (c), 550 C (d).
Figure 6 Details of 42 \ 2h \ 47 region of 1045 steel XRD
diffractograms after tempering at 250 C, 400 C, 550 C along
with a’ (martensite) pattern for as-quenched condition.
alignment of the martensite crystals can be seen
within the packets of laths.
Figure 6 shows the region within 42 \ 2h \ 47 of
the XRD diffractograms for the austenitized/quen-
ched/tempered samples. A simulation procedure
was performed in PowderCell software [34] to obtain
the pattern for as-quenched martensite. This simula-
tion was based on crystallographic information from
Villars and Calvert [35] and lattice parameters
expected for a 0.45 wt%. C steel, i.e., c = 0.291 nm
and a = 0.285 nm obtained from Bhadeshia and
Content courtesy of Springer Nature, terms of use apply. Rights reserved.
J Mater Sci (2023) 58:1905–1924
Honeycombe [36]. The simulated pattern, which is
also shown in Fig. 6, indicates the (101) and (110)
dissociated peaks which is not the case in the
experimental diffractogram, a sign of martensite
tetragonality depletion. This is expected due to a
lattice relaxation caused by a reduction of carbon
contained in the martensite structure and the for-
mation of carbides of the binary Fe–C system, during
tempering in stages. Transition e-carbide (Fe2.4C) and
cementite (iron carbide, Fe3C) are the most common
in the precipitation sequence during tempering
martensite in steels [28]. It should be noticed that no
peaks related to retained austenite were detected
(43.9, for an a parameter of 0.357 nm, also obtained
from Bhadeshia and Honeycombe [36]). Considerable
remaining retained austenite would decrease the
hardness and could distort the thermal conductivity
identification [22, 23]. Thus, XRD analysis is consis-
tent with martensitic structures with carbide forma-
tion and tetragonality loss in the matrix
[13, 24, 28, 36]. Besides carbide formation, tempering
Table 3 Microhardness measurements for the studied 1045 steel
specimens
Specimen N Q&T250 Q&T400 Q&T550
Microhardness [HV] 194.8 435.4 363.3 280.2
J Mater Sci (2023) 58:1905–1924 1915

leads to the rearrangement or elimination of crystal

defects and the relief of stress in martensite [36]. This
is confirmed in the Vickers test results shown in
Table 3 (details of microstructural phenomena
evolving martensitic steel tempering can be found in
detail in several publications, e.g: Abbaschian and
Reed-Hill [33], Bhadeshia and Honeycombe [36],
Jung et al. [37]). The normalized hardness is in
agreement with the expected values in the literature
[38], presenting a higher standard deviation due to
higher microstructural inhomogeneity caused by
ferrite and perlite regions. The as-quenched sample
presented an average hardness of 549.8 HV. As
expected, the hardness values decrease according to
tempering temperature rise, which enhances stress
relaxation. The microhardness of the specimen tem-
Figure 7 Experimentally measured temperatures for all heat-
pered at 400 C is in line with the values found in the
treated samples.
literature [33, 36, 38], while the specimen tempered at
250 C and 550 C is slightly lower and higher than different heat-treatment conditions, which led to
the literature indication (approximately 500 and variations in their thermal properties. Using these
200 HV, respectively) [33, 36, 38]. temperature measurements and following the
methodology described in subsection ‘‘Accuracy
Simultaneous estimation of thermal analysis’’, the temperature-dependent thermal prop-
conductivity and specific heat erties of 1045 steel were determined for each heat-
treatment condition shown in Table 2. Figures 8 and
After performing a mesh independence study, an
9 show, respectively, the point estimates of k and cp
11-element mesh was used in COMSOL to numeri-
for all temperature intervals. Tables 4 and 5 give the
cally solve the direct problem given by Eqs. (1a, 1b,
estimates accompanied by their 99% confidence
1c, 1d), seeking to achieve reasonably low computa-
tional costs when estimating k(T) and cp(T).
Inverse problem solutions are obtained by com-
bining prior knowledge with measurements. Gradi-
ent-based methods, such as the L-M algorithm, can be
efficient in finding well-defined minima since ade-
quate staring values are available. This implies that
the initial guess provided must not be very far from
the wide region of the minimum [25]. To guarantee
the convergence of the simultaneous estimation pro-
cedure, the thermal property values at room tem-
perature found in Incropera et al. [39] were used as
initial guesses, and the feasible search space was set
with limits from one-half and one to one-half times
the starting values. Suitable initial values are more
critical for k due to its lower sensitivity compared to
cp. Estimates of a less temperature-sensitive thermal
parameter are more significantly impacted by mea-
surement errors and input data.
Figure 8 Thermal conductivity of 1045 steel as a function of
Figure 7 shows the temperatures measured in all
temperature. Comparison of the estimates of this study for
heat-treated samples studied. The differences different heat-treatment conditions with values available in the
between these measured data are due to their research literature.

Content courtesy of Springer Nature, terms of use apply. Rights reserved.

1916 J Mater Sci (2023) 58:1905–1924

Wilzer et al. [23] determined k(T) and cp(T) of 1045

Figure 9 Specific heat of 1045 steel as a function of temperature.
Comparison of the estimates of this study for different heat-
treatment conditions with values available in the research
literature.
intervals, which were calculated according to the
Bonferroni method [40]. For both thermal properties,
reasonably significant differences are observed
between the estimated values for the heat-treatment
conditions investigated. One can observe that k has
sparser and noisier estimates in relation to cp because
of its lower sensitivity. Also in Figs. 8 and 9, the
results are compared with linear functions from
previously published research. Grzesik and Nieslony
[32] provided data for normalized steel (N), while
steel after quenching, deep cryogenic treatment, and
tempering at 600 C (Q&DCT&T600). In general, the
curves agree reasonably well, with the slopes of the
lines differing to some degree. This may be because
both thermal properties do not follow a perfect linear
relationship with temperature, and linear regressions
end up biasing the temperature dependence assess-
ment. Wilzer et al. [22] measured constant thermal
properties of samples under different heat-treatment
conditions, as follows: Q&DCT&T200:
-1 -1
k = 40.3 W m K and cp = 440 J kg-1 K-1;
Q&DCT&T400: k = 49.3 W m-1 K-1 and cp-
-1 -1
= 450 J kg K ; and Q&DCT&T600:
k = 48.2 W m-1 K-1 and cp = 420 J kg-1 K-1. These
results are partially in agreement with those obtained
in the present study. This is because although there is
good accordance in terms of magnitude, there is
divergence concerning the connection between ther-
mal properties and tempering temperature. In this
work, continuously increasing tempering tempera-
ture was found to increase both k and cp. By contrast,
the measurements made by Wilzer et al. [22] at room
temperature do not behave exactly in the same way.
It should be mentioned that the variations observed
are mainly due to differences in the chemical com-
position of the 1045 steel studied here and those
investigated in other studies. These differences
modify the CCT curves, consequently leading to
different resulting microstructure and thermome-
chanical properties. Different heat-treatment

Table 4 Thermal conductivity

of 1045 steel under different Interval Assigned temperature [C] Thermal conductivity, k [W m-1 K-1]
heat-treatment conditions, for N Q&T250 Q&T400 Q&T550
all temperature intervals
1 24 52.2 ± 6.5 40.8 ± 5.5 45.0 ± 6.2 46.9 ± 6.4
2 32 51.4 ± 6.7 36.7 ± 5.6 45.2 ± 5.5 48.9 ± 6.2
3 40 48.9 ± 6.4 39.7 ± 5.7 42.5 ± 5.7 45.7 ± 5.8
4 48 49.3 ± 6.3 40.0 ± 5.6 41.3 ± 5.3 42.8 ± 5.7
5 56 48.2 ± 6.6 37.5 ± 5.7 43.2 ± 5.6 41.5 ± 5.8
6 64 51.2 ± 6.5 34.7 ± 5.4 39.7 ± 5.5 43.1 ± 5.7
7 72 49.7 ± 6.5 39.7 ± 6.1 39.5 ± 5.7 40.4 ± 5.5
8 80 48.3 ± 6.4 38.7 ± 6.0 42.8 ± 5.6 43.8 ± 6.0
9 88 50.9 ± 6.8 35.6 ± 5.4 40.3 ± 5.6 42.3 ± 5.7
10 96 47.3 ± 6.7 38.7 ± 5.5 40.4 ± 5.6 41.5 ± 5.6
11 104 48.5 ± 6.7 38.2 ± 5.7 38.7 ± 5.5 37.3 ± 5.5
12 112 46.2 ± 6.8 39.8 ± 5.8 39.3 ± 5.4 40.9 ± 5.7
13 120 46.2 ± 7.0 33.2 ± 5.6 39.6 ± 5.5 42.5 ± 5.8
14 128 45.0 ± 6.9 35.0 ± 5.7 37.8 ± 5.6 39.8 ± 5.6
15 136 43.8 ± 7.0 33.2 ± 5.6 34.4 ± 5.5 36.4 ± 5.7

Content courtesy of Springer Nature, terms of use apply. Rights reserved.

J Mater Sci (2023) 58:1905–1924 1917

Table 5 Specific heat of 1045

steel under different heat- Interval Assigned temperature [C] Specific heat, cp [J kg-1 K-1]
treatment conditions, for all N Q&T250 Q&T400 Q&T550
temperature intervals
1 24 449.5 ± 35.1 415.1 ± 31.5 423.1 ± 31.2 424.5 ± 33.2
2 32 452.4 ± 35.3 418.8 ± 33.0 426.2 ± 32.3 430.0 ± 31.8
3 40 456.5 ± 35.5 423.4 ± 31.9 430.6 ± 32.9 438.9 ± 32.9
4 48 459.7 ± 34.8 423.9 ± 32.8 436.5 ± 33.3 441.5 ± 34.4
5 56 461.0 ± 36.3 421.3 ± 34.4 440.4 ± 35.4 446.0 ± 33.9
6 64 465.0 ± 36.1 428.9 ± 32.6 441.6 ± 31.6 447.0 ± 33.6
7 72 468.7 ± 36.9 428.3 ± 35.3 445.1 ± 34.1 445.7 ± 35.0
8 80 472.8 ± 36.7 436.7 ± 34.3 448.5 ± 34.0 450.3 ± 35.3
9 88 475.1 ± 35.1 445.4 ± 32.3 453.1 ± 32.2 456.0 ± 33.5
10 96 478.7 ± 37.1 450.4 ± 33.9 457.2 ± 32.7 463.6 ± 33.8
11 104 482.7 ± 34.9 454.4 ± 32.8 460.7 ± 30.9 468.5 ± 32.9
12 112 487.7 ± 36.3 459.5 ± 35.3 464.9 ± 34.8 473.0 ± 32.8
13 120 491.0 ± 37.3 464.3 ± 33.9 469.3 ± 34.1 472.7 ± 34.3
14 128 495.4 ± 36.1 468.6 ± 35.1 473.2 ± 35.0 475.9 ± 36.0
15 136 501.6 ± 37.5 473.3 ± 35.8 477.1 ± 35.3 486.2 ± 35.9

parameters (e.g., austenitizing temperature, soaking
time, quenching medium, etc.) and distinct thermal
characterization techniques may also explain these
discrepancies.
The maximum k values for the heat-treatable car-
bon steel were found to occur at the lowest temper-
atures of the range studied. The thermal conductivity
of low-alloy steels, which is primarily electronic,
decreases with increasing temperature due to pho-
non–electron interaction. Once this thermal scattering
phenomenon saturates, k becomes practically inde-
pendent of temperature [41]. On the other hand,
values of cp for the quenched and tempered 1045 steel
assume higher values as the temperature increases.
This positive relationship between specific heat and
temperature for metal alloys is mainly attributed to
lattice vibrations (phonons). Secondarily, electronic
(exciting electrons) and magnetic (Gibbs energy)
contributions also affect the stored energy [42].
Relationship between thermal properties
and microstructural aspects
The main objective of Q&T is to improve the
mechanical properties of steels, like hardness,
strength, and wear resistance. Thus, the effects of
such heat treatment on these types of properties are
much more commonly investigated than the effects of
Q&T on thermal properties. In most carbon steels,
grain size, microscopic structure, and carbon content
define the resulting properties. In general, strength
and hardness increase with increasing carbon con-
tent; and silicon, manganese, and nitrogen yield a
considerable impact on strength, as does grain size
[13, 43]. For ferrite–pearlite microstructure (Fig. 5a),
ferrite grain size is a relevant factor concerning both
toughness and strength. This is because strength in
ferrite–pearlite steels is more significantly controlled
by the ferritic matrix, which has a strength lower than
that of perlite. For martensitic microstructure
(Fig. 5b–d), strength is also impacted by lath
martensite packet size. It should be mentioned that
hardness does not increase steadily with carbon in
martensitic steels due to the formation of retained
austenite [43].
Kuepferle et al. [18], Wilzer et al. [22], Wilzer et al.
[23], Klein et al. [24], and Uher [44] have already
shown that the thermal conductivity of carbon steels
is strongly affected by alloying degree, microstruc-
tural imperfections, and impurities. Since the last two
aspects are modified by Q&T heat treatment, some
impact on k is therefore expected. Investigations by
Wilzer et al. [22] and Wilzer et al. [23] on non-alloyed
and high-alloyed martensitic steel and by Geller et al.
[45] on high-speed steel showed that the chemical
composition of the martensite matrix most consider-
ably controls thermal conductivity. Thus, the most
important is the composition of the tempered
martensite, which in the case of 1045 steel is similar to
pure iron due to its low content of alloying elements
[22, 23]. Reed and Clark [46] and Terada et al. [47]
indicated that in general alloying elements produce a

Content courtesy of Springer Nature, terms of use apply. Rights reserved.

1918
relevant drop in k of pure iron. This is because car-
bide precipitation after Q&T yields scatter sources for
phonons and electrons, decreasing thermal conduc-
tivity. For non-alloyed martensitic steels, e.g., 1045
steel, this effect is limited due to the depletion of
alloying elements in the martensitic matrix [22].
In general, steels have their thermal conductivity
decreased with increase in alloy content. Peet et al.
[48] concluded that silicon, vanadium, and copper
have a slight impact on the thermal conductivity of
steels. Manganese, nickel, molybdenum, and chro-
mium showed to be much more significant. Stainless
steels, for example, have a much lower thermal con-
ductivity compared with most steels due to a higher
content of these last four alloying elements. Chro-
mium enhances the scattering of free electrons,
leading to a reduction in heat conduction due to
electronic contribution [49]. The influence of man-
ganese and molybdenum resembles that of chro-
mium when they are in solid solution [47]. Shanina
et al. [50] and Gavriljuk et al. [51] also indicated that
chromium, manganese, and molybdenum decrease
the concentration of conduction electrons in iron.
Wang and Li [52] studied hot work steels to assess
the influence of different alloying elements on k of
ferrite. They suggested that the effects of silicon and
aluminum are much more significant than those of
chromium, manganese, and molybdenum. In this
context, the low content of silicon and the presence of
aluminum may explain why the thermal conductivity
determined for the normalized sample is higher than
that found in Grzesik and Nieslony [32]. It is
important to mention that is difficult to accurately
analyze the effect of an alloying element on k of steels
separately. This is because the effect of an alloying
element can be modified or even superposed in the
presence of another element.
At room temperature, the k values determined here
for non-alloyed 1045 steel under all heat-treatment
conditions range from approximately 41 (Q&T250) to
52 (N) W m-1 K-1. These values are significantly
lower than that for pure iron, which is about
80 W m-1 K-1 [39]. The difference in the thermal
conductivities may be mainly associated with two
mechanisms [22, 53]: (i) 1045 steel is complete
martensitic after being quenched and tempered, and
therefore has a high dislocation density because of
the martensitic structure; (ii) although non-alloyed,
tempered 1045 carbon steel has a martensite matrix
containing up to 2 wt% of dissolved alloying
Content courtesy of Springer Nature, terms of use apply. Rights reserved.
J Mater Sci (2023) 58:1905–1924
elements, which act as impurities. Both mechanisms
are known to be reducing factors in thermal con-
ductivity [46].
Diffusion of alloying elements in martensitic
structures is restrained under ambient conditions, but
is stimulated by tempering [54]. At tempering tem-
peratures up to 100 C, carbon atoms can segregate
into lattice defects prior to carbide formation. Carbide
precipitation exerts an important influence on the
thermal conductivity of heat-treatable steels [24].
Above 200 C, precipitation of transition e-carbide
(Fe2.4C) and loss of martensite tetragonality initiate
[55]. Tempering at temperatures above 350 C leads
to a more substantial formation and coarsening of
carbides, decreasing the interfacial barriers, even
though the carbides act as additional imperfections
[56]. In this case, the tempering temperature is suffi-
ciently high to enable the formation of cementite (iron
carbide) with composition and crystal structure Fe3C
[56, 57]. The formation of Fe3C-cementite is expected
when tempering low-alloy steels [58]. Martensite
tetragonality depletion and carbide formation were
evidenced in the microstructural analysis of the
quenched and tempered specimens (subsection
‘‘Microstructural characterization’’).
Tempering yields an increase in the thermal con-
ductivity of steel. This is associated with the reduc-
tion of scattering sources in the iron-based matrix
[49]. Carbide precipitation during tempering with-
holds a certain amount of alloy content. This results
in a depletion of alloying elements in solid solution
and a subsequent increase in electronic thermal con-
ductivity. Additionally, as carbide formation
depends on carbon diffusion, higher k values are
expected for steels tempered at higher temperatures
due to solid carbon loss. Thermal conductivity also
tends to increase at high tempering temperatures due
to the bonding of alloying elements in the iron car-
bides formed and to the decrease in the martensite
matrix distortion. These two mechanisms, which are
more substantially stimulated when tempering at
400/550 C, enhance the electron density and
decrease the phonon and electron dispersion [24].
The k estimates determined in this study are in line
with this. However, no great difference is noticed
between k values for the specimens tempered at 400
and 550 C. This suggests that increasing tempering
temperature has a limited influence on increasing
thermal conductivity. This occurs due to the complete
formation of carbon in Fe3C precipitate that has
J Mater Sci (2023) 58:1905–1924
almost achieved its equilibrium composition because
of a low-alloy concentration. This saturation phe-
nomenon is facilitated because Fe3C carbide contains
a similar matrix composition to that of substitutional
alloying elements since they have low atomic diffu-
sivity [22]. Carbon depletion and substitutional
depletion in the matrix are both primarily responsible
for impacting the thermal conductivity of this heat-
treated carbon steel. Therefore, one can consider that
k of quenched and tempered 1045 steel is mainly
controlled by the composition of the tempered
martensite matrix.
Investigations by Ray and Mohanty [59] on carbon
steel and by Mesquita et al. [60] on chromium-based
steel indicated that silicon content has a considerable
impact on the carbide precipitation stages. They
showed that alloy steels containing less than 0.3 wt%
of silicon tend to form mainly Fe3C precipitate. This is
an important fact because the studied 1045 steel has a
relatively low silicon content (0.15 wt%). Thus, the
formation of Fe3C is thermodynamically favorable
and the negative influence of silicon on the marten-
site matrix thermal conductivity is significantly
reduced. In this study, this is significant when tem-
pering at 550 C since this heat-treatment condition
encloses the diffusion of substitutional elements.
Inward diffusion of alloying elements, mainly chro-
mium and manganese, into Fe3C formed with the
martensite matrix occurs until thermodynamic equi-
librium is reached [57].
Raju et al. [17] investigated the on-heating phase
transformations of reduced activation ferritic–
martensitic steel, and no relevant differences were
detected in the specific heat of different martensitic
structures formed from different austenitizing tem-
peratures. Wang et al. [61] studied the quenching of
1080 carbon steel and revealed that cp of martensite
remains practically constant until transformation into
austenite. This fact concerning the steady heat-ca-
pacity behavior of martensite observed by these
researchers is mainly because the steel structure is
completely hardened after being cooled from the
austenite phase. Raju et al. [17] and Jeyaganesh et al.
[62] indicated that tempered martensite has higher cp
values than untempered martensitic structures. They
also showed that the heat-capacity characteristics of
tempered ferritic-martensitic steels are very slightly
increased with increasing tempering temperature
and time. Wilzer et al. [22] and Wilzer et al. [23] did
not indicate greatly significant changes in the
Content courtesy of Springer Nature, terms of use apply. Rights reserved.
1919
volumetric heat capacity by tempering three
martensitic steels. In this study, the specific heat
values of specimens tempered at three distinct tem-
peratures differed only about 4%. This mild influence
of tempering temperature is due to the reduction of
the stored energy in the martensite matrix, with
thermal activation followed by structural reorgani-
zation [62].
The normalized specimen studied here has a
comparatively higher specific heat compared to the
different quenched and tempered microstructures.
This can be attributed to a higher amount of undis-
solved precipitated carbides in martensitic
microstructures in relation to normalized
microstructures [17].
Multiplying the density by the specific heat of a
material gives the volumetric specific heat, which is
an input quantity in the heat diffusion equation. In
this sense, it will be also important to investigate the
influence of density on k and cp. Wilzer et al. [22]
showed that the density of the martensite matrix of
quenched 1045 steel increases with tempering due to
carbon loss and defect density reduction. However,
density varied less than 1% with the tempering
temperature ranging from 200 to 600 C. For this
reason, the densities of all metallic specimens studied
here were considered equal and constant.
Since the thermal properties of ferromagnetic
materials are significantly modified by changes in the
working temperature [63], it is fundamental to
investigate the relationship between temperature-
dependent thermal properties and microstructural
aspects of iron-based materials. Therefore, the pre-
sent work is relevant because it combines heat
transfer and materials science to study the effects of
tempering temperature on k and cp of quenched and
tempered 1045 steel. Additionally, the results of this
study provide suitable information for the design of
engineering systems dealing with heat-treated carbon
steels.
Validation of results
Taking both thermal properties as linearly dependent
on temperature, curves describing k(T) and cp(T) were
determined by linear regressions through the point
estimates. Table 6 gives the correlating equations
obtained for all heat-treatment conditions, with T be-
ing introduced in C. These equations, which are
valid up to 140 C, were further used to numerically
1920
Table 6 Thermal properties of 1045 steel as a linear function of
temperature for different heat-treatment conditions
Heat-treatment Thermal property equations
condition
k [W m-1 K-1] cp [J kg-1 K-1]
N 53.1 - 0.055 9 T 437.3
Q&T250 40.7 - 0.040 9 T 397.5
Q&T400 46.3 - 0.061 9 T 411.7
Q&T550 48.2 - 0.056 9 T 414.9
calculate the temperature field of each sample. Dis-
crepancies no greater than 1.6 C were noted between
numerical and experimental temperatures. This
residual represents approximately 1.1% of the tem-
perature range covered, which is satisfactory, par-
ticularly when dealing with a low-cost experimental
setup tested in a transient state. Unaccounted heat
loss effects involved in the IHCP modeling impact
the accordance between the temperatures. The insu-
lation blocks made of ceramic fiber have non zero
thermal properties, so some heat loss from the sam-
ple to the insulation is naturally expected. By
assuming a linear relationship between thermal
properties and temperature, errors related to fitting
linear regressions also contribute to biasing the
numerical temperature calculations.
Also as part of the validation exercise, the
repeatability of the simultaneous estimation proce-
dure was verified by inverse heat flux estimation. The
heat flux applied at x = 0 was retrieved using the
estimated linear curves and the measured tempera-
tures. The heating load on each metal sample was
estimated using Beck’s nonlinear technique with 10
future time steps [25]. In Fig. 10, one can see that the
retrieved heat flux densities agree reasonably well
with the heat flux calculated from experimental
electrical parameters (power, resistance, and current).
One can also observe that the retrieved heat flux
histories become smaller as time proceeds and tem-
perature rises due to heat loss to the environment.
This condition and the differences between calculated
and measured temperatures demonstrate that the
actual heat flux supplied to the samples is a function
of time, not constant as assumed. However, the small
temperature residuals and the good agreement
between the heat fluxes indicate that the mathemat-
ical formulation has minimal inconsistencies.
Content courtesy of Springer Nature, terms of use apply. Rights reserved.
J Mater Sci (2023) 58:1905–1924
? 0.449 9 T
? 0.541 9 T
? 0.477 9 T
? 0.497 9 T
Figure 10 Comparison between the heat flux load calculated
from experimental parameters and the heat flux histories estimated
from the thermal property curves and temperature measurements.
Table 7 lists the relative uncertainties related to all
error sources (numerical and experimental) consid-
ered in the accuracy analysis. The total uncertainty
(Utotal) associated with simultaneously estimating k
and cp as a function of temperature is 3.7%. This
satisfactorily low value implies that the IHCP tech-
nique provides k and cp with sufficient accuracy to
enable reliable thermal analyses involving quenched
and tempered 1045 steel.
Conclusions
In this study, the temperature-dependent thermal
properties of 1045 steel were simultaneously esti-
mated after quenching and tempering at 250, 400, and
550 C. An inverse heat conduction approach was
used to determine k and cp estimates that were then
associated with microstructural aspects. Thermal
conductivity, which is lower for complete martensite,
is increased by tempering. Although a reasonable
rise, increasing tempering temperature has a limited
effect on k. Variations in k were associated with two
factors: (i) for quenched martensite matrix: dissolved
alloying elements and high dislocation density; (ii)
for tempered martensite matrix: carbon loss, which is
followed by tetragonality depletion, and carbide
precipitation, which are also influenced by dissolved
alloying elements. The thermal conductivity of this
heat-treated carbon steel was shown to be impacted
not only by the chemical composition, alloying
J Mater Sci (2023) 58:1905–1924 1921

Table 7 Uncertainties associated with the simultaneous estimation of k and cp

Error source Individual Average value of the evaluated Relative

uncertainty parameter uncertainty [-]

Electrical resistance measurement 10–1 X 333.3 X 3:00  104

Electric current measurement 10–2 A 0.34 A 2:94  102
Roughness measurement 10–2 lm 0.51 lm 1:96  102
Weight measurement 10–2 g 194.91 g 5:13  105
Dimension measurement 0.05 mm 36.67 mm 1:36  103
Hardness measurement 10–1 HV 318.4 HV 3:14  104
PARDISO tolerance 10–8 C 140 C 7:14  1011
BDF tolerance 10–2 C 140 C 7:14  105
Thermocouple calibration
(PT-100 sensor resolution) 10–2 C 140 C 7:14  105
Temperature reading
(Reference junction compensation, voltage signal measurement, 10–1 C 140 C 7:14  104
and data fluctuation)
Data acquisition resolution 10–2 C 140 C 7:14  105
Thermal insulation deficiency 1.6 C 140 C 1:14  102
L-M tolerance 10–3 C 140 C 7:14  106

elements, and hardness but also by the tempering
temperature and related microstructural characteris-
tics. Values of cp were shown to vary mildly with the
different conditions of water quenching and tem-
pering. This was attributed mainly to the decrease of
the stored energy in the martensite matrix. Regarding
both thermal properties, the estimates for the nor-
malized condition have values higher than those for
quenched and tempered conditions, besides being in
good agreement with literature data.
1045 steel is a widely used non-alloyed carbon steel
and provides a reasonably large range of properties
when quenched and tempered. Although less deci-
sive than some other aspects, Q&T heat treatment
ends up playing a significant role when it comes to
the thermal properties of heat-treatable martensitic
steels. The thermal property equations determined in
this study are useful for the engineering design of
temperature-varying applications, since considering
thermal properties as constants rather than functions
of temperature can lead to inaccurate and out-of-re-
ality thermal analyses/simulations.
Future research work should focus on applying the
proposed technique to different metallic materials
subjected to heat treatments other than quenching
and tempering. Additionally, as the current experi-
mental setup provides reliable k and cp estimates up
to 140 C, obtaining an experimental setup capable of
handling higher temperature ranges would be
important.
Acknowledgements
The authors are thankful for the financial support
provided by the Brazilian agencies CNPq, CAPES,
and FAPEMIG.
Data availability
The data that support the findings of this study are
available from the corresponding author upon rea-
sonable request.
Declarations
Conflicts of interest The authors have no conflicts
of interest to disclose.
Ethical statement This study does not involve
human participants, animal experimentation or bio-
logical material.

Content courtesy of Springer Nature, terms of use apply. Rights reserved.

1922
References
[1] Cross JO, Opila RL, Boyd IW, Kaufmann EN (2015)
Materials characterization and the evolution of materials.
MRS Bull 40:1019–1034. https://doi.org/10.1557/mrs.2015.
271
[2] Kempen PJ, Kircher MF, de la Zerda A, Zavaleta CL, Jokerst
JV, Mellinghoff IK, Gambhir SS, Sinclair R (2015) A cor-
relative optical microscopy and scanning electron micro-
scopy approach to locating nanoparticles in brain tumors.
Micron 68:70–76. https://doi.org/10.1016/j.micron.2014.09.
004
[3] Charles HK, Weiner JA, Blum NA (1985) Materials char-
acterization and analysis: applications to microelectronics.
Johns Hopkins APL Tech Dig Appl Phys Lab 6:237–249
[4] Ebnesajjad S, (2014) Chapter 4—Surface and material
characterization techniques surface. In: Treatment of mate-
rials for adhesive bonding, 2nd ed, 39–75. William Andrew
Publishing, Oxford. https://doi.org/10.1016/B978-0-323-26
435-8.00004-6
[5] Minor AM, Dehm G (2019) Advances in in situ nanome-
chanical testing. MRS Bull 44:438–442. https://doi.org/10.
1557/mrs.2019.127
[6] Buck W, Rudtsch S (2006) Thermal properties—measure-
ment methods for materials properties. In: Czichos H Saito T
Smith L (eds.) Springer handbook of materials measurement
methods, 399–428. Springer https://doi.org/10.1007/978-3-
540-30300-8.
[7] Martı́nez A, Carmona M, Cortés C, Arauzo I (2020) Char-
acterization of thermophysical properties of phase change
materials using unconventional experimental technologies.
Energies 13:4687. https://doi.org/10.3390/en13184687
[8] Gustafsson SE, Karawacki E, Khan MN (1979) Transient
hot-strip method for simultaneous measuring thermal con-
ductivity, thermal diffusivity of solids and fluids. J Appl
Phys 12:1411–1421. https://doi.org/10.1088/0022-3727/12/
9/003
[9] Parker WJ, Jenkins JR, Butler PC, Abbott BL (1961) Flash
method of determining thermal diffusivity, heat capacity and
thermal conductivity. J Appl Phys 32:1679–1684. https://doi.
org/10.1063/1.1728417
[10] Ramos NP, Carollo LFS, Lima e Silva SMM (2020) Contact
resistance analysis applied to simultaneous estimation of
thermal properties of metals. Meas Sci Technol. https://doi.
org/10.1088/1361-6501/ab8e6a
[11] Ramos NP, De Melo Antunes M, Guimarães G, Lima e Silva
SMM (2022) Simultaneous Bayesian estimation of the
temperature-dependent thermal properties of a metal slab
using a three-dimensional transient experimental approach.
Content courtesy of Springer Nature, terms of use apply. Rights reserved.
J Mater Sci (2023) 58:1905–1924
Int J Therm Sci 179:107671. https://doi.org/10.1016/j.ijthe
rmalsci.2022.107671
[12] Gojic M, Kosec L, Matkovic P (1998) The effect of tem-
pering temperature on mechanical properties and
microstructure of low alloy Cr and CrMo steel. J Mater Sci
33:395–403. https://doi.org/10.1023/A:1004375914591
[13] Santofimia MJ, Zhao L, Petrov R, Sietsma J (2008) Char-
acterization of the microstructure obtained by the quenching
and partitioning process in a low-carbon steel. Mater Charact
59:1758–1764. https://doi.org/10.1016/j.matchar.2008.04.
004
[14] Klein S, Weber S, Theisen W (2015) Effect of heat treatment
on phase structure and thermal conductivity of a copper-
infiltrated steel. J Mater Sci 50:3586–3596. https://doi.org/
10.1007/s10853-015-8919-y
[15] Zhou L, Tang G, Ma X, Wang L, Zhang X (2018) Rela-
tionship between microstructure and mechanical properties
of M50 ultra-high strength steel via quenching-partitioning-
tempering process. Mater Charact 146:258–266. https://doi.
org/10.1016/j.matchar.2018.10.009
[16] Babasafari Z, Pan AV, Pahlevani F et al (2021) Effect of
silicon and partitioning temperature on the microstructure
and mechanical properties of high-carbon steel in a
quenching and partitioning heat treatment. J Mater Sci
56:15423–15440. https://doi.org/10.1007/s10853-021-0627
0-w
[17] Raju S, Jeya Ganesh B, Rai AK, Mythili R, Saroja S,
Mohandas E, Vijayalakshmi M, Rao KBS, Raj B (2009)
Measurement of transformation temperatures and specific
heat capacity of tungsten added reduced activation ferritic–
martensitic steel. J Nucl Mater 389(3):385–393. https://doi.
org/10.1016/j.jnucmat.2009.02.030
[18] Kuepferle J, Wilzer J, Weber S et al (2015) Thermo-physical
properties of heat-treatable steels in the temperature range
relevant for hot-stamping applications. J Mater Sci
50:2594–2604. https://doi.org/10.1007/s10853-015-8829-z
[19] Nunura CRN, dos Santos CA, Spim JA (2015) Numerical—
experimental correlation of microstructures, cooling rates
and mechanical properties of AISI 1045 steel during the
Jominy end-quench test. Mater Des 76:230–243. https://doi.
org/10.1016/j.matdes.2015.03.031
[20] Wei Zhu H, Xu K, Qin S, Ren Xiao F, Liao B (2018) Effect
of heat treatment on microstructure and properties of 1045
steel modified with (NbTi)C nanoparticles. Mater Sci Eng A
728:175–182. https://doi.org/10.1016/j.msea.2018.05.019
[21] Esfahani AK, Babaei M, Sarrami-Foroushani S (2021) A
numerical model coupling phase transformation to predict
microstructure evolution and residual stress during quench-
ing of 1045 steel. Math Comput Simul 179:1–22. https://doi.
org/10.1016/j.matcom.2020.07.016
J Mater Sci (2023) 58:1905–1924
[22] Wilzer J, Lüdtke F, Weber S, Theisen W (2013) The influ-
ence of heat treatment and resulting microstructures on the
thermophysical properties of martensitic steels. J Mater Sci
48:8483–8492. https://doi.org/10.1007/s10853-013-7665-2
[23] Wilzer J, Küpferle J, Weber S, Theisen W (2014) Temper-
ature-dependent thermal conductivities of non-alloyed and
high-alloyed heat-treatable steels in the temperature range
between 20 and 500 C. J Mater Sci 49:4833–4843. https://d
oi.org/10.1007/s10853-013-7665-2
[24] Klein S, Mujica Roncery L, Walter M, Weber S, Theisen W
(2017) Diffusion processes during cementite precipitation
and their impact on electrical and thermal conductivity of a
heat-treatable steel. J Mater Sci 52(1):375–390. https://doi.
org/10.1007/s10853-016-0338-1
[25] Beck JV, Blackwell B, Clair C (1985) Inverse heat con-
duction: III-posed problems. Wiley. ISBN 0–47108–319- 4
[26] Deng X, Ju D (2013) Modeling and simulation of quenching
and tempering process in steels. Phys Procedia 50:368–374.
https://doi.org/10.1016/j.phpro.2013.11.057
[27] Atkins M (1980) Atlas of continuous cooling transformation
diagrams for engineering steels. ASM International, Mate-
rials Park. ISBN: 087170093X
[28] Bhadeshia HKDH, Chintha AR, Lenka S (2019) Critical
assessment 34: Are v (Hägg), g and e carbides transition-
phases relative to cementite in steels? Mater Sci Technol
35(11):1301–1305. https://doi.org/10.1080/02670836.2019.
1625526
[29] Ramos NP, De Melo Antunes M, Lima e Silva SMM (2021)
An experimental and straightforward approach to simulta-
neously estimate temperature-dependent thermophysical
properties of metallic materials. Int J Therm Sci 166:106960.
https://doi.org/10.1016/j.ijthermalsci.2021.106960
[30] Taylor JR. (1997) An introduction to error analysis: the study
of uncertainties in physical measurement. Science Books.
ISBN 978–0–935–70275–0
[31] D’Alessandro G, de Monte F, Amos DE (2019) Effect of
heat source and imperfect contact on simultaneous estima-
tion of thermal properties of high-conductivity materials.
Math Probl Eng 2019:1–15. https://doi.org/10.1155/2019/
5945413
[32] Grzesik W, Nieslony P (2004) Prediction of friction and heat
flow in machining incorporating thermophysical properties
of the coating—chip interface. Wear 256:108–117. https://d
oi.org/10.1016/S0043-1648(03)00390-9
[33] Abbaschian R, Reed-Hill RE (2009) Physical metallurgy
principles—SI version. Cengage Learning, Stamford ISBN:
0495438510
[34] Kraus W, Nolze G (1996) PowderCell 2.3 for Windows,
POWDER CELL—a program for the representation and
manipulation of crystal structures and calculation of the
Content courtesy of Springer Nature, terms of use apply. Rights reserved.
1923
resulting X-ray powder patterns. J Appl Cryst 29:301–303. h
ttps://doi.org/10.1107/S0021889895014920
[35] Villars P, Calvert L, (1991) Pearson’s handbook of crystal-
lographic data for intermetallic phases. ASM International,
Materials park. ISBN: 0871704161
[36] Bhadeshia H, Honeycombe R (2017) Steels: microstructure
and properties. Butterworth-Heinemann, Oxford. ISBN:
008100270X
[37] Jung M, Lee SJ, Lee YK (2009) Microstructural and
dilatational changes during tempering and tempering kinetics
in martensitic medium-carbon steel. Metall Mater Trans A
40:551–559. https://doi.org/10.1007/s11661-008-9756-2
[38] Costa e Silva ALV, Mei PR (2010) Aços e Ligas Especiais,
São Paulo: blucher. ISBN: 852120518X
[39] Incropera FP, Dewitt DP, Bergman TL, Lavine AS. (2006)
Fundamentals of Heat and mass transfer. John Wiley & Sons.
ISBN 978–0–471–45728–2
[40] Seber GAF, Wild CJ (2003) Nonlinear regression. Wiley-
lnterscience. ISBN 978–0–471–47135–6
[41] Klemens PG, Williams RK (1986) Thermal conductivity of
metals and alloys. Int Met Rev 31:12–19. https://doi.org/10.
1179/imtr.1986.31.1.197
[42] Chuang Y, Schmid R, Chang YA (1985) Magnetic contri-
butions to the thermodynamic functions of pure Ni Co, and
Fe. Metall Mater Trans A 16:153–165. https://doi.org/10.
1007/BF02816041
[43] ASM International Handbook Committee (1991) ASM
handbook vol. 4: heat treating. ASM International. ISBN:
0–87170–379–3
[44] Uher C, (2004) Thermal conductivity of metals. In: Tritt TM
(eds) Thermal conductivity. Physics of solids and liquids.
Springer. https://doi.org/10.1007/0-387-26017-X_2
[45] Geller YA, Moiseev VF, Koltunov AA (1963) Heat con-
ductivity of high-speed cutting steels. Met Sci Heat Treat
5(9):493–497. https://doi.org/10.1007/bf00648958
[46] Reed RP, Clark AF (1983) Materials at low temperatures.
ASM International. ISBN: 0–87170–146–4.
[47] Terada Y, Ohkubo K, Mohri T, Suzuki T (2002) Effects of
alloying additions on thermal conductivity of ferritic iron.
ISIJ Int 42(3):322–324. https://doi.org/10.2355/isijinternatio
nal.42.322
[48] Peet MJ, Hasan HS, Bhadeshia HKDH (2011) Prediction of
thermal conductivity of steel. Int J Heat Mass Transf
54:2602–2608. https://doi.org/10.1016/j.ijheatmasstransfer.2
011.01.025
[49] Wilzer J, Küpferle J, Weber S, Theisen W (2015) Influence
of alloying elements, heat treatment, and temperature on the
thermal conductivity of heat treatable steels. Steel Res Int
86:1234–1241. https://doi.org/10.1002/srin.201400294
1924
[50] Shanina BD, Gavriljuk VG, Konchits AA, Kolesnik SP
(1998) The influence of substitutional atoms upon the elec-
tron structure of the iron-based transition metal alloys.
J Phys Condens Matter 10:1825–1838. https://doi.org/10.1
088/0953-8984/10/8/015
[51] Gavriljuk VG, Shanina BD, Berns H (2000) On the corre-
lation between electron structure and short range atomic
order in iron-based alloys. Acta Mater 48:3879–3893. http
s://doi.org/10.1016/S1359-6454(00)00192-0
[52] Wang G, Li Y (2019) Effects of alloying elements and
temperature on thermal conductivity of ferrite. J Appl Phys
126:125118. https://doi.org/10.1063/1.5115441
[53] Saha DC, Biro E, Gerlich AP, Zhou Y (2020) Martensite
tempering kinetics: effects of dislocation density and heating
rates. Mater Charact 168:110564. https://doi.org/10.1016/j.
matchar.2020.110564
[54] Porter DA, Easterling KE, Sherif MY (2009) Phase trans-
formations in metals and alloys. CRC Press. ISBN: 978–1–
003–01180–4. https://doi.org/10.1201/9781003011804.
[55] Ohmori Y, Tamura I (1992) Epsilon carbide precipitation
during tempering of plain carbon martensite. Metall Mater
Trans A 23:2737–2751. https://doi.org/10.1007/
BF02651753
[56] Ning D, Dai CR, Wu JL, Wang YD, Wang YQ, Jing Y, Sun J
(2021) Carbide precipitation and coarsening kinetics in low
carbon and low alloy steel during quenching and subse-
quently tempering. Mater Charact 176:111111. https://doi.
org/10.1016/j.matchar.2021.111111
[57] Fonstein N (2015) Advanced high strength sheet steels:
physical metallurgy, design, processing, and properties.
Springer International Publishing. ISBN: 978–3–
319–19165–2. https://doi.org/10.1007/978-3-319-19165-2.
[58] Krauss G (2017) Tempering of lath martensite in low and
medium carbon steels: assessment and challenges. Steel Res
Int 88:1–18. https://doi.org/10.1002/srin.201700038
Content courtesy of Springer Nature, terms of use apply. Rights reserved.
J Mater Sci (2023) 58:1905–1924
[59] Ray SK, Mohanty ON (1983) Tem investigation of carbide
precipitation in low carbon steels containing silicon. Trans
Japan Inst Met 24:81–87. https://doi.org/10.2320/matertran
s1960.24.81
[60] Mesquita RA, Barbosa CA, Morales EV, Kestenbach HJ
(2011) Effect of silicon on carbide precipitation after tem-
pering of H11 hot work steels. Metall Mater Trans A
42:461–472. https://doi.org/10.1007/s11661-010-0430-0
[61] Wang KF, Yang HTY (1997) Experimental and computa-
tional study of the quenching of carbon steel. J Manuf Sci
Eng Trans ASME 119:257–265. https://doi.org/10.1115/1.
2831102
[62] Jeyaganesh B, Raju S, Murugesan S, Mohandas E, Vijay-
alakshmi M (2009) A study on the effect of thermal ageing
on the specific-heat characteristics of 9Cr-1Mo-0.1C
(mass%) steel. Int J Thermophys 30:619–634. https://doi.
org/10.1007/s10765-009-0558-6
[63] Cezairliyan A, Ho CY, Anderson AC (1988) Specific heat of
solids. Cindas data series on material properties, 1–2.
Hemisphere Publishing Corporation. ISBN: 978–0–
891–16834–8
Publisher’s Note Springer Nature remains neutral with
regard to jurisdictional claims in published maps and
institutional affiliations.
Springer Nature or its licensor (e.g. a society or other
partner) holds exclusive rights to this article under a
publishing agreement with the author(s) or other rightsh-
older(s); author self-archiving of the accepted manuscript
version of this article is solely governed by the terms of
such publishing agreement and applicable law.
Terms and Conditions
Springer Nature journal content, brought to you courtesy of Springer Nature Customer Service Center GmbH (“Springer Nature”).
Springer Nature supports a reasonable amount of sharing of research papers by authors, subscribers and authorised users (“Users”), for small-
scale personal, non-commercial use provided that all copyright, trade and service marks and other proprietary notices are maintained. By
accessing, sharing, receiving or otherwise using the Springer Nature journal content you agree to these terms of use (“Terms”). For these
purposes, Springer Nature considers academic use (by researchers and students) to be non-commercial.
These Terms are supplementary and will apply in addition to any applicable website terms and conditions, a relevant site licence or a personal
subscription. These Terms will prevail over any conflict or ambiguity with regards to the relevant terms, a site licence or a personal subscription
(to the extent of the conflict or ambiguity only). For Creative Commons-licensed articles, the terms of the Creative Commons license used will
apply.
We collect and use personal data to provide access to the Springer Nature journal content. We may also use these personal data internally within
ResearchGate and Springer Nature and as agreed share it, in an anonymised way, for purposes of tracking, analysis and reporting. We will not
otherwise disclose your personal data outside the ResearchGate or the Springer Nature group of companies unless we have your permission as
detailed in the Privacy Policy.
While Users may use the Springer Nature journal content for small scale, personal non-commercial use, it is important to note that Users may
not:

1. use such content for the purpose of providing other users with access on a regular or large scale basis or as a means to circumvent access
control;
2. use such content where to do so would be considered a criminal or statutory offence in any jurisdiction, or gives rise to civil liability, or is
otherwise unlawful;
3. falsely or misleadingly imply or suggest endorsement, approval , sponsorship, or association unless explicitly agreed to by Springer Nature in
writing;
4. use bots or other automated methods to access the content or redirect messages
5. override any security feature or exclusionary protocol; or
6. share the content in order to create substitute for Springer Nature products or services or a systematic database of Springer Nature journal
content.
In line with the restriction against commercial use, Springer Nature does not permit the creation of a product or service that creates revenue,
royalties, rent or income from our content or its inclusion as part of a paid for service or for other commercial gain. Springer Nature journal
content cannot be used for inter-library loans and librarians may not upload Springer Nature journal content on a large scale into their, or any
other, institutional repository.
These terms of use are reviewed regularly and may be amended at any time. Springer Nature is not obligated to publish any information or
content on this website and may remove it or features or functionality at our sole discretion, at any time with or without notice. Springer Nature
may revoke this licence to you at any time and remove access to any copies of the Springer Nature journal content which have been saved.
To the fullest extent permitted by law, Springer Nature makes no warranties, representations or guarantees to Users, either express or implied
with respect to the Springer nature journal content and all parties disclaim and waive any implied warranties or warranties imposed by law,
including merchantability or fitness for any particular purpose.
Please note that these rights do not automatically extend to content, data or other material published by Springer Nature that may be licensed
from third parties.
If you would like to use or distribute our Springer Nature journal content to a wider audience or on a regular basis or in any other manner not
expressly permitted by these Terms, please contact Springer Nature at

onlineservice@springernature.com