Professional Documents
Culture Documents
https://doi.org/10.1038/s41560-020-00728-y
Aluminium-based battery technologies have been widely regarded as one of the most attractive options to drastically improve,
and possibly replace, existing battery systems—mainly due to the possibility of achieving very high energy density with low
cost. Many reports have demonstrated primary or rechargeable Al-based battery chemistries in both aqueous and non-aqueous
electrolytes. However, the practical realization of these battery chemistries has been difficult over a long period of time (170
years). In fact, no Al-based battery has been shown with the required stability or touted energy density. Typically, the perfor-
mance of Al-based batteries is overstated in the literature due to imprecise considerations that do not fairly evaluate practically
achievable energy densities. Here we provide accurate calculations of the practically achievable cell-level capacity and energy
density for Al-based cells (focusing on recent literature showing ‘high’ performance) and use the results to critically assess
their future deployment.
T
here is an increasing demand for battery-based energy stor- The vast majority of studies on non-aqueous rechargeable Al batter-
age in today’s world. Li-ion batteries have become the major ies have been performed as AIBs, with only a few studies dedicated
rechargeable battery technology in energy storage systems to AAB rechargeables, since rechargeable AABs have invariably
due to their outstanding performance and stability. However, their shown inferior cycle life and reversibility than AIBs8. In the past few
relatively high cost and safety issues still need to be addressed1,2. years, AIBs have received increased attention as a possible solution
For primary batteries, Zn–MnO2 alkaline cells are by far the most for fast charge/discharge applications with ultra-high efficiency and
well-established, though there is a desire to find alternatives with energy density9.
improved capacity3. Al-based batteries are among the most attrac- Many of the studies on AABs or AIBs reported in the literature
tive alternatives to compete with existing cell chemistries in both end by claiming high energy density materials and/or successful
primary (non-rechargeable) and secondary (rechargeable) applica- demonstrations of the technology have been achieved. However,
tions due to the high natural abundance of Al, its wide availability, such claims have invariably failed to consider the comprehensive
ease in processing (including recycling), low cost and high theo- implementation of the reported chemistry in practical full cells.
retical volumetric (8056 mAh cm–3) and gravimetric (2981 mAh g–1) There is a need in the literature for a critical analysis of the realis-
capacity. A comparison of some of the most important properties tic deployment of Al anodes across different chemistries. Here we
for metals that are attractive options as energy storage solutions is provide a critical assessment of the electrochemical performances
provided in Fig. 1a. of state-of-the-art Al batteries reported in the literature. Using all
Metallic Al has been used as an anode material in all four major of the available information about full cell materials and chemistry,
battery categories—aqueous primary and secondary as well as we first carry out an accurate estimation of the practically achiev-
non-aqueous primary and secondary. The deployment of Al began able gravimetric capacity and energy density. This is followed by a
in the1850s where it was coupled with a carbon cathode in the Buff substantive discussion on their expected operational life.
cell4. After that, Al anodes were investigated in different aqueous
media from saline cells to Leclanche-type dry cells5. In 1962, the Al–air batteries
concept of the Al–O2/air battery was introduced for the first time The most common AABs typically consist of an Al (pure or alloyed)
and in 1988 Al–Cl2 rechargeable cells were developed in molten salt anode, air cathode (comprised of a gas diffusion layer and catalyst
electrolytes6. Though conceptually interesting, many of these cells layer), and a KOH-based electrolyte that incorporates additives to
showed a propensity towards corrosion and passivation, which led suppress corrosion and H2 evolution, such as ZnO (ref. 10) (Fig. 1b).
to a significant amount of research into overcoming such hurdles At the anode, Al metal is oxidized to Al(OH)4–, which subsequently
in both aqueous and non-aqueous cells. One such solution was the forms solid Al(OH)3 layers on the electrode surface as the solubility
invention of so-called Al-ion batteries (AIBs) in 20117 and research limit is reached (Eqs. (1) and (2)).
in this area over the past several years has mostly focused on under-
standing their reaction mechanisms and searching for suitable Al þ 4OH� $ AlðOHÞ�
4 þ3e
�
E0 ¼ �2:34 V vs: SHE
cathodes. ð1Þ
Today, the dominant Al-based primary batteries have been
Al-air batteries (AABs) and the dominant secondary batteries have AlðOHÞ� �
4 $ AlðOHÞ3 þOH ð2Þ
been AIBs. For primary battery applications, non-aqueous electro-
lytes are likely not economically feasible due to the high cost of ionic At high pH, the passivation film can be dissolved, exposing the
liquid (IL)-based electrolytes compared to aqueous electrolytes. reacting Al, but that also leads to high corrosion rates and H2 gas
Because of this, most AABs utilize a KOH-based aqueous electrolyte. evolution (Eqs. (1), (3) and (4))11.
Department of Chemical Engineering, University of South Carolina, Columbia, South Carolina, USA. ✉e-mail: mustainw@mailbox.sc.edu
a Standard potential
(V vs. SHE) Li+ + e– Li Na+ + e– Na K+ + e– K Zn2+ + 2e– Zn Al3+ + 3e– Al
–3.05 V –2.71 V –2.93 V –0.76 V –1.67 V
9000 25
e– e– e– e–
Discharge Discharge
OH– O2
H2 OH– AlCl4-
(GDL + catalyst layer)
O2
Graphite cathode
Al2Cl7- AlCl4-
(glassy carbon)
OH–
AlCl4-
Air cathode
Al anode
Al anode
OH– AlCl4-
O2
H2 OH– Al2Cl7-
Al(OH)4- O2 AlCl4-
AlCl4-
OH– AlCl4-
OH– O2
Al(OH)3
Al2Cl7-
Fig. 1 | Attractive properties and chemistries of Al as an anode material for use in AABs and AIBs. a, Comparison of gravimetric and volumetric
capacities as well as cost, abundance, standard potentials, cation radius and charge density of cations for Li, Na, K, Zn and Al20,50. b, Schematic illustration
of the discharge process in an Al–air battery with aqueous KOH electrolyte showing the movement of hydroxide ions to the anode causing undesired
corrosion and passivation phenomena. c, Schematic of the discharge process in an Al-ion battery with non-aqueous ionic liquid electrolyte including a
representation of intercalated chloroaluminate anions in the graphite cathode.
2H2 O þ 2e� $ H2 þ 2OH� Eo ¼ �0:83 V vs: SHE ð3Þ theoretical cell voltage of an AAB is 2.74 V, but mainly due to cor-
rosion and the passivating layer, the operating cell voltage has been
shown to be between 1.2 V and 1.6 V. The second problem is a high
2Al þ 2OH� þ 6H2 O $ 3H2 þ 2AlðOHÞ�
4 ð4Þ rate of corrosion when the oxide layer is removed, especially in alka-
line electrolytes.
At the cathode, the oxygen reduction reaction (ORR) occurs Because the cathode reaction utilizes ambient air, there is
(Eq. (5)). only a need to carry one of the reacting species, meaning that they
have the promise to achieve very high energy densities. In fact, the
O2 þ 2H2 O þ 4e� $ 4OH� E0 ¼ 0:40 V vs: SHE ð5Þ maximum theoretical gravimetric capacity of an AAB is second
only to a Li–air battery (Fig. 1a). In the literature, studies focused
Combined, these reactions yield the overall cell reaction in on AABs have been quick use the high capacity and theoretical
Eq. (6). open-circuit voltage to tout a high theoretical gravimetric energy
density of 4140 Wh kg–113. However, the reported energy densities
4Al þ 3O2 þ 6H2 O $ 4AlðOHÞ3 E0 ¼ 2:74 V vs: SHE have been much lower than 1000 Wh kg–1 and even these values
ð6Þ have often failed to consider the mass of all of the components
of the cell: anode, cathode, electrolyte, separator and packaging—
Al as an anode in aqueous media has two main problems. First is meaning that reported values are often not an accurate representa-
the formation of a passivating oxide layer, which reduces the oper- tion of what is actually achievable in full cells and true cell-level
ating cell voltage and increases both the charge transfer and mass discharge capacities and energy densities are reduced significantly
transfer resistance. As shown theoretically by Nørskov. et al.12, this when properly assessed. This is shown in Table 1 using recent
causes the open circuit potential of the Al anode to drop from the studies utilizing novel chemistries on either the electrodes or the
Nernstian value of –2.34 V to –1.87 V vs. SHE at pH = 14.6, which electrolyte that claim either record discharge capacities or energy
limits the cell operating voltage and practical energy density. The densities.
23
Perspective
Perspective NATure Energy
1,000 support new cell designs. For example, one recent report innova-
AIB theoretical
tively mitigates Al corrosion at open circuit through oil displace-
48 ment. In this case, an accurate estimate of energy density was made
100 True intercalation AIB practical limit 44 38 by the authors (for example, 680 Wh kg–1) for the AAB18 (~6 times
30
32 13 49 33 35
29
42 16
17
lower than the theoretical value); however, the components required
23
27
31 7 41
43
15
to facilitate the continuous flow of electrolyte and the storage and
26 34
25 36 37 handling of the oil were not considered. Of course, these would add
10 28
Aqueous AAB a significant amount of weight (a reasonable estimate could not be
Aqueous AIB made) to the AAB pack, limiting the practical application of this
14 Non-aqueous AAB type of battery. What is hopefully clear from the above discussion
Non-aqueous AIB
1 is that even in the best batteries there remains a huge gap between
1 10 100 1,000 10,000
the theoretical cell-level energy density that is often marketed for
Reported energy density (Wh kg–1) AABs (4140 Wh kg–1) with reported values (77–2300 Wh kg–1) and
b there is another large gap between reported values and what is actu-
1,000
Li-ion
ally practically-achievable or demonstrated (2.52–35880 Wh kg–1)
LiFeS2 as shown in Fig. 2a.
Cell-level energy density (Wh kg–1)
48
38
Al-ion batteries
100 Zn–MnO2
13
32
33
42
44
41
35 The most common AIB configuration uses an Al foil anode, a
49
16 15 17 23
43
31 30 29 25
graphite cathode19, and an acidic room temperature non-aqueous
7
27 26
37
28
Lead–acid IL electrolyte (for example, AlCl3:[EMIm]Cl > 1; [EMIm]Cl is
34 36
10 1-Ethyl-3-methylimidazolium chloride), as illustrated in Fig. 1c.
Aqueous primary AIBs operate under similar principles as Li-ion batteries, with the
14 Aqueous secondary main idea that it is possible to intercalate a reacting species inside of
Non-aqueous primary
Non-aqueous secondary the carbon. The main assertion for these cells is that AIBs are able to
1
0 1 10 100 1000 10,000 100,000 1000,000
replace the standard one-electron process realized with lithium with
a three-electron process at the anode. However, the reaction is not a
Battery operational life (h)
simple dissolution of Al to Al3+ (theoretical capacity of 2980 mAh g–1)
followed by intercalation of Al3+ into the carbon. The redox reaction
Fig. 2 | Evaluation of practical energy density and operational lifetime in the most common AIBs is much more complicated and involves a
of Al anode batteries. a, Illustration of reported versus practical energy large number of additional reactants from the electrolyte, as shown
density of AABs and AIBs as well as a comparison of achievable versus in Eq. (7).
theoretical energy density of Al batteries. b, Comparison of cell-level
energy density and estimated battery operational life of state-of-the-art Al þ 7AlCl� � �
4 $ 4Al2 Cl7 þ 3e ð7Þ
Al-based batteries with well-established battery technologies. The
numbers indicate the corresponding references. Therefore, in reality, AIBs require 8 Al atoms per 3 electrons, not
1 Al atom per 3 electrons, which has a huge effect on the theoretical
capacity and energy density of these systems. Considering the total
More specifically, one recent study using a three-dimensional number of Al atoms (and Cl atoms) that are needed to drive this
(3D) printed Al anode prepared by laser sintering14 reported an reaction, the theoretical capacity of the anode reaction is actually
achieved capacity of 239 mAh g–1 and a gravimetric energy density 67 mAh g–1, not the commonly reported 2980 mAh g–1 (Eq. (16)).
of 227 Wh kg–1. However, in the study, all of the experimental data AIBs are further limited by a lack of suitable cathode chemistries
was normalized only to the 11 mg of Al that was deposited, and the (for intercalation or conversion reactions)20. Though it is interesting
mass of the rest of the battery components (that is, current collec- from a fundamental perspective to understand how Al intercalates
tors, gelled electrolyte membrane, polyimide sheets as cell packag- into carbon, the low capacities that have been reported across the
ing, anode substrate and cathode active material) were neglected. board for these materials are all caused by the same main drawback:
When the total mass of the battery (assuming a generous capacity Al3+ does not participate in noninteractive diffusion in the AlCl3:IL
ratio of the negative electrode to the positive electrode, N/P, of 1) electrolyte due to its low ionic radii (54 pm) and high valence state21.
is considered, the resulting capacity and cell-level energy density This forces Al to coordinate with neighbouring anions (for example,
are much lower, 2.66 mAh g–1 and 2.52 Wh kg–1, respectively. Other Cl, F) to form complexes within the room temperature IL. The car-
studies utilizing non-liquid electrolytes (gelled or paper based)15–17 bon, therefore, intercalates an AlCl4– complex rather than Al3+, as
have shown relatively low Al corrosion, which is desirable, and shown in Eq. (8).
have reported high achievable capacity (900–2336 mAh g–1) and
high energy density (900–1402 Wh kg–1). However, these values Cn ½AlCl4 þ e� $ Cn þ AlCl�
4 ð8Þ
were all based solely on the mass of the Al anode. Furthermore,
in some cases, the operating discharge time was limited—even as Such a large intercalating ion introduces structural limita-
low as 24 minutes—due to premature passivation accompanied by tions that force the repeating unit to be very large, estimated to be
a significant capacity decay16. When the mass of all of the battery 8–72 carbons for every intercalated AlCl4– anion22 (unlike LiC6 in
components — anode and cathode active materials, separator/elec- Li-ion batteries). Therefore, the gravimetric energy density of the
trolyte, current collectors (Al foil on the anode side and, typically, overall cell is significantly hindered by the diluting carbon mass.
Ag or Cu foil on the cathode side), and battery shells (related to In addition, the intercalation of such a large species leads to large
graphite-based cathodes, the average voltage for all cases was assumed to be 1.4 V except the multi-ions system, which was 3.5 V. eAverage discharge capacity of initial and cut-off capacity was considered for the studies where the capacity retention (last cycle) was not close to
100%. Note: average discharge capacity was only used to calculate battery operational life.
NATure Energy Perspective
Battery life assessment Related to AABs, in addition to poor achievable life their practi-
In addition to energy density, the durability of Al batteries was ana- cal energy densities remain ~10–15 times lower than their theoreti-
lysed in terms of battery operational life and compared with four cal values. In addition to the limitations of Al, the low energy density
well-known aqueous and non-aqueous battery technologies. The is caused by the ORR at the AAB cathode which is kinetically slow,
details for the battery operational life calculations are provided in and product instability leads to the formation of a barrier layer with
Methods. Of course, the lifetime of commercial batteries can range a gel-like consistency comprised of aluminum–oxygen-organic spe-
widely depending on the current rate that is dictated by the applica- cies. This causes generally poor performance (that is, low discharge
tion. For the purpose of fair comparison, this study considered a voltage even at low rates). Moreover, AABs are not cost effective,
moderate rate (0.1 to 1 mA) for primary commercial batteries and since all modern high performing, long life ORR cathodes rely on
250–300 cycles per year were assumed for secondary commercial expensive electrocatalysts. Therefore, there should be more of a
batteries to be comparable with testing conditions of Al batteries in focus on other types of battery chemistries than AAB, though it is
the literature. Therefore, there is a specific range that can be calcu- questionable whether primary battery technologies with metal Al
lated for battery life and energy density. In Fig. 2b, it can be seen that anodes (even using corrosion inhibitors or alloying with other met-
aqueous primary AABs have considerably lower energy density and als) can ever be deployed.
operational life compared to Zn–MnO2 alkaline batteries—their Hopefully, this Perspective benefits the battery research com-
main competitor in the commercial space. Non-aqueous primary munity by presenting a more realistic picture regarding the status
AABs can operate for longer times and higher voltages compared of Al battery technologies. There are many attractive features for
to aqueous AABs, and the achievable energy density is higher for Al-based batteries and there are obvious possible advantages over
non-aqueous AABs (Table 1). Though the cell-level energy density existing batteries, including Li-ion. However, there is clearly a size-
of the highest performing non-aqueous primary AAB is compara- able gap between the current state of rechargeable AIBs and existing
ble with industry-leading LiFeS2, the operational lifetime remains Li-ion technologies. Today, it appears that this gap may inevitably
orders of magnitude lower. prove insurmountable. For that statement to be wrong, it is likely
In rechargeable aqueous AIBs, utilizing AlCl3-based water-in-salt necessary for the community to completely rethink the chemistry
electrolytes might be promising—presently delivering a 79 Wh kg–1 of existing Al batteries, revisit the cathode materials and electrolytes
cell-level energy density. Batteries with Al(OTF)3-based aqueous that would enable the use of Al in the future, and carefully evaluate
electrolytes have shown energy densities that are comparable with the cell balancing for practicality by adjusting the N/P ratio.
lead–acid batteries. However, the operational life for aqueous AIBs
is significantly lower than lead–acid batteries. Also hampering com- Methods
mercial deployment is the fact that Al(OTF)3 based aqueous elec- AAB cell-level energy density. The methodology to calculate cell-level capacity,
trolytes are ~130 times more expensive than AlCl3-based aqueous energy density and battery operational life is provided in this section. One of the
essential points in our analysis was to consider the mass of every component in a
electrolytes and are double the cost of non-aqueous AlCl3-[EMIm] cell: anode, cathode, electrolyte, separator and packaging, though packaging was not
Cl electrolytes44. For non-aqueous AIBs with an Al/AlCl3-[EMIm] extensively discussed in the main text. The average voltage was estimated based on the
Cl/graphite chemistry, the achievable lifetimes are much lower than discharge curves in the original paper, if not mentioned explicitly in the given study.
the incumbent technology, Li-ion batteries, which have substan- For AABs, the amount of charge passed for the Al anode (Q) was calculated based on
the reported capacity and the mass of the Al used. Then, the mass of consumed Al was
tially longer operational life than any other non-aqueous AIB.
estimated by dividing Q by the theoretical capacity of Al. It was assumed that the N/P
ratio is equal to 1. The mass (mi) of the cathode and electrolyte were estimated based
Outlook on the information provided in each study. The mass of the separator and packaging
In this Perspective, the recent development of Al battery technol- were assumed to take 5% and 30% of the total mass, respectively, based on available
ogy was highlighted from a practical perspective and a quantitative estimations for these components of AABs18. This allowed the cell-level capacity and
energy density to be calculated using Eqs. (10) and (11).
analysis of current energy density values in Al battery technology
was performed. It was realized that in aqueous alkaline electro- Ccell�level ¼
Q
ð10Þ
lytes, anodes that are principally comprised of Al metal have seri- manode þ mcathode þ melectrolyte þ mseparator þ mpackage
ous corrosion and gassing issues in addition to poor solubility of
Al(OH)4–—leading to passivation as either Al(OH3) or Al2O3. For Energy density ¼ Ccell�level ´ Vavg ð11Þ
Al-based batteries with non-aqueous acidic IL electrolyte, a huge
Where Vavg is the average is the average discharge voltage.
amount of supporting electrolyte needs to be carried which signifi-
cantly reduces the gravimetric and volumetric energy density. AIB cell-level energy density. For AIBs, a slightly modified version of the AAB
A comparison of battery performance of aqueous and method was used in the calculations that can be generalized easily for any AIB
non-aqueous AIBs clarifies that aqueous systems can deliver high system that aims to become commercialized. Again, it was assumed that the N/P
capacities but short cycle life and poor capacity retention. In con- ratio is 1, though here a common Al mass (1 g) was assumed. Then, the amount
of charge passed for the anode was calculated based on the reported conditions.
trast, non-aqueous AIBs have a higher operation and cycle life, The charge passed and the reported cathode capacity were used to determine the
but have limited capacity. They also rely on IL electrolyte, which is cathode mass. Next, using the electrolyte molecular weights, concentration or
likely a non-starter for most applications, since they are expensive density of dissolved reactant, r value (ratio of AlCl3:[EMIm]Cl in the electrolyte,
and require highly specialized, expensive cell materials to avoid cor- needed in the chemistries that involve AlCl3–[EMIm]Cl electrolyte) and number
rosion (Ti, Ta, Mo)—though it may be possible to develop coatings of electrons, the mass of electrolyte was estimated and it was found that the
approximate ratio of the electrolyte mass to the cathode mass is ~3 in most cases,
to enable stainless steel hardware, and more stable current collec- which is consistent with the calculations done by Elia et al.25. Finally, we estimated
tors. Furthermore, ILs have not been produced at the required scale the mass of the separator and packaging to be 5% and 10%, respectively, based on
and may require non-ambient cell construction and materials pro- available estimations for these components of AIBs25.
cessing. Another issue with non-aqueous AIBs is mass transport,
because ion movement occurs in the opposite direction of diffusion. AIB theoretical energy density. For comparison sake, in addition to the achieved
energy density, it is important to calculate the maximum theoretical energy density
Finally, the fundamental chemistry of Al in existing ILs requires a for an AIB battery, which is given by Eq. (12)46.
significant electrolyte mass to be carried, which limits the achiev-
CA CC
able energy density values. For example, the practical energy density Theoretical energy density ¼ ´V ð12Þ
CA þ CC
of an Al/AlCl3–[EMIm]Cl/Graphite AIB is 27–50 Wh kg–1. The con-
tribution of significant electrolyte mass in this battery system is an Where CA and CC are the theoretical capacities for the anode and cathode,
inevitable fact, making its future deployment doubtful. respectively, and V is the cell voltage.