Professional Documents
Culture Documents
net/publication/264809538
CITATIONS READS
22 384
5 authors, including:
Some of the authors of this publication are also working on these related projects:
Development of lower temperature operating sodium beta-alumina batteries for grid scale energy storage View project
All content following this page was uploaded by Jin Yong Kim on 16 April 2015.
This journal is ª The Royal Society of Chemistry 2013 J. Mater. Chem. A, 2013, 1, 14935–14942 | 14935
View Article Online
Results
Planar button cell conguration C-rate (xed-capacity cycling tests)
The button cell consists of the battery case (cathode and anode The electrochemical test results of cells that were tested at 280 C
case), alpha-alumina (99.5% purity) xture, and BASE (3 cm2 at two different C-rates (C/3 and 1 C) with a cycling capacity of 90
active area) as reported in our previous publication.10 mA h (20–80% SoC) are shown in Fig. 1. These cells contained the
Composite yttria-stabilized zirconia (YSZ)/b00 alumina (BASE) cathode granules with a Ni/Na molar ratio of 1.82. No degradation
discs were fabricated adopting the vapour phase conversion in capacity was observed in either C/3 or 1 C tests up to 60 cycles
process, which has been reported in the literature.12,13 The (refer to Fig. 1a). The specic discharge energy density of cells
volume ratio of b00 alumina to YSZ in the composite BASE was cycled with 90 mA h was 150 W h kg1, which varied slightly for
7 : 3. The thickness of the BASE disc aer conversion and creep- different C-rates (Fig. 1b). The overall energy efficiencies at C/3
attening was 500 mm. Before loading any active materials, a and 1 C were 96% and 93% in Fig. 1c, respectively. Even though
BASE disc was glass-sealed to an a-alumina xture. The anode energy efficiency drops slightly with C-rate, the cell tested at 1 C
side of the BASE disc was further treated with aqueous lead still revealed high efficiency, over 90%. The voltages at the end of
acetate (Pb[CH3COO]2) solution and heat treated at 400 C charge (EOC) and the end of discharge (EOD) are shown in Fig. 1d.
under nitrogen in order to improve the wettability of the Although the cells tested at C/3 and 1 C did not show any capacity
sodium melt.14 degradation over 60 cycles (Fig. 1a), the cell tested at 1 C shows a
Molybdenum discs and springs were used as current faster increase in EOC voltage (0.8 mV per cycle) compared to the
collectors for the cathode, due to the electrochemical stability of cell tested at C/3 (0.2 mV per cycle), indicating a faster increase in
metallic molybdenum under the operating conditions of the cell polarization and cell degradation.
cathode.15 Current collectors for the anode were made of The cells were disassembled aer 60 cycles in the glove box
stainless steel discs and springs. and the fracture surfaces of the cathodes were examined using
A button cell was then assembled using compressive sealing SEM/EDS (refer to Fig. 2). The brighter spots appearing in back-
with a gold O-ring (0.5 mm thick) for the cathode side and a scattered electron images (Fig. 2a and c) represent Ni particles
silver O-ring (0.5 mm thick) for the anode side in an atmo- according to EDS point analysis. Corresponding images of Ni
sphere-controlled glove box purged with argon. Aer being mapping are shown in Fig. 2b and d. The typical Ni particle size
assembled, the button cell was tested at 280 C in air. for the cell tested at C/3 was 1–2 mm (Fig. 2a and b), similar to that
of raw Ni powder. However, a signicant Ni particle growth
Battery tests
Cell tests were carried out using an Arbin potentiostat (MSTAT
8000) controlled by vendor supplied soware (MITS Pro). The
cell was initially charged up to 2.8 V to obtain a full capacity at a
constant current of 10 mA and discharged back to 30 mA h
residual capacity (20% state of charge, SoC) at the same
current. Three different test schedules were used to identify the
effects of various parameters. First, a xed-capacity cycling test
was conducted with a xed cycling capacity window (60% of
theoretical capacity) to analyze the inuence of C-rate, Ni/NaCl
ratio, and voltage at the end of charge (EOC). Second, a deep-
cycling test was also performed between cutoff voltages of 2.4 V
and 2.7 V in order to examine the effects of large-capacity
cycling on cell degradation. Finally, a controlled discharge test
(partial discharge test) was carried out between cutoff voltages
Fig. 1 Battery testing data on cells with Ni/NaCl ¼ 1.82. (a) Capacity plot for the
of 2.5 V and various high limits to understand the effects of SoC
button cells over 60 cycles. (b) Specific energy density (W h kg1) for cells at C/3
at EOC on cell performance. The same current of 30 mA (C/3, and 1 C rates. (c) Energy efficiency plot for button cells at C/3 and 1 C rates. (d)
10 mA cm2) was used for charging and discharging except for Voltages for EOC and EOD on 3 cm2 button cells at C/3 (30 mA) and 1 C (90 mA)
high-rate tests at 1 C (90 mA, 30 mA cm2). rates.
14936 | J. Mater. Chem. A, 2013, 1, 14935–14942 This journal is ª The Royal Society of Chemistry 2013
View Article Online
Fig. 2 (a and c) SEM images (backscattering, 1000) for cathode granules from
the cell (Ni/NaCl ¼ 1.82) tested at C/3 and 1 C rates, respectively. (b and d) Nickel Ni/NaCl ratios (xed-capacity cycling tests)
mapping on the SEM images shown in (a) and (c). To examine the effects of Ni/NaCl ratios on cell degradation,
cells with three different molar ratios of Ni to NaCl (1.82, 1.7
and 1.5) in cathode granules were tested. Cell tests were
(10 mm) was observed for the cell tested at 1 C (Fig. 2c and d). To
observe the particle size of NaCl, Na mapping was also conducted
on the fracture surfaces of cathode materials, as shown in Fig. 3.
NaCl crystals are represented by brighter areas in Na mapping
images (Fig. 3a and c) due to the higher Na density of NaCl
compared to NaAlCl4. Some dark areas shown in the Na mapping
are due to the uneven fracture surfaces. Despite the difference in
C-rates, a similar size distribution of NaCl particles (30–50 mm)
was observed on the cells tested at C/3 and 1 C. This result implies
that C-rate made no signicant impact on the growth of NaCl
particles even though there was some growth of NaCl, which had
an initial size of 10 mm. Fig. 4 Voltage profile of the 30th cycle for 1 C (90 mA) tests (1 C charge and C/3
discharge) with the voltage at EOC fixed at 2.675 V (black) and 2.8 V (red),
respectively. The charging process consists of two steps, a constant-current
EOC voltages (xed-capacity cycling tests) (90 mA) charging process followed by a constant-voltage charging process once
To identify the effects of EOC voltage on cell degradation, two the voltage reaches the voltage limits.
Fig. 5 SEM images of cathode granules from the cells cycled with different EOC
voltage limits. EOC voltage limit of 2.675 V: (a) Ni particles (1000) and (c) NaCl
Fig. 3 (a and c) Na mapping for the cross-section of cathode granules from cells particles (300). EOC voltage limit of 2.8 V: (b) Ni particles (1000) and (d) NaCl
tested at C/3 and 1 C rates, respectively. (b and d) SEM images of NaCl particles. particles (300).
This journal is ª The Royal Society of Chemistry 2013 J. Mater. Chem. A, 2013, 1, 14935–14942 | 14937
View Article Online
performed for each cathode formula under two different test Ni/NaCl ratios (deep cycling)
schemes: (1) xed-capacity cycling tests at 30 mA (C/3) with a
Fig. 8 shows results of deep-cycling tests conducted between 2.4
cycling capacity window of 90 mA h (20–80% SoC) and (2)
and 2.7 V at a constant current of 30 mA. The initial capacity of
deep-cycling tests at 30 mA between 2.4 V (EOD) and 2.7 V
the cells run with the deep-cycling schedule was 120 mA h
(EOC).
(80% of the theoretical capacity) for all Ni/NaCl ratios. Unlike
The electrochemical behaviours of cells tested with the xed-
the cells tested with the xed cycling capacity (90 mA h, 60%
capacity cycling schedule are shown in Fig. 6. As expected,
of theoretical capacity), the cells tested for deep capacity cycles
overpotentials during charging and discharging slightly revealed clear capacity fades (Fig. 8b). The voltage proles vs.
increase with decrease in the Ni/NaCl ratio (Fig. 6a) since the
Published on 23 September 2013. Downloaded by PNNL Technical Library on 13/03/2014 16:56:30.
SoC collected during the 30th cycle clearly indicate that capacity
lower Ni content in the cathode leads to higher ohmic resis-
fade is mainly due to the reduced charging capacity (Fig. 8a).
tance. No capacity fade was observed for the three cells since
The sizes of Ni and NaCl particles were also measured from
none of the cells reached the cutoff voltages of 2.8 V (for
the fracture surfaces of cathodes aer 60 cycles using SEM
charging) and 1.8 V (for discharging). The EOC and EOD volt-
(Fig. 9). As described earlier, the bright spots in Fig. 9a, c and e
ages vs. cycle numbers are shown in Fig. 6b. The cell degrada-
(back scattering, 1000 magnication) represent metallic Ni
tion rate indicated by an increase in EOC voltage is higher for a
cell with a Ni/Na ratio of 1.5, while cells with Ni/Na ratios of
1.82 and 1.7 reveal quite stable performance. As shown in
Fig. 2a, 7a and c, the average Ni particle sizes measured from
SEM analysis aer 60 cycles were 2, 2, and 5 mm for Ni/NaCl
ratios of 1.82, 1.7, and 1.5, respectively. No signicant differ-
ences in the size of NaCl crystals were observed, and the average
sizes of NaCl particles were 30–50 mm for all three cells, as
shown in Fig. 7b and d.
Fig. 8 (a) Voltage profile of the 30th charge and discharge process for Ni/NaCl ¼
1.82 (black), 1.7 (red), and 1.5 (green) in the cells under voltage cutoff cycles. The
cutoff voltage for charge and discharge was 2.7 and 2.4 V, respectively. See text
for more detailed experimental conditions. (b) Capacity plot for cells tested for
Ni/NaCl ¼ 1.82 (black), 1.7 (red), and 1.5 (green).
Fig. 6 (a) Voltage profile of the 30th cycle at C/3 rate for Ni/NaCl ¼ 1.82 (black),
1.7 (red), and 1.5 (green). (b) Voltages for EOC and EOD on 3 cm2 button cells with
C/3 (30 mA) for Ni/NaCl ¼ 1.82 (black), 1.7 (red), and 1.5 (green), respectively.
Fig. 9 (a, c, and e) SEM (backscattering, 1000) images of Ni particles of the cells
Fig. 7 (a and c) SEM (backscattering, 1000) images of Ni particles of the cells cycled between 2.4 and 2.7 V with Ni/NaCl ¼ 1.82, 1.7, and 1.5, respectively. (b, d,
cycled for Ni/NaCl ¼ 1.7 and 1.5, respectively. (b and d) SEM images of NaCl and f) SEM images of NaCl particles for cells cycled between 2.4 and 2.7 V
particles for cells cycled between 2.4 and 2.7 V (backscattering, 300) for cells (backscattering, 300, 250, and 200) for cells with Ni/NaCl ¼ 1.82, 1.7, and
with Ni/NaCl ¼ 1.7 and 1.5, respectively. 1.5, respectively.
14938 | J. Mater. Chem. A, 2013, 1, 14935–14942 This journal is ª The Royal Society of Chemistry 2013
View Article Online
This journal is ª The Royal Society of Chemistry 2013 J. Mater. Chem. A, 2013, 1, 14935–14942 | 14939
View Article Online
Table 1 Summary of effect on Ni and NaCl particle sizes for different Ni/NaCl ratios, C-rates and capacitiese
applied, wider cycling capacity and higher SoC at EOC were the cathode so that the available active surface area of Ni
achieved while the SoC at EOD remained similar due to the particles can still be sufficient even aer particle growth. On the
same EOD voltage limit of 2.5 V. Since the EOC voltage itself was other hand, all the cells that experienced fast degradation
not a main factor in increasing the Ni particle size as mentioned revealed more growth of NaCl particles (refer to Experiments
above, the increase in Ni particle growth with higher EOC 7–9). The cells, in which the size of NaCl is larger than 50 mm,
voltage seems to be closely related to higher SoC at EOC. In exhibit faster degradation.
summary, the current density, Ni/NaCl ratio and SoC at EOC are To the best of our knowledge, the detailed electrochemical
parameters that inuence the growth of Ni particles. reaction mechanisms of each ingredient such as NaCl, Ni,
The effects of experimental parameters on NaCl particle NiCl2, and NaAlCl4 have not been clearly reported. Even though
growth were also examined. As can be seen in the results of the overall electrochemical reaction shown in eqn (1) is quite
Experiments 1–6, the current density and the Ni/NaCl ratio did simple, a detailed mechanism of cell reactions can be proposed
not seem to inuence the growth of NaCl crystals signicantly based on correlations between particle growth (Ni and NaCl)
when the cells were tested with the same cycling capacity of and cell performances. A schematic diagram of proposed
90 mA h. The most signicant NaCl particle growth was obvious charging and discharging processes for a Na–NiCl2 battery is
when the cells were deep cycled between 2.4 and 2.7 V (Exper- shown in Fig. 11.
iments 7–9). In this case, a lower Ni/NaCl ratio leads to faster For the charging process, the electrochemical reduction of
NaCl growth. Na+ (reaction (A) in Fig. 12) takes place on the anode aer Na+
However, the effects of SoC at the EOC on NaCl growth are ions move through the BASE. At the same time, the electro-
marginal, unlike Ni particle growth, as shown in Experiments chemical oxidation of Ni (reaction (B) in Fig. 12) occurs on the
10 and 11. For the particle growth of NaCl, it can be concluded surface of Ni particles by consuming chloride anions to form a
that the Ni/NaCl ratio inuences it signicantly only when the NiCl2 layer. In reactions (A) and (B), Na+ and Cl ions should be
cycling capacity window is large (deep cycling).
The particle growth of both Ni and NaCl in cathodes is
considered one of the most important factors that affect the
degradation of Na–NiCl2 battery performance. The larger the
particles of active ingredients contained in the cathode, the less
active surface area is available for electrochemical reactions.
Therefore, the particle growth of Ni and NaCl can lead to an
increase in cell polarization due to the reduced active area. Even
though growth of both particles can cause degradation in cell
performance, our controlled experiments show that only the
growth of NaCl particles leads to signicant fade. In our
experiments, Ni particles have grown up to 10 mm mainly due to
the high current density, low Ni/NaCl ratio, and high SoC at
EOC. However, this Ni growth did not induce signicant cell Fig. 12 Schematic diagram of charge and discharge processes for a Na–NiCl2
degradation. This is possibly related to the use of excess Ni in battery.
14940 | J. Mater. Chem. A, 2013, 1, 14935–14942 This journal is ª The Royal Society of Chemistry 2013
View Article Online
supplied from secondary electrolyte NaAlCl4 ((A) and (B)) since Experiment 7 had a larger capacity window (or cycling capacity),
direct contact between solid NaCl and Ni is quite limited. it can be concluded that the major parameter in the cell
Therefore, the depletion of NaCl in the catholyte can occur degradation is the growth of NaCl particles caused by cycling
locally, leading to phase shiing of the catholyte to a partially with a large capacity window.
Lewis-acidic condition to form an AlCl3-rich phase such as It was observed that larger Ni particles were correlated with
NaAl2Cl7.7 Since the phase diagram of NaCl–AlCl3 has a stable higher C-rate, higher SoC at the EOC, and lower Ni/NaCl ratio.
liquid phase for NaCl–AlCl3 close to 1 at 280 C,9 NaCl particles However, it is not clear how Ni particles grow during the cell
should also be dissolved in Lewis-acidic melts to form neutral cycling. A few mechanisms for Ni particle growth in different
melts (reaction (C) in Fig. 12). During the discharge process, Na+ applications, such as solid oxide fuel cells (SOFCs)17,18 and
Published on 23 September 2013. Downloaded by PNNL Technical Library on 13/03/2014 16:56:30.
and Cl ions, produced from the oxidation of liquid sodium on catalysis,19 have been proposed: (1) Ostwald ripening, where Ni
the anode and the reduction of NiCl2 on the cathode, increase atoms migrate from smaller particles to larger particles and (2)
the NaCl–AlCl3 ratio in the catholyte, and excessive NaCl in the vapor transport between Ni particles. However, neither of the
melts is precipitated out as solid NaCl crystals in order to above mechanisms could be directly adopted for the case of the
maintain the neutral catholyte. Based on this analysis, detailed Na–NiCl2 battery, due to a signicantly lower operating
battery reaction mechanisms for the charge process can be temperature (280 C) compared to SOFCs (800 C) and catal-
proposed as follows. ysis (500 C). Any direct migration of Ni particles, Ni atoms, and
Ni vapors is unlikely to happen in a ZEBRA battery.
1Na+ (from NaAlCl4) + 1e / Na (l, anode) (A) One feasible mechanism of Ni particle growth is Ostwald
ripening which occurs by the migration of Ni2+ ions through
1/2Ni(s, cathode) + 1Cl (from NaAlCl4) / dissolved NiCl2 in the melts. When NiCl2 layers are produced on
1/2NiCl2 (s, cathode) + 1e (B) the surfaces of Ni particles during the charge process, it is likely
to facilitate Ostwald ripening through dissolved Ni2+ species in
the melts. However, the Ni particle growth through Ostwald
2NaAlCl4(l) Na+ Cl / ripening should be much slower than the NaCl growth
NaAl2Cl7(l) (A + B in the catholyte) mentioned earlier because of the lower solubility of NiCl2
compared to NaCl in the melts. The detailed study of NiCl2
solubility in NaCl-saturated NaAlCl4 (basic melts) has been
previously reported for the temperature range between 200 and
NaAl2Cl7(l) + NaCl(s, cathode) / 400 C.7 The reported level of Ni in the basic melts was lower
2NaAlCl4 (l) , in the catholyte (C) than 5 ppm with the presence of additives at a temperature
lower than 300 C.7 A slower Ni particle growth was also
Apparently, larger NaCl particles with smaller surface areas reported at lower temperatures where the solubility of NiCl2 is
can slow down the dissolution of NaCl (reaction (C)) in the further limited.7 In the case of higher rate tests, the more acidic
melts. This is in good agreement with our experimental obser- melts (AlCl3-rich phase) can present locally on the Ni surface
vation that higher polarization (and faster degradation) was due to high current densities. In a locally acidic environment,
associated with larger NaCl particles. The growth of NaCl the level of Ni2+ could be increased by forming nickel tetra-
particles over cycles can be explained by Ostwald ripening, chloroaluminate according to the reaction below under the
which is basically a thermodynamically driven process because battery operating conditions.20
larger particles are more energetically stable than smaller
particles due to the reduced surface energy.16 For charging and NiCl2 + 2AlCl3 / Ni(AlCl4)2 (2)
discharging processes, NaCl particles undergo dissolution and
precipitation, respectively. When supersaturated NaCl is The process of Ostwald ripening will be pronounced with an
precipitated out from melts during discharge, it has a tendency enhanced Ni2+ level in the acidic melts. This is consistent with
to grow on the surfaces of larger particles. Over testing cycles, our experimental results (Experiments 2–4) where a faster
the population of smaller NaCl particles will be decreased while growth of Ni particles was observed for higher C-rate tests. To
relatively larger NaCl particles grow even larger, and overall the maintain the basicity of the melts, it is necessary to have an
average NaCl particle size will increase. It is not difficult to excess amount of solid NaCl in the cathode at all times. At high
expect that the disappearance of smaller NaCl particles in the SoC at EOC, most of the small NaCl particles in the cathode
cathode is greater for tests cycled with a larger capacity window. should be dissolved in the melts, and melts can be more acidic
For instance, the average NaCl particle size increased from 30– due to the sluggish dissolution of larger NaCl particles. As a
50 mm (with a cycling capacity of 90 mA h) to 50–70 mm (with a consequence of the acidic environment, a faster growth of Ni
cycling capacity of 121 mA h). However, high SoC at EOC (more particles is expected for the battery tests with higher SoC at
consumption of NaCl) has a marginal inuence on the NaCl EOC, which is in good agreement with the experimental results
growth. In Experiments 7 and 11 listed in Table 1, both cells (Experiments 7, 8, 9, 11, and 12). For cells having lower Ni ratios
were charged to a similar SoC (96–98%), but they revealed quite (Experiments 6, 8, and 9), the available active surface area of the
different NaCl growth and cell degradation. Since the major Ni electrode is relatively small. Thus, a higher current density
difference between the two experiments is that the cell of on Ni particles is expected for a cell with a lower Ni/NaCl ratio
This journal is ª The Royal Society of Chemistry 2013 J. Mater. Chem. A, 2013, 1, 14935–14942 | 14941
View Article Online
despite the same overall cell current density. A faster growth of 3 K. B. Hueso, M. Armand and T. Rojo, Energy Environ. Sci.,
Ni particles observed for the cells having lower Ni ratios can also 2013, 6, 734–749.
be attributed to a higher local current density. 4 Z. G. Yang, J. L. Zhang, M. C. W. Kintner-Meyer, X. C. Lu,
D. W. Choi, J. P. Lemmon and J. Liu, Chem. Rev., 2011,
Conclusions 111, 3577–3613.
5 A. VanZyl, Solid State Ionics, 1996, 86–88, 883–889.
The Na–NiCl2 battery was tested to determine the effects of 6 B. Dunn, H. Kamath and J. M. Tarascon, Science, 2011, 334,
various parameters such as C-rate, Ni/NaCl ratio, cutoff voltage, 928–935.
and capacity windows on particle growth and cell performance 7 J. Prakash, L. Redey and D. R. Vissers, J. Electrochem. Soc.,
Published on 23 September 2013. Downloaded by PNNL Technical Library on 13/03/2014 16:56:30.
degradation. For Ni, higher current density, higher SoC at EOC, 2000, 147, 502–507.
and lower Ni/NaCl ratio are the main parameters that lead to 8 X. C. Lu, G. L. Li, J. Y. Kim, J. P. Lemmon, V. Sprenkle and
faster particle growth. In the case of NaCl, signicant growth Z. G. Yang, J. Power Sources, 2012, 215, 288–295.
was detected only when the cells were cycled with a large cycling 9 C. Robelin, P. Chartrand and A. D. Pelton, J. Chem.
capacity window. With a large cycling window, the growth of Thermodyn., 2004, 36, 683–699.
NaCl particles was accelerated with a decrease in the Ni/NaCl 10 G. S. Li, X. C. Lu, C. A. Coyle, J. Y. Kim, J. P. Lemmon,
ratio. Even though the growth of both Ni and NaCl can inuence V. L. Sprenkle and Z. G. Yang, J. Power Sources, 2012, 220,
the cell degradation, our results indicate that the growth of 193–198.
NaCl is a dominant factor affecting cell degradation. The use of 11 J. L. Sudworth, J. Power Sources, 2001, 100, 149–163.
excess Ni seems to play a role in tolerating the negative effects 12 Kuan-Zong Fung, Jan-Fong Jue and Anil Vasudeo Virkar, U.S.
of particle growth of Ni since the available active surface area of Pat., 6117807, 2000.
Ni particles can still be sufficient even aer particle growth. 13 G. Parthasarathy, N. Weber and A. V. Virkar, ECS Trans.,
2007, 6, 67–76.
Acknowledgements 14 K. C. Patil, G. V. Chandras, M. V. George and C. N. R. Rao,
Can. J. Chem., 1968, 46, 257–265.
This work is supported by the Laboratory-Directed Research and 15 B. V. Ratnakumar, A. I. Attia and G. Halpert, J. Power Sources,
Development Program of Pacic Northwest National Laboratory 1991, 36, 385–394.
(PNNL) and the Office of Electricity Delivery and Energy Relia- 16 L. Ratke and P. W. Voorhees, Growth and Coarsening:
bility's storage program. PNNL is a multiprogram laboratory Ripening in Material Processing, Springer, Germany,
operated by Battelle Memorial Institute for the Department of 2002.
Energy under Contract DE-AC05-76RL01830. 17 S. P. Jiang, J. Mater. Sci., 2003, 38, 3775–3782.
18 D. Simwonis, F. Tietz and D. Stover, Solid State Ionics, 2000,
References 132, 241–251.
19 J. Sehested, J. A. P. Gelten, I. N. Remediakis, H. Bengaard
1 J. L. Sudworth, J. Power Sources, 1994, 51, 105–114. and J. K. Norskov, J. Catal., 2004, 223, 432–443.
2 X. C. Lu, G. G. Xia, J. P. Lemmon and Z. G. Yang, J. Power 20 T. Javadi and A. Petric, J. Electrochem. Soc., 2011, 158, A700–
Sources, 2010, 195, 2431–2442. A704.
14942 | J. Mater. Chem. A, 2013, 1, 14935–14942 This journal is ª The Royal Society of Chemistry 2013