Articles

https://doi.org/10.1038/s41560-021-00797-7

Regulating electrodeposition morphology in

high-capacity aluminium and zinc battery anodes
using interfacial metal–substrate bonding
Jingxu Zheng 1, David C. Bock2, Tian Tang 1, Qing Zhao 3, Jiefu Yin 3, Killian R. Tallman 2
,
Garrett Wheeler 2, Xiaotun Liu3, Yue Deng 1, Shuo Jin3, Amy C. Marschilok 2,4,5,
Esther S. Takeuchi2,4,5, Kenneth J. Takeuchi4,5 and Lynden A. Archer 1,3 ✉

Although Li-based batteries have established a dominant role in the current energy-storage landscape, post-Li chemistries (for
example, Al or Zn) are emerging as promising candidates for next-generation rechargeable batteries. Electrochemical cells
using Al or Zn metal as the negative electrode are of interest for their potential low cost, intrinsic safety and sustainability.
Presently, such cells are considered impractical because the reversibility of the metal anode is poor and the amount of charge
stored is miniscule. Here we report that electrodes designed to promote strong oxygen-mediated chemical bonding between Al
deposits and the substrate enable a fine control of deposition morphology and provide exceptional reversibility (99.6–99.8%).
The reversibility is sustained over unusually long cycling times (>3,600 hours) and at areal capacities up to two orders of
magnitude higher than previously reported values. We show that these traits result from the elimination of fragile electron
transport pathways, and the non-planar deposition of Al via specific metal–substrate chemical bonding.

C
reating conformal metal coatings through electrochemical
methods is important in multiple fields, including semicon-
ductor manufacturing1, energy storage2, metal plating3 and so
on. Although the governing chemistry is in principle simple—sol-
vated metal cations in a typically liquid electrolyte capture electrons
from an electrically conductive substrate, and are thereby reduced
into solid metallic form, for example, Mn+ (sol.) + ne− → M(s)
—the control over metal electrodeposition morphology remains
challenging2,4. The search for fundamental solutions has recently
re-emerged as an area of scientific and technological interest because
of the role that the stable electrodeposition of metals in closed electro-
chemical cells plays in the stability and safety of rechargeable batteries
that utilize metal anodes, for example, Al, Li, Zn, Na and so on, to
achieve a higher energy and the low-cost storage of electrical energy.
The working principle of such electrochemical cells is as follows:
the reversible plating and stripping of metal at the anode enables
the reversible storage and release of electrical energy. In prac-
tice, however, repeated formation and dissolution of the metallic
phase is confounded by multiple problems that result in a porous,
non-planar deposition morphology5,6. A number of mechanisms
have been proposed to explain these observations. The list includes
intrinsic crystal anisotropy7, uneven solid–electrolyte interphase
that drives heterogeneous growth4, concentration of electric field
lines at ‘hot spots’ for growth8 and growth limited by mass trans-
port9. Porous metal deposits created in one plating cycle also have
a strong propensity to only partially dissolve in the next, to form
the so-called ‘dead’ and/or ‘orphaned’ metals6,10–12. Over repeated
cycling, the accumulated metal deposits bridge the interelectrode
space to short-circuit the cell10,13. All of these processes are obviously
fatal for the stable long-term operation of an electrochemical cell,
but ohmic heat generated by shorting poses added safety concerns
associated with thermal runaway14.
In theory, metal anodes that take advantage of micro- and/or
nanopatterned conductive substrates that guarantee full access to
electron and ion transport pathways throughout the plating/strip-
ping processes can surmount all of these challenges5,15. In conven-
tional battery configurations, ion transport occurs via interfacial
contact between solid metal deposits and liquid electrolyte, which
means that control of the electrolyte volume and interface chem-
istry are sufficient to ensure a full ionic access. In contrast, elec-
tron transport relies solely on physical, solid–solid contact between
the metal deposits and a conductive substrate. The hypothesis that
guides the current study is that the fragility of the electron trans-
port pathways is the fundamental barrier to fully reversible metal
electrodeposition processes (Supplementary Fig. 1)6,10. Earlier
work showed that, although electrodeposition of metals, such as
Cu in patterned trenches for integrated circuits1,16 and Li and Na
in three-dimensional (3D) metal or carbon foams5,6, constrain the
size of the electrodeposits, the transport length L and therefore the
electron transport timescale τ are broadly distributed, as the metal
is free to deposit in coarse, non-planar morphologies5.
In this study, we investigated the electrodeposition of metals on
patterned substrates designed to strongly coordinate with the metal
deposit via surface chemical bonding (Fig. 1). Hypothetically,
strong metal–substrate bonding increases the driving force for
nucleation to yield a uniform distribution of small metal nuclei.
The uniformly distributed nuclei enable compact, uniform deposi-
tion, which produces deposits with a controlled thickness, L, dur-
ing the growth phase. The straightforward effect of such control
is that the electron transport timescale τ ≈ L2/D, which sets the

Department of Materials Science and Engineering, Cornell University, Ithaca, NY, USA. 2Energy and Photon Sciences Directorate, Brookhaven National
1

Laboratory, Upton, NY, USA. 3Robert Frederick Smith School of Chemical and Biomolecular Engineering, Cornell University, NY, USA. 4Department of
Chemistry, State University of New York at Stony Brook, Stony Brook, NY, USA. 5Department of Materials Science and Chemical Engineering,
State University of New York at Stony Brook, Stony Brook, NY, USA. ✉e-mail: laa25@cornell.edu

398 NatUre Energy | VOL 6 | April 2021 | 398–406 | www.nature.com/natureenergy

Nature Energy Articles
a b c

Nano-Al L Micro-Al
Carbon
Nano-Al
Carbon Carbon

Al–O–C covalent bond

Al atom/cluster

Fig. 1 | Metal–substrate bonding-induced regulation of electrodeposition. Strong surface bonding between Al and a conductive fibrillar carbon substrate
facilitates electronic transport and influences the morphological evolution of the Al anode. a, At the onset of electrodeposition, Al forms the interfacial
Al–O–C chemical bonds with the surface of carbon fibres. b, Guided by the chemical bonding interaction, Al grows laterally to form a uniform, compact
deposition layer that comprises nanoscale grains on the carbon fibres. c, After the available carbon surface is fully covered, the subsequent Al deposits
form microsized particles embedded among the carbon fibres. L denotes the electron transport length.

effective exchange current density at the electrode, can be read-

ily manipulated. Here 1/D is the electron mobility15. We therefore Table 1 | Metallic Al foil anode plating/stripping throughput per
hypothesized that the presence of a strong metal–substrate bonding cycle for Al-ion electrochemical cells
provides a mechanism to narrow the distribution to produce com-
pact deposits17, which allows us to control τ and thereby maintain Reference Year Areal capacity Lifetime Configuration
of Al deposition (h)
a high reversibility. We experimentally tested this hypothesis using
(mAh cm–2)
Al metal as a model system. On substrates for which interfacial
chemical bonding was not detected, Al persistently formed coarse, 13 2015 0.10 240 Pouch
heterogeneous electrodeposits, regardless of the geometry of the 41 2015 0.14–0.42 500 Coin
substrates. By contrast, the Al deposition morphology on carbon 42 2016 0.15 200 Pouch
fibres treated by ionic liquid and AlCl3 was highly uniform across
multiple length scales. The uniform Al deposition was enabled by 43 2016 0.08 120 Pouch
Al–O–C bonds formed on the substrate surface, as evidenced by 44 2017 0.1–0.3 500 Coin
X-ray photoelectron spectroscopy. Such a precise regulation on 45 2017 0.12 600 Pouch
morphology was shown to markedly improve the electrochemical
23a 2017 1.25 600 Coin
performance of the Al electrodes.
46 2018 0.008–0.08 200 Pouch
Electrodeposition of Al metal in conventional battery anodes 47 2018 0.015–0.06 40 Coin
The rationale for choosing Al is straightforward—intrinsically safe, 48a 2018 0.4 1000 Coin
low cost and energy dense Al batteries are considered prospective
49 2019 0.0675 150 Pouch
alternatives to Li-ion batteries (Supplementary Table 1)17 in appli-
cations such as backup storage for electric power generation from 50 2019 0.03–0.08 600 Coin
intermittent, renewable sources, where cost and scalability are criti- 51 2020 0.18 30 Pouch
cal. Additionally, the reversible plating/stripping of Al was demon- 24a 2020 0.1–0.5 170 Coin
strated quite recently in 201118, which motivated a strong revival of
This work a
2021 8 3600 Coin
research interest in Al metal anodes for rechargeable batteries. As
a final point, we note that innovations in the design of reversible This worka 2021 0.4–0.5 100 Anode-free pouch
cathode materials compatible with Al metal anodes have produced As representative references, we selected typical studies that disclose the area-specific
batteries of exceptional promise13, albeit at impractically low areal parameters. aModified Al metal anodes were used in these studies.
specific capacities (0.01–0.18 mAh cm–2), that is, nearly two orders
of magnitude lower than that of state-of-the-art Li-ion batteries
(1–3 mAh cm–2) (Table 1). We assessed the performance of metallic Al anodes at elevated
The electrode areal specific capacity is an important—but areal capacities (for example, 0.8–8 mAh cm–2) by plating/strip-
oftentimes overlooked—battery parameter. Fundamentally, using ping Al metal on a conventional, planar stainless-steel substrate.
an impractically low areal capacity means that the intrinsic anode Surprisingly, the electrochemical cells fail quickly (<20 hours) by
failure modes discussed above might be obscured by an interelec- short-circuiting at these high capacities (Supplementary Fig. 2). To
trode spacing that is too large. From an applications perspective, the evaluate the plating/stripping reversibility of Al at practical areal
disproportionately large contribution from inactive battery parts to capacities—without disruptions due to battery shorting—we fol-
the overall weight and volume leads to a lower overall cell energy lowed a practice first reported in Lin et al. and intentionally placed
density and an elevated cost per unit of energy stored19–21. The slow multiple layers of separator in the interelectrode space13. The plat-
progress in addressing this issue raises concern about the actual ing/stripping reversibility (Coulombic efficiency (CE)) is thereby
potential of Al-based batteries as next-generation energy storage measured as approximately 85%, which means that 15% of the
systems19. Building Al metal anodes that sustain prolonged cycling Al deposits are irreversibly lost in each cycle (Fig. 2a and voltage
at practical areal capacities is therefore a critical step towards com- profile in Supplementary Fig. 3). As depicted in Supplementary
mercially relevant Al batteries22,23. Fig. 4, a practical Al anode would require a CE of 99%, preferably

NatUre Energy | VOL 6 | April 2021 | 398–406 | www.nature.com/natureenergy 399

Articles Nature Energy

a b c
Time (h)
0 10 20 30 40 50 60
Al plating/stripping CE (%) 110

100 Short

90

80 d

70

60 100 µm
0 2 4 6 8 10 12 14 16
Cycle no.

e f g
e– e– e– e–

100 µm
Al GF SS Al GF SS

Fig. 2 | The propensity of Al anodes to exhibit heterogeneous growth on conventional substrates. a, CE measured in Al plating/stripping reactions at a
capacity of 3.2 mAh cm–2. Two layers of a GF separator were needed to prevent shorting. b, SEM analysis of Al electrodeposits formed on a planar steel foil
electrode at 4 mA cm–2 and for a capacity of 0.8 mAh cm–2. c,d, Elemental maps of Al (c) and Fe (d) from an EDS analysis of the electrode surface shown
in b. e, Schematic diagram showing the failure mechanism of Al electrodes. A higher areal deposition capacity and the accumulation of dead metal in the
GF membrane over multiple plating/striping cycles result in the proliferation of non-planar metal deposits in the interelectrode space. f, SEM image of Al
electrodeposits formed on a non-planar Ni foam electrode at a rate of 4 mA cm–2 and capacity of 1.0 mAh cm–2. g,h, Elemental maps of Al (g) and Ni (h)
from an EDS analysis of the electrode surface shown in f. SS, inert stainless-steel electrode.

above 99.5%. We note also that even with multiple separators, the
3.2 mAh cm–2 cell fails after only ~60 hours. Together, (1) the rapid
battery shorting and (2) the low plating/stripping reversibility at
elevated areal capacities impose fundamental, pressing challenges
for practical Al batteries.
To diagnose the underlying problems, we performed scanning
electron microscopy (SEM) to characterize the electrodeposit
morphology. The SEM results in Fig. 2b show that the Al deposits
exhibit a heterogeneous, non-planar morphology. Consistent with
prior reports24, the Al deposits were as large as 20–50 μm, sparsely
distributed on the substrate and isolated from each other. This led
to large and broadly distributed L and τ values, which produced
sluggish, incomplete metal stripping, as illustrated in Fig. 2b–d. The
low plating/stripping efficiency and the battery shorting in the Al
anodes are therefore concluded to be consequences of the local-
ized, aggressive growth of the Al electrodeposits normal to the sub-
strate plane. We here follow the literature21 and loosely refer to the
non-planar Al electrodeposit growth as dendritic, but show below
that it is unusually aggressive and possesses previously unreported
characteristics that appear to arise from the strong affinity of Al for
the separator (Fig. 2e).
The speed with which our Al electrodes fail by short-circuiting—
even at conventional electrode areal capacities—exceeds typical val-
ues for Li anodes21, which implies that the materials physics that
control the non-planar electrodeposition of Al are quite different.
It also explains the rather low areal capacity limits explored in pre-
vious studies. We note further that whereas deposition of Li in a
3D porous material, such as a Ni foam, can substantially extend
the time-to-failure due to short-circuiting25, deposition of Al in a
Ni foam substrate (Fig. 2f–h) has, at most, a negligible impact on
the deposition morphology. Consequently, using Ni foam as the
anode substrate exhibited a rather limited improvement on the elec-
trochemical performance (Extended Data Fig. 1). The first clue to
understanding the aggressive non-planar deposition of Al is that
glass fibre (GF) (SiO2) membranes, rather than organic polymers,
are used as the separator. This choice stems from the fact that typi-
cal polyolefin separators are chemically unstable in the imidazolium
chloride + AlCl3 electrolyte melt. Early literature reports show that
there is a possibility for a strong chemical affinity between metallic
Al and SiO2 via an interfacial reaction: 4Al + 3SiO2 → 2Al2O3 + 3Si
(refs 26,27), which implies that, in addition to the other factors that
may drive non-planar metal electrodeposition, a chemical driving
force favours the growth of Al electrodeposits into the separator.
This growth mode was confirmed using SEM and energy dispersive
spectroscopy (EDS) analysis on either a stainless steel or a Ni foam
substrate (Supplementary Figs. 5 and 6), which explains the speed
with which Al is able to short-circuit cells even when multiple layers
of separator membranes are used or when the deposition is per-
formed using a Ni foam substrate.
Uniform Al deposition enabled by O-mediated metal–
substrate bonding
This discovery poses a perhaps obvious question: could the pro-
pensity of Al to grow into the separator be countered by creating
metal–substrate bonding that is strong enough to chemically usher
Al into a patterned substrate? This question can be theoretically
reasoned by an analysis of the nucleation process in electrodeposi-
tion, which plays a pivotal role in determining the size, uniformity

400 NatUre Energy | VOL 6 | April 2021 | 398–406 | www.nature.com/natureenergy

Nature Energy Articles
a b

50

40

No. of particles
30

20

10

0
500 nm 50 100 150 200 250 300 350
Particle diameter (nm)

c d

500 nm 500 nm

e f

Al

5 µm 5 µm
C

g h

C(002)
Intensity (a.u.)

Al(200)
C(100) Al(111)
90 Al(200)
Al(111) C(100)

60

25 30 35 40 45 50
C(002) 30
2θ (°)

Fig. 3 | Microstructure of Al metal deposits formed on a substrate with strong metal–substrate bonding. a,c–f, SEM analysis of the Al deposit
morphology at varying areal capacities: 0.2 (a), 1 (c), 3 (d), 3 (e) and 4 (f) mAh cm–2. Inset (e): SEM and corresponding EDS mapping of a selected region
in e. b, Particle size distribution at 0.2 mAh cm–2 (average = 139 nm). g, 2D XRD pattern of Al deposits on carbon fibres. h, Corresponding integrated XRD
line scan for the material in g. a.u., arbitrary units.

and distribution of electrodeposits8. On the application of a volt-
age, the initial process is the desolvation of the coordinated species
in the electrolyte as they approach the negatively charged electrode
surface28,29. In this desolvation process, the metal-containing spe-
cies are converted into a cationic form by losing the coordinating
ions and/or molecules. The subsequent step is the electrochemical
adsorption of the species: Mn+ + ne− + ∗ → M∗, where * refers
to the adsorption site and M* to the adsorbed metal. This reaction
can be dissociated into two steps: (1) electrochemical reduction of
the metal, Mn+ + ne− → M(free), and (2) binding with the sub-
strate, M(free) + ∗ → M∗, which directly depends on the interaction
between the metal and the substrate; the nucleation overpotential is
determined by the free energy change ΔG of the latter30. Classical
nucleation theory then implies that strong metal–substrate bonding
results in a smaller critical nucleate size, rc = −2γ/ΔG, where γ is the
energy penalty for creating a unit area of the metal surface and ΔG

NatUre Energy | VOL 6 | April 2021 | 398–406 | www.nature.com/natureenergy 401

Articles Nature Energy

a b c
Data Data 24,000 Data
12,000 Peak 1 Peak 1 Peak 1
Peak 2 3,000 Background Background
Intensity (c.p.s.)

Intensity (c.p.s.)

Intensity (c.p.s.)
Peak 3 Fit 18,000 Fit
Peak 4
8,000 Background
Fit 2,000
12,000

4,000 1,000
6,000
d e f
Data Data 24,000 Data
12,000 Peak 1 Peak 1 Peak 1
Peak 2 3,000 Peak 2 Peak 2
Intensity (c.p.s.)

Intensity (c.p.s.)

Intensity (c.p.s.)
Peak 3 Peak 3 Background
Background 18,000
Peak 4 Fit
Peak 5 Fit
8,000 2,000
Background
Fit 12,000

4,000 1,000
6,000

292 290 288 286 284 282 80 78 76 74 72 538 536 534 532 530 528
Binding energy (eV) Binding energy (eV) Binding energy (eV)

Fig. 4 | XPS characterization of the metal–substrate interaction. a–f, C 1s (a,d), Al 2p (b,e) and O 1s (c,f) spectra of Al samples in carbon fibre electrodes
in the as-deposited state (a–c) and the sputtered state (d–f). The peaks that correspond to the Al–O–C bond are shaded by grey and indicated in the plots
by vertical black lines. On sputtering, surface materials are removed from the samples by the incident Ar+ ion beam. Peak assignments are tabulated in
Supplementary Table 2. c.p.s., cycles per second.

is the Gibbs free energy associated with the formation of the con-
densed metal phase during the earliest stages of nucleation.
The implementation of this concept in a battery anode is not
straightforward, because the nucleation occurs on an electroni-
cally conducting substrate, which means it does not strictly become
homogeneous. As discussed above, the formed metal nuclei may
also make strong bonds with the separator, substrate or both, which
will affect their shape. Additionally, to achieve a high electrode
reversibility, the geometrical and chemical integrity of the substrate
must remain intact over repeated cycles of metal plating/stripping.
On this basis, a screening framework for promising substrate mate-
rial is proposed: the substrate should be electrically conductive,
mechanically robust and capable of forming interfacial metal–sub-
strate bonding. It is important that the chemical reaction between
the substrate and the metal is limited within the interface, rather
than penetrates into the bulk. A bulk reaction, owing to the crys-
tallographic and volume changes associated, inevitably gives rise to
morphological instabilities of the substrate over cycling.
In screening for potential substrate materials, free-standing
interwoven carbon fibres (SEM image in Supplementary Fig. 7) are
singled out because previous studies show that these materials form
multiple types of interfacial covalent bonds, which include Al–C,
Al–O–C and so on, with Al metal during electroplating31,32. The cal-
culated binding energy of Al–O–C is as high as a few electronvolts31,
which means that the driving force for Al to grow on the substrate
surface is rather large. This characteristic was previously utilized
to electroplate a uniform Al precoating layer on carbon, which
improved the Al/C wettability in subsequent metallurgical manu-
facturing of Al/C composite materials31. This carbon-based mate-
rial is also of specific interest here because the bonding occurs via
interfacial interaction at oxygen-enriched defective sites on the sur-
face rather than via bulk phase transition. Of note is that the Raman
spectrum of the interwoven carbon fibres shows both the D band
and G band (Supplementary Fig. 8). This means that the carbon
contains a considerable amount of defects, which reportedly pro-
motes the bonding with Al (ref. 31). X-ray photon electron spectros-
copy (XPS) characterization (Extended Data Fig. 2) shows that the
exposure of carbon fibres to the imidazolium chloride + AlCl3 elec-
trolyte remarkably increases the level of oxygen enrichment without
altering the microstructure of the substrate. On these bases, in a
non-planar architecture made of interwoven carbon fibres, the pos-
sible chemical metal–substrate bonding between C and Al mediated
by oxygen is then hypothetically able to promote uniform Al growth
on individual carbon fibres (Fig. 1), as opposed to the macroscopic
outer surface of the architecture, as shown in Supplementary Figs.
5 and 6.
Figure 3 reports the main results of the present study. The Al
metal electrodeposition morphology on interwoven carbon fibres
was investigated using SEM. In stark contrast to the Al deposited
on planar stainless steel and on non-planar Ni foam, for which
the deposit sizes are quite variable with average values of L on the
order tens of micrometres, the Al electrodeposits on the carbon
fibre surface with L typically in the range 100–200 nm in the nucle-
ation stage (Fig. 3a,b); this that for a fixed D, τ could be reduced
by a factor of 10,000 or more. As the deposition capacity increases,
the nanoscale Al crystals grow laterally (Fig. 3c–e and EDS maps
in Supplementary Fig. 9), that is expand along the carbon sur-
face to generate a conformal Al coating layer of thickness on the
order of 102 nm (for example, 160 nm, see Supplementary Fig.
10) that appear to strictly follow the surface curvature (Fig. 3d).
Supplementary Fig. 10 also shows the morphology of the Al deposit
layer that faces the carbon fibre—nanoscale Al grains are seen to
merge into a compact layer. The Al deposition morphology is highly
uniform across a macroscopic area, as shown in Supplementary Fig.
11, without any observable evidence of aggressive and/or dendritic
growth. Microstructural characterization of the morphology after
thermal annealing shows that surface diffusion is not as effective a
mechanism for smoothing the Al electrodeposits (Supplementary
Fig. 12); rather, the initial uniform nucleation promoted by metal–
substrate bonding plays a dominant role. These results confirm that
the strong metal–substrate bonding is able to (1) guide a metal elec-
trodeposition precisely using the surface topography of a strongly
interacting non-planar architecture and (2) promote the uniform
nanoscale nucleation and lateral growth of metal deposits to create
a compact film.
We increased the deposition capacity beyond 3 mAh cm–2 and
observed that the growth of Al enters a second stage (Fig. 1): growth
of microscale Al crystals on the compact nanoscale Al layer (Fig. 3f).

402 NatUre Energy | VOL 6 | April 2021 | 398–406 | www.nature.com/natureenergy

Nature Energy
a 0 2,000 4,000 6,000
Al CE(%)
105
100
95 CE = 99.8%
90
0 1,000 2,000
b
0 200 400 600
105
Al CE(%)
100
95 CE = 99.6%
90
0 1,000 2,000
c 0 50 100 150 200 250
105
Al CE(%)
100
95 CE = 99.6%
90
0 1,000 2,000
Fig. 5 | Electrochemical cycling behaviour of structured Al electrodes in galvanostatic plating/stripping experiments. a–c, Al plating/stripping efficiency
measured at areal capacities of 0.8 (a), 3.2 (b) and 8 (c) mAh cm–2. d, Voltage profiles measured during Al plating/stripping at a very high areal capacity
of 8 mAh cm–2. The current densities used for the respective measurements are 4 (a) and 1.6 (b,c) mA cm–2.
X-ray diffraction (XRD) (Fig. 3g,h) verified that only Al crystals
were present on the carbon substrate fibres. The microscale Al
deposits gradually filled the space among the carbon fibres (mor-
phology at 8 mAh cm–2 shown in Supplementary Fig. 13). The
microscale Al deposits were intimately connected to the compact
nanoscale Al layer and to each other. This observation confirms
that the metal–substrate bonding has a decisive influence on the
Al deposition morphology. After the substrate surface was fully
occupied by the compact nanoscale Al layer, the newly deposited Al
crystals resumed the intrinsic growth mode at the micrometre scale.
As a more direct control experiment, we artificially precoated the
carbon fibres with an interphase that blocked the possible Al–O–C
bonding, and evaluated the Al electrodeposition morphology on
this ‘deactivated’ carbon fibre matrix (Supplementary Fig. 14). The
SEM characterization shows that the Al deposition was no longer
uniform over the surface of the carbon fibres. This negative control
experiment validates our hypothesis that oxygen-enriched surface
chemistry plays an indispensable role to realize the control over
deposition morphology.
Moreover, we devised physical simulation experiments to more
quantitatively understand the role played by a large driving force
at the nucleation stage and its impact on the ultimate deposition
morphology (Extended Data Fig. 3, Supplementary Note 1 and
Supplementary Fig. 15). It was shown that with a nucleation driven
by a large overpotential of 2 V, uniform Al growth was achieved
at the subsequent stage of deposition under conventional con-
stant current conditions. This set of results confirms that a large
driving force imposed in the initial nucleation process serves as a
determinant factor in achieving uniform Al deposition. However,
this physical approach is fundamentally associated with chemical
and hydrodynamic instabilities, which makes it unsuitable for bat-
tery applications (detailed discussion after Supplementary Fig. 15).
Further molecular-level simulation of this process could be obtained
by joint density function theory methods33 that capture the detailed
kinetic steps near a solid–liquid interface, for example, desolvation,
bonding and interphase formation33–35.
To more thoroughly interrogate the nature of the metal–sub-
strate bonding, we performed X-ray photoelectron spectroscopy.
Figure 4a–c shows the XPS results of the as-deposited sample
(3 mAh cm–2), which are consistent with the conclusion that no
possible metal–substrate bonding is detected on the surface. This
observation is expected because the substrate architecture is fully
covered by the compact deposition layer, which means that the
metal–substrate interface is not exposed. Figure 4d–f shows the XPS
results of the sample after Ar+ sputtering, during which the material
NatUre Energy | VOL 6 | April 2021 | 398–406 | www.nature.com/natureenergy
Articles
8,000 Cycle no.
d
0.6
no.1
no.50 (500 h)
0.4 no.125 (1,250 h)
3,000 Time (h)
no.200 (2,000 h)
Potential (V)
800 Cycle no. no.360 (3,600 h)
0.2
0.0
3,000 Time (h)
300 350 Cycle no.
–0.2
0 2 4 6 8
3,000 Time (h) Capacity (mAh/cm2)
on the sample surface was etched away. In this case, obvious new
peaks emerged (C 1s, 283.5 eV; Al 2p, 74.7–74.9 eV; O 1s, 531.9 eV),
which are ascribed to Al–O–C bonding. XPS peak assignments are
tabulated in Supplementary Table 2. The observation is consistent
with that of previous work31. The metal–substrate bonding is also
evidently mediated by the surface oxidation species of the carbon.
In contrast, XPS measurements of Al electrodeposits on control
substrates suggest that the Al–O–C bonding and other metal–sub-
strate bonding do not exist on stainless steel and Ni foam (Extended
Data Fig. 4).
Electrochemical performance of regulated metal anodes
To evaluate the reversibility of the Al deposited at these high areal
capacities, we measured the plating/stripping CE of the Al elec-
trodes in electrochemical cells. The CE reveals the ratio between the
amount of metal that can be stripped and the amount that is origi-
nally plated. The results are reported in Fig. 5. In stark contrast to
the low CE and rapid battery failure observed on a planar stainless
steel substrate, anodes produced by Al electrodeposition in a pat-
terned substrate composed of carbon fibres manifested a high level
of reversibility (99.4–99.8%, stripped morphology in Supplementary
Fig. 16) over a broad range of capacities (0.8–8 mAh cm–2). The
results also showed that the cells maintain stable operations for
more than 2,000 hours (Fig. 5a–c). The voltage profile in Fig. 5d
confirms that the plating/stripping reaction was stable, as evidenced
by the narrowing overpotential over the 3,600 hours of cycling, and
does not show any trend towards deterioration. It is also notewor-
thy that the CE at 8 mAh cm–2 was not compromised by the pres-
ence of microscale Al deposits, owing to what we suspect is a robust,
interconnecting electronic pathway in the conductive non-planar
architecture that guarantees a good electrochemical access to the
metal and prevents the formation of dead or orphaned Al. When
paired with a cathode, this improvement could be clearly reflected
in proof-of-concept full cells at an elevated capacity of ~1 mAh cm–2,
which is 1–2 orders of magnitudes higher than the values reported
in the literature (Supplementary Figs. 17 and 18). Consistent with
the plating/stripping measurement, batteries using Al foil as the
anode fail rapidly in <100 cycles. By contrast, ‘anode-free’ full bat-
teries with a carbon-fibre anode without any pre-stored Al showed
a stable cycling and excellent capacity retention over thousands of
cycles. These results of the full battery tests further confirm the high
reversibility of Al plating/stripping on carbon fibres when subject
to cycling.
As an extreme test of the approach, we performed plating/strip-
ping experiments at an unprecedentedly high current density, that
403
Articles
a b
5 µm
c d
Shorting-induced
0.5
endless charging
0 90
Potential (V)
–0.5
0.4 mAh cm–2 plating
on inert substrate
–1.0
2nd cycle
–1.5
0.0 0.2 0.4 0.6 0.8
Areal capacity (mAh cm–2)
Fig. 6 | Ultrahigh current density plating/stripping of Al. a,b, SEM analysis of Al plated at a current density of 40 mA cm–2 on stainless steel (a) and on
carbon fibres (b). c, Voltage profile showing that the Al||stainless steel cells fail through short-circuiting by the second cycle. d, CE versus cycle index for
the Al plating/stripping process on carbon fibres with metal–substrate bonding.
is, 40 mA cm–2. SEM characterization revealed that diffusion-limited
ramified electrodeposit structures with classical fractal and/or den-
dritic morphologies formed on the planar electrode (Fig. 6a). In
contrast, nanoscale, compact Al electrodeposition was maintained
on the carbon-based fibres (Fig. 6b). As expected, the cells using
the planar stainless-steel foil failed quickly, after <5 minutes, due
to the short-circuiting process (Fig. 6c). This can be compared with
the electrochemical results (Fig. 6d), which showed high levels of
reversibility and long lifetimes (for example, CE = 99.96% over
60,000 cycles at a current density of 40 mA cm–2 and an areal capac-
ity of 0.4 mAh cm–2) with the regulation produced by the metal–
substrate bonding.
The stable plating/stripping behaviours of Al under the regula-
tion of metal–substrate bonding can be understood in a quantitative
manner. As Al forms a compact layer on the surface of the carbon
fibres, the electron transport length scale is maintained as small. As
a quantitative estimation, the characteristic relaxation time τ of the
nanoscale Al deposits formed in the patterned substrate is four orders
of magnitudes smaller than that of the dendritic Al (L ≈ 160 versus
20 μm) formed in the planar case. This means that the nanoscale
Al can be stripped around four orders of magnitude faster. In addi-
tion to considerations based on transport, the non-planar and/or
dendritic metal deposit is associated with a stronger morphologi-
cal instability—as stripping occurs at the bottom of the deposits,
the whole structure detaches from the electrode6. Our observations,
therefore, suggest that an Al anode in which the metal is deposited
in a patterned substrate capable of forming strong bonds with Al are
of potential immediate interest for applications in Al batteries that
operate under various conditions. In Table 1, we compare the elec-
trochemical performance achieved for our Al anodes with those of
404
Nature Energy
5 µm
105
Al CE(%)
100
95
40 mA; 0.4 mAh cm–2
0 20,000 40,000 60,000
Cycle no.
105
Al CE(%)
100
95
40 mA; 1 mAh cm–2
90
0 200 400 600 800 1,000 1,200
Cycle no.
prior literature reports. It is apparent that even without taking into
account the intrinsic advantages of Al metal, which include safety,
cost, ease of manufacturing and high earth-crust abundance, Al bat-
teries based on the proposed anode design show enormous promise.
To explore the broader utility of our concept, we studied the elec-
trodeposition of Zn, another metallic battery anode material that
has aroused enormous research interest recently36–38, in a patterned
substrate composed of interwoven carbon fibres (Supplementary
Fig. 19). The galvanostatic plating/stripping results demonstrate
that, unlike Al, Zn has a low plating/stripping efficiency and slow
kinetics on carbon fibres, which indicates that the substrate archi-
tecture alone is insufficient to achieve highly reversible deposition.
We artificially introduced a strong metal–substrate interaction by
coating the substrate with graphene, which we recently reported
strongly coordinates with Zn by an epitaxial mechanism owing to
a minimized interfacial energy36. The plating/stripping reversibility
and lifetime are improved remarkably by this coating. SEM results
showed that the presence of the strong metal–substrate interac-
tion via the graphene interphase can effectively promote the uni-
form deposition of nanoscale, plate-like Zn metal with a minimized
electron transport length scale, in comparison with the microscale
Zn deposits observed on bare carbon fibres (Supplementary Fig.
19). Moreover, the effectiveness of the metal–substrate bonding
can be demonstrated on planar substrates, which is a more con-
ventional configuration in commercial batteries. We evaluated the
electrochemical plating/stripping of Al on stainless steel coated
by carbon nanotubes with and without carboxylic side groups.
As shown in Supplementary Fig. 20, the planar substrates coated
with carboxylic-functionalized carbon nanotubes manifested stable
cycling and a high CE, in comparison with those of bare stainless
NatUre Energy | VOL 6 | April 2021 | 398–406 | www.nature.com/natureenergy
Nature Energy
steel and even non-planar Ni foam. On this basis, we conclude that
the concept to regulate electrodeposition morphology can be readily
extended to other electrodeposition systems of different chemistries
or geometries by rationally designing an artificial metal–substrate
interphase that provides strong interactions.
Conclusion
In summary, we report that specific interfacial metal–substrate
bonding can be used to achieve a fine control of metal electrodeposit
morphology and a uniform, compact and exceptionally reversible
metal deposition. Using metallic Al as an example, we first showed
how such bonds can be used to overcome the metal’s natural affin-
ity for the separator and to prevent aggressive, non-planar electro-
deposition. Extension of the concept to create patterned Al anodes
revealed that it is possible to achieve highly reversible metal anodes
with combinations of areal specific capacity and cycle life that are
one or more orders of magnitude higher than those in previous
literature reports. Our findings confirm that the reversibility of
metal anodes requires a continuous access to ionic and electronic
transport pathways in the electrode and is strongly correlated with
control of the electron transport length scale. Successful extension
to Zn metal anodes indicates that the concept may be generalized
to achieve a high reversibility in other metallic anodes in batteries.
Methods
Materials. [EMIm]Cl (1-ethyl-3-methylimidazolium chloride, 98%), AlCl3
(99.99%), Whatman glass fibre filters (GF/B), 0.25 mm Zn foil (99.9%),
carboxymethyl cellulose and ZnSO4·7H2O were purchased from Sigma Aldrich.
Al foil (0.25 mm thick, 99.99%) and tantalum foil (0.127 mm thick) were bought
from Alfa Aesar. Ni foam was manufactured by MTI. Stainless steel 304 was
used as a conventional, planar substrate for electrodeposition. Deionized water
was obtained from a Milli-Q water purification system. Graphene dispersion in
N-methyl-2-pyrrolidone (4 wt%), few-layer graphene dispersion in water (4 wt%),
carboxylic-functionalized carbon nanotubes and graphitized carbon nanotubes
were purchased from ACS Material. Free-standing interwoven carbon fibres (Plain
carbon cloth 1071 HCB) were purchased from Fuel Cell Store.
Preparation of electrolytes. Al anode electrolyte. In an Ar-filled glovebox
(<0.1 ppm H2O, <1 ppm O2), AlCl3 powder was added slowly into [EMIm]
Cl in a glass vial with 500 r.p.m. stirring to form a yellowish liquid electrolyte.
The molar ratio was ionic liquid:AlCl3 = 1:1.3, which resulted in a Lewis acidic
electrolyte in which reversible Al plating/stripping was possible. Al foil was placed
in the as-prepared ionic liquid–AlCl3 electrolyte to remove impurities such as
hydrochloric acid in the electrolyte39. After resting for 7 days, the electrolyte turned
colourless and was used as an electrolyte for Al cells.
Zn anode electrolyte. ZnSO4·7H2O was slowly dissolved into the deionized water in
a glass vial to prepare the 2 M ZnSO4 electrolyte for Zn cells. The transparent, clear
electrolyte was rested for 1 day before use.
Characterization of materials. Field-emission scanning electron microscopy
(FESEM) was carried out on Zeiss Gemini 500 SEM equipped with a Bruker EDS
detector to study the electroplate/strip morphology. An accelerating voltage of 5 kV
was used for imaging, and 10 kV was used for the EDS measurement. 2D-XRD
was performed on a Bruker D8 General Area Detector Diffraction System with a
Cu Kα X-ray source. The incident beam angle and the detector angle were both
set at 18°. Raman spectroscopy was performed on a Renishaw InVia Confocal
Raman microscope using a 785 nm laser. XPS experiments were carried out in a
ultrahigh vacuum chamber equipped with a SPECS Phoibos 100 MCD analyser.
A non-monochromatized Al Kα (hν = 1,486.6 eV) X-ray source was used for
the analysis. The accelerating voltage was 15 kV and the current was 20 mA.
The chamber had a base pressure of 2 × 10−9 torr. Electrodes were pressed onto a
conductive copper tape and mounted on the sample holder. The charge correction
for the data was done to by adjusting the C 1s binding energy to 284.8 eV for C–C
and C–H bonds40. The XPS analysis regions measured were 65–90 eV for Al 2p,
520–545 eV for O 1s and 265–300 eV for C 1s. A pass energy of 20 eV, step size of
0.05 eV and scan number of 10 were applied to measure each individual region.
Ar sputtering was performed at room temperature with a pressure of 2 × 10−5 torr
using an energy of 1.5 keV.
Electrochemical measurements. The galvanostatic plating/stripping performances
of metals in coin-type electrochemical cells (CR2032) were tested on Neware
battery test systems at room temperature. The diameter of the electrodes in the
plating/stripping measurements was 0.5 inch (1.27 cm). Electrodes are separated
NatUre Energy | VOL 6 | April 2021 | 398–406 | www.nature.com/natureenergy
Articles
by one layer of GF/B (Whatman), unless otherwise specified. Stainless steel and
carbon cloth were washed by ultrasonication in deionized water and acetone.
In each coil cell, ~120 μl of electrolyte was added by pipette to make sure the
electrodes and the separator were wetted.
In the Al metal plating/stripping CE measurements, the electrochemical
cells had two electrodes: an Al source electrode and a substrate electrode (for
example, Al||stainless steel, Al||carbon fibres and so on). In the measurement,
a certain amount of Al was plated on the substrate electrode of interest, and
a reverse potential was applied to strip the deposited Al metal back to the Al
source electrode, CE = plating stripping capacity
capacity on the substrate × 100%. In the Al||stainless
steel cells, a commercial Al foil was used as the Al source. In the Al||carbon fibre
cells, Al electrodeposited on carbon fibre was used as the Al source because the
battery short was caused by uncontrollable Al growth during both the plating on
the substrate electrode and on the Al source electrode. As a result, both of the
electrodes needed to be carbon fibres to measure the true stabilizing effect of using
carbon fibres as the deposition substrate.
Full cells were assembled to further evaluate the Al plating/stripping behaviour
in the anode. In ‘anode-free’ full cells constructed using carbon fibres, no Al metal
was prestored in the anode. This means that Al metal solely comes from battery
charging, 4Al2Cl7– + 3e– → Al(s) + 7AlCl4–. To offset the initial capacity loss possibly
owing to side reactions, as can be seen in Figs. 5 and 6, carbon fibre electrodes were
subjected to a formation process: Al was plated and striped on carbon fibres at 40
mA cm–2, 0.4 mAh cm–2 for 50 cycles. After the formation process, the carbon fibre
electrode was charged to 0.7 V versus Al3+/Al to fully remove any electrochemically
active Al metal, and was harvested for the subsequent assembly of full cells. In
control full cells, commercial 0.25-mm-thick Al foil was used as the anode. This
means that a large amount of extra Al metal was prestored in the anode, as can be
quantified by an N:P ratio of 100:1 (70 mAh cm–2 anode, 0.7 mAh cm–2 cathode).
The graphite-based cathodes that offered a high areal capacity were designed
according to principles and methods discussed in a prior study20. Non-planar,
interwoven carbon fibre electrodes were used as the matrix to ensure a fast
transport and the physical integrity of the thick, high-areal-capacity cathodes.
Carboxymethyl cellulose was added into a water dispersion of few-layer graphene
to form a viscoelastic fluid. A strain was applied to periodically agitate the fluid to
drive the active materials into the porous medium, that is, the interwoven carbon
fibres. The graphene-infused carbon fibre electrodes (diameter: 1/4 inch; 0.64 cm)
were dried at 60 °C overnight in an oven. Tantalum foils were used as the spacer to
prevent possible corrosion in the cathode.
Data availability
The datasets generated during the current study are available in the article and its
Supplementary Information.
Received: 20 August 2020; Accepted: 15 February 2021;
Published online: 5 April 2021
References
1. Josell, D., Wheeler, D., Huber, W. H. & Moffat, T. P. Superconformal
electrodeposition in submicron features. Phys. Rev. Lett. 87, 016102 (2001).
2. Zachman, M. J., Tu, Z., Choudhury, S., Archer, L. A. & Kourkoutis, L. F.
Cryo-STEM mapping of solid–liquid interfaces and dendrites in
lithium-metal batteries. Nature 560, 345 (2018).
3. Baker, H. The metal of the future. Nature 42, 537 (1890).
4. Tikekar, M. D., Choudhury, S., Tu, Z. & Archer, L. A. Design principles for
electrolytes and interfaces for stable lithium-metal batteries. Nat. Energy 1,
16114 (2016).
5. Zheng, J. et al. Physical orphaning versus chemical instability: Is dendritic
electrodeposition of Li fatal?. ACS Energy Lett. 4, 1349–1355 (2019).
6. Deng, Y. et al. On the reversibility and fragility of sodium metal electrodes.
Adv. Energy Mater. 9, 1901651 (2019).
7. Shi, F. et al. Strong texturing of lithium metal in batteries. Proc. Natl Acad.
Sci. USA 114, 12138–12143 (2017).
8. Biswal, P., Stalin, S., Kludze, A., Choudhury, S. & Archer, L. A. On the
nucleation and early-stage growth of Li electrodeposits. Nano Lett. 19,
8191–8200 (2019).
9. Bai, P., Li, J., Brushett, F. R. & Bazant, M. Z. Transition of lithium growth
mechanisms in liquid electrolytes. Energy Environ. Sci. 9, 3221–3229 (2016).
10. Fang, C. et al. Quantifying inactive lithium in lithium metal batteries. Nature
572, 511–515 (2019).
11. Zheng, J. et al. Regulating electrodeposition morphology of lithium: towards
commercially relevant secondary Li metal batteries. Chem. Soc. Rev. 49,
2701–2750 (2020).
12. Zheng, J. & Archer, L. A. Controlling electrochemical growth of metallic zinc
electrodes: toward affordable rechargeable energy storage systems. Sci. Adv. 7,
eabe0219 (2021).
13. Lin, M.-C. et al. An ultrafast rechargeable aluminium-ion battery. Nature 520,
324–328 (2015).
405
Articles
14. Higashi, S., Lee, S. W., Lee, J. S., Takechi, K. & Cui, Y. Avoiding short circuits
from zinc metal dendrites in anode by backside-plating configuration. Nat.
Commun. 7, 11801 (2016).
15. Zhu, C., Usiskin, R. E., Yu, Y. & Maier, J. The nanoscale circuitry of battery
electrodes. Science 358, eaao2808 (2017).
16. Reid, J. Copper electrodeposition: principles and recent progress. Jpn J. Appl.
Phys. 40, 2650 (2001).
17. Elia, G. A. et al. An overview and future perspectives of aluminum batteries.
Adv. Mater. 28, 7564–7579 (2016).
18. Jayaprakash, N., Das, S. K. & Archer, L. A. The rechargeable aluminum-ion
battery. Chem. Commun. 47, 12610–12612 (2011).
19. Faegh, E., Ng, B., Hayman, D. & Mustain, W. E. Practical assessment of the
performance of aluminium battery technologies. Nat Energy 6, 21–29 (2021).
20. Zheng, J. et al. Nonplanar electrode architectures for ultrahigh areal capacity
batteries. ACS Energy Lett. 4, 271–275 (2018).
21. Albertus, P., Babinec, S., Litzelman, S. & Newman, A. Status and challenges in
enabling the lithium metal electrode for high-energy and low-cost
rechargeable batteries. Nat. Energy 3, 16–21 (2018).
22. Levy, N. R. & Ein-Eli, Y. Aluminum-ion battery technology: a rising star or a
devastating fall? J. Solid State Electrochem. 24, 2067–2071 (2020).
23. Chen, H. et al. Oxide film efficiently suppresses dendrite growth in
aluminum-ion battery. ACS Appl. Mater. Interfaces 9, 22628–22634 (2017).
24. Zhao, Q., Zheng, J., Deng, Y. & Archer, L. Regulating the growth of
aluminum electrodeposits: towards anode-free Al batteries. J. Mater. Chem. A
8, 23231–23238 (2020).
25. Chi, S. S., Liu, Y., Song, W. L., Fan, L. Z. & Zhang, Q. Prestoring lithium into
stable 3D nickel foam host as dendrite-free lithium metal anode. Adv. Funct.
Mater. 27, 1700348 (2017).
26. Roberts, S. & Dobson, P. J. Evidence for reaction at the Al–SiO2 interface.
J. Phys. D 14, L17 (1981).
27. Bauer, R. S., Bachrach, R. Z. & Brillson, L. J. Au and Al interface reactions
with SiO2. Appl. Phys. Lett. 37, 1006–1008 (1980).
28. Bard A. J. & Faulkner L. R. Electrochemical Methods: Fundamentals and
Applications (John Wiley and Sons, 2002).
29. Yang, H. et al. The rechargeable aluminum battery: opportunities and
challenges. Angew. Chem. Int. Ed. 58, 11978–11996 (2019).
30. Pande, V. & Viswanathan, V. Computational screening of current collectors
for enabling anode-free lithium metal Batteries. ACS Energy Lett. 4,
2952–2959 (2019).
31. So, K. P. et al. Improving the wettability of aluminum on carbon nanotubes.
Acta Mater. 59, 3313–3320 (2011).
32. So, K. P., Biswas, C., Lim, S. C., An, K. H. & Lee, Y. H. Electroplating
formation of Al–C covalent bonds on multiwalled carbon nanotubes.
Synth. Met. 161, 208–212 (2011).
33. Choudhury, S. et al. Designing solid–liquid interphases for sodium batteries.
Nat. Commun. 8, 898 (2017).
34. Jadhav, A. L., Xu, J. H. & Messinger, R. J. Quantitative molecular-Level
understanding of electrochemical aluminum–ion intercalation into a
crystalline battery electrode. ACS Energy Lett. 5, 2842–2848 (2020).
35. Zhao, Q. et al. Solid electrolyte interphases for high-energy aqueous
aluminum electrochemical cells. Sci. Adv. 4, eaau8131 (2018).
36. Zheng, J. et al. Reversible epitaxial electrodeposition of metals in battery
anodes. Science 366, 645 (2019).
37. Wang, Z. et al. A metal–organic framework host for highly reversible
dendrite-free zinc metal anodes. Joule 3, 1289–1300 (2019).
38. Zheng, J. et al. Spontaneous and field-induced crystallographic reorientation
of metal electrodeposits at battery anodes. Sci. Adv. 6, eabb1122 (2020).
39. Vaughan, J. & Dreisinger, D. Electrodeposition of aluminum from aluminum
chloride–trihexyl(tetradecyl)phosphonium chloride. J. Electrochem. Soc. 155,
D68–D72 (2008).
40. Barr, T. L. & Seal, S. Nature of the use of adventitious carbon as a binding
energy standard. J. Vac. Sci. Technol. A 13, 1239–1246 (1995).
41. Paranthaman, P. M. A high performance hybrid battery based on aluminum
anode and LiFePO4 cathode. Chem. Commun. 52, 1713–1716 (2016).
406
Nature Energy
42. Zhang, X., Tang, Y., Zhang, F. & Lee, C. S. A novel aluminum–graphite
dual-ion battery. Adv. Energy Mater. 6, 1502588 (2016).
43. Wang, S. et al. A novel aluminum-ion battery: Al/AlCl3-[EMIm]Cl/Ni3S2@
graphene. Adv. Energy Mater. 6, 1600137 (2016).
44. Chen, H. et al. A defect-free principle for advanced graphene cathode of
aluminum-ion battery. Adv. Mater. 29, 1605958 (2017).
45. Chen, H. et al. Ultrafast all-climate aluminum–graphene battery with
quarter-million cycle life. Sci. Adv. 3, eaao7233 (2017).
46. Liu, Z. et al. Carbon nanoscrolls for aluminum battery. ACS Nano 12,
8456–8466 (2018).
47. Zhang, E. et al. A novel aluminum dual-ion battery. Energy Storage Mater. 11,
91–99 (2018).
48. Wu, F., Zhu, N., Bai, Y., Gao, Y. & Wu, C. An interface-reconstruction effect
for rechargeable aluminum battery in ionic liquid electrolyte to enhance
cycling performances. Green Energy Environ. 3, 71–77 (2018).
49. Zhang, E. et al. Unzipped carbon nanotubes for aluminum battery.
Energy Storage Mater. 23, 72–78 (2019).
50. Xu, H. et al. Low-cost AlCl3/Et3NHCl electrolyte for high-performance
aluminum-ion battery. Energy Storage Mater. 17, 38–45 (2019).
51. Tan, B. et al. Synthesis of RGO-supported layered MoS2 with enhanced
electrochemical performance for aluminum ion batteries. J. Alloy Compd 841,
155732 (2020).
Acknowledgements
We thank F. Zhu, L. Luo, R. Luo, M. Pfeifer and X. Ren for valuable discussions. This
work was supported as part of the Center for Mesoscale Transport Properties, an Energy
Frontier Research Center supported by the US Department of Energy, Office of Science,
Basic Energy Sciences, under award no. DE-SC0012673. This work made use of the
Cornell Center for Materials Research Shared Facilities which are supported through
the NSF MRSEC program (DMR-1719875). The pouch cell assembly line used for the
anode-free electrodes with a 9 cm2 area was supported under award 75039 from the New
York State Energy Research and Development Authority (NYSERDA) and award 76890
from the New York State Department of Economic Development (DED), which were
provided as matching funds to the Center for Mesoscale Transport Properties under
award no. DE-SC0012673. E.S.T. acknowledges support as the William and Jane Knapp
Chair of Energy and the Environment.
Author contributions
L.A.A. directed the research. J.Z. and L.A.A. conceived and designed this
work. J.Z. and L.A.A. wrote the paper with inputs from other authors. J.Z. and T.T.
performed the electrodeposition, electrochemical measurements and structure
characterizations. J.Z., Q.Z., G.W. and D.C.B. assembled the pouch cells. D.C.B. and
K.R.T. collected and interpreted the XPS data with input from A.C.M.; Q.Z., J.Y., X.L.,
Y.D. and J.S. analysed the data. All the authors contributed to the review and editing
of the manuscript.
Competing interests
The authors declare no competing interests.
Additional information
Extended data is available for this paper at https://doi.org/10.1038/s41560-021-00797-7.
Supplementary information The online version contains supplementary material
available at https://doi.org/10.1038/s41560-021-00797-7.
Correspondence and requests for materials should be addressed to L.A.A.
Peer review information Nature Energy thanks the anonymous reviewers for their
contribution to the peer review of this work.
Reprints and permissions information is available at www.nature.com/reprints.
Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in
published maps and institutional affiliations.
© The Author(s), under exclusive licence to Springer Nature Limited 2021
NatUre Energy | VOL 6 | April 2021 | 398–406 | www.nature.com/natureenergy
Nature Energy Articles

Extended Data Fig. 1 | Electrochemical plating/stripping behavior of Al metal on nonplanar nickel foam substrate. a, Coulombic efficiency obtained at
0.8 mAh, 4 mA/cm2. Voltage profiles of Al plating/stripping: (b) 0.8 mAh, 4 mA/cm2;(c) 3.2 mAh, 1.6 mAh/cm2 and (d) 8.0 mAh, 1.6 mA/cm2. The results
mean that the improvement made by using a nonplanar, inert architecture is very limited, particularly at practical capacities, that is 3.2 and 8 mAh/cm2.

NatUre Energy | www.nature.com/natureenergy

Articles Nature Energy

a b C=O C-O
c
Data Data Data
45,000
Intensity (cps)

Peak 1 Peak 1 Peak 1

Peak 2 Peak 2 Peak 2
Peak 3 Peak 3 Peak 3
30,000 Background
Fit
Background
Fit
Background
Fit

15,000

0
290 288 286 284 290 288 286 284 290 288 286 284
Binding Energy (eV) Binding Energy (eV) Binding Energy (eV)
Extended Data Fig. 2 | XPS spectrum of interwoven carbon fibers. a, Pristine, (b) after exposure to IL+AlCl3 electrolyte, (c) after exposure to dimethyl
carbonate as a negative control. Peak assignments: 284.8 eV (C-C, C-H), 286 eV (C-O), 287.5–288 eV (C=O, O-C-O), 289 eV (O=C-O). The intensities at
286 and 287.5 ~ 288 eV suggest that the exposure to IL+AlCl3 notably increases the level of oxygen enrichment on carbon fibers.

NatUre Energy | www.nature.com/natureenergy

Nature Energy Articles

Extended Data Fig. 3 | SEM images and EDS mapping of Al deposition morphology obtained using a sequential, two-step protocol. In Step I, a small
capacity (0.05 mAh/cm2) of Al is deposited at fixed overpotential, η values (for example, η = 0.05, 0.3, 1.0, 2.0, 3.0 V); In Step II, a greater areal capacity
(that is 0.45 mAh/cm2) of Al is galvanostatically deposited at a current density (that is 4 mA/cm2). SEM images of the Al deposition morphology obtained
using this protocol, with the η value equal to (a) 0.05 V, (b) 0.3 V, (c) 1.0 V, (d) 2.0 V, and (e) 3.0 V. f–o, the corresponding EDS mapping results. See
Supplementary Note 1 for a detailed discussion.

NatUre Energy | www.nature.com/natureenergy

Articles Nature Energy

a Al on carbon fiber d Al on steel g Al on Ni foam

40,000 40,000
Data Data Data
12,000 Al-O-C
Intensity (cps)

Intensity (cps)

Intensity (cps)
Peak 1 Peak 1 Peak 1
Peak 2 Peak 2 Peak 2
Peak 3 30,000 Peak 3 30,000 Peak 3
Peak 4 Peak 4 Peak 4

8,000 Background Background Background

Fit
20,000 Fit
20,000 Fit

4,000 10,000 10,000

C 1s Data Data Data
12,000
Intensity (cps)

Intensity (cps)

Intensity (cps)
Peak 1 Peak 1 Peak 1
Peak 2
Peak 3
16,000 Peak 2
Peak 3
16,000 Peak 2
Peak 3
Peak 4 Peak 4 Peak 4

8,000 Peak 5 Background Background

Background
Fit 12,000 Fit
12,000 Fit

4,000
8,000 8,000
292 290 288 286 284 282 292 290 288 286 284 282 292 290 288 286 284 282

b Binding Energy (eV)

e Binding Energy (eV)
h Binding Energy (eV)
24,000 Data 60,000 Data Data
Al-O-C
Intensity (cps)

Intensity (cps)

Intensity (cps)
Peak 1 Peak 1 Peak 1
Background Background 36,000 Background
18,000 Fit Fit Fit

40,000
12,000 24,000
20,000
6,000 12,000
O 1s 24,000 Data 60,000 Data Data
Intensity (cps)

Intensity (cps)

Intensity (cps)
Peak 1 Peak 1 Peak 1
Peak 2 Peak 2 36,000 Background
18,000 Background
Fit
Background
Fit
Fit

40,000
12,000 24,000
20,000
6,000 12,000
538 536 534 532 530 528 538 536 534 532 530 528 538 536 534 532 530 528

c Binding Energy (eV) f Binding Energy (eV) i Binding Energy (eV)

Data 12,000 Data Data
Al-O-C
Intensity (cps)

Peak 1
Intensity (cps)

Intensity (cps)

Peak 1 Peak 1
3,000 Background Background 9,000 Background
Fit Fit Fit
8,000
2,000 6,000
4,000
1,000 3,000
Al 2p
Data 12,000 Data Data
Intensity (cps)

Peak 1
Intensity (cps)

Intensity (cps)

Peak 1 Peak 1
3,000 Peak 2 Peak 2 Peak 2
Peak 3 Background 9,000 Background
Background Fit Fit

2,000
Fit 8,000
6,000
1,000
4,000 3,000
80 78 76 74 72 80 78 76 74 72 80 78 76 74 72
Binding Energy (eV) Binding Energy (eV) Binding Energy (eV)

Extended Data Fig. 4. | XPS spectrum of Al electrodeposits on carbon fibers, stainless steel and nickel foam. XPS of Al deposited on (a)–(c) carbon
fibers, (d)–(f) stainless steel, and (g)–(i) Ni foam. (a)(d)(g) C 1 s spectra; (b)(e)(h) O 1 s spectra; (c)(f)(i) Al 2p spectra. Upper panels and lower panels
show spectra before and after Ar+ sputtering, respectively. After sputtering, the Al-O-C bonding was observed on samples where Al was deposited on
carbon fibers. On other samples, no characteristic metal-substrate covalent bonding is observable.

NatUre Energy | www.nature.com/natureenergy