THERMAL CONDUCTIVITY OF OXIDE SCALE AND ITS COMPONENTS IN THE RANGE FROM 0 ºC TO 1300 ºC:

GENERALIZED ESTIMATES WITH ACCOUNT FOR MOVABILITY OF PHASE TRANSITIONS

Emmanuil Beygelzimer1, Yan Beygelzimer2

1
OMD-Engineering LLC, Dnipro, Ukraine
2
Donetsk Institute for Physics and Engineering named after A.A. Galkin, National Academy of Sciences of Ukraine, Kyiv, Ukraine

*Corresponding author: emmanuilomd@gmail.com, tel.: +380 (50) 368-63-42, 49000, Volodymyr Monomakh Street, 6 of. 303, Dnipro, Ukraine.

ABSTRACT

The data of different authors on the thermal conductivity of wüstite Fe1-xO, magnetite Fe3O4, hematite Fe2O3 and pure iron are systematized. The generalized values are
described by piecewise smooth functions containing as varying parameters the temperatures of magnetic and polymorphic (for iron) transformations as well as the
thermodynamic stability boundary (for wüstite). At polymorphic transformation a finite break of the thermal conductivity function is envisaged, at other critical points
only a break of its temperature derivative is acceptable. The proposed formulas are presented in two forms: the general form that allow varying the values of critical
temperatures and the particular form corresponding to their basic values: the boundary of thermodynamic stability of wüstite - 570 ºС, the Curie points of magnetite - 575
ºС, of hematite - 677 ºС, of iron – 770 ºС and polymorphic transformation temperature of iron - 912ºС. To calculate the thermal conductivity of oxide scale as a whole, it
is proposed to take into account separately metallic iron and composite matrix of iron oxides. Model computations using the proposed formulas shows that the true thermal
conductivity of oxide scale (without pores), depending on the temperature may be from 3 to 6 W·m-1·K-1 in the absence of metallic iron and up to 15 W·m-1·K-1 if free
iron is released in the eutectoid decay of wüstite. The effective thermal conductivity of oxide scale, taking into account its real porosity, may be lower by 15-35%. The
obtained dependencies are recommended for use in mathematical simulation of production and processing of steel products in the presence of oxide scale on their surface.

Keywords: thermal conductivity; wüstite; magnetite; hematite; oxide scale; Curie point

INTRODUCTION their basic values, adopted for a particular form of approximating formulas, are
given in [1]. As a key assumption, it is considered that if the critical temperature
This report completes a series of publications of the authors on the mathematical changes within the real range of its movability, the values of thermal conductivity
description of the thermophysical properties of wüstite Fe1-xO (x ≈ 0.05...0.15), at this critical point and at two distant reference points on different sides of it
magnetite Fe3O4, hematite Fe2O3, and metallic iron Fe, as well as oxide scale in remain unchanged. In other words, it is assumed that as the critical point changes,
general in the temperature range from 0 °С to 1300 °С. The thermal expansion the temperature dependence graph of the thermal conductivity shifts parallel to the
coefficient [1], density [2] and heat capacity [3] have been described earlier. In this temperature axis, remaining fixed at the extreme reference points. Accordingly,
paper, we look at the thermal conductivity coefficient, which, as usual, is when the critical temperature varies, the other parameters of the approximating
understood as the coefficient of proportionality between the vector of heat flux functions change automatically.
density and the temperature gradient. Conjugation of approximating formulas between intervals separated by critical
When approximating the known data on thermophysical properties, the authors set points was performed taking into account the accepted "behavior model" of
themselves two main tasks (see [1]): 1) to consider the movability of phase thermal conductivity in critical states [1]: with a break of the function during the
transitions in order to use it for adaptation of mathematical models, and 2) to polymorphic transformation and with a break of the first derivative function (but
generalize the results of various studies in order to recommend the most without breaking the thermal conductivity coefficient itself) at the Curie points and
characteristic values of the properties at different temperatures for a given at the boundary of thermodynamic stability.
substance. The first task was solved by including the temperatures of phase The type of approximating formulas was chosen according to the value of specific
transitions in the approximating functions as varying parameters. The second thermal resistance k [m‧K·W-1], the inverse of the thermal conductivity coefficient
problem (generalization) was solved by choosing the type of approximating λ [W·m-1·K-1] (it turned out to be more convenient, the idea was borrowed from
function and the coordinates of the reference points through which the graph of [5].):
this function should pass. In this respect, an essential feature of the studies of
1
thermal conductivity presented in this article is related to the fact that the empirical 𝑘= (1)
data on this characteristic have the largest scatter among all thermophysical 𝜆
properties. For example, for any of the considered iron oxides, the data of different Therefore, the data on the thermal conductivity coefficient known from the
authors on the thermal conductivity may differ by 2-4 times. In this regard, the technical literature was first converted into the values of specific thermal resistance
choice of coordinates of reference points for the thermal conductivity coefficient and only then approximated. Accordingly, the ordinates of all reference points for
is to a large extent conditional, because it is possible that new empirical data that the approximating function were expressed in terms of the thermal resistance. The
will be obtained in the future, or the thermal conductivity values in some specific reverse recalculation into the thermal conductivity coefficient was performed at
conditions may not correspond to the choice made by the authors. However, the the very end after obtaining the approximating function for the resistivity.
authors hope that for each of the considered substances the chosen type of Taking into account the large scatter of currently available empirical data (see
approximating functions correctly reflects the nature of the dependence of thermal Introduction), the coordinates of the reference points chosen by the authors should
conductivity on temperature, and the accumulation of new empirical data may be considered as "basic". In the case of new data, the values of these coordinates
require only a correction of the coordinates of reference points, but not a revision can be adjusted without changing the general formulas proposed.
of the approximating functions. When processing the information on iron oxides we focused on the data obtained
on polycrystalline samples, and with minimal porosity (or recalculated to zero
METHODS porosity, as, for example, in [4]). Data on single crystals were used as additional
information.
Approximation of data on the thermal conductivity of the structural components of
oxide scale was carried out according to the general methods described in detail in
the first part [1]. In accordance with these methods, the critical temperatures are
included in the approximating functions in the form of formal parameters. Such
critical temperatures are taken as: Curie points (for Fe3O4, Fe2O3, Fe), polymorphic
transformation temperature (for Fe), thermodynamic stability boundary (Chaudron
point for Fe1-xO). The real ranges of movability of these critical temperatures and

DOI: 1
RESULTS AND DISCUSSION

Magnetite Fe3O4

Data on the thermal conductivity coefficient of Fe3O4 available in the technical

literature are shown in Fig. 1. The approximation was performed separately in two
temperature intervals: below and above the Curie point. The values of the thermal
conductivity of magnetite from [5] were not taken into account, since they are
drastically different from other data. The resulting generalized formulas are
summarized in Table 1 and contain the Curie point (T1) as a formal parameter.
At the basic value of the Curie point of magnetite T1 = 848 K (575 ºC), formulas
from Table 1 are reduced to the particular form:
 in the range of 273 K ≤ T ≤ 848 K
𝜆Fe3O4 = (0,10136 + 2,9321 ∙ 10−4 ∙ 𝑇 − 0,10165 ∙ 𝑇 −2 )−1 (2)
 in the range of 848 K < T ≤ 1573 K
𝜆Fe3O4 = 2,857 (3)
where λFe3O4 [W·m ·K ] is the thermal conductivity coefficient of magnetite; T
-1 -1

[K] is the design temperature.

The graph built by formulas (2)-(3) is shown in Fig. 1. Another graph (denoted
“Approx 2”) in the same figure built by the general formulas from Table 1 with the
same basic value of the Curie point of magnetite, but with other values of reference
points coordinates, namely: T0 = 200 K; k0 = 0.14 m·K·W-1, k1 = 0.28 m·K·W-1, T2
= 1600 K; k2 = 0.24 m·K·W-1 (if we take into account the data of measurements in
work [5] and the tendency to a slight rise in the thermal conductivity coefficient
with increasing temperature above the Curie point, noted in [6; 10]). However, it
should be emphasized that this graph is given only to demonstrate the possibility
Fig. 1. Thermal conductivity coefficient of magnetite (Fe3O4) according to
of the proposed approach; to date, in the absence of other data, for generalized
empirical data of Grosu [6], Kamilov [7], Noda [4], Molgaard (separately at
estimates of the thermal conductivity of magnetite the authors recommends using
pressures 44 and 88 MPa) [8], Slack [9], Akiyama [5], Takeda [10], Faik [11],
the basic values of the coordinates of the reference points given in Table 1.
Dortman [12], Koenig [13, p. 154, array 4]. The conventional symbols in the
According to the formulas of Table 1, the shift of the Curie point of magnetite
legend of the graph: M - monocrystal, P - polycrystal, R - raw ore, S - sintered
within the real limits of its movability can change its thermal conductivity by no
sample. The experimental points are conventionally connected by straight lines.
more than 6-7%. For example, at 500 ºC the calculated value of λFe3O4 is “Approx” denotes the graph built according to the formulas (2)-(3) (with the basic
2.99 W·m-1K-1 if the magnetic transition occurs at T1 = 823 K (550 ºC), and value of the Curie point of magnetite T1 = 848 K (575 ºC) and coordinates of the
3.17 W·m-1K-1 if the Curie point is T1 = 900 K (627 ºC). reference points from Table 1), “Approx 2” is the graph built according to the
formulas from Table 1 with the basic value of the Curie point but with alternative
coordinates of the reference points – see text.

Table 1. Formulas for the thermal conductivity coefficient of Fe3O4 in two temperature intervals (T1 [K] is the Curie point of magnetite)
Temperature interval [K] 273 ≤ 𝑇 ≤ 𝑇1

Approximating function [W·m-1K-1] 𝜆Fe3O4 = (𝑎0 + 𝑎1 𝑇 𝑛 + 𝑎2 𝑇 𝑚 )−1

Constants 𝑛 = 1.0 𝑚 = -2.0

Basic coordinates of the reference points T0 = 200 K k0 = 0.16 m·K·W-1 k1 = 0.35 m·K·W-1
𝑘1 𝑇0𝑚 + 𝑘0 𝑇1𝑛 − 𝑘0 𝑇1𝑚 − 𝑘1 𝑇0𝑛
𝑎0 =
𝑇0𝑛 𝑇1𝑚 − 𝑇1𝑛 𝑇0𝑚 − 𝑇1𝑚 + 𝑇0𝑚 + 𝑇1𝑛 − 𝑇0𝑛

𝑘0 𝑇1𝑚 − 𝑘1 𝑇0𝑚 − 𝑎0 (𝑇1𝑚 − 𝑇0𝑚 )

Coefficients to be calculated 𝑎1 =
𝑇0𝑛 𝑇1𝑚 − 𝑇1𝑛 𝑇0𝑚

𝑘0 𝑇1𝑛 − 𝑘1 𝑇0𝑛 − 𝑎0 (𝑇1𝑛 − 𝑇0𝑛 )

𝑎2 =
𝑇0𝑚 𝑇1𝑛 − 𝑇1𝑚 𝑇0𝑛

Temperature interval [K] 𝑇1 < 𝑇 ≤ 1573

−1
Approximating function [W·m-1K-1] 𝜆Fe3O4 = (𝑏0 + 𝑏1 (𝑇 − 𝑇1 ))
Basic coordinates of the reference points T2 = 1600 K k2 = 0.35 m·K·W-1

𝑏0 = 𝑘1
Coefficients to be calculated 𝑘2 − 𝑘1
𝑏1 =
𝑇2 − 𝑇1

2
Wüstite Fe1-xO

Data on the thermal conductivity coefficient of wüstite available in the technical

literature are shown in Fig. 2. Taking into account the results of [5], the
approximation was performed separately in two temperature intervals: before and
after the Chaudron point (the authors of [5] correlated the kink in the temperature
dependence of the thermal conductivity of wustite not with the Chaudron point,
but with its Tammann temperature (825 K/552 ºC), equal to half of the melting
point in Kelvin degrees). Table 2 shows the formulas in general form with the
Chaudron point T1 as a formal parameter. At the basic value of T1 = 843 K (570
°С) these formulas take the particular form (the corresponding graph is shown in
Fig. 2):
 in the range of 273 K ≤ T ≤ 843 K
𝜆FeO = (2.7054 ∙ 10−2 + 9.4008 ∙ 10−3 ∙ 𝑇 0.5 − 3.6455 ∙ 𝑇 −2 )−1 (4)
 in the range of 843 K < T ≤ 1573 K
−1
𝜆FeO = (0.3 − 7.9260 ∙ 10−5 (𝑇 − 843)) (5)
where λFeO [W·m ·K ] is the thermal conductivity coefficient of wüstite; T [K] is
-1 -1

the design temperature.

Fig. 2. Thermal conductivity coefficient of wüstite according to Akiyama [5],

Takeda [10], Li (separately for heating and cooling) [14, p. 2104], Slowik [15].
“Approx” - graph built by approximating formulas (4)-(5).

Table 2. Formulas for calculating the thermal conductivity coefficient of wüstite in two temperature intervals (T1 [K] is the Chaudron point)

Temperature interval [K] 273 ≤ 𝑇 ≤ 𝑇1

Approximating function [W·m-1K-1] 𝜆FeO = (𝑎0 + 𝑎1 𝑇 𝑛 + 𝑎2 𝑇 𝑚 )−1

Constants n = 0.5 m = -2.0

Basic coordinates of the reference points T0 = 200 K k0 = 0.16 m·K·W -1
k1 = 0,3 m·K·W-1

𝑘1 𝑇0𝑚 + 𝑘0 𝑇1𝑛 − 𝑘0 𝑇1𝑚 − 𝑘1 𝑇0𝑛

𝑎0 =
𝑇0 𝑇1 − 𝑇1𝑛 𝑇0𝑚 − 𝑇1𝑚 + 𝑇0𝑚 + 𝑇1𝑛 − 𝑇0𝑛
𝑛 𝑚

𝑘0 𝑇1𝑚 − 𝑘1 𝑇0𝑚 − 𝑎0 (𝑇1𝑚 − 𝑇0𝑚 )

Coefficients to be calculated 𝑎1 =
𝑇0𝑛 𝑇1𝑚 − 𝑇1𝑛 𝑇0𝑚

𝑘0 𝑇1𝑛 − 𝑘1 𝑇0𝑛 − 𝑎0 (𝑇1𝑛 − 𝑇0𝑛 )

𝑎2 =
𝑇0𝑚 𝑇1𝑛 − 𝑇1𝑚 𝑇0𝑛

Temperature interval [K] 𝑇1 < 𝑇 ≤ 1573

−1
Approximating function [W·m-1K-1] 𝜆FeO = (𝑏0 + 𝑏1 (𝑇 − 𝑇1 ))

Basic coordinates of the reference points T2 = 1600 K k2 = 0.24 m·K·W-1

𝑏0 = 𝑘1
Coefficients to be calculated 𝑘2 − 𝑘1
𝑏1 =
𝑇2 − 𝑇1

Hematite Fe2O3 −1
𝜆Fe2O3 = (0.25 − 2.6667 ∙ 10−4 (950 − 𝑇)) (6)
Known from the technical literature data on the thermal conductivity of hematite  in the range of 950 K < T ≤ 1573 K
are shown in Fig. 3. The approximating formulas are summarized in Table 3 and −1
contain the Curie point of hematite as a formal parameter T1. At the basic value of 𝜆Fe2O3 = (0.25 + 6.1538 ∙ 10−5 (𝑇 − 950)) (7)
T1 = 950 K (677 ºC) for hematite, these formulas are reduced to the particular form where λFe2O3 [W·m-1·K-1] is the thermal conductivity coefficient of hematite; T [K]
(the corresponding graph is shown in Fig. 3): is the design temperature.
 in the range of 273 K ≤ T ≤ 950 K According to the formulas in Table 3, hematite has about the same sensitivity of
the thermal conductivity to Curie point movability as that of magnetite, namely no

3
more than 6%. For example, at Curie point T1 = 943 K (670 ºC) the calculated Iron Fe
value of λFe2O3 at this temperature is 4.00 W·m-1·K-1; if the magnetic transition
occurs at T1 = 998 K (725 ºC), then the calculated value of λFe2O3 at the same Experimental data of different authors on the thermal conductivity coefficient of
temperature of 670 ºC is 4.23 W·m-1·K-1. iron are systematized in Fig. 4. Approximating formulas in general form are given
in Table 4 and contain the following defining parameters: T [K] is design
temperature; T1 [K] is the Curie point of iron; T2 [K] is the temperature of
polymorphic transformation (γ→α transition at cooling).
Calculations by formulas of Table 4 show that shifting the Curie point within its
real movability range can change the thermal conductivity coefficient of iron by
no more than 4%. For example, the calculated value of λFe at 750 ºC is
29.6 W·m-1·K-1 if the Curie point is T1 = 1032 K (759 ºC) and 30.7 W·m-1·K-1 if T1
= 1046 K (773 ºC).
The sensitivity of the thermal conductivity coefficient of iron to the variations of
the polymorphic transformation temperature T2 is quite obviously determined by
the jump of λFe at point T2 from 30.0 W·m-1·K-1 in α-phase (below T2) to
27.7 W·m-1·K-1 in γ-phase (above T2).
At the basic values of iron Curie point T1 = 1043 K (770ºC) and polymorphic
transformation T2 = 1185 K (912ºC) formulas from Table 4 are reduced to the
particular form (corresponding graph shown in Figure 4):
 in the range of 273 K ≤ T ≤ 1043 K
𝜆Fe = [7.7 ∙ 10−3 + 9.2122 ∙ 10−6 ∙ 𝑇 1.11 + 6.4624
(8)
∙ 10−3 𝑒 −0.014∙(1043−𝑇) ]−1
 in the range of 1043 K < T ≤ 1185 K
𝜆Fe = [3.3295 ∙ 10−2 + 1.5051 ∙ 10−3 𝑒 −0.04∙(𝑇−1043) ]−1 (9)
 in the range of 1185 K < T ≤ 1573 K:
𝜆Fe = [2.7804 ∙ 10−2 + 1.6359 ∙ 1010 ∙ 𝑇 −4 ]−1 (10)
where λFe [W·m ·K ] is the thermal conductivity coefficient of iron; T [K] is the
-1 -1

Fig. 3. Thermal conductivity coefficient of hematite according to empirical data design temperature.
from Noda [4], Takeda [10], Akiyama [5], Molgaard (separately at pressures 44
and 88 MPa) [8], Muta [6], Dortman [12], Koenigsberger (cited from [4]) and
Smoke [16, p. 64]. Notation: M - monocrystal (a, c - crystal axes), P - polycrystal,
S - sintered. The experimental points are conventionally connected by straight
lines. The graph built by formulas (6)-(7) is shown (denoted as “Approx”)

Table 3. Formulas for calculating the thermal conductivity coefficient of hematite in two temperature intervals (T1 [K] is the Curie point)

Temperature interval [K] 273 ≤ 𝑇 ≤ 𝑇1

−1
Approximating function [W·m-1K-1] 𝜆Fe2O3 = (𝑎0 + 𝑎1 (𝑇1 − 𝑇))

Basic coordinates of the reference points T0 = 200 K k0 = 0.05 m·K·W-1 k1 = 0.25 m·K·W-1
𝑎0 = 𝑘1

Coefficients to be calculated 𝑘0 − 𝑘1
𝑎1 =
𝑇1 − 𝑇0

Temperature interval [K] 𝑇1 < 𝑇 ≤ 1573 К

−1
Approximating function [W·m-1K-1] 𝜆Fe2O3 = (𝑏0 + 𝑏1 (𝑇 − 𝑇1 ))

Basic coordinates of the reference points T2 = 1600 K k2 = 0.29 m·K·W-1

𝑏0 = 𝑘1

Coefficients to be calculated 𝑘2 − 𝑘1
𝑏1 =
𝑇2 − 𝑇1

4
 Fig. 4. Thermal conductivity coefficient of pure iron according to Fulkerson [17], Powell [18, p. 35], Adcock [19, p. 156, set 3], Fieldhouse
[19, p. 156, set 6], Mikrukov [20], Lucks [19, p. 156, set 12], Jaeger [19, p. 156, set 15], Honda [19, p. 156, set 24], Maurer [19, c. 156, sets
34 and 35], Ranque [19, c. 156, set 47], Taylor [19, c. 156, set 62], Akiyama [5], Armstrong [21], Eucken [19, c. 156, sets 7, 8 and 9],
Shanks [22], Laubitz [23] and the generalized recommendations of Touloukian [16, p. 15]. “Approx” is the graph calculated by (8)-(10).

Oxide scale as a whole

Fig. 5 compares temperature dependences of thermal conductivity coefficient of

different structural components of oxide scale calculated according to the above
formulas for basic values of critical temperatures. According to these generalized
data the thermal conductivity of iron oxides in the temperature range from 0 to
1300 °С is approximately 10 times lower than that of iron itself. At temperatures
from 600 to 1300 °С thermal conductivities of all iron oxides are approximately
equal, and in the range from 0 to 600 °С thermal conductivity of hematite is
approximately 1.5-3 times higher than that of wüstite and magnetite. For oxide
scale in general, two aspects should be considered: 1) the effect of each solid
component of scale on its overall thermal conductivity and 2) the porosity of oxide
scale.
In order to assess the effect of the individual components of the oxide scale on its
overall thermal conductivity, we first consider a composite material consisting of
only two components. The component with the higher thermal conductivity will
be assigned an ordinal number "1". In other words, we will assume that:
𝜆1 ≥ 𝜆2 (11)
where λ1 and λ2 are thermal conductivity coefficients of the components.
Let us denote the volume fractions of the components by symbol φ with the
corresponding index, whereby:
𝜑1 + 𝜑2 = 1 (12) Fig. 5. Comparison of thermal conductivity coefficient of wüstite, magnetite,
hematite and iron depending on temperature. Calculations were made by
formulas (4)-(5), (2)-(3), (6)-(7) and (8)-(10), respectively.

5
 Table 4. Formulas for the thermal conductivity coefficient of iron (T1 [K] is the Curie point, T2 [K] is the temperature of polymorphic transformation)
Temperature interval [K] 273 ≤ 𝑇 ≤ 𝑇1
−1
Approximating function [W·m-1K-1] 𝜆Fe = (𝑎0 + 𝑎1 𝑇 𝑛 + 𝑎3 𝑒 −𝑎4 (𝑇1−𝑇) )

Constants n = 1.11 a0 = 0.0077 a4 = 0.014

Basic coordinates of the reference points T0 = 200 K k0 = 0.011 m·K·W -1
k1 = 0.0348 m·K·W-1

𝑊1 (𝑘1 − 𝑎0 ) + 𝑎0 − 𝑘0
𝑎1 =
𝑊1 𝑇1𝑛 − 𝑇0𝑛
Coefficients to be calculated

𝑎3 = 𝑘1 − 𝑎0 − 𝑎1 𝑇1𝑛

Auxiliary parameter 𝑊1 = 𝑒 −𝑎4 (𝑇1 −𝑇0 )

Temperature interval [K] 𝑇1 < 𝑇 ≤ 𝑇2

−1
Approximating function [W·m-1K-1] 𝜆Fe = (𝑏0 + 𝑏3 𝑒 −𝑏4 (𝑇−𝑇1 ) )

Constants b4 = 0.04
Basic coordinates of the reference points k2(2) = 0.0333 m·K·W-1

𝑘1 − 𝑘2(2)
𝑏3 =
Coefficients to be calculated 1 − 𝑊2

𝑏0 = 𝑘1 − 𝑏3

Auxiliary parameter 𝑊2 = 𝑒 −𝑏4(𝑇2 −𝑇1 )

Temperature interval [K] 𝑇2 < 𝑇 ≤ 1573

Approximating function [W·m-1K-1] 𝜆Fe = (𝑑0 + 𝑑1 𝑇 𝑝 )−1

Constants p = -4
Basic coordinates of the reference points T3 = 1600 K k2(3) = 0.0361 m·K·W-1 k3 = 0.0303 m·K·W-1

𝑘3 − 𝑘2(3)
𝑑1 = 𝑝 𝑝
𝑇3 − 𝑇2
Coefficients to be calculated
𝑝
𝑑0 = 𝑘2(3) − 𝑑1 𝑇2

It is known that the total thermal conductivity of a composite lies between two 𝜆 𝜑2
=1− (15)
limit values which correspond to the cases of parallel and sequential arrangement 𝜆1 𝜆1 𝜑
− 1
of components relative to the direction of the heat flow [24]. In the "parallel 𝜆1 − 𝜆2 3
connection" model, i.e., when the components are arranged in such a way that the
boundary surface between them is parallel to the heat flow, the composite thermal For further analysis it is convenient to enter relative parameters:
conductivity λ is determined as the arithmetic mean value (this is the upper  the relative heat transfer coefficient of the second component:
estimate): 𝜆2
𝑌2 = ≤ 1,0 (16)
𝜆 = 𝜆1 𝜑1 + 𝜆2 𝜑2 (13) 𝜆1

In the " sequential connection" model, i.e., when the components are arranged one  relative thermal conductivity coefficient of the composite material
after the other across the heat flow, the thermal resistances are averaged (this is the 𝜆
lower estimate): 𝑌= ≤ 1,0 (17)
𝜆1
1 𝜑1 𝜑2 In these relative parameters, the above calculated models of the total thermal
= + (14)
𝜆 𝜆1 𝜆2 conductivity of the composite are written as follows:
For composite materials in which one of the components is present in the form of  parallel connection model (upper estimate, analogous to (13)):
chaotically arranged closed inclusions in a matrix of the second component, in
practice the formula of Odelevski [25; 26, p. 24] is applied, which gives a result 𝑌 = 𝜑1 + 𝑌2 (1 − 𝜑1 ) (18)
intermediate between models (13) and (14):
 sequential connection model (lower estimate, analogous to (14)):

6
 1) Oxide volume fractions excluding metallic iron
1 − 𝜑1 −1
𝑌 = [𝜑1 + ] (19) 𝜓FeO 𝜓Fe3O4 𝜓Fe2O3
𝑌2 𝜔FeO = ;𝜔 = ; 𝜔Fe2O3 = ;
1 − 𝜓Fe Fe3O4 1 − 𝜓Fe 1 − 𝜓Fe
 Odelevski model (analogous to (15)): 𝜓xO (21)
𝜔xO =
1 − 𝜑1 1 − 𝜓Fe
𝑌 = 1−
1 𝜑
− 1
(20) where the symbol ω with the index denotes the volume fraction of the
1 − 𝑌2 3 corresponding oxide in the solid material of oxide scale without metallic iron (the
index "xO" means the oxide of the alloying element); the symbol ψ with the index
Fig. 6 shows the graphs built by formulas (18)-(20). They give a visual assessment
denotes the volume fraction of the corresponding component in the solid material
of the influence of the relative thermal conductivity and the volume fraction of the
of oxide scale, including metallic iron. The indicated volume fractions meet the
components at different calculated models of the total thermal conductivity of the
conditions of rationing:
composite. It can be seen that if the thermal conductivity coefficients of the
individual components are close to each other (Y2 = 0.9 corresponds to this case in
𝜓FeO + 𝜓Fe3O4 + 𝜓Fe2O3 + 𝜓xO + 𝜓Fe = 1 (22)
Fig. 6), then all models give approximately the same composite thermal
conductivity result. If the thermal conductivities of the components differ by an
and
order of magnitude (the case Y2 = 0.1 in Fig. 6), then the range of results between
the limiting models becomes very large, and the calculated estimate of the total
thermal conductivity of the composite significantly depends on the selected model. 𝜔FeO + 𝜔Fe3O4 + 𝜔Fe2O3 + 𝜔xO = 1 (23)
It is indicative that the sequential model turns out to be insensitive to a small
volume fraction of the first component, and within its content up to 20% (φ1 < 0.2) 2) Cumulative thermal conductivity coefficient of the oxide matrix (according
is determined by the least thermally conductive component. to the model of sequential connection):
𝜔FeO 𝜔Fe3O4 𝜔Fe2O3 𝜔xO −1
𝜆𝑜𝑥 = [ + + + ] (24)
𝜆FeO 𝜆Fe3O4 𝜆Fe2O3 𝜆xO
where in the right part, the symbol λ with an index indicates the thermal
conductivity coefficient of the corresponding oxide (for example, for iron oxides,
according to the formulas given earlier in this article);
3) Total heat transfer coefficient of oxide scale without pores (according to
Odelevski model):

1 − 𝜓Fe
𝜆𝑠𝑐 = 𝜆Fe [1 − ] (25)
𝜆Fe 𝜓
− Fe
𝜆Fe − 𝜆𝑜𝑥 3
where λFe is the thermal conductivity coefficient of iron (e.g., from the formulas of
Table 4 or (8)-(10)); ψFe is the volume fraction of metallic iron in the oxide scale;
λox is the total thermal conductivity coefficient of the oxide matrix (see (24)).
It is easy to see that the proposed chain of formulas "works" in the absence of
metallic iron in the oxide scale. Then ψFe = 0, and the successive application of
formulas (21)-(25) gives the equality obvious in this case: λsc = λox.
As an example, the graphs of λsc, which are calculated by the formulas (21)-(25)
for the four hypothetical compositions of oxide scale listed in Tables 5, are shown
in Fig. 7. In the first three cases, the same content of components in the entire
temperature range (which in the first approximation can be regarded as
corresponding to the conditions of rapid cooling) is conventionally assumed, and
in the fourth - variable content depending on the temperature (simulating the
conditions of slow cooling). In all cases, the oxides of the alloying elements are
Fig. 6. Calculated models of the relative thermal conductivity of the composite as not taken into account. It can be seen that the main changes in the thermal
a function of the volume fraction of the component with the highest thermal conductivity of scale are associated with metallic iron. Thermal conductivity of
conductivity: 1 - parallel connection model; 2 - sequential connection model; 3 - scale consisting only of iron oxides (composition 1) varies in a narrow range from
Odelevski model. Y2 is the relative thermal conductivity of the second component 3 to 6 W·m-1K-1. Increasing the proportion of metallic iron up to 12% leads to an
(see (16)). For Y2 = 0.9 the model numbers are not shown, since the graphs are increase in the thermal conductivity of scale up to 13-15 W·m-1K-1. This is
practically the same. especially evident for composition 4, when wüstite decays into magnetite and iron
upon cooling.

Applied to our case of surface oxide scale, two of the considered models may be Table 5. Oxide scale hypothetical compositions for the graphs in Fig. 7.
relevant: sequential bonding and Odelevski one. Indeed, the oxides are Volume fraction of component ψ
characterized by a layered arrangement on the surface (with increasing oxygen Composition
Fe1-хO Fe3O4 Fe2O3 Fe
content as the distance from the metal increases, for example, an inner layer of
Fe1-xO, above it - Fe3O4 and an outer layer of Fe2O3). Since the heat flux is usually 1 0,8 0,15 0,05 0
directed perpendicular to the surface, this arrangement of oxides corresponds to a 2 0,5 0,35 0,1 0,05
sequential pattern of overall thermal conductivity of the oxide scale. The situation 3 0,2 0,55 0,15 0,1
is quite different when metallic iron is present in the oxide scale, for example, due
to eutectoid decay of wüstite. In this case, the iron particles can be regarded as 4 changes with temperature according to Fig. 8
chaotically arranged closed inclusions, which corresponds to the Odelevski model.
On this basis, to calculate the thermal conductivity of oxide scale in the general
case, we propose a "combined" model, in which the influence of metallic iron is
accounted for by the Odelevski formula, and oxides - by the model of sequential
connection. In the notations we have adopted, this combined model is written in
the form of the following chain of formulas:

7
 Fig. 9. Relative coefficient of influence of oxide scale porosity η on the value of
its effective thermal conductivity coefficient (calculation by formula (26)). Dotted
Fig. 7. Graphs of temperature dependence of thermal conductivity of scale lines indicate data corresponding to the level of porosity of furnace and secondary
calculated by formulas (21)-(25). Numbers in the curves are the numbers of scale during hot rolling.
calculated compositions of oxide scale according to Table 5.
CONCLUSION

In order to obtain generalized formulas for calculating the thermal conductivity

coefficient for iron oxides and pure iron, the empirical data of different authors are
presented in the form of the inverse value - the specific heat resistance. As
reference points for approximation we used two nodal points beyond the target
range from 0 to 1300 ºС and the critical points connected with sharp change of
properties: for magnetite Fe3O4 and hematite Fe2O3 - Curie points, for wüstite
Fe1-xO - the boundary of its thermodynamic stability (Chaudron point), for pure
iron Fe - Curie point and the temperature of polymorphic α↔γ transformation. As
a result, generalized formulas were obtained that allow the values of critical
temperatures and coordinates of the reference points to be varied. Taking into
account the large scatter of empirical data, the accepted basic values of thermal
conductivity at the reference points may be subject to correction in specific studies.
In this case, the proposed approach makes it possible to automatically consider the
new coordinate values in the estimates of the temperature dependence of the
thermal conductivity coefficient while preserving the general form of the
approximating functions.
The influence of different components of scale on its total thermal conductivity is
proposed to consider in a combined way: metallic iron - by the Odelevski formula
for chaotically located closed inclusions, and oxides - by the model of a sequential
arrangement of layers perpendicular to the direction of the heat flow. Model
Fig. 8. Hypothetical variable oxide scale composition for simulations (number 4 in calculations show that in the absence of metallic iron in oxide scale composition,
Fig. 7). its total thermal conductivity without considering pores can vary in a narrow range
from 3 to 6 W·m-1K-1, depending on the temperature and the percentage of oxides.
The influence of oxide scale porosity on its thermal conductivity coefficient can With an increase in fraction of metallic iron in the oxide scale up to 12% (e.g., as
be considered according to a theoretically justified formula recommended in [27, a result of eutectoid decomposition of wüstite), its thermal conductivity coefficient
p. 249] for porous materials with small, uniformly distributed air inclusions: can increase to 15 W·m-1K-1.
According to the adopted assessment it is shown that with a pore fraction of 5%
2⁄ (which is typical for air scale) its effective thermal conductivity decreases by 15%,
𝜆′𝑠𝑐 = 𝜆𝑠𝑐 (1 − 𝜂 3) (26)
and with a pore fraction of 20% (which is typical for furnace scale during hot
where 𝝀′𝒔𝒄 is the effective thermal conductivity coefficient of oxide scale with pores rolling) - by 35%.
taken into account; λsc is its true thermal conductivity coefficient without The proposed methods are recommended for use in the mathematical modeling of
considering pores (for example, by (21)-(25)); η is the porosity of oxide scale in the production and processing of steel products in the presence of surface oxide
fractions of one, (ratio of the pore volume to the total volume with pores). scale.
Fig. 9 shows the ratio of 𝝀′𝒔𝒄⁄𝝀𝒔𝒄 as a function of oxide scale porosity, calculated
from (26). The calculation for hot rolling conditions shows a decrease in thermal REFERENCES
conductivity due to porosity for furnace scale (η ≥ 0.2) by at least 35%, for air scale
1. E. Beygelzimer, Y. Beygelzimer. Generalized estimates for thermal expansion
(η ≈ 0.05) - by 15%.
of oxide scale in the range from 0 ºC to 1300 ºC with account for movability of
phase transitions in its component, 2021, 10 p.
https://arxiv.org/ftp/arxiv/papers/2110/2110.08528.pdf
2. E. Beygelzimer, Y. Beygelzimer. Generalized estimates for the density of
oxide scale in the range from 0 °C to 1300 °C, 2021, 9 p.
https://arxiv.org/ftp/arxiv/papers/2110/2110.09791.pdf

8
3. E. Beygelzimer, Y. Beygelzimer. Heat capacity of oxide scale in the range process, part 2: determination of material properties of oxide scales on steel under
from 0 °C to 1300 °C: Generalized estimates with account for movability of phase spray-water cooling conditions: Steel research, 61 (7), 1990, 302-311.
transitions, 2021, 8 p. https://arxiv.org/ftp/arxiv/papers/2110/2110.11101.pdf. https://doi.org/10.1002/srin.199000353
4. Y. Noda, K. Naito. The Thermal Conductivity and Diffusivity of MnxFe3-xO4 16. Y. S. Touloukian (Ed.) Recommended Values of the Thermophysical
(0 ≤ x ≤ 1.5) from 200 to 700 K: Netsu So Kutei, 5 (1), 1978, 11-18. Properties of Eight Alloys, Major Constituents and Their Oxides. NBS Sub-
5. T. Akiyama, H. Ohta, R. Takahashi, Waseda Y., Yagi J.-i. Measurement and Contract No. CST-7590, NASA Order R-45. Purdue University. Lafayette, Ind.,
Modeling of lron Ore Thermal Conductivity for Dense Iron Oxide and Porous Iron 1966, 549 p.
Ore Agglomerates in Stepwise Reduction: ISIJ International, 32 (7), 1992, 829- 17. W. Fulkerson, J. P. Moore, D. L. McElroy. Comparison of the Thermal
837. Conductivity, Electrical Resistivity, and Seebeck Coefficient of a High-Purity Iron
6. Y. Grosu, A. Faik, I. Ortega-Fernández, B. D'Aguanno. Natural Magnetite for and Armco Iron to 1000 ºC: Journal of Applied Physics, 37, 1966, 2639.
thermal energy storage: Excellent thermophysical properties, reversible latent heat https://doi.org/10.1063/1.1782098
transition and controlled thermal conductivity: Solar Energy Materials & Solar 18. R. W. Powell, C. Y. Ho, P. E. Liley. Thermal Conductivity of Selected
Cells, 161, 2017, 170-176. https://doi.org/10.1016/j.solmat.2016.12.006 Materials. US Department of Commerce National Standard Reference Data Series,
7. I. K. Kamilov, G. M. Shakhshaev, Kh. K. Aliev, G. G. Musaev, M. M. 1966, Nov. 25, 180 p.
Khamidov. Some features of the behavior of the thermal conductivity of ferrites in 19. Y. S. Touloukian, R. W. Powell, C. Y. Ho, P. G. Klemens. Thermophysical
the vicinity of magnetic phase transitions: Zh. Eksp. Theor. Phys, 41 (2), 1975, Properties of Matter. Vol. 1. Thermal Conductivity. Metallic Elements and Alloys.
290-296. IFI/Plenum. New York-Washington, 1970, 1588 p.
8. J. Molgaard, W. W. Smeltzer. Thermal Conductivity of Magnetite and 20. V. E. Mikryukov. Thermal conductivity and electrical conductivity of metals
Hematite: J. Appl. Physics, 42 (9), 1971, 3644-3647. and alloys. Moscow: State publishing house for ferrous and non-ferrous
https://doi.org/10.1063/1.1660785 metallurgy, 1959, 260 p. (in Russian).
9. G. A. Slack. Thermal Conductivity of MgO, Al2O3, MgAl2O4, and Fe3O4 21. L. D. Armstrong, T. M. Dauphinee. Thermal Conductivity of Metals at High
Crystals from 3° to 300° K: Physical review, 126 (2), 1962, 427-441. Temperatures: Canadian Journal of Research, Sec. A., 25, 1947, 357-374.
https://doi.org/10.1103/PhysRev.126.427 https://doi.org/10.1139/cjr47a-030
10. M. Takeda, T. Onishi, Sh. Nakakubo, Sh. Fujimoto. Physical Properties of 22. H. R. Shanks, A. H. Klein, G. C. Danielson. Thermal Properties of Armco
Iron-Oxide Scales on Si-Containing Steels at High Temperature: The Japan Iron: Journal of Applied Physics, 38 (7), 1967, 2885-2892.
Institute of Metals. Materials Transactions, 50 (9), 2009, 2242-2246. https://doi.org/10.1063/1.1710018
https://doi.org/10.2320/matertrans.M2009097 23. M. J. Laubitz. The Unmatched Guard Method of Measuring Thermal
11. A. Faik. Development of an innovative thermal energy storage (TES) solution Conductivity At High Temperatures: Canadian Journal of Physics, 41, 1963, 1663-
for middle-size concentrated solar power (CSP) plant: Conference Internationale 1678. https://doi.org/10.1139/p73-163
Derbi, June 14th, 2017. Perpignan, 2017. 18 p. 24. M. O. W. Richardson (Ed.) Polymer engineering composites: Applied Science
12. N. B. Dortman (Ed.). Physical properties of rocks and minerals (petrophysics). Publishers Ltd., London, 1977.
Geophysics Handbook. Moscow: Nedra, 1984, 455 p. (in Russian). 25. V. I. Odelevski. Calculation of the generalized conductivity of heterogeneous
13. Y. S. Touloukian, R. W. Powell, C. Y. Ho, P. G. Klemens. Thermophysical systems. Zh Tech Physics, 21 (6), 1951, 667–685 (in Russian).
Properties of Matter. Vol. 2. Thermal Conductivity. Nonmetallic Solids. 26. G. N. Dulnev, Yu. P. Zarichnyak. Thermal Conductivity of Mixtures and
IFI/Plenum. New York-Washington, 1971, 1286 p. Composite Materials: A Reference Book. Leningrad: Energiya, 1974, 264 p. (in
14. M. Li, M. Endo, M. Susa. Temperature Dependence of Thermal Diffusivity Russian).
and Conductivity of FeO Scale Produced on Iron by Thermal Oxidation: ISIJ 27. A. Missenard. Conductivité thermique des solids, liquids, gaz et de leurs
International, 57 (12), 2017, 2097–2106. mélanges. Paris: Editions Eyrolles, 1965, 464 p.
https://doi.org/10.2355/isijinternational.ISIJINT-2017-301
15. J. Slowik, G. Borchardt, Ch. Kohler, R. Jeschar, R. Scholz. Influence of oxide
scales on heat transfer in secondary cooling zones in the continuous casting