Content uploaded by Nathalie Pereira
Author content
All content in this area was uploaded by Nathalie Pereira on Nov 11, 2014
Content may be subject to copyright.
Investigation of stainless steel pickling liquor as a
precursor for high Capacity battery electrode
materials
Sheel Sanghvi, Nathalie Pereira,*Anna Halajko and Glenn G. Amatucci
Stainless steel and battery manufacturing industries can be united for mutual economic and environmental
benefits by utilizing stainless steel pickling liquor waste product as a precursor for high energy iron fluoride
based positive electrode materials in batteries. This study analyzes the feasibility, environmental, and
economic cost of the ferric fluoride (FeF
3
) synthesis approach through the use of recycled pickling liquor
from the stainless steel fabrication industry. This new synthesis method is determined as more
environmentally friendly than the current methods of disposing spent stainless steel pickling liquor and
producing ferric fluoride separately. X-ray diffraction analysis demonstrated the synthesis occurs in two
steps: the conversion of spent pickling liquor to produce the crystallohydrate b-FeF
3
$3H
2
O followed by
its dehydration into anhydrous FeF
3
. Materials obtained from pickling solutions were found
electrochemically active with improved cycling stability and similar capacity compared to commercial
FeF
3
. Pickling solutions synthesized with nickel and chromium to better replicate stainless steel industry
spent pickling waste showed improved electrochemical performance.
Introduction
Transportable power sources are enabling technologies for an
enormous breadth of devices that bring signicant benetto
society. Although almost all applications require high perfor-
mance, the sheer scale of battery production demands appro-
priate environmental responsibility and economic feasibility.
Metal uoride nanocomposites have been shown as an
appealing approach to positive electrode materials for primary
and secondary batteries based on their high energy densities
compared to the current state of the art.
1
More specically, iron-
based materials are desirable based on iron's low toxicity,
abundance and therefore potential lower cost. However, the
uoride component can contribute a fairly signicant process-
ing cost and possible impact to the environment especially if
processed in a non responsible manner.
The most common process of producing FeF
3
is shown by
the reaction in eqn (1). A constant stream of anhydrous
hydrogen uoride (HF) is passed over anhydrous ferric chloride
(FeCl
3
) in a heated, oxygen-free reactor.
2
FeCl
3
+HF/FeF
3
+ HCl (1)
Such method of production of iron uorides poses a signif-
icant environmental issue through an unnecessary, extra
exploitation of energy and precursor materials.
A possible alternative to produce FeF
3
is to utilize pickling
liquor, a waste product of the stainless steel industry. The
stainless steel manufacturing industry requires a nal process
known as pickling to etch the surface of as processed stainless
steel and restore the protective effects of chromium and other
surface oxides. Aer the pickling process is completed, the acid
is rinsed offthe surface of the steel and stored in waste
containers as spent pickling liquor (SPL). As discussed in detail
below, such SPL typically contains nitric acid, hydrouoric acid,
iron and various other metals.
Large amounts of spent liquor need to be disposed of each
year by stainless steel manufacturers leading to a non insig-
nicant cost to the product. Although some stainless steel
manufacturers attempt to regenerate the acids or recover the
metals from the spent liquor, most processes are not fully
effective leading to appreciable environmental and nancial
costs. A frequently employed method to treat spent liquor is
through lime neutralization.
Both processes of producing new FeF
3
and disposing spent
liquor introduce certain environmental impacts. A quantitative
metric needs to be enacted to compare if the impact of these
process outweigh the impact of producing FeF
3
from SPL. This
is accomplished through the use of the Waste Reduction Algo-
rithm (WAR) GUI created by the United States Environmental
Protection Agency. The WAR GUI produces a numerical score
for a chemical process called the potential environmental
impact (PEI) to describe the impact a process will have. This is
dened in further detail below.
Energy Storage Research Group, Dept. of Materials Science and Engineering, Rutgers
University, North Brunswick, New Jersey 08902, US. E-mail: npereira@rci.rutgers.edu
Cite this: RSC Adv.,2014,4,57098
Received 30th September 2014
Accepted 23rd October 2014
DOI: 10.1039/c4ra11539b
www.rsc.org/advances
57098 |RSC Adv.,2014,4,57098–57110 This journal is © The Royal Society of Chemistry 2014
RSC Advances
PAPER
The PEI values obtained from the WAR GUI for each of the
processes will be plugged into eqn (2) to estimate the environ-
mental impact of the fabrication of the FeF
3
from the spent
liquor compared to the combination of the spent liquor
disposal and the current commercial FeF
3
fabrication process.
Appropriate satisfaction of this equation would justify the
environmental benet of converting spent pickling liquor into
FeF
3
.
Env. Impact
New FeF
3
+ Env. Impact
SPL Disposal
>
Env. Impact
FeF
3
from SPL
(2)
However, before any environmental calculations can be
completed, a chemical pathway behind the conversion process
must be demonstrated. A possible pathway to convert SPL into
iron uorides is the precipitation of b-FeF
3
$3H
2
O from SPL
followed by dehydration into FeF
3
. Other research groups have
independently studied different points of this pathway. Tjus
et al. and Sartor et al. conrm that b-FeF
3
$3H
2
O crystals can be
separated from spent liquor.
3,4
Quite recently, Myung et al. and
Liu et al. produced the anhydrous form of ferric uoride from
the crystal hydrate.
5,6
This paper aims to unite all points to form
one complete method of an environmentally sound supply
stream to show that the waste pickling liquor from the stainless
steel industry can be transformed to high value added compo-
nent product of high energy density iron uoride for battery
applications. This study also encompasses the evaluation of the
electrochemical activity of the products synthesized from pick-
ling liquors with a comparison to a commercial FeF
3
material.
Finally, the economics of the synthesis of FeF
3
from spent
liquor are considered and compared to costs associated to the
disposal of spent liquor and the fabrication of FeF
3
using
current methods.
Experimental
Sample solutions of spent pickling liquor were synthesized by
dissolving iron nitrate hydrate Fe(NO
3
)
3
$9H
2
O (Alfa Aesar),
chromium nitrate hydrate Cr(NO
3
)
3
$9H
2
O (Sigma Aldrich), and/
or nickel nitrate hydrate Ni(NO
3
)
2
$6H
2
O (Sigma Aldrich) in a
solution of hydrouoric acid (HF) (48%, Sigma Aldrich) and
deionized water. Solutions were stored at room temperature for
one week to allow for crystallization to fully occur. The produced
crystals were rst ltered with lter paper (Whatman 542) then
washed with three 10 mL aliquots of ethanol (Sigma Aldrich).
Finally, the water adhering to the surface of the crystals was
removed by drying overnight at 70 C in air with <1% relative
humidity.
The dehydration of the crystals was completed in a Lindberg/
Blue-M tube furnace at 150–400 C under owing argon for
various intervals of time. Additionally, differential scanning
calorimetry (TA-Instruments Q10) was performed using pans
hermetically sealed in a helium environment subsequently
heated at a rate of 5 C min
1
from room temperature to 450 C.
X-ray diffraction was completed using a Bruker D8 Advance
diffractometer with a 0.022qstep size, a 1.9 s per step count, a
15–602qangle span and a CuKaX-ray source. Furthermore,
using the soware TOPAS, X-ray diffraction patterns were tted
by Rietveld Renement to estimate crystal size. Scanning elec-
tron microscopy was performed on a Zeiss Sigma Field Emis-
sion SEM with an Oxford INCA PentaFETx3 Energy Dispersive
Spectroscopy (EDS) attachment. The secondary electron
detector was used with a voltage of 5.00 kV and a working
distance of 7.9 millimeters.
In order to evaluate the electrochemical performance of the
materials obtained from pickling liquor and compare to
commercial FeF
3
(Advanced Research Chemicals), nano-
composites were fabricated by high-energy mechanomilling.
7
The iron uorides either synthesized from pickling liquor or
obtained from a commercial source were milled with 15 wt%
activated carbon (ASupra, Norit) for 1 h in helium.
Electrodes were fabricated according to the Bellcore devel-
oped process.
8
The active materials were mixed in acetone
(Aldrich) with poly(vinylidene uoride-co-hexauoropropylene)
binder (Kynar 2801, Elf Atochem), carbon black additive (Super
P, MMM) and dibutyl phthalate plasticizer (Aldrich). Aer
plasticizer extraction in anhydrous ether (Aldrich), the elec-
trodes typically consisted of 57.25% active material, 12% carbon
additive, and 31% binder. Two-electrode coin cells were
assembled using a lithium foil counter electrode and glass ber
separators (GF/D, Whatman) saturated with 1 M LiPF
6
in
ethylene carbonate–dimethyl carbonate electrolyte (50 : 50 in
vol.%) from BASF. The cells were cycled at 60 C under a
constant current of 7.5 mA g
1
based on the weight of the
nanocomposite, between 1.5 and 4.5 V.
Various sources of information were consulted to determine
the material ow and energy consumption for each production
stream. Using such data, an environmental impact estimate was
calculated using the WAR GUI soware to generate a PEI score.
The PEI is composed of a sum of eight categories: human
toxicity potential by ingestion, human toxicity potential by
dermal contact or inhalation, aquatic toxicity potential, terres-
trial toxicity potential, global warming potential, ozone deple-
tion potential, photochemical oxidation potential, and
acidication potential. Each category is assigned a specic
weight depending on how important that environmental
concern is to the chemical process. For this study, all categories
were assumed to have equal weight. Furthermore, for each of
the eight categories, each chemical is assigned a normalized
value which is computed from toxicology information in the
program's database and literature. The normalized values are
added together to create the PEI. The specic PEI of interest for
this paper is calculated as a summation of the impact of
chemicals produced and energy consumed.
Many assumptions are considered for all PEI calculations in
this study. For example, the crystals produced from SPL by the
process described later in this paper are assumed to be
composed primarily of FeF
3
with nominal amounts of CrF
3
.
Additionally, except for the SPL composition, if the amount of
reactants or products in a specic chemical process is given as a
range of values, then the smallest amounts of each reactant or
product are used for calculations. This will minimize the envi-
ronmental impact of the chemical process and provide the best
case scenario. Another assumption is that non-descriptive items
This journal is © The Royal Society of Chemistry 2014 RSC Adv.,2014,4,57098–57110 | 57099
Paper RSC Advances
such as solid waste or heavy metals are not included in PEI
calculations, which causes the mass balance to be slightly off.
Also, the impact of transportation between facilities was not
included in PEI calculations because of the multiple options
and distances for production facilities. Next, the energy impact
was combined to include all utilities such as heating, electricity,
refrigeration, etc. and gas was assumed to be the only source to
generate the energy. This provides a uniform basis for energy
calculations. Finally, the impact of process water was ignored
because water has an insignicant impact and can be used in
varying quantities for different processes.
Results
1. Disposal process of spent pickling liquor from the
stainless steel industry
Spent liquor typically has the composition of the “Industrial
SPL”shown in Table 1. The spent liquor is disposed of by lime
neutralization. Calcium hydroxide or slaked lime is introduced
into a tank containing the liquor to raise the pH of the solution
and precipitate ferric hydroxide and calcium uoride. The solid
waste from this reaction is removed by ltering and then, it is
pressed dry and landlled. The aqueous waste, which contains
calcium and nitrate ions, is either released into a stream or sent
to a wastewater treatment plant for further processing. Fig. 1
shows the inputs and outputs of this process.
2. Current commercial synthesis process of FeF
3
Anhydrous ferric uoride is rarely fabricated at industrial levels
because the hydrated, chloride form of the metal salt is
preferred. However, it is possible to manufacture anhydrous
ferric uoride by passing a stream of anhydrous hydrogen
uoride over anhydrous ferric chloride in an oxygen-free envi-
ronment.
2
Fig. 2 below shows this process along with the
production of the precursors.
The precursors for ferric uoride production are hydrogen
uoride and ferric chloride. Hydrogen uoride is synthesized by
reacting concentrated sulfuric acid with the mineral uorspar
in a heated reactor.
9
It is then dehydrated by condensation and
distillation. A co-product of this reaction is calcium sulfate,
which can enter other markets rather than being discarded as
waste.
10
The other precursor for this reaction, ferric chloride,
can be synthesized in its hydrated form by recycling spent steel
pickling liquor (contains hydrochloric acid) with scrap iron.
However, it is necessary to dehydrate the hydrated ferric chlo-
ride to provide a suitable precursor for ferric uoride
production. The dehydration methods known are not practical.
Dehydration by thermal treatment produces a low yield, while
dehydration with a dehydrating agent such as thionyl chloride
uses hazardous reactants and is costly. The issues associated
with both dehydration processes prevent the use of ferric
chloride produced by recycling steel pickling liquor. Instead,
anhydrous ferric chloride is produced by oxidizing red hot iron
with chlorine gas.
10
This process has a large yield and generates
hydrogen chloride as a waste product. The produced anhydrous
hydrogen uoride (HF) is passed over anhydrous ferric chloride
(FeCl
3
) in a heated, oxygen-free reactor to produce FeF
3
.
2
3. Synthesis of FeF
3
from Fe-only pickling liquor
Production of ferric uoride from spent pickling liquor can be
split into two steps: the synthesis of b-FeF
3
$3H
2
O from SPL and
the dehydration of b-FeF
3
$3H
2
O to FeF
3
. Two practices
employed in industry to prepare b-FeF
3
$3H
2
O are evaporation
or membrane ltration in conjunction with crystallization.
During evaporation, the spent liquor is heated to its boiling
point to remove excess acids and water while increasing the iron
concentration to 80–120 kg m
3
of SPL.
11,12
This concentration
is required to form a super saturated solution that will later be
crystallized. Membrane ltration produces results similar to
evaporation. First, microltration eliminates large solids from
the solution and then, nanoltration separates metal ions from
the free acids. The portion retaining metal ions is recirculated
through the system multiple times to increase the iron
concentration.
3
The supersaturated solution is then held at
approximately 50 C to allow for the nucleation and growth of b-
FeF
3
$3H
2
O crystals. Additional hydrouoric acid can also be
added to supplement crystal growth.
Unlike evaporation or membrane ltration, low temperature
crystallization has not yet been employed by industry. Pickling
liquor is slowly cooled to about 40 C to precipitate ice and
FeF
3
$3H
2
O crystals, which can then be separated by
suction ltration.
4
These three possible routes to synthesize b-
FeF
3
$3H
2
O are highlighted in the upper portion of Fig. 3.
However, because the solutions in this paper are synthetically
prepared to specic concentrations, these techniques are not
studied in detail. They are only mentioned to describe the
conversion process from pickling liquor to FeF
3
in its entirety.
The b-FeF
3
$3H
2
O crystals are then converted to FeF
3
by
thermal decomposition as highlighted in the lower portion of
Fig. 3. The specic temperatures for decomposition are dis-
cussed herein. In this study, sample pickling liquors containing
iron as the only metallic species were prepared to establish a
Table 1 Spent pickling liquor compositions
Compositions of spent pickling liquor
Pickling liquor [Fe
3+
](gL
1
) [Cr
3+
](gL
1
) [Ni
2+
](gL
1
) [NO
3
](gL
1
)[F
](gL
1
)
Industrial SPL
22
30–45 5–10 3–5 150–180 60–80
Fe SPL A 43.2 ——143.9 44.1
Fe SPL B
23
78.9 ——258.7 80.7
Fe–Cr–Ni SPL
4
42.2 10.8 6 195 60
57100 |RSC Adv.,2014,4,57098–57110 This journal is © The Royal Society of Chemistry 2014
RSC Advances Paper
comparative baseline. The solution was initially composed of
the composition Fe SPL A in Table 1, but this process resulted in
a very low yield of fabrication. The composition was therefore
altered to the composition Fe SPL B in Table 1 which improved
yield to about 27% mol of precipitated b-FeF
3
$3H
2
Oaer one
week of reaction based on original concentration of Fe in the
solution. The fabricated crystals were singled phase and iden-
tied as b-FeF
3
$3H
2
O by X-ray diffraction analysis as shown in
Fig. 4.
DSC analysis was performed to study the characteristics of
decomposition during dehydration. Fig. 5 reveals the trans-
formation of b-FeF
3
$3H
2
O to FeF
3
occurs in a single, large
endothermic decomposition step at approximately 140.47 C
while the small bump at 97.32 C could be associated to the
evaporation of adsorbed water. As a result, heat treatments were
performed at various temperatures above 140 C under owing
Ar with the goal to dehydrate b-FeF
3
$3H
2
O into FeF
3
.Arst
sample was heat-treated at 200 C for two hours, which resulted
in the decomposition of b-FeF
3
$3H
2
O into a hydrate of lower
water content FeF
3
$0.33H
2
O and small amounts of anhydrous
FeF
3
(Fig. 6). A second sample was then heat-treated at higher
temperature. At 400 C, dehydration was driven further and the
sample consisted mostly of anhydrous FeF
3
with a very small
amount of FeF
3
$0.33H
2
O (Fig. 6). The residual FeF
3
$0.33H
2
O
phase could result from reaction of FeF
3
with ambient air upon
transfer from the tube furnace to a He-lled glove box. Finally, a
third sample was submitted to two successive 2 hour heat-
treatments at 150 C and 400 C leading to successful dehy-
dration with complete transformation into single phase FeF
3
(Fig. 7). Table 2 shows the lattice parameters of the FeF
3
materials synthesized from SPL B are consistent with JCPDS
values and close to that obtained from a commercial source. In
addition, the material we fabricated is of lower crystallinity than
FeF
3
obtained from a commercial source with 38 nm crystallite
size compared to 62 nm for the commercial sample.
Aer heat treatment at 400 C, the sample was observed to be
of an orange-brown color rather than the characteristic light
green color of commercial FeF
3
. It is possible an oxide layer has
formed on the surface of the powder in spite of the lack of
evidence by XRD analysis. The absence of oxide by XRD most
likely stems from amounts below detection limit. To further
identify if this oxide layer exists at the surface, temperature was
increased and a sample was submitted to three consecutive
heat-treatments at 150 C, 400 C and 450 C respectively. Fig. 7
Fig. 1 Material flow chart for spent pickling liquor neutralization.
Fig. 2 Material flow chart of ferric fluoride production.
This journal is © The Royal Society of Chemistry 2014 RSC Adv.,2014,4,57098–57110 | 57101
Paper RSC Advances
shows that heating from 400 C to 450 C results in the growth
of Fe
2
O
3
. Additionally, the color of the sample changes to a
slightly darker shade of orange-brown. Finally, because the
position of the FeF
3
peaks do not shiaer heat treatment,
there is indication that surface oxidation is occurring rather
than oxidation within the crystal lattice. Small amounts of
Fig. 3 Material flow chart of ferric fluoride production from spent pickling liquor.
Fig. 4 XRD pattern of crystals precipitated from iron only pickling solution at room temperature. Peaks indexed are associated to b-FeF
3
$3H
2
O
(JCPDS 00-032-0464).
57102 |RSC Adv.,2014,4,57098–57110 This journal is © The Royal Society of Chemistry 2014
RSC Advances Paper
surface oxidation are expected to be a benet to the ultimate
electrochemical processes.
4. Synthesis of FeF
3
and CrF
3
from Fe–Cr–Ni pickling liquor
In order to more accurately replicate pickling liquors obtained
from the stainless steel fabrication process, chromium and
nickel metal species were added to generate pickling liquors of
composition Fe–Cr–Ni SPL from Table 1. The composition is
derived from a metals ratio of 36 : 5:9 Fe : Ni : Cr and a metals
concentration of 60g L
1
.
4
The solution is prepared and stored
for one week to allow for crystallization to occur. Aer crystal-
lization was complete, a yield of nearly 29% was obtained,
assuming both FeF
3
$3H
2
O and CrF
3
$3H
2
O precipitate out of
solution. This value is similar to that obtained with pure iron
solutions.
X-ray diffraction analysis (Fig. 8) of the crystals obtained
aer crystallization reveals the presence of b-FeF
3
$3H
2
O and,
CrF
3
$3H
2
Oora-FeF
3
$3H
2
O. Since CrF
3
$3H
2
O is isostructural to
the a-FeF
3
$3H
2
O phase, distinction between both phases is
difficult.
13
However, it should be noted that a-FeF
3
$3H
2
Oisa
disordered metastable phase whose crystal structure reorients
into b-FeF
3
$3H
2
O over an extended period of time or at elevated
temperatures and no alpha phase was identied in our previous
precipitations without the presence of Cr.
14
Therefore, we
believe it could be possible to distinguish between both phases
aer heat treatment.
DSC analysis, shown in Fig. 9, reveals that the decomposition
of the mixture of b-FeF
3
$3H
2
O with CrF
3
$3H
2
Oora-FeF
3
$3H
2
O
into FeF
3
occurs in two closely spaced endothermic decompo-
sition steps at approximately 146.96 C and 166.45 C respec-
tively. The rst step correlates to the decomposition of b-
FeF
3
$3H
2
O while the second step could correlate to the
decomposition of CrF
3
$3H
2
Oora-FeF
3
$3H
2
O. A small bump
before the main peak at 96.88 C can be associated to the
evaporation of adsorbed water.
Fig. 5 DSC scan of b-FeF
3
$3H
2
O heated in helium from RT to 450 C
at a rate of 5 C min
1
.
Fig. 6 XRD pattern of the initial b-FeF
3
$3H
2
O sample after 2 hour heat-treatment at 200 C (bottom), and at 400 C (top). Peaks indexed by stars
are consistent with FeF
3
(JCPDF 01-088-2023) and peaks indexed by diamonds are consistent with FeF
3
$0.33H
2
O (JCPDS 01-076-1262).
This journal is © The Royal Society of Chemistry 2014 RSC Adv.,2014,4,57098–57110 | 57103
Paper RSC Advances
As performed with the pure b-FeF
3
$3H
2
O samples reported
above, samples were heated above the decomposition temper-
atures obtained by DSC. As with the optimized pure Fe uoride
hydrates, the samples were heat-treated at 150 C for 2 hours to
allow for the initial dehydration step, then at 400 C for 5 hours
to form the nal product. The resulting XRD scan reported in
Fig. 10 reveals single phase FeF
3
. Lattice parameters (Table 2)
are similar to that obtained with the Fe-only pickling solution
and consistent with JCPDS values for rhombohedral FeF
3
. The
synthesized material shows broader diffraction peaks than the
ones obtained with the commercial FeF
3
consistent to a lower
crystal size of 41 nm obtained by Rietveld renement.
Since X-ray diffraction does not present any evidence for any
distinct Cr-based phase, we can hypothesize on the possibility
of the formation of a chromium–iron solid solution during
crystallization. The later would be consistent with the Hume–
Rothery rules determined from the similarity in atomic radii of
chromium (0.62 ˚
A) and iron (0.65 ˚
A) ions in the 3+ state.
15
This
would also be consistent with a previous report of Cr
0.5
Fe
0.5
F
3
solid solution uorides.
16
As shown in Table 3, quantitative EDS
analysis of the sample synthesized from Fe–Cr–Ni pickling
liquor conrmed the material comprises iron, chromium,
uorine, and oxygen however no nickel was detected. The
Fe : Cr elemental ratio is calculated from the relative atomic
percents of Fe and Cr resulting in a value of approximately
3.747 : 1. This is incorporated into the chemical formula of
ferric uoride to give a composition of Fe
0.79
Cr
0.21
F
3
. Field
emission electron microscopy (FESEM) analysis of the nal
product revealed a wide distribution of crystal shape and sizes
(Fig. 11). Furthermore, even though X-ray analysis does not
indicate the presence of any oxide phase, through visual iden-
tication of the color change it can be concluded that an oxide
coating has formed on the surface of the iron–chromium uo-
ride particles as similarly mentioned above for the pure FeF
3
sample.
5. Electrochemical performance
As demonstrated above, anhydrous iron uoride has been
successfully synthesized from iron based pickling liquors
through successive heat-treatments under argon. The samples
Fig. 7 XRD scan of the b-FeF
3
$3H
2
O sample submitted to two consecutive heat-treatments at 150 C and 400 C (bottom), and followed by a
third heat-treatment at 450 C for two hours (top). Peaks indexed by stars are associated to FeF
3
(JCPDS 01-071-3710) and peaks indexed by
open circles to Fe
2
O
3
(JCPDS 00-089-0599).
Table 2 Lattice parameters obtained by Reitveld refinement using the
software TOPAS and compared to theoretical values
Source Crystal Structure a(
˚
A) c (˚
A)
Theoretical ( JCPDS 01-088-2023) Rhombohedral 5.1941(9) 13.334(9)
Commercial FeF
3
Rhombohedral 5.206(3) 13.287(3)
FeF
3
from Fe SPL B Rhombohedral 5.208(0) 13.352(8)
FeF
3
from Fe–Cr–Ni SPL Rhombohedral 5.201(2) 13.314(2)
57104 |RSC Adv.,2014,4,57098–57110 This journal is © The Royal Society of Chemistry 2014
RSC Advances Paper
tested for electrochemical activity were synthesized through a
succession of two heat-treatment at 150 C for two hours and at
400 C for ve hours. Both the materials obtained from Fe-only
and Fe–Ni–Cr liquors were evaluated and compared to
commercial FeF
3
aer formation into nanocomposites. Fig. 12
presents the second cycle voltage proles obtained at 60 C with
a current of 7.5 mA g
1
based on the weight of the
nanocomposite. Both materials fabricated from pickling liquors
are electrochemically active and exhibit discharge proles
similar to that of commercial FeF
3
, except for the slightly lower
capacity of the Fe–Cr–Ni sample. All capacities obtained excee-
ded 75% of theoretical values inclusive of the FeF
3
/Li
0.5
FeF
3
/LiFeF
3
insertion reaction above 3 V and the LiFeF
3
/3LiF +
Fe conversion reactions at approximately 2 V. No distinct elec-
trochemical signatures originating from CrF
3
or NiF
2
could be
abstracted from the voltage prole.
Discharge capacity per gram of FeF
3
was plotted as a func-
tion of cycle number in Fig. 13. Both samples fabricated from
pickling solution showed improved capacity retention
compared to the commercial material. Best results were
obtained with the sample derived from the Fe–Cr–Ni solution.
6. Environmental calculations
6.1. Disposal of spent pickling liquor. The process of spent
pickling liquor disposal was summarized in Fig. 1 while Table 4
presents the quantities of input as well as the output materials
involved in the disposal process. The values in Table 4 are
calculated supposing stoichiometric reactions. However, an
excess of the hydroxide ion is included in this table to ensure
mass balance even though it forms water by neutralizing the
acids. Using these assumptions, the estimated environmental
Fig. 8 XRD pattern of crystals produced from the Fe, Ni, and Cr sample pickling liquors. Peaks marked by point-up triangles are associated to b-
FeF
3
$3H
2
O (JCPDS 00-032-0464) and peaks marked by point-down triangles are associated to CrF
3
$3H
2
O (JCPDS 00-017-0316) or a-FeF
3
-
$3H
2
O (JCPDS 00-024-0071).
Fig. 9 DSC scan of hydrated crystal sample from Fe–Cr–Ni SPL
heated from RT to 450 C at a rate of 5 C min
1
.
This journal is © The Royal Society of Chemistry 2014 RSC Adv.,2014,4,57098–57110 | 57105
Paper RSC Advances
impact for SPL disposal is 745 PEI per L of SPL. Since values for
this section are given in units of PEI per L of SPL and the theme
of units used throughout this paper is PEI per kg of FeF
3
, the
above value can be converted into 23 281 PEI per kg of FeF
3
.
This is accomplished with a conversation factor that 1 L of SPL
has the potential to yield 0.032 kg of FeF
3
(or (Fe,Cr)F
3
). The
derivation of this factor is explained in detail in Section 6.3
below. Major contributing elements to this calculation are
associated to the calcium and nitrate ions remaining in the
wastewater, which mainly have aquatic toxicity effects. The
energy contribution is assumed to be minimal because the
neutralization process does not require relatively demanding
mixing or ltering technology.
6.2. Production of anhydrous ferric uoride via current
commercial synthesis process. The commercial production of
anhydrous FeF
3
was presented in Fig. 2 while Table 5 shows the
material ow used to estimate the environmental impact of
production of ferric uoride. The estimated environmental
impact for ferric uoride production is 1.273 PEI per kg of FeF
3
.
Fig. 10 XRD patterns of crystals produced from the Fe- and the Fe–Cr–Ni pickling liquors heated at 150 C for two hours and then at 400 C for
five hours, compared to a commercial FeF
3
material. Peaks indexed indicate FeF
3
(JCPDS 01-071-3710) while system peak is present in the Fe–
Cr–Ni sample as shown by the star.
Table 3 Atomic percent information obtained from EDS of the final
(Fe,Cr)F
3
product
Element Atomic percent
C 4.12
O 7.85
F 58.74
Cr 6.17
Fe 23.12
Fig. 11 SEM image of the final (Fe,Cr)F
3
product at a 652
magnification.
57106 |RSC Adv.,2014,4,57098–57110 This journal is © The Royal Society of Chemistry 2014
RSC Advances Paper
The chemical component, which has a value of 1.2662 PEI per
kg of FeF
3
, greatly outweighs the energy component, which has
a value of 0.0068 PEI kg
1
. This occurs because of the high
toxicity of the outputs, specically hydrogen chloride gas which
is released into the environment. Specic assumptions for this
PEI calculation are that uorspar is primarily calcium uoride
and the synthesis of ferric uoride from its precursors is stoi-
chiometric with a two percent loss of ferric chloride.
17
6.3. Production of (Fe,Cr)F
3
from spent pickling liquor. As
described above, production of (Fe,Cr)F
3
from spent pickling
liquor can be split into two steps: crystallization of the hydrate
from SPL and dehydration to (Fe,Cr)F
3
. The composition of SPL
used in this paper allows for crystallization to favorably occur
without any added chemicals or energy. The various techniques
of crystallization mentioned above incorporate an amount of
energy with an environmental cost comparatively negligible to
the rest of the system. Therefore, the only contributing factor to
environmental cost from the crystallization step is disposal of
the effluent remaining aer crystallization. The exact compo-
sition of this liquid is unknown. However, by calculating
backwards from the estimated crystal yield and crystal compo-
sition, the remaining liquid composition is estimated to be 30.7
gL
1
[Fe
3+
], 7.7 g L
1
[Cr
3+
], 6 g L
1
[Ni
2+
], 43.6 g L
1
[F
], and
195 g L
1
[NO
3
]. Slight uctuations in this calculation should
not have a major effect on PEI. The worst case scenario for
disposing this effluent is through lime neutralization. Table 6
below shows the stoichiometrically calculated material ow
used to estimate the PEI. The estimated environmental impact
for this method of crystallization effluent disposal is 727.7 PEI
per L of SPL.
The dehydration step of the (Fe,Cr)F
3
crystallohydrate
follows a simple thermal decomposition process. Argon gas is
introduced into a furnace to remove ambient air and the
furnace is incrementally heated to 400 C for extended periods
of time. The main source of environmental impact for this step
is the energy required to heat the furnace. The environmental
impact of the process gas is assumed to be negligible because it
is oen a byproduct of other gas rening processes. The impact
of heating is estimated from the required enthalpy for trans-
formation given by the DSC shown in Fig. 9. The enthalpy is
approximately 16.313 kJ per gram of crystallohydrate. Thus, the
potential environmental impact for dehydration is 0.16 PEI per
kg of (Fe,Cr)F
3
.
Before the combined PEI for the entire conversion process
can be calculated, the units per L of SPL need to be adjusted into
is per kg of (Fe,Cr)F
3
. This is accomplished by calculating the
amount of metal uoride hydrates present in the proposed
solution and then adjusting the value according to the yield
obtained through experimental trials. Finally, the hydrated
form of the crystal must be converted to its dehydrated form.
Therefore, 1 L of SPL has the potential to yield 0.032 kg of
(Fe,Cr)F
3
. The new PEI for the crystallization step is 22 740 PEI
per kg of (Fe,Cr)F
3
. The combined PEI for the whole conversion
process is roughly the same value because thermal decompo-
sition does not contribute much to environmental impact.
Discussion
Crystallization from iron only pickling solutions proved that an
appreciable yield of nanocyrstalline b-FeF
3
$3H
2
O could rela-
tively rapidly be produced. The crystallization process efficiency
could be further improved by optimization of temperature
control, crystallization time, crystal seed introduction, and HF
concentration. Forsberg and Rasmuson demonstrated that the
Fig. 12 Second cycle voltage profile of the samples synthesized from
pickling liquors compared to commercial FeF
3
performed at 60 C and
7.5 mA g
1
based on the weight of the nanocomposite.
Fig. 13 Discharge capacity (per gram of active FeF
3
)versus cycle
number of the samples synthesized from pickling liquors compared
commercial FeF
3
performed at 60 C and 7.5 mA g
1
based on the
weight of the nanocomposite.
Table 4 Inputs and outputs for SPL disposal used in the PEI calculation
Inputs Outputs
SPL (L) Ca(OH)
2
(g) Fe(OH)
3
(g) Cr(OH)
3
(g) Ni(OH)
2
(g) CaF
2
(g) Ca(NO
3
)
2
(g) OH
(g)
1 233.43 82.66 21.4 9.48 123.2 258.04 53.64
This journal is © The Royal Society of Chemistry 2014 RSC Adv.,2014,4,57098–57110 | 57107
Paper RSC Advances
crystal growth of iron uoride trihydrate can be optimized at a
temperature of 50 C, which in turn results in a higher yield.
18
Additionally, basic crystal growth kinetics indicate that if the
sample pickling liquor was allowed to crystalize for a longer
time period, then a larger yield of b-FeF
3
$3H
2
O is produced
because there is more time to reach thermodynamic equilib-
rium. Furthermore, the same result can be achieved by the
addition of seed crystals which provide sites of reduced surface
energy for growth. Finally, we performed preliminary experi-
ments indicating adding additional concentrated HF to sample
pickling liquor would result in not only a dramatically increased
yield, but also a shorter crystallization time. The tradeoffto this
approach is that precipitate composition could change depen-
dent on the altered solubility of the metal ions in solution.
Thermal treatment at 400 C was demonstrated to be effec-
tive in converting b-FeF
3
$3H
2
O into FeF
3
, but with a potential
slight oxidation of the surface. The synthesized FeF
3
material
successfully demonstrated electrochemical activity similar to
that of a commercial source and approached theoretical values.
In addition, the material obtained from prickling liquor also
provided improved capacity retention upon cycling compared to
the commercial sample. Such results may stem from the pres-
ence of an oxide surface layer based on previous results that
demonstrated the benets of the introduction of oxygen into
the iron uoride.
19
Using a similar method, nanoparticles of FeF
3
with a small
fraction of CrF
3
in solid solution were fabricated from Fe,Cr,Ni-
pickling liquors. Such material was observed to also be elec-
trochemically active, although of slightly lower capacity than
FeF
3
, but of further improved capacity retention. Further
optimization of this positive electrode material is possible
through renement of the techniques of Badway et al. and
Yabuuchi et al. which could result in a higher capacity battery
material with improved cycling capabilities.
20,21
The ability to form solid solutions with various contents of
Cr provides tuneability that is important to describing the
potential environmental impact of the conversion process as
the composition of the waste generated is directly related to the
composition of the material crystallized. If SPL is composed
mostly of a high concentration of chromium, then the potential
environmental impact for disposal or conversation would be
much higher because of chromium's much enhanced toxicity
compared to Fe. However, for the sake of this paper, the
composition of the waste generated by converting SPL into
(Fe,Cr)F
3
through crystallization is assumed to be 30.7 g L
1
[Fe
3+
], 7.7 g L
1
[Cr
3+
], 6 g L
1
[Ni
2+
], 43.6 g L
1
[F
], and 195 g
L
1
[NO
3
]. From this information the estimated environmental
impact for converting SPL into (Fe,Cr)F
3
can be calculated. It
should be noted that this calculation is solely based on the
disposal of the effluents produced from a conversion process
which yields 29% of (Fe,Cr)F
3
. Any additional processing steps
are ignored because of the very little amount of energy or added
chemicals required to complete them.
The worst case scenario for disposal of the conversion
effluent is through lime neutralization. Table 7 compares the
potential environmental impact of this process with the current
practices of SPL disposal and new FeF
3
production. The current
practices generate a slightly increased environmental cost,
indicating that the worst case scenario for converting SPL into
iron uorides is only marginally more environmentally sound.
Table 5 Inputs and outputs for ferric fluoride production used in the PEI calculation
10,17,24
Inputs Amount (kg t
1
FeF
3
) Outputs Amount (kg t
1
FeF
3
)
Fluorspar 1117.2–1170.4 CaSO
4
(marketable coproduct) 1969.8
H
2
SO
4
1383.2–1436.4 CaSO
4
(nonmarketable
coproduct)
2.662–26.62
Fe (scrap) 504.6 CaF
2
3.19–37.24
Cl
2
962.4 SO
42
0.372–10.64
Fluoride 0.0372–0.532
Si 0.0532–0.532
CO
2
2.934–17.604
H
2
1.071–19.071
Fe 0.07335–0.7.335
Zn 0.007335–2.2005
Heavy metals <0.0007335–0.8802
Solid waste 7.335–51.345
HCl 969.0
Total energy 6.17–10.60 GJ t
1
FeF
3
FeF
3
, anhydrous 1000
Table 6 -Inputs and outputs for crystallization effluent disposal used in the PEI calculation
Inputs Outputs
SPL (L) Ca(OH)
2
(g) Fe(OH)
3
(g) Cr(OH)
3
(g) Ni(OH)
2
(g) CaF
2
(g) Ca(NO
3
)
2
(g) OH
(g)
1 201.48 58.75 15.25 9.48 89.53 258.04 53.43
57108 |RSC Adv.,2014,4,57098–57110 This journal is © The Royal Society of Chemistry 2014
RSC Advances Paper
However, this margin can be dramatically increased through
optimization of the conversion process as all our calculated
values were based on an assumption of <30% yield. Simple
adaption of the crystallization techniques mentioned above of
iron-only pickling solutions to t Fe,Cr,Ni-pickling liquors will
have a signicant impact on the yield. Table 8 shows the rela-
tionship between crystallization yield and environmental
impact. This table is created by following the procedure high-
lighted in Section 6.3 above and assumes additional environ-
mental costs associated with achieving the higher
crystallization yield are insignicant. It shows that the
increased efficiency produces more (Fe,Cr)F
3
with less pickling
liquor to be discarded as waste, generating a lower PEI thereby
indicating a very signicant lower environmental impact. This
clearly demonstrates the signicant impact of yield on the
environmental and likely economic costs and should be the
focus of more detailed research. The optimization further
highlights how SPL conversion is the more environmentally-
friendly technology.
A best case scenario would be to effectively create a closed
loop process to recycle the conversion effluent and minimize
environmental impact. One approach proposed by a group at
Complutense University of Madrid recommends the use of KF
and KOH as a reagent to selectively precipitate iron and chro-
mium in uoride salts. Followed by hydrolysis of the precipi-
tates and neutralization of the effluent, this process produces
Fe(OH)
3
, Cr(OH)
3
, and Ni(OH)
2
as solids and a solution of K
+
,
F
, and NO
3
ions that can later be treated by membrane
ltration to recover valuable acids.
22
The focus of their process
is to recycle acids back towards pickling and recover nickel as a
marketable good. A slight modication of their process through
the use of HF as a reagent would shifocus towards the
production of FeF
3
$3H
2
O and CrF
3
$3H
2
O, precursors necessary
for battery material production. Subsequent dehydration of the
crystal hydrates would produce desired positive electrode
material. This process would remove much of the metal ions in
solution and produces a waste composed of hydrouoric and
nitric acid that needs to be reconstituted with new acids before
being returned into fresh pickling liquor stream. The only
contributing factors to the environmental impact would be the
production of the acids and energy, which have a negligible
impact when compared to what is saved from the lack of SPL
disposal.
While we demonstrated that spent pickling liquor similar to
that obtained by stainless steel manufacturers can be trans-
formed into a metal uoride material that can be used as a
positive electrode material for lithium batteries and that such
process is more environmentally friendly than current practices
for the disposal of the spent liquor and iron uoride fabrica-
tion, we are well aware that costs as well as demand will drive
the adoption of the proposed process. As a result we decided to
briey touch on the subject of costs without going into detail, as
it would fall out of the scope of this manuscript. From the
stainless steel industry perspective, even if the spent pickling
liquor were donated tremendous savings would be associated to
the elimination of the disposal process. From the battery
manufacturer, savings would be associated to lower costs for an
iron uoride material of potentially improved electrochemical
performance. Lower iron uoride fabrication costs would derive
from the elimination of the anhydrous HF-based process that
requires capital-intensive installations due to the corrosive HF
gas, as well as high utility and maintenance costs. Overall, the
proposed process would not only be more environmentally
sound but it would also bring about tremendous savings to both
industries. The battery manufacturers' economical benets
would also increase as the conversion process is optimized.
Finally, ultimate cost savings would be optimized by collocating
the facilities for the conversion of the SPL near stainless steel
plants where spent liquor is produced in order to reduce
transport costs, both economic and environmental.
Conclusion
It has been demonstrated that it is possible to convert spent
pickling liquor into a viable battery material through environ-
mentally benecial process of crystallization and dehydration.
Anhydrous FeF
3
of similar electrochemical performance to a
Table 7 Comparison of the potential environmental impact of each of scenario assuming a yield of 29% for the conversion of (Fe,Cr)F
3
$3H
2
Oto
(Fe,Cr)F
3
Scenario Process
Potential environmental
impact PEI per kg of battery material
Current industry practices SPL disposal 23 281
New FeF
3
production 1.27
Total 23 282.27
Worst case scenario Conversation of SPL into (Fe,Cr)F
3
22 740
Table 8 Effect of increasing crystallization yield on quantity of
material and environmental impact
Crystallization
yield (%)
kg of (Fe,Cr)F
3
/L
of SPL Total PEI
29% 0.032 22 740
40% 0.044 16 400
50% 0.055 13 010
60% 0.066 10 750
70% 0.077 9140
80% 0.088 7940
90% 0.099 6990
This journal is © The Royal Society of Chemistry 2014 RSC Adv.,2014,4,57098–57110 | 57109
Paper RSC Advances
commercially derived FeF
3
has been established using compo-
sitions similar to the effluent of the stainless steel pickling
process. Even though this material shows improved capacity
retention compared to commercial FeF
3
, much can be done to
further the performance of the material such as forming
nanocomposites with a conductive matrix in situ during the
crystallization process, incorporation of oxygen and optimiza-
tion of transition metal substitution, as demonstrated in the
literature. The use of the stainless steel effluent as a source raw
material for the fabrication of iron based transition metal
uorides may be of great benets to both industries and the
environment in the future as the process represents a decrease
in overall environmental impact, and economic cost as a waste
product is transformed into a value added product. For the
stainless steel industry, savings would come in the form of
dramatically lower disposal costs, while the battery manufac-
turers' economical benets would stem from lower material
cost.
Acknowledgements
The authors would like to acknowledge the Rutgers Energy
Institute and the Rutgers University Energy Storage Research
Group for nancial support.
References
1 N. Pereira, F. Badway, M. Wartelsky, S. Gunn and
G. G. Amatucci, J. Electrochem. Soc., 2009, 156, A407–A416.
2 T. Mahmood, C. Lindahl and R. Davis, US Pat., US4938945,
1988.
3 K. Tjus, R. Bergstrom, U. Fortkamp, K. Forsberg and
A. Rasmuson, Development of a Recovery System for Metals
and Acids from Pickling Baths using Nanoltration and
Crystallization, IVL Swedish Environmetnal Research
Institute Ltd., Stockholm, 2006.
4 M. Sartor, D. Buchloh, F. Rogener and T. Reichardt, Chem.
Eng. J., 2009, 153,50–55.
5 S. Myung, S. Sakurada, H. Yashiro and Y. Sun, J. Power
Sources, 2013, 223,1–8.
6 L. Liu, H. Guo, M. Zhou, Q. Wei, Z. Yang, H. Shu, X. Yang,
J. Tan, Z. Yan and X. Wang, J. Power Sources, 2013, 238,
501–515.
7 F. Badway, N. Pereira, F. Cosandey and G. G. Amatucci, J.
Electrochem. Soc., 2003, 150, A1209–A1218.
8 A. Gozdz, C. Schmutz, M. Tarascon and P. Warren, US Pat.,
US5418091, 1995.
9 P. Patnaik, Handbook of Inorganic Chemicals, McGraw-Hill,
New York, 2003.
10 Large Volume Inorganic Chemicals –Ammonia, Acids and
Fertilizers, European Commission Institute for Prospective
Technological Studies Sustainability in Industry, 2007.
11 A. Krepler, US Pat., US4252602, 1981.
12 J. B. Stephenson, J. C. Hogan and R. S. Kaplan, Environ. Prog.,
1984, 3,50–53.
13 F. H. Herbstein, M. Kapon and G. M. Reisner, Z. Kristallogr.,
1958, 171, 209–224.
14 D. G. Karraker and P. K. Smith, Inorg. Chem., 1991, 31, 1118–
1120.
15 B. Adamczyk, A. Hess and E. Kemnitz, J. Mater. Chem., 1996,
6, 1731–1735.
16 I. Plitz, F. Badway, J. Al-Sharab, A. DuPasquier, F. Cosandey
and G. G. Amatucci, J. Electrochem. Soc., 2005, 152, A307–
A315.
17 Commercial Source.
18 K. Forsberg and A. Rasmuson, J. Cryst. Growth, 2006, 296,
213–220.
19 W. Zhang, L. Ma, H. Yue and Y. Yang, J. Mater. Chem., 2012,
22, 24769–24775.
20 F. Badway, N. Pereira, F. Cosandey and G. G. Amatucci, J.
Electrochem. Soc., 2003, 150, A1318–A1327.
21 N. Yabuuchi, M. Sugano, Y. Yamakawa, I. Nakai,
K. Sakamoto, H. Muramatsu and S. Komaba, J. Mater.
Chem., 2011, 21, 10035–10041.
22 J. Hermoso, J. Dufor, J. L. Galvez, C. Negro and F. Lopez-
Mateos, Ind. Eng. Chem. Res., 2005, 44, 5750–5756.
23 Stainless Steel Manufacturer.
24 Additional Information submitted during information exchange
on Large Volume Inorganic Chemicals –Solids and Other
Industry, European Commission Institute for Prospective
Technological Studies Sustainability in Industry, 2005.
57110 |RSC Adv.,2014,4,57098–57110 This journal is © The Royal Society of Chemistry 2014
RSC Advances Paper