Performances of metal fluoride added carbon anodes with pre-electrolysis for electrolytic synthesis of NF3

Article abstract of DOI:10.1016/j.electacta.2010.10.053

Metal fluoride added carbon anodes treated by pre-electrolysis were investigated for electrolytic production of nitrogen trifluoride (NF ₃) in molten NH₄F·KF·4HF at 100 °C. The conditions for pre-electrolysis were first optimized using a graphite sheet anode as a model anode. The formation of fluorine-graphite intercalation compounds (fluorine-GICs) with semi-covalent C-F bonds, (C_xF) _n, on the MgF₂ and CaF₂ added carbon anode surface was accelerated by pre-electrolysis at potentials less than 4.0 V. Critical current densities (CCD) on the MgF₂ added carbon anodes pre-electrolyzed under various conditions were determined, and the highest CCD was 290 mA cm^-2 obtained for that pre-electrolyzed at 3.5 V for 500 C cm^-2. This anode was successfully used in the electrolysis at 100 mA cm^-2 for 290 h and the maximum NF₃ current efficiency was 55%. From these results, it was concluded that the metal fluoride added carbon anode treated by pre-electrolysis has a high potential for electrolytic production of NF₃ at higher current density.

Full text of DOI:10.1016/j.electacta.2010.10.053

Electrochimica Acta 56 (2011) 4425–4432
Contents lists available at ScienceDirect
Electrochimica Acta
journal homepage: www.elsevier.com/locate/electacta
Performances of metal fluoride added carbon anodes with pre-electrolysis for
electrolytic synthesis of NF3
Tomohiro Isogaia,d, Kazuhiro Hirooka , Tetsuro Tojo , Hitoshi Takebayashi ,
Morihiro Saito , Minoru Inaba
a
a
c
c
b
a,b,1
a,b,∗,1
, Akimasa Tasaka
Department of Applied Chemistry, Graduate School of Engineering, Doshisha University, 1-3 Miyakodani, Tatara, Kyotanabe, Kyoto 610-0321, Japan
Department of Molecular Chemistry and Biochemistry, Faculty of Science and Engineering, Doshisha University, 1-3 Miyakodani, Tatara, Kyotanabe, Kyoto 610-0321, Japan
Advanced Carbon Technology Center, Toyo Tanso Co., Ltd., 5-7-12 Takeshima, Nishiyodogawa-ku, Osaka 555-0011, Japan
b
c
d
Process Technology Department, Chemical Division, Daikin Industries, Ltd., 1-1 Nishihitotsuya, Settsu, Osaka 566-8585, Japan
a r t i c l e i n f o
a b s t r a c t
Article history:
Received 3 August 2010
Received in revised form 10 October 2010
Accepted 16 October 2010
Available online 28 March 2011
Metal fluoride added carbon anodes treated by pre-electrolysis were investigated for electrolytic pro-
◦
duction of nitrogen trifluoride (NF3) in molten NH4F·KF·4HF at 100 C. The conditions for pre-electrolysis
were first optimized using a graphite sheet anode as a model anode. The formation of fluorine-graphite
intercalation compounds (fluorine-GICs) with semi-covalent C–F bonds, (CxF)n, on the MgF2 and CaF2
added carbon anode surface was accelerated by pre-electrolysis at potentials less than 4.0 V. Critical
current densities (CCD) on the MgF2 added carbon anodes pre-electrolyzed under various conditions
Keywords:
Nitrogen trifruoride
Metal fluoride added carbon anode
Fluorine-graphite intercalation compounds
Semi-covalent C–F bond
Graphite fluoride
−2
were determined, and the highest CCD was 290 mA cm obtained for that pre-electrolyzed at 3.5 V for
−
2
−
2
5
00 C cm . This anode was successfully used in the electrolysis at 100 mA cm for 290 h and the maxi-
mum NF3 current efficiency was 55%. From these results, it was concluded that the metal fluoride added
carbon anode treated by pre-electrolysis has a high potential for electrolytic production of NF3 at higher
current density.
©
2010 Elsevier Ltd. All rights reserved.
1
. Introduction
on the carbon surface. On the other hand, fluorine-graphite inter-
calation compounds (fluorine-GICs) with semi-covalent C–F bonds,
(CxF)n (x > 3), which have a higher surface energy and a higher
electric conductivity, are also formed on the carbon anode during
electrolysis. It has been reported that the presence of fluorine-GICs
on the carbon anode surface effectively suppresses the anode effect
[2–25].
A large amount of nitrogen trifluoride (NF ) is currently pro-
3
duced and consumed as a dry etchant and a cleaning gas for CVD
processes in the electronic industry in Japan. Two methods are
known for electrolytic production of NF : One is the electrolysis
3
of molten NH F·2HF with nickel anode, and the other is the elec-
4
trolysis of molten NH F·KF·4HF with carbon anode [1]. The former
Because the formation of fluorine-GICs is catalyzed by metal flu-
orides such as LiF, AlF , MgF , NiF and so on [15], several attempts
4
method has an advantage that pure NF3 free from carbon tetraflu-
3
2
2
oride (CF ) can be obtained. However, a relatively high corrosion
rate of the nickel anode is critical problems, which raise the cost
using metal fluorides have been examined to suppress the anode
effect for several decades. It has been reported that addition of
metal fluorides into the melts is effective to suppress the anode
effect [16–18]. However, the concentration of the additive near the
anode surface decreases with time, because it progressively pre-
cipitates as sludge on the bottom of cell. Therefore, it is necessary
to add the metal fluoride periodically into the electrolyte in order
to maintain the concentration of the additive. Furthermore, the
repetitive addition of the metal fluoride to the electrolyte spoils
the performance of electrolysis because of an increase in the vis-
cosity of the electrolyte, and needs frequent removal of the sludge
from the bottom of cell. Groult et al. investigated detail the electro-
chemistry of modified carbon anodes prepared by impregnation
4
of NF . On the other hand, the latter method using carbon anode is
3
free from corrosion of the anode; however, it has drawbacks such
as the occurrence of the anode effect, disintegration of the anode,
and contamination of NF3 with a small amount of CF . Of these,
4
the most serious problem of the latter method to be solved is the
occurrence of the anode effect.
The anode effect is caused by formation of a graphite fluoride
layer with covalent C–F bonds, (CF)n, which has a very low surface
energy and reduces the effective surface area of the anode [2–5],
with solutions containing a precursor salt, such as Al(NO3) and
∗ Corresponding author. Tel.: +81 774 65 6592; fax: +81 774 65 6841.
3
Ni(NO3) [7,11,12]. Furthermore, the LiF impregnated anodes are
E-mail address: atasaka@mail.doshisha.ac.jp (A. Tasaka).
2
1
ISE member.
prepared by immersing into the pore of row carbon material into
0
013-4686/$ – see front matter © 2010 Elsevier Ltd. All rights reserved.
doi:10.1016/j.electacta.2010.10.053

4
426
T. Isogai et al. / Electrochimica Acta 56 (2011) 4425–4432
content to less than 0.02 wt% (ca. 20 mmol dm 3) [33], prior to
−
molten LiF under high pressure employing Hot Isostatic Pressing
Process. It has been reported to be much effective and useful than
the addition of LiF into the electrolyte [20–25].
electrochemical measurements. Anodic polarization curves were
obtained at a sweep rate of 200 mV s using an automatic polar-
ization system (Hokuto Denko Ltd., HZ-3000). The pre-electrolysis
was conducted at a constant current density of 5 mA cm or at
constant potentials of 2.3, 3.5, 4.0, 6.0, and 8.0 V with a poten-
tio/galvanostat HA-303 (Hokuto Denko Ltd.) and a coulomb meter
HF-201 (Hokuto Denko Ltd.). The electrolytic productions of NF3
were carried out galvanostatically at current densities in the range
−
1
Recently, metal fluoride added carbon anodes were developed
and investigated from viewpoints of surface energy and current
efficiency for NF production in molten NH F–KF–HF systems [26].
−
2
3
4
The results indicated that metal fluoride added carbon anodes are
−2
successfully used in the electrolysis at 10 mA cm for 200 h and the
maximum current efficiencies of NF3 are almost equal to pristine
carbon (Pristine-C) anode. In addition, it has been reported that pre-
electrolysis of a carbon anode at potentials lower than 2.3 V, where
fluoride anions cannot discharge, is effective for forming fluorine-
GICs with semi-covalent C–F bonds and results in an improvement
of CCD [27,28].
−
2
of 5–300 mA cm
Denko Ltd.).
with a potentio/galvanostat HA-303 (Hokuto
XRD and XPS analyses were conducted using RINT-2500 (Rigaku
Corp. Co., Ltd.) with Cu-K␣ radiation and using AXIS-165 (Shimadzu
Corp. Co., Ltd.) with Mg-K␣ radiation (1253.6 eV), respectively. For
XPS measurements, gold thin film was vapor-deposited on the
sample, and the binding energy of the gold 4f7/2 level at 84.0 eV
was used for calibration. Raman spectra were obtained with a T-
In the present study, anodic behavior of MgF2 and CaF2
added carbon (MgF /C and CaF /C) anodes were studied in the
2
2
NH F·KF·4HF melt. The conditions for pre-electrolysis were first
4
+
optimized using a graphite sheet anode as a model anode in the
64000 spectrometer (Horiba Jobin Yvon Inc.) using an Ar -ion laser
NH F·KF·3HF melt containing MgF or CaF . The effects of the pre-
(514.5 nm, 100 mW) as an excitation source. SEM images were
obtained with a cold field emission scanning electron microscope
S-4300 (Hitachi High-Technologies Corp.).
4
2
2
electrolysis conditions, such as anode potential and charge passed,
on the critical current density (CCD) on MgF /C anodes were inves-
2
tigated in the NH F·KF·4HF melt. Electrolysis was then conducted
The anode gas was analyzed with a gas chromatograph/mass
spectrometer (GCMS-QP2010, Shimadzu Corp. Co., Ltd.) and gas
chromatograph (G-2800T, Yanaco Analytical Systems Inc.) after
gaseous hydrogen fluoride and fluorine gas were removed by
passing through a tube filled with NaF pellets (Morita Chemi-
cal Industries, 99.9%) and activated alumina balls. Porapak-R and
Molecular Sieve 5A were used as column packings. Helium was
used as a carrier gas.
4
in the NH F·KF·4HF melt with the MgF /C anode treated with pre-
4
2
electrolysis under the optimized condition.
2
. Experimental
Molten fluorides of NH F·KF·4HF and NH F·KF·3HF were used
4
4
as electrolytes. They were prepared from KF·HF (Morita Chemical
Industries, 99.9%), NH F·HF (Morita Chemical Industries, 99.9%),
4
and extra pure anhydrous hydrogen fluoride (Morita Chemical
Industries, 99.99%) in sealed nickel vessels.
3. Results and discussion
In order to obtain the anodic polarization curves, MgF2 (5 wt%)
3.1. Anodic behavior of metal fluorides added carbon anodes
and CaF2 (4 wt%) added isotropic dense carbons (MgF /C and
2
CaF /C, Toyo Tanso Co., Ltd.) [26] and pristine isotropic dense car-
In order to optimize the conditions for pre-electrolysis, the
anodic behavior of the MgF /C and CaF /C were investigated by
2
−
3
bon (Pristine-C, FE-5, Toyo Tanso Co., Ltd., density: 1.69 g cm
,
2
2
polosity: 11.7 vol.%, flexural strength: 103 MPa) [20] with a surface
cyclic voltammetry. Fig. 1 shows anodic polarization curves of
2
◦
area of 1.0 cm were used as anodes. A nickel sheet with a surface
MgF /C and CaF /C in molten NH F·KF·3HF at 100 C and at a sweep
2
2
4
2
−1
area of ca. 40 cm was used as a cathode and the NH F·KF·4HF melt
rate of 100 mV s . In the first anodic sweep, a small peak was
observed at 1.8 V, which is due to the electrolysis of residual water
in the melt, i.e., the formation of graphite oxide, CzO(OH)y, oxygen,
and/or small amount of fluorine-GICs having semi-covalent C–F
bonds. It was followed by a plateau up to 4 V, where fluorine-GICs
having semi-covalent C–F bonds, (CxF)n (x > 3), with a higher sur-
face energy and a higher electric conductivity were formed [2–14].
The anodic current increased from 4.0 V, where the discharge of
fluoride ions took place. After peaking at 6.5–7.0 V, the current
decreased acutely because of the occurrence of the anode effect,
where insulating graphite fluoride with covalent C–F bonds, (CF)n
having an extremely low surface energy was formed on the carbon
4
◦
at 100 C was used as an electrolyte. A box-type cell made from PTFE
(
polytetrafluoroethylene) with a capacity of 0.8 dm3 described in
previous papers [27,29] was used.
In order to optimize the pre-electrolysis conditions, a graphite
2
sheet (Toyo Tanso Co., Ltd.) with a surface area of 0.5 or 1.0 cm was
2
used as the anode. The nickel sheet with a surface area of ca. 60 cm
◦
was used as the cathode. Here NH F·KF·3HF melt at 100 C was used
4
as an electrolyte instead of NH F·KF·4HF melt in order to suppress
4
the exfoliation of the graphite sheet anode during electrolysis. An
excess amount of MgF2 or CaF2 (Sigma–Aldrich Corp.) was added
in the electrolyte to be saturated. A beaker-type PTFE cell with a
3
capacity of 0.3 dm described in a previous paper [30] was used.
surface [2–5]. The peak current density at 6.5–7.0 V on the MgF /C
2
In order to investigate the CCDs and the current efficiency for
anode was higher than that on the CaF /C anode, so that we gave
2
NF3 formation, MgF /C and the Pristine-C samples with a surface
priority to former anode in following sections. Significant changes
were observed in the second sweeps. The peak and plateau in the
potential range of 1.8–4.0 V disappeared and the current due to the
discharge of fluoride ions started to flow at potentials higher than
5.0 V. A rapid decrease in current was also observed at potentials
higher than ca. 6.5 V.
From these results, it is concluded that the anodic reactions on
metal fluoride added carbon anode can be divided into four regions:
electrolytic oxidation of water in the melt (Region I), formation of a
fluorine-GIC film (Regions I and II), electrochemical fluorination of
ammonium cations and/or ammonia on the carbon anode covered
with a fluorine-GIC film (Region III), and the occurrence of the anode
effect (Region IV). These results agreed with the results in a previous
report on LiF impregnated carbon anode [27].
2
2
area of 10 cm were used as anodes. A cylindrical nickel cell with a
capacity of 1.5 dm3 [29,31] was used. The anode was located at the
center of the cell and the cell wall was utilized as the cathode. A
2
surface area of cathode is not less than 300 cm . The NH F·KF·4HF
4
◦
melt at 100 C was used as an electrolyte.
◦
A Cu/CuF electrode (402 ± 3 mV vs. H at 100 C) was employed
2
2
as a reference electrode in all electrochemical measurements [32],
but the potentials are referred to as volts vs. the hydrogen elec-
trode in the present study. The cells were placed in a dry nitrogen
chamber to remove moisture from the atmosphere. Since the chem-
icals contained water to some extent, electrolysis for dehydration
was conducted using an auxiliary carbon anode and the Ni counter
−2
electrode at ca. 10 mA cm for about ten days to reduce the water

T. Isogai et al. / Electrochimica Acta 56 (2011) 4425–4432
4427
5
4
3
2
1
00
00
00
00
00
0
Graphite
(002)
Graphite
(004)
(
a) MgF2/C
(a) MgF2
GIC
I
II
III
IV
8 V
6
4
V
V
1
2
1
st
nd
0th
2.3 V
0
1
2
3
4
5
6
7
8
9
Potential vs. RHE / V
Before
5
4
3
2
1
00
00
00
00
00
0
10
20
30
40
50
60
(
b) CaF /C
2
2θ/ degree
I
II
III
IV
Graphite (002)
Graphite (004)
GIC
(
b) CaF2
1
st
nd
0th
2
8
V
1
0
1
2
3
4
5
6
7
8
9
6
4
V
V
Potential vs. RHE / V
Fig. 1. Anodic polarization curves of (a) MgF2 and (b) CaF2 added carbon anodes in
molten NH4F·KF·3HF at 100 ◦C and at a sweep rate of 100 mV s
−1
2.3 V
.
10
20
30
40
50
60
3
.2. Formation of fluorine-GICs on graphite sheet anodes in
2θ/ degree
NH F·KF·3HF melts containing metal fluorides
4
Fig. 2. X-ray diffraction patterns of the graphite sheet anodes before and after poten-
In order to investigate the catalytic effect of metal fluorides on
GIC formation in detail, a graphite sheet anode was used as a model
−2
tiostatic electrolysis for 500 C cm in molten NH4F·KF·3HF containing 1 wt.% (a)
MgF2 and (b) CaF2 at 100 ◦C.
anode in molten NH F·KF·3HF containing a colloidal suspension of
4
metal fluorides and those anodes after electrolysis were analyzed
by XRD. The potential dependence on fluorine-GIC formation was
investigated on the graphite sheet anode. Fig. 2 shows XRD patterns
of graphite sheet anodes after potentiostatic electrolysis at 2.3, 4.0,
−2
5
00 C cm in the melt containing 1 wt.% MgF with a graphite sheet
2
anode. Fig. 3(a) and (b) shows XRD patterns of the graphite sheet
anode afterelectrolysis. Theintensityofthediffractionfromstage-7
fluorine-GICs increased with increasing charge passed. Further-
−
2
6
1
.0, and 8.0 V for 500 C cm
in molten NH F·KF·3HF containing
4
◦
wt.% MgF2 and CaF2 at 100 C. The stage number, n, was deter-
Graphite (002)
Graphite (004)
mined from the c-axis repeat distance of fluorine-GICs, Ic, using the
following equation [34]:
(
(
(
a) 2.3 V / 100 C cm-2
b) 2.3 V / 500 C cm-2
c) 3.5 V / 500 C cm-2
IC (nm) = 0.605 + 0.335(n − 1),
(1)
◦
◦
The diffraction lines at 2ꢀ = 26.5 and 54.7 , which are assigned
to hexagonal graphite (0 0 2) and (0 0 4) planes [27], were com-
monly observed for all samples. On the samples after kept at
Stage-7 GIC
Stage-6 GIC
2
.3 V (Region II) in both electrolytes, additional diffraction lines
◦
◦
assigned to fluorine-GICs were observed at 2ꢀ = 22.5 , 34.06 ,
◦
◦
◦
◦
3
7.92 , 40.82 , 50.8 , and 53.86 , which are assigned to (0 0 7),
(a)
(
0 0 1 0), (0 0 1 1), (0 0 1 2), (0 0 1 5), and (0 0 1 6) planes, respec-
tively, of stage-7 fluorine-GIC [27]. In contrast, these lines were
not observed after kept at 4.0, 6.0 (Region III), and 8.0 V (Region IV).
The other small diffraction lines were attributed to the residue of
(
b)
c)
(
molten NH F·KF·3HF. These results suggest that the stage number
4
1
0
20
30
40
50
60
of fluorine-GIC depends on anode potential and that the graphite
sheet anode should be polarized in Region II to enhance the forma-
tion of the fluorine-GICs.
2θ/ degree
Fig. 3. X-ray diffraction patterns of the graphite sheet anodes after electrolysis at
.3 V for (a) 100 C cm and (b) 500 C cm and (c) at 3.5 V for 500 C cm in molten
NH4F·KF·3HF containing 1 wt.% MgF2 at 100 C.
In order to investigate the effect of charge passed on the forma-
tion of fluorine-GICs, electrolysis was conducted at 2.3 V for 100 and
−2
−2
−2
2
◦

4
428
T. Isogai et al. / Electrochimica Acta 56 (2011) 4425–4432
more, Fig. 3(c) shows X-ray diffraction pattern of the graphite sheet
C4 C3 C2 C1
−2
anode polarized at 3.5 V for 500 C cm in molten NH F·KF·3HF con-
4
◦
taining 1 wt.% MgF2 at 100 C. As shown in Fig. 3(c), diffraction
lines observed at 2ꢀ = 31.0 , 35.4 , and 50.8 are assigned to (0 0 8),
C 1s level
◦
◦
◦
(
[
0 0 9), and (0 0 1 3) planes, respectively, of stage-6 fluorine-GIC
27]. In addition, week diffraction line at around 2ꢀ = 12.6 , which is
◦
assigned to (0 0 3) plane of stage-7 fluorine-GIC and/or (0 0 4) plane
of stage-6 fluorine-GIC, respectively, of stage-7 fluorine-GIC were
commonly observed. These results mean that higher potentials
in Region II accelerate formation of fluorine-GICs on the surface,
which gives a fluorine-GIC with a lower stage number.
3.3. Surface analysis of MgF2 added carbon anode after
pre-electrolysis
−
2
300
295
290
285
280
MgF /C anodes were polarized at 3.5 V for 500 C cm
as
2
◦
pre-electrolysis in molten NH F·KF·4HF at 100 C, and the pre-
Binding energy / eV
4
electrolyzed surfaces were analyzed by XPS and XRD. Fig. 4 shows
F2
F1
XPS spectra of the C 1s and F 1s levels of the MgF /C anode after
2
the pre-electrolysis. A large peak at 284.5 eV and a shoulder at ca.
2
86 eV were observed on the C 1s spectrum. They can be divided in
F 1s level
four symmetrical peaks at 284.8, 285.8, 288.0 and 289.5 eV, which
is denoted as C , C , C , and C , respectively, as shown in the upper
1
2
3
4
panel of Fig. 4 [35,36]. The binding energy and the full width at half
maximum (FWHM) of every component are shown in Table 1. The
C1 peak at 284.8 eV is assigned to the C–C bonds of the graphite bulk
and the C peak at 285.8 eV to the C–O and/or C–H bonds, which was
2
present even on the original graphite before pre-electrolysis. Peak
C at 288.0 eV and peak C at 289.5 eV are assigned to semi-covalent
3
4
C–F bonds and the covalent C–F bonds of graphite fluorides, respec-
tively, which were formed during the pre-electrolysis. The peaks
at 294 and 297 eV are assigned to K 2p3/2 and K 2p1/2, respec-
tively, from the residue of the melt. On the other hand, a large peak
appeared at ca. 687 eV on the F 1s spectrum and it can be divided in
two symmetrical peaks at 686.7 and 688.1 eV. The former and lat-
695
690
685
680
Binding energy / eV
ter peaks are assigned to the semi-covalent (F ) and the covalent
Fig. 4. XPS spectra of C 1s and F 1s level for the MgF2/C anode polarized at 3.5 V for
1
−
2
◦
5
00 C cm in molten NH4F·KF·4HF at 100 C (MgF2/C-PE).
C–F bonds (F ), respectively. These results revealed that fluorine-
2
GICs were produced during pre-electrolysis at 3.5 V, together with
a smaller amount of graphite fluoride.
3.4. Effect of pre-electrolysis condition on critical current
densities of MgF2 added carbon anodes
Fig. 5 compares X-ray diffraction patterns of MgF /C anodes
2
before and after pre-electrolysis at 3.5 V. Two broad diffraction lines
◦
◦
were observed at 2ꢀ = 25 and 43 and are assigned to the graphite
0 0 2) and (1 0) planes, respectively [37–39]. Some sharp diffrac-
To optimize the condition for pre-electrolysis of MgF /C anodes,
2
◦
(
CCDs on those were determined in molten NH F·KF·4HF at 100 C.
4
◦
◦
◦
◦
◦
◦
tion lines observed at 2ꢀ = 27.4 , 35.4 , 40.6 , 43.9 , 53.7 , and 56.4 ,
are assigned to the (1 1 0), (1 0 1), (1 1 1), (2 1 0), and (2 2 0) planes,
respectively, of MgF2 in the carbon anode. The diffraction lines
also observed at 2ꢀ = 34.9 , 38.6 , and 45.7 are assigned to (0 0 9),
0 0 1 0), and (0 0 1 2) planes, respectively, of stage-6 fluorine-GIC
27]. The results obtained by XRD are in accordance with those by
XPS. From these results, it is concluded that the addition of MgF2,
not only in the melt, but also in the carbon anode effectively cat-
alyzes the formation of fluorine-GICs during the pre-electrolysis at
The composition of electrolyte used in this section was slightly
different from that in Sections 3.2 and 3.3. As the vapor pres-
sure of HF over the NH F·KF·4HF melt is higher than that over
4
◦
◦
◦
the NH F·KF·3HF melt, the graphite sheet anode was completely
4
−
2
(
[
expanded only after 30 C cm by formation of fluorine-GICs and
electrolysis was terminated [25]. Conversely, as the presence of
KF in the melts reduces the current efficiency of NF3 formation
on the LiF impregnated carbon anode, it is considered that the
NH F·KF·4HF melt is suitable for electrolytic synthesis of NF . Fig. 6
4
3
3
.5 V.
shows galvanostatic polarization curves of a Pristine-C and MgF /C
2
Table 1
Binding energy (BE), full width at half maximum (FWHM), component and assignment of each peak separated from XPS spectra of C 1s and F 1s levels.
Component
BE (eV)
Assignment
MgF2/C-PE (Fig. 4)
BE (eV)
MgF2/C-PE-AE (Fig. 9)
BE (eV)
FWHM (eV)
FWHM (eV)
C1s
C1
C2
C3
C4
284.8
285.8
288.0
289.5
C–C
C–O, C–H
Semi-ionic C–F bond
Covalent C–F bond
284.8
285.8
288.0
289.5
1.57
1.48
2.56
2.76
284.7
285.8
–
1.41
1.72
–
289.5
4.54
F1s
F1
F2
686.7
688.1
Semi-ionic C–F bond
Covalent C–F bond
686.4
688.0
2.21
2.88
–
–
2.77
688.0

T. Isogai et al. / Electrochimica Acta 56 (2011) 4425–4432
4429
100
MgF2
b) 3.5V / 500C cm-2
(
N F
Electrolyte
GIC
NF3
N2
O2
2
4
N F
80
2
2
N O
2
6
4
2
0
0
0
(
a) Without pre-electrolysis
0
10
20
30
40
50
60
0
50
100
150
200
250
2θ / degree
Time / h
Fig. 5. X-ray diffraction patterns of MgF2/C anodes (a) before pre-electrolysis and
Fig. 7. Changes in current efficiencies for the constituents in anode gas with time
−2
◦
(
b) after pre-electrolysis at 3.5 V for 500 C cm in molten NH4F·KF·4HF at 100 C.
on MgF2/C-PE anode at 100 mA cm−2 in molten NH4F·KF·4HF at 100 ◦C.
1
0
−
2
−2
−2
charge passed, i.e., 75 mA cm for 100 C cm and 190 mA cm
9
8
7
6
5
4
3
2
1
0
−2
for 500 C cm . As shown in Fig. 3(a) and (b), the amount of stage-7
fluorine-GIC increased with increasing the charge passed for pre-
electrolysis. Therefore, it is concluded that the charge passed during
pre-electrolysis is one of the most important parameters for sup-
pressing the anode effect.
−
2
Furthermore, pre-electrolysis at 3.5 V for 500 C cm
icantly increased CCD to 290 mA cm
signif-
−
2
on the MgF /C anode.
2
This fact suggests that stage-6 fluorine-GIC was generated on
the MgF2/C anode by pre-electrolysis at 3.5 V similarly to the
graphite sheet anode. As shown in Fig. 3(c), the amount of fluorine-
GICs increased and the stage number of fluorine-GICs decreased
with increasing the anode potential not less than 4.0 V for pre-
electrolysis at a given charge passed. Therefore, it is concluded
that the anode potential for pre-electrolysis is another important
parameter for suppressing the anode effect and that best condition
Pristine carbon anode
Without pre-electrolysis
With pre-electrolysis at 2.3 V for 100 C cm-
2
2
With pre-electrolysis at 2.3 V for 500 C cm-
With pre-electrolysis at 3.5 V for 500 C cm
-2
0
50
100
150
200
250
300
Current density / mA cm-2
−
2
for pre-electrolysis in present study is at 3.5 V for 500 C cm on
the MgF2/C anode.
Fig. 6. Galvanostatic polarization curves of Pristine-C and MgF2/C anodes with and
without pre-electrolysis in molten NH4F·KF·4HF at 100 ◦C.
anodes treated with and without pre-electrolysis under various
conditions. The values of CCD obtained from the anodic polarization
curves are summarized in Table 2 together with those found in the
literature [17,18,20,22,27,28]. Without pre-electrolysis, the CCD on
3.5. Electrolytic production of NF3 using MgF /C-PE
2
Since the electrolytic NF3 production using nickel anode and
fluorine (F ) production using carbon anode are operated at
2
−2
−2
the MgF /C anode was 42 mA cm , which was slightly higher than
100 mA cm in the industry, it is very important to develop carbon
2
−2
−2
that of the Pristine-C anode (30 mA cm ). Pre-electrolysis at 2.3 V
was effective for improving CCD, and CCD increased with increasing
anodes that can maintain at least 100 mA cm in the electrolytic
production of NF3.
Table 2
Summary of CCDs in electrolytic production of NF3 and F2.
Electrolyte
Anode
Pre-electrolysis
–
Stage number of GIC
CCD (mA cm−2)
30
References
[27]
NH4F·KF·4HF
Pristine carbon
2
2
2
2
.3 V/250 C
.3 V/500 C
.3 V/1000 C
.3 V/500 C, 4.0 V/500 C
50
50
50
7, 6
5
122
LiF impregnated carbon
MgF2 added carbon
–
55
[24]
–
2
2
3
42
78
190
290
Present study
.3 V/100 C
.3 V/500 C
.5 V/500 C
7
7
6
KF·2HF
Pristine carbon
–
–
–
–
–
27
>60
>60
>60
>40
[21]
[20]
[24]
LiF impregnated carbon
LiF–NaF impregnated carbon
AlF3–NaF impregnated carbon
LiF–CaF2 added carbon

4
430
T. Isogai et al. / Electrochimica Acta 56 (2011) 4425–4432
1
0
9
8
7
6
5
4
3
2
1
0
C4
C2 C1
C 1s level
0
50
100
150
200
250
300
300
295
290
285
280
Time / h
Binding energy / eV
Fig. 8. Variation of the potential of MgF2/C-PE anode during electrolysis at
−2
◦
1
00 mA cm in molten NH4F·KF·4HF at 100 C.
F2
Electrolysis was carried out at 100 mA cm−2 using the MgF /C
F 1s level
2
−
2
anode with pre-electrolysis at 3.5 V for 500 C cm
in molten
◦
NH F·KF·4HF at 100 C (MgF /C-PE). Fig. 7 shows the changes in
4
2
current efficiency for the constituents in the anode gas with time.
The anode gas was composed of NF , N , O , N F , N F , and N O.
3
2
2
2
2
2
4
2
Although N gas was the main product in the beginning of electrol-
2
ysis, the current efficiency for NF3 formation increased with time,
reached its maximum (55.5%) at ca. 50 h and then became almost
constant. The maximum current efficiencies for NF3 formation on
Pristine-C [26], the MgF /C, and the MgF /C-PE are summarized
2
2
in Table 3 together with the current efficiencies for other gaseous
products. Although the current density on MgF /C-PE was ten times
2
higher than those on Pristine-C and MgF /C, the current efficiency
for NF3 formation on the former was almost the same as those on
the latter two.
2
695
690
685
680
Binding energy / eV
Fig. 8 shows the variation of the anode potential during elec-
−
2
−2
Fig. 9. XPS spectra of the MgF2/C anode with pre-electrolysis at 3.5 V for 500 C cm
trolysis at 100 mA cm using MgF /C-PE in molten NH F·KF·4HF
2
4
after electrolysis at 100 mA cm for 300 h in molten NH4F·KF·4HF at 100 ◦C (MgF2/C-
−2
◦
at 100 C. The anode potential was ca. 5.5 V for 50 h in the begin-
ning, then increased gradually to 6.2 V, and stagnated at 6.2 V from
PE-AE).
1
00 to 290 h. At 290 h, the anode potential rose suddenly up to
ysis in Fig. 9.SEM images of MgF /C, MgF /C-PE, and MgF /C-PE-AE
over 10 V, where the anode effect took place and electrolysis was
terminated. It is well known that the anode effect is enhanced in
inadequately dehydrated melt [18]. It is not easy to eliminate suf-
ficiently a trace of water in the melt containing fluorides such as
2
2
2
are shown in Fig. 11. The morphology of MgF /C was highly rugged
2
and had many macropores, which are characteristic to an isotropic
dense carbon. No significant difference appeared in the texture of
NH F. Therefore, it is expected that electrolysis can be continued at
4
−
2
1
00 mA cm for longer duration without the anode effect, when
well dehydrated (less than 0.006 wt% (ca. 6 mmol dm ) [33]) melt
is used as an electrolyte.
−
3
MgF2
electrolyte
3
.6. Surface analysis of MgF /C-PE-AE.
2
Fig. 9 shows XPS spectra of the C 1s and F 1s level of the MgF /C
2
used in electrolysis for 300 h in Fig. 8 (MgF /C-PE-AE). The result of
2
peak fitting and binding energy and the full width at half maximum
(
FWHM) of every component is shown in Table 1. Compared with
XPS spectra for the pre-electrolyzed MgF /C anode before elec-
2
trolysis (MgF /C-PE) in Fig. 4, the C3 peak on the C 1s spectrum
2
and the F1 peak on the F 1s spectrum, which are assigned to the
semi-covalent C–F bonds of fluorine-GICs, disappeared. These facts
suggest that the layer of fluorine-GICs formed in pre-electrolysis
changed to graphite fluoride with the covalent C–F bonds.
10
20
30
40
50
60
2
θ
/ degree
Fig. 10 shows X-ray diffraction pattern of the anode after
electrolysis (MgF /C-PE-AE). When compared with the XRD pat-
tern in Fig. 5, no diffraction line assigned to fluorine-GICs was
observed. This fact well agrees with the results of the XPS anal-
2
Fig. 10. X-ray diffraction patterns of MgF2/C anode with pre-electrolysis at 3.5 V for
−2
−2
5
00 C cm after electrolysis at 100 mA cm for 300 h in a molten NH4F·KF·4HF at
◦
100 C (MgF2/C-PE-AE).

T. Isogai et al. / Electrochimica Acta 56 (2011) 4425–4432
4431
Table 3
The current efficiencies for the constituents in the anode gas using various carbon anodes in molten NH4F·KF·4HF at 100 C.
◦
Current density (mA cm−2)
Current efficiency (%)
Anode material
N2
O2
NF3
N2F2
N2F4
N2O
Overall
Pristine carbon [26] (Pristine-C)
MgF2 added carbon without pre-electrolysis [26] (MgF2/C)
MgF2 added carbon with pre-electrolysis (MgF2/C-PE)
10
10
100
23.6
28.0
12.6
8.22
8.28
24.1
51.4
53.2
55.7
3.51
1.62
0.12
4.56
3.27
2.64
3.15
1.51
0
94.4
96.0
95.2
the carbon surface after pre-electrolysis as shown in Fig. 11(b). In
contrast, the surface morphology of MgF /C-PE-AE was character-
2
ized by shiny mirror-like surface, as shown in Fig. 11(c). Groult et al.
have reported that activation of carbon at 40 V in KF·2HF modified
strongly the surface morphology by electropolishing with sparks
and heat [13,14]. Based on this fact, it is considered that smoothing
and cleaning of the anode surface may be induced by occurrence of
the anode effect.
4. Conclusions
Metal fluoride added carbon anodes treated by pre-electrolysis
were investigated for electrolytic production of nitrogen trifluo-
◦
ride (NF ) in molten NH F·KF·4HF at 100 C. The conditions for
3
4
pre-electrolysis were first optimized using a graphite sheet anode
as a model anode. The formation of fluorine-graphite intercalation
compounds (fluorine-GICs) with semi-covalent C–F bonds, (CxF)n,
on the MgF and CaF added carbon anode surface was accelerated
2
2
by pre-electrolysis at potentials less than 4.0 V. The stage number
of fluorine-GICs was dependent on the condition of electrolysis
such as charge passed and anode potential, and was gradually
decreased with the increase in charge passed. Critical current den-
sities (CCD) on the MgF2 added carbon anodes pre-electrolyzed
under various conditions were determined, and the highest CCD
−
2
was 290 mA cm
obtained for that pre-electrolyzed at 3.5 V for
−
2
5
1
00 C cm . This anode was successfully used in the electrolysis at
00 mA cm for 290 h and the maximum NF3 current efficiency
−
2
was 55%. The anode effect took place after electrolysis for 290 h,
because a trace of water in the melt enhanced formation of the (CF)n
layer. From these results, it was concluded that the metal fluoride
added carbon anode treated by pre-electrolysis has a high potential
for electrolytic production of NF3 at high current densities.
Acknowledgments
This work was supported by a grant to the Research Center for
Advanced Science and Technology at Doshisha University from the
Ministry of Education, Japan and scholarship contribution of Mitsui
Chemicals, Inc.
References
[
[
[
1] A. Tasaka, J. Fluorine Chem. 128 (2007) 296.
2] A. Tasaka, PhD Thesis, Kyoto University, 1970.
3] M. Nishimura, A. Tasaka, K. Nakanishi, N. Watanabe, Denki Kagaku (Electro-
chemistry) 38 (1970) 294.
[
[
4] A. Tasaka, N. Watanabe, Yoyu-en (Molten Salts) 13 (1970) 152.
5] H. Groult, S. Durand-Vidal, D. Devilliers, F. Lantelme, J. Fluorine Chem. 107
(
2001) 247.
[
[
6] H. Groult, J. Fluorine Chem. 119 (2003) 173.
7] H. Groult, D. Devilliers, M. Vogler, C. Hinnen, P. Marcus, F. Nicolas, Electrochim.
Acta 38 (1993) 2413.
[
[
8] H. Groult, D. Devilliers, S. Durand-Vidal, F. Nicolas, M. Combel, Electrochim.
Acta 44 (1999) 2793.
9] H. Touhara, F. Okino, Carbon 38 (2000) 241.
[
10] A. Tasaka, H. Itoh, T. Isogai, H. Aio, T. Tojo, J. Electrochem. Soc. 138 (1991)
21.
4
Fig. 11. SEM images of MgF2/C anodes (a) without pre-electrolysis and with pre-
[11] H. Groult, D. Devilliers, M. Vogler, M. Chemla, J. Appl. Electrochem. 24 (1994)
−2
electrolysis at 3.5 V for 500 C cm (b) before (MgF2/C-PE) and (c) after electrolysis
870.
at 100 mA cm for 300 h in a molten NH4F·KF·4HF at 100 ◦C (MgF2/C-PE-AE).
−2
[12] H. Groult, D. Devilliers, J. Fluorine Chem. 87 (1998) 151.

4
432
T. Isogai et al. / Electrochimica Acta 56 (2011) 4425–4432
[
[
13] H. Groult, F. Lantelme, I. Crassous, C. Labrugère, A. Tressaud, C. Belhomme, A.
Colisson, B. Morel, J. Electrochem. Soc. 154 (2007) C331.
14] I. Crassous, H. Groult, F. Lantelme, D. Devilliers, A. Tressaud, C. Labrugère,
M. Dubois, C. Belhomme, A. Colisson, B. Morel, J. Fluorine Chem. 130 (2009)
[26] A. Tasaka, M. Kodama, T. Tojo, H. Takebayashi, M. Inaba, Abstract of Interna-
tional Conference on Fluorine Chemistry’04 Kyoto, Kyoto, Japan, May 9–11,
2004, p. 63.
[27] A. Tasaka, M. Yamanaka (Miki), E. Morimoto, S. Nagamine, A. Mimoto, T.
Fujukawa, M. Abe, A. Kobayashi, H. Takebayashi, T. Tojo, M. Inaba, J. New Mat.
Electrochem. Syst. 9 (2006) 297.
1
080.
[
[
15] T. Nakajima, in: T. Nakajima (Ed.), Fluorine–Carbon and Fluoride–Carbon Mate-
rials, Chemistry, Physics and Application, Marcel Dekker Inc., New York, USA,
[28] A. Tasaka, Electrochemistry 75 (2007) 934.
1
995, p. 3.
16] N. Watanabe, M. Ishii, S. Yoshizawa, J. Electrochem. Soc. Jpn. 29 (1961)
17.
[29] A. Tasaka, H. Kobayashi, M. Negami, M. Hori, T. Osada, K. Nagasaki, T. Ozaki, H.
Nakayama, K. Katamura, J. Electrochem. Soc. 144 (1997) 192.
[30] Y. Shodai, M. Inaba, K. Momota, T. Kimura, A. Tasaka, Electrochim. Acta 49
(2004) 2131.
[31] A. Tasaka, K. Mizuno, A. Kamata, K. Miki, J. Fluorine Chem. 57 (1992) 121.
[32] A. Mimoto, T. Miyazaki, J. Yamashita, S. Nagamine, M. Inaba, A. Tasaka, J. Elec-
trochem. Soc. 153 (2006) D149.
1
[
[
[
17] D. Devilliers, B. Teisseyre, M. Chemla, Electrochim. Acta 35 (1990) 153.
18] T. Nakajima, T. Ogawa, N. Watanabe, J. Electrochem. Soc. 134 (1987) 8.
19] T. Tojo, J. Hiraiwa, M. Dohi, Y. Chong, N. Watanabe, J. Fluorine Chem. 54 (1991)
1
36.
[
[
20] T. Tojo, J. Hiraiwa, M. Dohi, Y. Chong, J. Fluorine Chem. 57 (1992) 93.
21] T. Tojo, J. Hiraiwa, N. Watanabe, Kagaku Denki (Electrochemistry) 66 (1998)
[33] A. Tasaka, M. Hori, H. Harada, H. Inoue, M. Ohashi, Sci. Eng. Rev. Doshisha Univ.
34 (1993) 153.
5
63.
[34] T. Mallouk, N. Bartlett, J. Chem. Soc. Chem. Commn. (1983) 103.
[35] A. Tressaud, E. Durand, C. Labrugere, J. Fluorine Chem. 125 (2004) 1639.
[36] T. Nakajima, in: T. Nakajima (Ed.), Fluorine–Carbon and Fluoride–Carbon Mate-
rials, Chemistry, Physics and Application, Marcel Dekker Inc., New York, USA,
1995, p. 14.
[37] T. Nakajima, M. Molinier, M. Motoyama, Carbon 29 (1991) 429.
[38] J.M. Jiménez Mateos, E. Romero, C. Gómez De Salazar, Carbon 31 (1993)
1159.
[
22] H. Takebayashi, J. Hiraiwa, T. Tojo, A. Tasaka, N. Watanabe, Yoyuen (Molten
Salts) 46 (2003) 51.
23] T. Tojo, Yoyuen (Molten Salts 48 (2005) 28.
24] N. Watanabe, Y.-B. Chong, M. Aritsuka, M. Kanemaru, J. Mikami, Proceedings
of the 4th Japan–China Bilateral Conference on Molten Salt Chemistry and
Technology, Kyoto, Japan, November 8–13, 1992, p. 111.
25] A. Tasaka, T. Kawagoe, A. Takuwa, M. Yamanaka, T. Tojo, M. Aritsuka, J. Elec-
trochem. Soc. 145 (1998) 1160.
[
[
[
[39] Y. Sato, T. Kume, R. Hagiwara, Y. Ito, Carbon 41 (2003) 351.