High solubility pathway for the carbon dioxide free production of iron

Article abstract of DOI:10.1039/c0cc01594f

We report a fundamental change in the understanding of iron oxide thermochemistry, opening a facile, new CO₂-free route to iron production. The resultant process can eliminate a major global source of greenhouse gas emission, producing the staple iron in molten media at high rate and low electrolysis energy.

Full text of DOI:10.1039/c0cc01594f

View Article Online / Journal Homepage / Table of Contents for this issue
COMMUNICATION
www.rsc.org/chemcomm | ChemComm
High solubility pathway for the carbon dioxide free production of ironw
Stuart Licht* and Baohui Wangz
Received 26th May 2010, Accepted 10th August 2010
DOI: 10.1039/c0cc01594f
We report a fundamental change in the understanding of iron
oxide thermochemistry, opening a facile, new CO -free route to
of ferric oxide, Fe
2 3
O , in 650 1C molten carbonate was
ꢀ
4.4
ꢀ14
;
2
reported as very low, a 10
mole fraction (Ksp = 4 ꢁ 10
ꢀ1
iron production. The resultant process can eliminate a major
global source of greenhouse gas emission, producing the staple
iron in molten media at high rate and low electrolysis energy.
DGsp = 240 kJ mol ) in lithium/potassium carbonate
mixtures. Furthermore, the solubility was reported as an
2 3 2 3
invariant of the fraction of Li CO and K CO in these
7
lithium/potassium carbonate mixtures. These low solubility
studies, of interest to the optimization of molten carbonate
fuel cells, have likely discouraged research into the electrowinning
of iron metal from ferric oxide in molten lithium carbonate.
We report that at higher temperature, molten lithium
carbonate is an effective medium for the electrolytic deposition
of iron metal. Unlike the current industry process, iron is
formed from molten lithium carbonate without the release
of greenhouse gas. The thermodynamic potentials for the
reduction of iron oxides to iron will be shown to be endothermic;
this decrease, with increasing temperature, will result in
considerable energy savings compared to lower temperature
electrolysis, when solar thermal and solar PV energy is
Along with control of fire, the carbothermal reduction of iron
is one of the founding technological pillars of civilization. Yet,
it is also one of the major global sources of greenhouse gas
release. In industry, iron is still produced by the greenhouse
gas intensive reduction of iron oxide by carbon–coke, and a
CO -free process to form this staple is needed. We report a
2
significant change in the value of the free energy and solubility
product (17 order of magnitude increase) of Fe O at elevated
2
3
temperatures leading to a carbon dioxide free process for the
production of iron. This process is of significance as it can
lead to elimination of one of the major global sources of
greenhouse gas release. The earliest attempt at electrowinning
iron from carbonate appears to have been in 1944 in the
unsuccessful attempt to electrodeposit iron from a sodium
carbonate, peroxide, metaborate mix at 450–500 1C, which
deposited sodium and magnetite (iron oxide), rather than
5
,6
applied, as in the STEP process.
Ferric oxide (99.6%, JT Baker) heated alone remains a red
powder. However, when heated in contact with solid Li CO
powder (Alfa, 99%), we observe it is highly soluble in molten
Li CO . Li CO (mp 723 1C) is transparent, and takes on a red
colour with dissolution of the Fe
Fe to 5 parts (by mass) Li CO
2
3
1
,2
iron. Later attempts have focused on the electrodepostion
,4
of iron from molten mixed chloride/fluoride electrolytes.
2
3
2
3
3
2
O
3
. A mixture of one part
A brief overview of iron history is in the ESI.w
2
O
3
2
3
contains some solid Fe
2
O
3
Here we show, a novel route to generate iron metal by the
electrolysis of dissolved iron oxide salts in molten carbonate
electrolytes, unexpected due to the reported insolubility of iron
oxide in carbonates. This process will prevent the extensive
release of carbon dioxide, which currently accompanies the
formation of iron metal from iron ores. We demonstrate this
at 750 1C, but is fully molten by 800 1C, for a mole fraction
solubility of 0.085 (2 M), and in accord with:
3
+
2ꢀ
3
Fe
2
O
3
- 2Fe + 3O ; Ksp = 3 ꢁ 10 ,
DGsp = ꢀ70 kJ molꢀ
1
(1)
This 17 order of magnitude increase in the Fe
2
O
3
solubility
CO
2 2 3
-free process with the reduction of either Fe O , available
constant sustains a significant concentration of iron in the
molten salt, and provides a route for the facile electrolytic
generation of iron metal. The cyclic voltammetry (CV) at 800
as the common mineral hematite, or Fe O , available as the
3
4
common mineral magnetite, to form iron metal at high
electrolysis current density and low electrolysis energy. The
low electrolysis energy can be driven by conventional
electrical sources, but is also consistent with a new STEP
1
C of this molten, 1 to 5 Fe
presented in Fig. 1, and exhibits a clear reduction peak at
0.8 V, on platinum (dark orange curve); the peak is more
2 3 2 3
O to Li CO electrolyte is
ꢀ
(
Solar Thermal Electrochemical Photo) process for electrolysis
5
without any evolution of carbon dioxide.
,6
pronounced at an iron electrode (light orange curve).
At constant current, iron is clearly deposited. The cooled
deposited product contains pure iron metal and trapped salt,
and changes to rust colour with exposure to water (Fig. 1,
photo top left). The net electrolysis is the redox reaction of
ferric oxide to iron metal and oxygen:
Ferric carbonate does not appear in nature due to the
3
+
relative acidity of the aqueous Fe cation (which forms ferric
hydroxide and evolves CO ). As recently as 1999, the solubility
2
Department of Chemistry, George Washington University, Ashburn,
Virginia 20147, USA. E-mail: slicht@gwu.edu; Tel: +1 703 726 8215
w Electronic supplementary information (ESI) available: A brief over-
o
, E = 1.28 V
Fe
2
O
3
- 2Fe + 3/2O
2
(2)
view of iron history, mechanism and calculated decrease of CO
2
The washed, electrolysis product weight is consistent with
>90% complete conversion of the 6 electron per Fe
coulombic reduction to iron, during 1 hour of electrolysis at
emissions with new STEP electrolysis, description of solar thermal
electrochemical photo energy conversion, and comparison of alkali
carbonate electrolytes. See DOI: 10.1039/c0cc01594f
zC hP irne as .e nt address: Northeast Petroleum University, Daqing, P. R.
2 3
O
ꢀ
2
either 200 or 20 mA cm of iron deposition. The measured
coulombic efficiency rises to a minimum of 97% on repetition,
7
004 Chem. Commun., 2010, 46, 7004–7006
This journal is c The Royal Society of Chemistry 2010

View Article Online
with care to prevent material loss during washing. The
extracted deposited mass contains pure iron metal, and
trapped salt, which is washed to remove the trapped salt by
(
1) grinding to a fine powder, (2) sonication in water for
0 minutes, (3) settling for 0.5 minute, then pouring off the
3
colloidal suspension and liquid from the precipitate. Steps 2
and 3 are repeated several times with the precipitate, which is
then observed to be reflective, grey and metallic. This is
followed by further washing, sonicating and decanting the
precipitate, respectively, in methanol, acetone and n-hexane
(
n-hexane wash is not necessary, but demonstrates that carbon
is not present), followed by 10 minutes of vacuum drying and
weighing. The product, dendritic iron crystals, is stored in
acetone, and responds uniformly to an external magnetic field.
The two principle ores of iron are hematite (Fe O ) and the
2
3
2
+/3+
mixed valence Fe
magnetite (Fe O ). We observe that
3 4
Fe O is also highly soluble in molten Li CO , and may be
3
3
4
2
reduced to iron with the net electrolysis reaction:
o
, E = 1.32 V
Fe
3
O
4
- 3Fe + 2O
2
(3)
Fe O4 is a black powder (Pflatz & Bauer 99.9%), which
3
forms a black liquid on dissolution in a molten Li CO . As
2
3
2 3
seen in Fig. 1 photo (right side), and as with the Fe O case,
the extracted cooled electrode, following extended electrolysis
and iron formation, contains trapped electrolyte. Following
ꢀ
2
1
hour of electrolysis at either 200 or 20 mA cm of iron
deposition, and the washing procedure described for the Fe
without the hexane step), the product weight is consistent
2
O
3
(
Fig. 1 Top: photographs of electrolysis product from in 800 1C
with a complete 8 electron per Fe
to iron.
3 4
O coulombic reduction
molten Li
hours on coiled 2.8 cm platinum wire (d = 0.5 mm). The 4.8 g mass
shown on the electrode contains 0.8 g iron metal (following grinding
and cleaning); (right side) following 0.5 A electrolysis in Fe for
2 3 2 3
CO : (left side) following 0.5 A electrolysis in Fe O for
2
2
The lower portion of Fig. 1 compares the cyclic voltammetry
of Fe O and Fe O in molten Li CO at 800 1C. Unlike the
2
3
3
4
2
3
3
O
4
single peak evident for Fe O , two reduction peaks appear for
3
2
2
1 hour on coiled 7.0 cm iron wire (d = 1.5 mm). Each case uses 16 g of
Fe O . Following the initial cathodic sweep (indicated by
3
4
the iron oxide in 80 g Li CO in a 75 ml alumina crucible. Bottom:
2
3
the left arrow), the CV at Pt exhibits two reduction peaks
black curve), which are more pronounced at an iron electrode
grey curve). The Fe
peaks observed at ꢀ0.6 V and ꢀ0.76 V
appear to be consistent with the respective reductions of
2 3
cathode restricted cyclic voltammetry in 800 1C molten Li CO ,
containing 1 : 5 by weight of either Fe O or Fe O . The measured
2 3 3 4
(
(
O
3 4
potential is dominated by the cathode overpotential through the use of
a 60-fold excess of anode surface area.
2
+
3+
Fe
2 3 3 4 2 3
and Fe . In either Fe O or Fe O , molten Li CO
temperature it is seen that the measured electrolysis potential
can fall below the thermodynamic potential. The latter is
calculated at unit activity, and the former may reflect a Nernst,
non-unit activity variation of the potential. The sharp drop in
electrolysis potential occurring at T > 850 1C parallels our
previous observation in iron-free molten Li CO in which the
electrolytes, the reduction occurs at a potential before we
observe any reduction of the electrolyte, and at constant
current, iron is deposited. In iron-free Li CO , electrolysis of
2
3
6
CO to C or CO has been detailed:
2
(
low temperature, molten carbonate) CO
2
- C + O
2
(4)
(5)
2
3
main electrolysis product switches from solid C to CO at these
6
temperatures.
(
T > 850 1C in Li CO ) CO - CO + 1/2O
2
3
2 2
In Fig. 2, our calculated thermodynamic electrolysis
potentials for the conversion of Fe O or Fe O to iron and
The advantages of a Li
CO electrolyte, are described in the ESI.w The measured
full cell electrolysis potential, E, of Fe or Fe in Li CO
3
2 3 2 3
CO , compared to Na CO or
K
2
3
2
3
3
4
O
2
, eqn (2) and (3), are plotted with the circle symbols.
2
O
3
3
O
4
2
The processes are each endothermic; the required electrolysis
potential decreases with increasing temperature. The
for a range of constant current densities, J, is presented in
Table 1. J is determined by the cathode surface area. The
subscript SA in J or E denotes a small (4-fold excess), and LA
a large (60-fold excess) surface area of anode. The measured
potentials are similar with smooth Pt or iron cathodes, and
with anodes of smooth Pt or Ni (nickel oxide, prepared as
McMaster 200 pure Ni sheet). E is stable during continuous
electrolysis at this range of J. Comparing large and small
thermodynamic electrolysis potentials for Fe
lower than for Fe , and in the temperature domain of
molten Li CO , each is considerably below their respective rest
2 3
O are B0.06 V
3
O
4
2
3
potentials at 25 1C of 1.28 and 1.32 V. As seen in the figure, the
experimental electrolysis potentials at a constant 0.050 A are
similar for Fe O and Fe O at 800 1C, but at T > 800 1C are
2
3
3
4
lower for the Fe O containing electrolyte. At high
2
relative anodes in Table 1, it is evident that anode (O
, rather
2
3
This journal is c The Royal Society of Chemistry 2010
Chem. Commun., 2010, 46, 7004–7006 7005

View Article Online
Fig. 2 Temperature variation of the electrolysis potential to iron
metal and oxygen from either Fe or Fe . As in ref. 5, the
2
O
3
3 4
O
calculated electrolysis potentials are determined from the thermo-
dynamic free energy components of the reactants and products as
E = ꢀDG(reaction)/nF using data from ref. 8–10. The measured
2 3 3 4
electrolysis potentials, in either Fe O or Fe O , are at a constant
2
.050 A between a 2.8 cm Pt cathode and a 12 cm Ni anode.
2
0
Fig. 3 The 2.7 V maximum power of the CPV can drive either two
1.35 V iron electrolyses at 800 1C (schematically represented), or three
Table 1 Experimental constant current full cell potentials at smooth
electrodes for the electrolysis of 1 : 5 Fe or Fe in 800 1C Li CO
2
O
3
3
O
4
2
3
0
.9 V iron electrolyses at 950 1C. The energy stored in the reduction of
5
ꢀ2
2 3
Fe O to Fe at 25 1C is 1.28 V. Hence, when 0.9 V is used to form iron
J
LA/mA cm
3
0.85 V
20
1.18 V
200
500
at 950 1C, there is a considerable energy savings, achieved through the
application of external heat, including solar thermal, to the system.
E(Fe
E(Fe
2
O
O
3
)
)
LA
1.40 V 1.70 V
0.60 V 0.90 V
200
3
4
LA
ꢀ2
J
SA/mA cm
7
18
70
E(Fe
E(Fe
2
O
O
3
)
)
SA
1.15 V 1.25 V 1.57 V 1.70 V
1.27 V 1.70 V
increasing temperature, provides a new route for iron
production. Iron is formed without an extensive release of
3
4
SA
CO
occurring iron oxide ores, hematite, Fe
Fe . Iron, a basic commodity, currently accounts for
2
in a process compatible with the predominant naturally
iron) overpotentials dominate. E is less, and J higher, for
textured electrodes or with increased mass transport (electrolyte
convection or electrode movement/rotation).
2 3
O
, and magnetite,
3
O
4
the release of one quarter of worldwide CO2 emissions by
industries. The endothermic nature of the new synthesis route
The low voltage iron carbonate electrolysis presented in this
study can be driven by conventional or renewable electricity,
provides an effective vehicle for the solar efficient, CO free,
2
respectively, with diminished or no CO
2
emission, as
production of iron as a STEP process.
delineated in the ESI.w The decrease in electrolysis potential
with increasing temperature presents a low energy electrolysis
opportunity for the STEP process. In this process solar
thermal provides heat to decrease the electrolysis potential,
and solar visible generates electronic charge to drive the
electrolysis. The synergistic use of both visible and infrared
sunlight provides high solar energy conversion efficiencies
Notes and references
1 L. Andrieux and G. Weiss, Comptes Rendus, 1944, 217, 615.
2
3
4
G. M. Haarberg, E. Kvalheim, S. Rolseth, T. Murakami,
S. Pietrzyk and S. Wang, ECS Trans., 2007, 3, 341.
S. Wang, G. M. Haarberg and E. Kvalheim, J. Iron Steel Res. Int.,
2008, 15, 48.
G. M. Li, D. H. Wang and Z. Chen, J. Mater. Sci. Technol.
(Shenyang, People’s Repub. China), 2009, 25, 767.
S. Licht, J. Phys. Chem. C, 2009, 113, 16283.
S. Licht, B. Wang, S. Ghosh, H. Ayub, D. Jinag and J. Ganley,
J. Phys. Chem. Lett., 2010, 1, 2363, online at: http://pubs.acs.org/
doi/abs/10.1021/jz100829s.
5
,6
(
STEP energy diagram is in the ESIw). Eqn (2) and (3)
provide, in molten carbonate electrolytes, low energy routes
for the carbon dioxide free formation of iron metal from
iron ores. We use a 37% solar energy conversion efficient
concentrator photovoltaic (CPV) as a convenient power
source to drive the low electrolysis energy iron deposition
5
6
7
L. Qinfeng, F. Borum, I. Petrushina and N. J. Bjerrum,
J. Electrochem. Soc., 1999, 146, 2449.
8 M. W. Chase, Jr., NIST-JANAF Thermochemical Tables,
without CO
Details of the CPV and STEP experimental configurations in
the Li CO electrolyte for CO reduction are given in ref. 6.
2
formation as schematically represented in Fig. 3.
4th Edition, J. Phys. Chem. Ref. Data, 1998, 35, 1.
J. D. Cox, D. Wagma and V. Medvedev, CODATA Key Values for
Thermodynamics, Hemisphere Pub. Corp., NY, 1984.
2
3
2
9
The unexpected solubility of iron oxides in lithium
carbonate electrolytes, coupled with facile charge transfer
and and a sharp decrease in iron electrolysis potentials with
1
0 Data from ref. 8 and 9 interactively available at: http://webbook.
nist.gov/chemistry/form-ser.html.
7
006 Chem. Commun., 2010, 46, 7004–7006
This journal is c The Royal Society of Chemistry 2010