AISTech 2019 — Proceedings of the Iron & Steel Technology Conference

6–9 May 2019, Pittsburgh, Pa., USA

DOI 10.1000.377.069

Effect of Carbon Bonding State and Concentration on Melting of Direct Reduced Iron

Geonu Kim1, Yilmaz Kacar1, and Petrus Christiaan Pistorius1*

1
Center for Iron & Steelmaking Research (CISR), Carnegie Mellon University, Department of Materials Science and
Engineering
5000 Forbes Ave, Pittsburgh, PA, USA, 15213
Phone: +1 (412)-268-1332
Email: geonuk@andrew.cmu.edu

Keywords: Direct-reduced iron, Carbon, Cementite, Graphite, Nitrogen, Energy

INTRODUCTION
Direct-reduced iron (DRI) is one of the major feedstocks in the electric arc furnace (EAF) steelmaking process. The most-
used DRI is produced by flowing reducing gases; unlike coal-based processes, the gas-reduced DRI can contain a high
percentage of carbon.1 Advantages of carbon-bearing DRI in iron and steelmaking have been studied. The carbon in DRI acts
as a reductant for iron oxides and a source of energy by combustion with injected oxygen (decreasing electrical energy
demand in the EAF), removes nitrogen from the liquid steel through the carbon boiling effect, and helps foam slag by CO
generation. This work summarizes reported DRI carbon effects based on practical plant data and analyzes the effect of carbon
bonding on the energy required for melting. Differences in the melting behavior of DRI depending on carbon concentration
and carbon bonding state (cementite or graphite) were investigated through laboratory experiments.

DISCUSSION

Nitrogen removal in EAF by carbon from DRI

Carbon in DRI plays important roles in EAF steelmaking, acting as a reductant and chemical energy source.1 Furthermore,
higher-carbon DRI forms CO that flushes dissolved nitrogen out of the steel by the carbon boiling effect and foams the
slag.2,3 While the beneficial effect of higher-carbon DRI in EAF steelmaking appears obvious, it is not clear what minimum
amount of carbon needs to be added in DRI to achieve these positive effects. DRI currently produced with the Midrex and
Energiron processes can have high carbon concentrations (up to 4%). However, a much lower carbon concentration seems
adequate for decreasing the nitrogen levels in the liquid steel. Figure 1 illustrates that more than 0.6% C at the flat bath stage
does not give significantly lower nitrogen at tap (round markers in Figure 1; these results are from an EAF steelmaking
process that used a combination of hot metal, scrap, and high-nitrogen DRI from the coal based Stelco-Lurgi/Republic-
National process).2 Results from a plant using higher-carbon gas-reduced DRI showed only a weak effect of %C in DRI on
tap nitrogen (triangular markers in Figure 1). The plant examples in Figure 1 have an implication for the optimal carbon
concentration in DRI: lower carbon inputs might be adequate to control dissolved nitrogen.

© 2019 by the Association for Iron & Steel Technology. 661

 90 Tap nitrogen (ppm) vs. Melt‐in carbon (%)

Tap nitrogen (ppm)
80
70 Tap nitrogen (ppm) vs. Carbon in DRI (%)
60
50
40
30
20
10
0
0 0.5 1 1.5 2 2.5
Melt‐in carbon (%) DRI carbon (%)
Carbon input (%)
Figure 1. Effect of carbon in EAF steelmaking on the nitrogen concentration at tap. Dashed line with round markers: mean
nitrogen content of tapped steel (with 95% confidence intervals) depending on melt-in carbon concentration.2 Dotted line
with triangular markers: mean nitrogen content of tapped steel (with 95% confidence intervals) for DRI with different carbon
concentrations (ArcelorMittal Lazaro Cardenas flat carbon electric steelmaking shop, with 100% DRI feed).4

Influence of DRI on energy requirement for EAF steelmaking

Table 1 is a correlation for the electrical energy demand of EAFs, from a European study of extensive process data from five
EAFs.5 DRI influences several parameters in the formula: the DRI contains gangue (requiring more slag formers, increasing
the energy demand), and carbon that is combusted with injected oxygen (supplying additional energy to the EAF). To reduce
the energy requirement, careful considerations of the specific EAF operational conditions are important to determine the
optimal DRI carbon concentration range.

Table 1. Equation for estimating electrical energy requirements of EAFs (in kWh per tonne).5
W G G / G G G T
375 400 ∙ 1 80 ∙ 50 ∙ 350 ∙ 1000 ∙ 0.3 ∙ 1600 1
kWh/t G G G G G ℃
t t M M M W W
∙ 8∙ 4.3 ∙ 2.8 ∙ NV ∙
min m /t m /t m /t kWh/t
G Furnace tap weight t Power-on time
G Weight of all ferrous materials t Power-off time
G / Total weight of DRI and HBI M Specific burner gas
G Weight of shredded scrap M Specific lance oxygen
G Weight of hot metal M Specific post-combustion oxygen
G Weight of slag formers NV Furnace specific factor (0.2-0.4)
T Tapping temperature W Energy losses
W Mean value of energy losses

In EAF steelmaking, the intrinsic properties of DRI that affect energy consumption are metallization, % Fe, % gangue, % C,
etc.3,6 Some references state that, compared with a scrap melting, an additional 100 – 200 kWh per ton is more required for
melting DRI;3,6 the equation in Table 1 indicates an additional 80 kWh/tonne. Lower metallization levels of DRI increase the
energy required because of the endothermicity of reduction of FeO to Fe; 6,7 each 1% of additional metallization (above 90%)
saved ~12 kWh per tonne of liquid steel.7 Decreasing the gangue percentage in DRI also reduces the energy requirement; a
larger slag quantity needs more input of energy and slag formers, and causes lower iron yield because the excessive slag

662 © 2019 by the Association for Iron & Steel Technology.

contains FeO and Fe droplets.6 Carbon in DRI is an efficient chemical energy source by combustion with the injected oxygen
in EAF (2C O → 2CO ∶ ∆H ℃ 118 kJ/mol O ).1,3,6–11 Based on data collected in three melt shops feeding EAFs
with 100% high-carbon DRI (one at Ternium Hylsa in Monterrey, and the others at Emirates Steel Industries in Abu Dhabi),
each 1% of carbon in DRI saved approximately 37 kWh per tonne of the liquid steel.12 While high-carbon DRI has multiple
benefits, use of excessively high %C may result in a high carbon level in the molten steel, requiring longer decarburization,1,9
and place an additional burden on DRI plant productivity.3,6 With regard to the energy consumption, the optimal carbon
concentration may differ from plant to plant; accurate understanding of DRI melting behavior along with the carbon
concentration and the carbon bonding type would provide a sound basis for optimizing the input carbon.

Contribution to EAF energy consumption: 𝐅𝐞𝟑 𝐂 𝐯𝐬. 𝐅𝐞 𝐆𝐫𝐚𝐩𝐡𝐢𝐭𝐞

The heat of formation of cementite (from Fe and C) is positive at 298 K (22.6 kJ/mol), indicating that melting Fe C-
containing DRI would have a lower EAF energy demand compared with melting a mixture of Fe and graphite (unbound
carbon). The size of this effect was assessed by calculating the enthalpy change for heating Fe-2%C from 25°C to 1600°C
(Table 2); calculations were performed with FactSage, using the FSstel database for liquid Fe-C.

Table 2. Energy requirement to heat 1 tonne of Fe-2%C from 25°C to 1600℃, for different carbon bonding states at 25℃
(calculated with Factsage13).
Situation Required heat transfer [kWh]
Melting Fe-graphite mixture 391
Melting Fe-Fe C mixture 380

The energy requirement is decreased by approximately 6 kWh/tonne for each 1% C (Table 2), if the carbon is in the form of
cementite instead of graphite. This difference is rather small compared with the total electrical energy input to EAF
steelmaking (around 400 kWh per tonne Fe), and might not be detectable in plant operations.

Higher and lower carbon transfer to EAF steel melt

There have been many studies showing that the carbon in DRI helps to reduce the nitrogen level in the liquid steel.2,4,10,11
Some previous laboratory tests indicated that DRI decreased the steel nitrogen content just by dilution, not by the carbon
boiling effect which would occur if carbon in DRI were transferred to the steel bath.14 In those experiments, the DRI pellets
remained buoyant in the slag melt until most of the carbon had been oxidized to CO, preventing the carbon from transferring
to the steel bath. However, in subsequent experiments, with higher-carbon DRI, evidence of strong interaction between the
DRI carbon and the metal bath was found.15 These laboratory-scale experiments were performed by melting slag oxides and
electrolytic iron at 1600℃ in an induction furnace, dropping high-%C or low-%C DRI into the melt, and then freezing the
reaction at the time of maximum CO evolution rate. Examples of quenched metal samples are shown in fig. 2; a considerable
portion of the metal melt (much more than the mass of the DRI pellet) was stirred into the slag by evolution of carbon
monoxide (for both higher-carbon and lower-carbon DRI): in this way, for feeding of both higher- and lower-carbon DRI,
direct interaction between the injected pellet with the metal bath has been identified.15 The low nitrogen contents achievable
in industry (Figure 1) similarly support the idea that carbon from DRI is transferred to the steel bath.

Figure 2. Metal samples from addition of single DRI pellets into steel-slag melts (quenched at the maximum reaction rate
between carbon from DRI and FeO from the slag).15 Both higher-carbon DRI (left) and lower-carbon DRI (right) cause
substantial mixing of metal into the slag.

© 2019 by the Association for Iron & Steel Technology. 663

Decarburization of DRI in EAF steelmaking
There have been several studies on the decarburization reaction kinetics and mechanisms of DRI in steelmaking
process.14,16-20 Upon addition, the DRI is surrounded by molten slag or steel; the cold DRI causes a shell of slag or steel to
solidify on the DRI surface. After the shell melts away (due to further heat transfer from the steel or slag), reactions between
the carbon in DRI and the slag-steel bath can occur.17 The first reaction stage is reduction of the remaining FeO inside the
DRI, by DRI carbon (FeO C → Fe CO ).14,16–18 For this reaction, previous work by Goldstein et al.14 showed 2 – 4
seconds of incubation time, and 20 - 30 seconds for the end of the reaction; in other work by Li et al.18 had 5 – 8 seconds and
10 – 20 seconds for the incubation period and the gas evolution period, respectively. This first stage is largely controlled by
heat transfer; the slag heat transfer characteristics (including temperature, thermal conductivity, viscosity and agitation) and
the thermal conductivity of DRI affect the incubation and reaction time.17,18 After the internal reaction, the next stage is the
reaction between the slag and the remaining DRI carbon ( FeO C → Fe CO ).16–18 The reaction rate of the
secondary decarburization stage is slower than that of the first stage, and mass transfer of FeO in the slag is the main rate
limiting factor.18 A higher slag temperature and relative movement of the slag and the pellet promote the reaction rate;18 the
carbon concentration of the DRI affects the maximum reaction rate.16

Effect of carbon bonding and concentration on melting

To determine whether the DRI melting behavior in an EAF would depend on the carbon concentration and the bonding state
(cementite or graphite), differential scanning calorimetry (DSC), confocal scanning laser microscopy (CSLM), and a DRI
injection into laboratory melts (steel and slag) were performed. The compositions of samples used in these tests are listed in
Table 3.
For the DSC measurement, various DRI pellet samples were heated until melted under constant helium flow (4 ml/min) with
heating rates of 60 and 80 ℃/min, and the consequent heat flow was recorded.
CSLM investigated the melting behavior at a ramping rate of 50℃/min to 1,500℃ under argon. The resulting optical images
were investigated to identify the melting start temperature and the sample appearance upon melting. These DSC and CSLM
results are summarized in table 4. As shown in the DSC data, the different kinds of the DRI pellets (containing graphite or
cementite, with different carbon concentrations) all had similar values of the enthalpy of melting, time for full melting, and
the melting onset temperature (around 1147℃, close to eutectic temperature of the iron-carbide or iron-graphite systems).
The CSLM results were similar to the DSC measurements; melting started in the range of 1077 – 1161 ℃ (around the
eutectic temperature) except for that of the lowest-carbon DRI (~0.78 wt.% C). Low-carbon DRI with less than 1wt.% total
carbon melted at a much higher temperature (>1400℃ than the high-carbon DRI (more than 2 wt.% C), as expected from the
phase diagram. Interestingly, there was a difference between the melting appearances of the graphitic and carbidic DRI
samples. The DRI with mostly cementite carbon tended to melt homogeneously; on the other hand, the DRI with graphite
carbon inside formed apparently round liquid regions in a few places that gradually expanded over time. Both transitions of
the graphitic and carbidic DRI took about 15 seconds to melting completion.

Table 3. Specifications of DRI samples (used for DSC, CSLM, and DRI injection experiments) quantified by XRD Rietveld
phase analysis with X’Pert high score+ software (PANanalytical, Almelo, The Netherlands).
wt.% of Composition
Sample Characteristic Total C (wt.%)
Fe 𝐶 C Fe FeO
DRI-DSC Exp (3.82%C-Graphite) 3.6 3.58 88.6 0 3.82
DRI-DSC Exp (3.82%C-Cementite) 45.7 0.77 49.0 0 3.82
DRI-DSC Exp (7.72%C-Grapahite) 0.3 7.7 92 0 7.72
DRI-Confocal Exp (4.80%C-Graphite) 0.0 4.8 95.2 0.0 4.8
DRI-Confocal Exp (4.50%C-Cementite) 61.5 0.4 38.1 0.0 4.5
DRI-Confocal Exp (3.44%C-Graphite) 5.1 3.1 91.9 0.0 3.44
DRI-Confocal Exp (3.59%C-Graphite) 8.8 3.0 88.3 0.0 3.59
DRI-Confocal Exp (2.30%C-Graphite) 5.9 1.9 92.2 0.0 2.3
DRI-Confocal Exp (0.78%C-Cementite) 11.6 0.0 88.4 0.0 0.78
DRI-Injection Exp (2.10%C-Graphite) 6.0 1.7 92.3 0.0 2.10

664 © 2019 by the Association for Iron & Steel Technology.

 DRI-Injection Exp (2.42%C-Graphite) 4.9 2.1 93.0 0.0 2.42
DRI-Injection Exp (8.01%C-Graphite) 3.2 7.8 84.9 0.0 8.01
DRI-Injection Exp (1.39%C-Cementite) 17.8 0.2 82.0 0.0 1.39
DRI-Injection Exp (2.75%C-Cementite) 41.2 0.0 58.7 0.1 2.75
DRI-Injection Exp (5.72%C-Cementite) 85.8 0.0 14.2 0.0 5.72

Table 4. Summary of DSC and CSLM experiments.

DSC experiment data
Melting
Heating rate of 60℃/min Heating rate of 80℃/min
Sample characteristic
Onset of ∆H Time Onset of ∆H Time
melting required to melting required to
[℃ μV ∙ s/mg melt [s] [℃ μV ∙ s/mg melt [s]

DRI-DSC Exp (3.82%C- 1159.33 22.28 75.0

1155.19 13.06 48.8
Graphite) 1153.24 17.99 65.5

DRI-DSC Exp (3.82%C- 1150.94 22.97 69.5

1156.65 9.43 48.8
Cementite) 1152.14 17.49 68.0
DRI-DSC Exp (7.72%C-
1148.95 21.24 68.0
Grapahite)
CSLM investigation data
Sample characteristic Onset of melting [℃ Melting behavior
DRI-Confocal Exp
1132 Liquid phase nucleation and growth
(4.80%C-Graphite)
DRI-Confocal Exp
1142 Homogeneously melted
(4.50%C-Cementite)
DRI-Confocal Exp
1085 Liquid phase nucleation and growth
(3.44%C-Graphite)
DRI-Confocal Exp
1077 Liquid phase nucleation and growth
(3.59%C-Graphite)
DRI-Confocal Exp
1161 Liquid phase nucleation and growth
(2.30%C-Graphite)
DRI-Confocal Exp
1450 Un-identified
(0.78%C-Cementite)

Melting behavior was tested with a laboratory induction furnace to simulate the decarburization and melting reaction of DRI
samples (with different carbon type and concentration) in a molten slag-steel system. Firstly, slag oxides (slag composition:
20 wt.% FeO, 24 wt.% SiO , 43 wt.% CaO, 8 wt.% MgO, and 5 wt.% Al O ) and electrolytic iron were melted at 1600℃ in
an MgO crucible. The closed furnace system was connected to an infrared gas analyzer with continuous Ar flow to track the
changes in the output gas compositions during the reaction. The DRI pellet was dropped through an opening in the top cover
(otherwise closed with a stopper). Typical results of carbon monoxide emission are given in fig. 3. A large excess of FeO
oxides was supplied in the slag relative to the amount of carbon in the DRI. There was the incubation period as shown in fig.
3, corresponding to literature reports of formation and subsequent melting of the frozen layer around DRI.17 The DRI pellets
used in these experiments were produced using hydrogen, and did not contain any FeO (100% metallization) (Table 3). Thus,
the first reaction step, the remaining FeO in DRI reacting with its inner carbon (FeO C → Fe CO , did not occur.

© 2019 by the Association for Iron & Steel Technology. 665

The incubation time (57 – 71 seconds; see table 5) was longer than in previous studies.14,18 Some of this is caused by the
delayed response of the infrared measurement to gas evolution.
Separate measurements with a step change of CO flow imposed on the system showed that the system behaves like a
continuously stirred tank reactor with a measured response time of 87 seconds (to 63% of final reading) at 1600℃. The time
constants for CO release (Table 5) are similar to the response time, meaning the reaction between the carbon in DRI and the
molten bath is very fast (once the induction period has passed) and not strongly affected by the carbon bonding state or
carbon amount. Like previous study by Sharifi et al.,16 the highest reaction rate increased as the initial carbon content of the
DRI increased. However, there is no obvious difference in reaction rate between the graphite and cementite DRI.

Table 5. Summary of the DRI injection experiments.

Sample characteristic Incubation Period [s] Reaction time constant [s] Highest reaction rate CO %
DRI-Injection Exp
64 89.6 5.89
(2.10%C-Graphite)
DRI-Injection Exp
71 121.0 11.26
(2.42%C-Graphite)
DRI-Injection Exp
69 125.2 14.04
(8.01%C-Graphite)
DRI-Injection Exp
60 94.2 6.91
(1.39%C-Cementite)
DRI-Injection Exp
57 99.5 9.78
(2.75%C-Cementite)
DRI-Injection Exp
64 134.4 12.21
(5.72%C-Cementite)

1
[C released as CO(g)]/ [Total C in DRI]

0.9
0.8
0.7
0.6
0.5
0.4
DRI‐Injection Exp (2.10%C‐Graphite)
0.3
DRI‐Injection Exp (1.39%C‐Cementite)
0.2
0.1
0
0 100 200 300 400 500 600
Time (s)
Figure 3. Typical CO product graphs from the DRI injection experiments (DRI samples dropped into the 1600℃ slag/steel
melt at 𝐭 𝟎𝐬). Carbon monoxide formed versus time, normalized with total carbon in DRI.

666 © 2019 by the Association for Iron & Steel Technology.

 CONCLUSIONS
In the EAF steelmaking process, both the higher and lower carbon DRI directly interact with the metal bath, providing
nitrogen removal by the carbon boiling effect. Plant data showed that even a limited amount of carbon (around 0.6% by
mass) gives a large decrease in tap nitrogen. Having the carbon in DRI as cementite rather than graphite decreases the energy
required for melting, but this difference is small compared with the total electrical energy input of EAF steelmaking.
Experimentally, DRI samples containing carbon in graphite or cementite form had similar heats of melting, the melting times,
and melting points (in DSC and CSLM experiments). Regarding the kinetics and mechanisms of the DRI melting in a
laboratory slag-steel melt, the heat transfer is most important rather than the type of carbon in DRI.
The results allow a better understanding of DRI melting. With regards to energy savings, melting temperature, melting
behavior, and kinetics, the chemical form of the DRI carbon is of secondary importance to the concentration of carbon in the
DRI.

ACKNOWLEDGEMENTS
Support by the members of the Center for Iron and Steelmaking Research is gratefully acknowledged. GK would also like to
thank his parents and sister who have supported him throughout his work.

REFERENCES

1. Memoli, F. Behavior and benefits of high-Fe3C DRI in the EAF. Iron Steel Technol. 2, 1928-1945 (2015, May).

2. Erwee, M. W. & Pistorius, P. C. Nitrogen in SL/RN direct reduced iron: origin and effect on nitrogen control in EAF
steelmaking. Ironmak. Steelmak. 39, 336–341 (2012).

3. Pretorius, E. & Oltmann, H. EAF Fundamentals. LWB Refract. Process Technol. Gr. 54 (2010).

4. Lule, R., Lopez, F., Torres, R., Espinoza, J. & Morales, R. D. The production of steels applying 100% DRI for nitrogen
removal, the experience of ArcelorMittal Lazaro Cardenas Flat Carbon. in AISTech 2009 Proceedings I, 489–498 (2009).

5. Kohle, S. et al. Improving the Productivity of Electric Arc Furnaces. no. 20803. European Commision, (2003).

6. Jones, J. Alternative Iron Feedstocks for EAF Steelmaking: Value In Use. in Electric Furnace Conference 55, 457–480
(1998).

7. Duarte, P. High-Carbon HBI and the Analysis of Formation of Iron Carbide and Behavior in the ENERGIRON ZR
Process. in AISTech 2018 Proceedings 101, 633–647 (2018).

8. Manenti, A. A. Economics and Value-in-Use of DRI in the USA. in AISTech 2015 Proceedings 1, 333–344 (2015).

9. Hunter, R. & Ravenscroft, C. Is Too Much Carbon A Problem? Direct from Midrex, 1st Quart. 4–8 (2014).

10. Hornby, S., Madias, J. & Torre, F. Myths and realities of charging DRI/HBI in electric arc furnaces. Iron Steel Technol.
13, 81–90 (2016, March).

11. Memoli, F., Picciolo, F., Jones, J. A. T. & Palamini, N. The use of DRI in a consteel® EAF process. Iron Steel Technol.
12, 72–80 (2015, February).

12. Duarte, P. & Scarnati, T. Advances in Energy Consumption and Environmental Improvements Using High Carbon DRI
in an EAF Shop. in International Workshop, Associazione Italiana di Metallurgia (2012).

13. Bale, C. W. et al. Reprint of: FactSage thermochemical software and databases, 2010–2016. Calphad Comput. Coupling
Phase Diagrams Thermochem. 55, 1–19 (2016).

14. Goldstein, D. A., Fruehan, R. J. & Ozturk, B. The Behavior of DRI in Slag-Metal Systems and Its Effect on the Nitrogen
Content of Steel. ISS Transactions. 26, 49–61 (1999).

© 2019 by the Association for Iron & Steel Technology. 667

15. He, Y. & Pistorius, P. C. Carbon transfer during melting of direct reduced iron. Iron Steel Technol. 14, 68–71 (2017,
January).

16. Sharifi, E. & Barati, M. The reaction behavior of direct reduced iron (DRI) in steelmaking slags: Effect of DRI carbon
and preheating temperature. Metall. Mater. Trans. B Process Metall. Mater. Process. Sci. 41, 1018–1024 (2010).

17. Barati, M. & Sharifi, E. Role of heat transfer in early stage decarburization of DRI in slag. Metall. Mater. Trans. B
Process Metall. Mater. Process. Sci. 43, 680–685 (2012).

18. Li, J. & Barati, M. Kinetics and mechanism of decarburization and melting of direct-reduced iron pellets in slag. Metall.
Mater. Trans. B Process Metall. Mater. Process. Sci. 40, 17–24 (2009).

19. Min, D. J. & Fruehan, R. J. Rate of reduction of FeO in slag by Fe-C drops. Metall. Trans. B 23, 29–37 (1992).

20. Sadrnezhaad, K. & Elliott, J. F. The Melting Rate of DRI Pellets in Steelmaking Slags. Iron steel Int. 53, 327–339
(1980).

668 © 2019 by the Association for Iron & Steel Technology.