Review Article

[Link]

Perspectives for electrochemical capacitors and

related devices
Patrice Simon 1,2,3 ✉ and Yury Gogotsi 4,5 ✉

Electrochemical capacitors can store electrical energy harvested from intermittent sources and deliver energy quickly, but their
energy density must be increased if they are to efficiently power flexible and wearable electronics, as well as larger equipment.
This Review summarizes progress in the field of materials for electrochemical capacitors over the past decade as well as out-
lines key perspectives for future research. We describe electrical double-layer capacitors based on high-surface-area carbons,
pseudocapacitive materials such as oxides and the two-dimensional inorganic compounds known as MXenes, and emerging
microdevices for the Internet of Things. We show that new nanostructured electrode materials and matching electrolytes are
required to maximize the amount of energy and speed of delivery, and different manufacturing methods will be needed to meet
the requirements of the future generation of electronic devices. Scientifically justified metrics for testing, comparison and
optimization of various kinds of electrochemical capacitors are provided and explained.

E
fficient energy-storage systems are required to power hybrid/ harvest higher power than batteries but contain lower energy den-
electric vehicles and the ever-increasing number of electronic sity, resulting in an operation time of tens of seconds to minutes.
gadgets, as well as storing energy from intermittent sources ECs store the charge via fast, surface-confined processes, which can
such as wind and sun. High-performance energy-storage systems are be electrostatic or faradic in nature5,7,10,12. Such surface storage causes
needed in our day-to-day lives in a connected environment, since we only minor, if any, volume change of the electrode during operation
all want our electronic devices, such as smartphones and tablets, to resulting in a long cycle life (>106 charge–discharge cycles), as well as
hold their charge and operate throughout the day. Electrochemical a fast charge rate, similar to discharge1. These key features, together
capacitors (ECs) play an increasing role in satisfying the demand with their energy and power densities, make them useful in applica-
for high-rate harvesting, storage and delivery of electrical energy, tions ranging from small devices for power electronics (power buf-
as we predicted in a review a decade ago1. Since then, the need for fer, memory-saving alternating current filtering13) to large-size cells
versatile, ubiquitous, high-power, high-energy-density storage has and modules for automotive transportation14–17. In the latter, they
only increased. are used for power delivery but also for energy harvesting, such as
The Ragone plot presented in Fig. 1 demonstrates the status of braking energy in cars and trams (recharging the EC device via a
power versus energy performance of several energy-storage sys- starter-alternator during vehicle braking steps), which requires high
tems available so far. Batteries lie in the high-energy and low-power power for a short time (a few seconds). Such applications cannot be
region, defining a time constant (operation time) ranging from one achieved with conventional batteries because of the limited power
to tens of hours2,3. They can deliver low power for long stretches and density as well as the charging rate. ECs are also used for grid energy
are used in applications ranging from power electronics to mobility storage for power quality and smoothing, power saving units and
and grid storage. The power and lifetime limitations of batteries are frequency regulation, for which the final device weight can reach
caused by the charge-storage mechanism, which involves transfor- several tonnes. Supercapacitors are also massively implemented in
mation of chemical bonds via electrochemical redox reactions in the electrical pitch control system of wind turbines. In summary,
the bulk of active materials. The high energy density of Li-ion bat- ECs are often used in complement to batteries for short-time (up to
teries (up to ~300 Wh kg–1) explains why they dominate the market, tens of seconds) energy delivery and harvesting, that is, when bat-
and this technology is likely to prevail for a long time3. However, teries fall short in delivering or receiving power15–17.
battery cycle life is limited to a few thousand cycles, owing to vol- Ragone plots (Fig. 1) are widely used in the literature to provide
ume changes in the materials upon cycling4. Lithium-ion batteries information on power versus energy density of devices, as well as
also suffer from a slower recharge rate compared with discharge, providing a benchmark for various technologies. The coloured areas
owing to Li-metal plating at the negative electrode. Although efforts in Fig. 1 correspond to performance obtained using the same charge
are ongoing to increase the lifetime and decrease the charging time and discharge current, whereas areas enclosed by dashed lines refer
of batteries, there are fundamental limitations related to solid-state only to discharge performance (with a recharge at low rate); this pre-
diffusion rate, phase transformations and volume changes on sentation highlights the power capability of ECs compared with bat-
charge/discharge. Also, the energy density of batteries quickly teries. Readers must bear in mind that such a plot gives only part of
decreases with size, limiting the use of micro-batteries for powering the information. More specifically, although Li-ion batteries can be
microscale and wearable devices. discharged in a few tens of seconds, the energy efficiency and cycle
ECs are another major family of energy-storage system with elec- life will be greatly affected, making ECs preferred for high-rate appli-
trical performance complementary to that of batteries1,5–12. They can cations when a long cycle life is required. Caution must be exercised

1
Materials Science Department—CIRIMAT, Université Paul Sabatier, Toulouse, France. 2Réseau sur le Stockage Electrochimique de l’Energie (RS2E), FR
CNRS, Toulouse, France. 3Institut Universitaire de France, Paris, France. 4A. J. Drexel Nanomaterials Institute, Drexel University, Philadelphia, PA, USA.
5
Department of Materials Science and Engineering, Drexel University, Philadelphia, PA, USA. ✉e-mail: simon@[Link]; gogotsi@[Link]

Nature Materials | VOL 19 | November 2020 | 1151–1163 | [Link]/naturematerials 1151

Review Article NaTuRE MaTERIalS

105 batteries performance towards the top right part of Fig. 1, which
3.6 ms 3.6 s highlights a trend: the boundary between electrochemical capaci-
tors — more specifically ECs based on redox pseudocapacitive
0.1 h
materials — and high-rate batteries is becoming increasingly
blurred. In the region marked by the star, the high-energy-density
104 challenge for ECs meets the high-power one for Li-ion batteries. For
the scientific communities, there is a foundation of common basic
science to share and develop (designing and controlling interfaces,
material architectures, creating analytical platforms, using model-
ling and artificial intelligence, and so on).
103
Specific power (W kg–1)

The next sections of this Review summarize progress in the field

of ECs over the past decade as well as showing key perspectives for
10 h future materials research, covering porous carbons, pseudocapaci-
tive materials, and the emerging fields of wearable energy storage
102 and of microdevices and fabric supercapacitors for the Internet of
Things (IoT).
Li-ion
Electrical double-layer capacitors
Na-ion, LTO Electrical double-layer capacitors (EDLCs) store charge electro-
10 statically through reversible adsorption of electrolyte ions onto
Solid state
(Li metal) electrochemically stable, high-surface-area carbon materials1,20.
1,000 h Charge separation occurs upon polarization at the electrode/elec-
Electrochemical
capacitors trolyte interface, producing what Helmholtz described in 1853 as
the double-layer capacitance, C (see also Fig. 2)1,21.
10–2 10–1 1 10 102 103
Specific energy (Wh kg–1) ε0 εr A C ε0 εr
C¼ or ¼ ð1Þ
d A d
Fig. 1 | Ragone plot. The plot shows the trends towards greater specific
power for batteries and specific energy for electrochemical capacitors where εr is the electrolyte dielectric constant, ε0 is the dielectric
(arrows), blurring the boundaries between the two as the trends approach constant of the vacuum, d is the effective thickness of the double
the star. Dashed lines represent zones where the cyclability of the device layer (charge separation distance) and A is the electrode surface
is altered in the case of symmetric cycling (same charging and discharging area. The absence of any faradic contribution in the charge-storage
rate at 100% depth of discharge). For Li-ion batteries, Li plating at the mechanisms explains the specific electrochemical signature, from a
negative electrode is mainly responsible for the decrease in cycle life and rectangular box-shaped cyclic voltammogram to a sloping profile
the limitation of charging rate. All the data for generating Fig. 1 come from during galvanostatic charge discharge (Fig. 2a). Although EDLCs
commercial devices (datasheets or real tests). The data for solid state are the main commercialized EC devices, they suffer from limited
batteries come from industry roadmaps. The diagonal dotted lines and energy density (equation (2)), and their challenge is to increase the
timescales represent characteristic operation timescales, obtained by amount of energy stored.
dividing the energy by the power. LTO, lithium titanium oxide (Li4Ti5O12).  
1 CV 2
W¼ ; ð2Þ
2 3; 600

when reporting performance metrics in Ragone plots, to avoid deliv- where C is the capacitance (in farads), V the cell voltage (volts) and
ering biased results18,19. Energy and power densities of commercial W the energy (watt hours).
devices are normalized by the weight (gravimetric) or the volume of Although high capacitance (>200 F g–1) could be achieved by
the cells (volumetric), which can be easily obtained from direct mea- adding redox functions1,12 (such as by doping carbon with hetero-
surement of cell weight or by cell volume calculation. Materials pre- atoms7,22), the limited cell voltage of aqueous electrolytes (typically
pared in the laboratory scale may not be assembled in a realistic cell less than ∼1.5 V) due to water electrolysis has so far hampered the
(with minimized contributions from the electrochemically inactive development of aqueous technology. Today, commercial EDLC cells
materials such as electrolyte, current collectors or packaging), so that use non-aqueous electrolytes — typically 1 M (C2H5)4N+,BF4– dis-
reporting performance metrics normalized to the total cell weight solved in acetonitrile or propylene carbonate — that yield an aver-
or volume makes little sense. Extrapolating materials-based perfor- age cell voltage of 2.7 V (refs. 23,24).
mance to a fully packaged cell may lead to unrealistic numbers that Self-discharge is an important parameter for ECs. It can originate
deliver the wrong message to the community11,18. Since the cell stack from charge recombination at the surface of the porous carbon elec-
includes current collectors, separators, active material films and elec- trode by diffusion of ions from the double layer to the electrolyte,
trolyte, energy and power densities normalized per dried cell stack or from the presence and consumption of redox impurities such as
weight or volume should be used (including everything except elec- surface functional groups and/or water traces in the organic elec-
trolyte weight). Benchmarking with other commercial devices can be trolyte20. Although the self-discharge of ECs is higher than that of
easily achieved by dismantling cells and reporting the same perfor- batteries owing to the different charge-storage mechanism (elec-
mance metrics: energy and power performance normalized per stack trostatic or surface redox versus bulk redox), commercial ECs have
volume or dried stack weight. Following these simple recommenda- now reached low leakage current (a few microamps per farad at the
tions makes it possible to report trustworthy research results. maximum voltage at room temperature)15. Such low values for the
Our dependence on electrical energy has only increased over leakage current correspond to several weeks to move from the max-
the past decade, and high expectations placed on batteries and ECs imum voltage to half voltage once the cell is charged. Additionally,
make the need to improve performance more relevant than ever. being high-power devices, supercapacitor banks may need a ther-
All the research and development efforts aim at pushing ECs and mal management system to dissipate the heat generated during

1152 Nature Materials | VOL 19 | November 2020 | 1151–1163 | [Link]/naturematerials

 NaTuRE MaTERIalS Review Article
a
3
20 mV s–1
1.0

0.5 2
Current (A g–1)

Voltage (V)
0.0

–0.5 1

–1.0

0
0.0 0.5 1.0 1.5 2.0 2.5 0 10 20 30 40 50 60
Voltage (V) Q (mAh g–1)

b c d

Small radius of curvature:

efficient charge storage

High capacitance
EMI-TFSI
bulk liquid

25 24%
PF–6 50
20 Anion
BMI+
ρion (nm–3)

15 Planar electrode

Capacitance (F g–1)
Cation
10
40
5
76%
0
0 2 4 6 8 10 12
z (nm) 0.7-nm pore
3.0
Relative ions per surface charge

30
Neat ionic liquid
2.5 0.005 0.006 0.007 0.008
Organic electrolyte ⟨CSpC⟩ (e)
2.0 Charge storage per carbon
1.5

1.0
Large radius of curvature:
0.5 inefficient charge storage

0.0 Low capacitance

0 2 4 6 8 10 12
z (nm) 0.7-nm pore

Fig. 2 | Carbon-based electrical double-layer capacitors. a, A typical CV (left) and a galvanostatic discharge plot (right) of a capacitive EDLC; Q is the
gravimetric capacity normalized to carbon weight. b–d, Charge-storage mechanisms. b, Snapshot of BMI,PF6 ionic liquid electrolyte between two graphite
electrodes under a polarization of 2 V (top). The long-range layered structure observed as a consequence of charge overscreening (middle) is suppressed
when adding an organic solvent (bottom). ρion, ion density. c, Formation of a superionic state as the consequence of the creation of co-ion pairs when the
ionic liquid EMI-TFSI is confined in pores with a size comparable to the ion size (top). The first solvation shell around an anion contains up to 24% anions
without polarization (middle). This results in increased ion population in the pores (bottom). d, The charge stored per carbon in nanopores of 3D porous
carbons increases with the decrease of the curvature radius, as the consequence of stronger/closer interactions of the confined ions with the carbon walls.
Adapted with permission from: c, ref. 53, SNL; d, ref. 55, American Chemical Society, with the help of Y. M. Liu. Panel b courtesy of Celine Merlet.

operation. However, compared with high-power batteries (using combined with modelling30,37–41, were adapted to tracking ion
Li4Ti5O12 anodes or LiFePO4 cathodes, for instance), the lower series fluxes and adsorption in nanopores. Electrochemical quartz-crystal
resistance of ECs decreases the polarization and heat generation25,26. microbalance experiments (not only in conventional gravimetric
Key advances have been made during the past decade in under- modes31,42 but also in modified dissipation43,44 or a.c.-gravimetric
standing the basic science driving the formation and charge of the modes45) and atomic force microscopy experiments46,47 revealed the
double layer in nanoporous carbon electrodes10,12,27,28, starting in 2006 electrolyte dynamics and ion adsorption during operation. These
with the pivotal discovery of the capacitance increase in subnano- research efforts led to revisiting the basic concept of the Helmholtz
metre carbon pores29. In situ characterization tools, such as nuclear double-layer formation at planar48,49 and nanoporous carbon elec-
magnetic resonance (NMR)28,30–32, neutron and X-ray scattering27,33, trodes. They also brought about a better understanding of the origin
mass spectrometry34, Raman and infrared spectroscopies35,36, of this capacitance increase. The increase is now attributed to the

Nature Materials | VOL 19 | November 2020 | 1151–1163 | [Link]/naturematerials 1153

 Review Article NaTuRE MaTERIalS

a b
1,000
Li-ion battery F2 O – N
+
Pseudocapacitor C C S O N
S F2 O
Electrochemical double-layer capacitor + –
– +
+ –
Specific energy (Wh kg–1)

O
– + O
100 + – + e– –
O – 2e–
– + O
N
– + –
+
O
– 2e– +
– +
+e– –e– + – –+ +e– –e–
F2 O – – + + N
C C S O – +
+
+ e– N
S F2 O
10 – – –
+ +
– + + +
– + + e– +
O O
O
N
O
Discharge Charge
O
1
1 10 100 1,000 10,000
Discharge time (s)
c d
e–
Electrolyte
90

60

Carbon Ions
Phase angle (degrees)
30

–30

AI electrolytic
–60 Graphene EDLC
Activated carbon EDLC

–90
e– Current collector e– 10–3 10–2 10–1 100 101 102 103 104 105
Frequency (Hz)

Fig. 3 | Current performance of carbon-based electrical double-layer capacitors and perspectives for improvements. a, Comparison of discharge energy
at various discharge times for a Li-ion battery (red), a pseudocapacitor (green) and an EDLC (blue) of similar volume. For discharge times less than ~10 s,
EDLCs can deliver higher energy density, while the discharge time should reach ~30 s for the current generation of pseudocapacitors. b, Concept of
biredox electrolyte, in which anions and cations are functionalized with redox active species to contribute to the charge-storage mechanism in addition
to double-layer storage. c, Surface modification of porous carbon electrodes (passive layer, for instance) to enlarge the operating voltage window. The
passive layer (orange colour) blocks the electron transfer at the electrolyte/carbon nterface (double red arrows), and allows for ion tranfer (double
green arrows). The electrical percolation is ensured by the carbon grains in the volume of the electrode (red arrows) and by direct contact of the carbon
electrode with the current collector. d, Ultra-high-rate graphene-based EDLC for a.c. to d.c. conversion, which can compete with aluminium electrolytic
capacitors for filtering applications. Adapted with permission from: a, ref. 57, The Electrochemical Society, IOP; d, ref. 13, AAAS. Panel b Credit: O. Fontaine

partial desolvation of ions resulting in ion accessibility to subnano- These results show that carbons with average pore size below 1 nm
metre pores27,28,30,31,39,42,43,50,51, leading to a specific organization of can suppress the diffuse layer and take advantage of the overscreening
the electrolyte confined in these nanopores51,52 as well as the cre- effect. A counter-ion charge-storage mechanism (adsorption of ions
ation of image charges40,53 on the carbon surface, and denser ion of the opposite charge of the carbon) may be preferred, rather than an
packing39,54,55. Figure 2b–d summarizes the current understanding of ion exchange mechanism, to limit ion fluxes and associated decrease
EDLC theory provided by these theoretical and computational stud- of ion transport31,42,48. While NMR studies have shown that the con-
ies. While, as a consequence of the charge overscreening, a relatively finement affects the ion dynamics by decreasing the self-diffusion
long-range layered structure is observed for a neat ionic liquid elec- coefficient in pores by two orders of magnitude28, ion transport is still
trolyte at planar graphite electrodes under polarization, the presence fast enough in these small and short pores to ensure low resistance
of solvent breaks the electrolyte organization (organic electrolyte; and high power performance1,12,29. Also, pores possessing ionophobic
see Fig. 2b). The confinement of a neat ionic liquid electrolyte in character have shown faster ion transport in non-aqueous electro-
pores with a size comparable to the ion size results in the formation lytes compared with ionophilic ones37. In neat ionic liquids, the cre-
of a superionic state with the creation of co-ion pairs, thus improv- ation of image charges on the carbon surface resulted in the creation
ing the capacitance (Fig. 2c). In nanopores of three-dimensional of co-ion pairs in nanopores, also called a superionic state, associated
(3D) carbons, the capacitance increase observed for small pore sizes with important capacitance enhancement53,56.
is the result of the absence of overscreening, thus reducing the effec- Interestingly, those scientific advances were transferred into real
tive double-layer thickness and causing stronger/closer interactions commercial EDLC products in the way that carbon manufactur-
of ions with carbon walls due to the confinement (Fig. 2d). ers modified their synthesis process to prepare activated carbons

1154 Nature Materials | VOL 19 | November 2020 | 1151–1163 | [Link]/naturematerials

NaTuRE MaTERIalS Review Article
with an average pore size in the nanometre range17. As a result, the redox density with liquid-like fast kinetics. The cation and anion
energy density of current commercial EDLCs (currently up to 10 of these biredox ionic liquids bear moieties that undergo very fast,
Wh kg–1)17 exceeds by a factor of 2 that of devices of the same size in reversible redox reactions, increasing the charge storage by transfer-
2005 (<4 Wh kg–1), and current EDLCs can deliver high power for ring electrons at the negative (reduction) and positive (oxidation)
tens of seconds, thus nicely replacing or complementing batteries electrodes64,65. The capacitance of the porous carbon electrode was
in a number of applications57 (Fig. 3a). However, the whole picture found to be twice as large for 0.5 M biredox salts in BMIm-TFSI
of the ion transport and adsorption in carbon nanopores is not yet (1-butyl-3-methylimidazolium bis(trifluoromethylsulfonyl)imide)
clear, and fundamental studies are still needed to better understand electrolyte compared with biredox-free solution65. The concept
the ionic fluxes inside large nanoporous networks with a variety of sounds appealing, and more work will be needed to (i) understand
pore shapes, connections between pores, and pore-wall structure the charge transfer mechanism including possible self-discharge
and properties. processes and (ii) synthesize biredox moieties with high capacity
Although important, fine-tuning the porous carbon structure and tuned electrode potentials to further improve the performance.
to take advantage of the electrolyte confinement effect may not be Increased voltage window and carbon capacitance was also recently
able to bring the energy density beyond 20 Wh kg–1, as required reported by using solvent-free ionic liquid electrolytes51,52, involving
for a major expansion of the market for EDLC devices. A way to ion reorganization inside the micro/mesoporous carbon network.
increase the carbon capacitance is to graft electroactive molecules These processes rely on local anion–cation exchange processes on
on porous carbon electrodes, thus adding a redox contribution to the angstrom scale, resulting in short time response.
the double-layer capacitance. In such an approach, pioneered by Alternatively, cell voltage improvement could be obtained by
Bélanger and co-workers, grafting anthraquinone58,59 from diazo- modifying the electrode/electrolyte interface via the design of passive
nium salts led to a considerable increase in capacitance in aqueous layers at the carbon surface, which would shift the electrolyte reduc-
electrolytes. However, stability remains a concern, as part of the tion and/or oxidation reactions to higher over-voltages (Fig. 3c).
redox molecules is adsorbed on the carbon and not grafted, and This is basically what is happening at graphite anodes during Li-ion
their further release into the electrolyte may lead to capacitance intercalation in Li-ion batteries, where a solid electrolyte interphase
decrease with time and reduced cycle life. (SEI) spontaneously forms on the graphite particle surface66. Once
Alternatively, increasing the cell voltage would be even more formed during the first cycle, the SEI layer prevents further electro-
rewarding as regards energy improvement (see equation (2)). Cell lyte reduction while allowing Li-ion intercalation into the graphite
voltage is driven by the electrochemical stability window of the underneath. In a similar way, designing specific, tailored proper-
electrolyte and the presence of electrochemically active surface ties of the carbon surface (hydrophobic or hydrophilic groups, for
functional groups at the porous carbon surface. Commercial cells instance) could yield an important over-voltage to increase the elec-
rated at 3 V have recently been developed by careful removal of trochemical stability window of the electrolyte. Those approaches
oxygen-containing terminations from the porous carbon surface15,16. based on modification of the carbon/electrolyte interface are cer-
Interestingly, model materials such as graphene or carbon nanotubes tainly appealing in terms of opportunities for performance improve-
with surfaces free from any functional groups have shown substan- ment but are highly challenging. The surface modification must not
tial improvement in cell voltage. Recent work from Kyotani and alter or hinder the access of the ions in the electrolyte to the porous
co-workers, using mesoporous carbon sheets prepared by chemical carbon structure. The modified carbon surface must be ionically
vapour deposition from CH4 on nanosized sacrificial templates made conductive to allow ion transfer from the electrolyte into the carbon
of alumina particles, demonstrated cell voltages up to 4.4 V (ref. 60). pores but electrically insulating to limit solvent oxidation and/or
The prepared mesoporous structure with curved graphene walls reduction at the carbon surface. The concept of the passive layer was
was free from reactive edge-sites and demonstrated a large voltage recently applied to Li-metal67 and Mg-metal68 anodes for Li-metal
window. These studies with model materials provide the scientific and Mg-metal batteries, respectively. The successful design of such
foundations for the design of ultra-stable porous carbon structures an artificial SEI-like layer (rather than the one formed spontane-
with improved performance. On the other hand, Antonietti’s group ously for graphite anodes) on the surface of porous carbon would
reported heteroatom‐doped carbons known as ‘noble carbons’, greatly boost the EDLC performance. Mastering the electrode/
obtained by carbonization of organic compounds61. These carbons electrolyte interface is also a key challenge for Li-ion batteries; this
show improved electrochemical stability versus oxidation, and could is particularly true with the current trend to develop all solid-state
lead to substantial energy gain, as long as their electrical conductivity Li-ion batteries based on inorganic ceramic electrolytes. There is
remains high enough for them to be used as active material. then a common interest in developing and sharing techniques for
The design of a high-voltage electrolyte (>5 V stability window) characterizing the electrode/electrolyte interface, as well as underly-
matching the requirements for energy-storage applications (high ing basic science12,27,31,32,34,52.
conductivity, large operation temperature range, low toxicity) is a In addition to increasing the voltage window, capacitance and
major challenge for battery and EC communities, and there is still energy density, substantial expansion of the temperature window
no certainty that it will become industrial reality. However, inter- has been achieved by using non-freezing mixed electrolytes (for
esting approaches to tailoring electrolytes for EC applications have example, eutectic mixtures of ionic liquids or mixed solvents) and
recently been proposed23. For instance, Balducci and co-workers minimizing diffusional limitations via carbon material design. As
proposed a combinatorial chemistry approach to select new solvent/ a result, ECs operating at temperatures down to −75 °C for space
salt couples by rational computational screening62,63. This identified applications69 or across the −50 °C to 100 °C range have been
the cyano-esters family as a new class of possible solvent, and espe- reported70,71.
cially the 3-cyanopropionic acid methyl ester. In combination with There is another growing trend in using high-power EDLCs to
the conventional (C2H5)4N+,BF4– salt, a cell operating voltage up to challenge the conventional aluminium electrolytic capacitors in
3.2 V was achieved, which confirms the potential of the method energy harvesting, back-up and even alternating-current (a.c.) filter-
and justifies further use of machine learning and artificial intelli- ing applications7,13,72. Whereas conventional porous-carbon-based
gence, owing to the large variety of combinations possible23. A new (activated carbon) EDLCs become resistive at frequencies above 1
concept, proposed by Rochefort and co-workers64 and by Fontaine Hz, non-porous carbons such as carbon black72 or vertically grown
and co-workers65, has emerged, in which the charge storage could graphene73 can operate at much higher frequencies (Fig. 3d). As
also be achieved in the electrolyte (Fig. 3b). It relies on the use of the capacitance of aluminium electrolytic capacitors is limited
biredox ionic liquids as salt in an electrolyte to achieve bulk-like to microfarads or millifarads, replacing them with EDLCs with a

Nature Materials | VOL 19 | November 2020 | 1151–1163 | [Link]/naturematerials 1155

Review Article NaTuRE MaTERIalS

Surface pseudocapacitance Intercalation pseudocapacitance

a c e
–1 –1
2 mV s 2 mV s 2 mV s–1
2

0.8 0.4
1

Current (A g–1)

Current (A g–1)
Current (A g–1)

0 0.0 0.0

–1
–0.8 –0.4

–2

0.0 0.2 0.4 0.6 0.8 1.0 1.5 2.0 2.5 3.0 –2.5 –2.0 –1.5 –1.0 –0.5 0.0
Potential (V vs NHE) Potential (V vs Li+/Li) Potential (V vs Ag)

Surface process Bulk process Bulk process

b d f
Hydrous grain boundary Nb2O5 in 1 M LiClO4 in PC Ti3C2 in 1 M LiTFSI in PC

Li+ Li+ (PC)0

10.7 Å
RuO2 nanocluster

a O Li O Li Ti C

Fig. 4 | Conceptual presentation of redox capacitance. a, CV of RuO2 electrode. b, The fast surface reactions at RuO2 particles explain the presence of
bumps on a capacitive-type CV, leading to a surface pseudocapacitance charge-storage mechanism. c, Electrochemical signature (CV) of T-Nb2O5 in
non-aqueous LiClO4 in propylene carbonate (PC) electrolyte, showing intercalation pseudocapacitance due to fast, non-diffusion-limited Li+ intercalation
into the bulk of the material. d, Structure of T-Nb2O5 with the intercalated Li ions. e, CV of a Ti3C2 MXene electrode in non-aqueous 1 M LiTFSI in PC
electrolyte, showing fast Li+ intercalation into the gaps between Ti3C2 layers. f, Structure of Ti3C2 MXene showing intercalated and desolvated Li ions
(noted as Li+(PC)0) between the Ti3C2 MXene layers. Adapted with permission from: a,b, ref. 12, SNL; e,f, ref. 86, SNL.

capacitance orders of magnitude higher is attractive. Higher volu- Pseudocapacitive and high-charge/discharge-rate
metric capacitance, extreme cyclability (over a million cycles at materials based on redox capacitance
a very high rate) and scalable device manufacturing must still be Unlike in EDLCs, the charge-storage mechanism in pseudocapaci-
achieved, and the phase angle should be closer to 90° across a wider tive materials is not purely electrostatic. It involves faradic processes
range of frequencies. Large-scale flexible supercapacitors that can — redox reactions1,5,20,74 that increase the capacitance and thus the
provide both high volumetric and high areal capacitance, as well as energy density of supercapacitors (Fig. 3a). The surface redox pseu-
a high-frequency (>100 Hz) capacitive response with a phase angle docapacitive mechanism was proposed in the 1970s by Conway,
approaching −90°, are expected to resolve the current impediment who explained that when reversible redox reactions occur at or near
for high-power applications. the surface of an appropriate electrode material, the electrochemi-
In summary, EDLCs are a class of ECs that store the charge elec- cal features are those of an EDLC but with considerably greater
trostatically without involving any redox reactions, providing high charge storage because of the redox reactions20. Materials such as
delivery/uptake power capability and ultra-long life (>106 cycles). MnO2 and RuO2 (Fig. 4a) were among the first reported examples
These key features make them suitable in a broad range of applica- of pseudocapacitive materials20,75. In these materials, the change in
tions including renewable energy and grid storage (power smooth- the redox state of the metal is balanced by the surface insertion/
ing, phase regulation, uninterruptible power supplies), as well as deinsertion of a cation from the electrolyte. The redox reactions
transportation (stop and start function, recovery of braking energy, are confined to the surface — or the near-surface volume — of the
power steering), where they offer specific added value compared material, differentiating it from the bulk faradic storage in battery
with batteries. Although electrolyte confinement in carbon nano- materials20,74,76,77. This results in a rectangular-shaped cyclic voltam-
pores came with substantial improvement in their capacitance and metry (CV) curve (Fig. 4a), similar to that of EDLCs.
energy density, alternative routes must be found to further improve Different from faradic reactions occuring in the bulk of (battery)
the performance. Efforts should be directed towards the design of materials, the pseudocapacitive mechanism (or redox capacitive
high-surface-area carbon structures with improved electrochemical mechanism) has the following characteristic features: (i) a charge
stability. Also, modification of the porous carbon/electrolyte inter- that changes with the potential; (ii) the absence of solid-state diffu-
face by preparing passive layers or designing specific properties sion limitation, owing to the surface or near-surface storage mecha-
(hydrophobicity or hydrophilicity) may offer more opportunities nism (surface pseudocapacitance; see Fig. 4b); and (iii) the absence
to enlarge the operation voltage range of EDLC devices — but this of phase change in the material during electrochemical polariza-
must be achieved without sacrificing power capability or cycle life. tion, which would result in sharp redox peaks in the CVs5,74,76–78.

1156 Nature Materials | VOL 19 | November 2020 | 1151–1163 | [Link]/naturematerials

NaTuRE MaTERIalS Review Article
These criteria result in symmetrical CV curves along the poten- Still, according to the results outlined above, the narrow voltage
tial axis, where the peaks, if present, are broad and exhibit a small window (typically about 1 V) limits the energy density (equation (2))
peak-to-peak voltage separation. for pseudocapacitive materials operating in aqueous electrolytes; this
The faradic nature of the charge-storage mechanism makes is an important drawback that hampers the industrial application of
the use of charge or capacity (in coulombs or amp hours) a suit- the technology. Possible remedies include the use of water-in-salt
able metric to characterize pseudocapacitive materials. Although electrolytes with a wider voltage window (2 V or more), as well
capacitance (in farads) may be given for comparison purposes, as asymmetric device design with, for example, a MXene negative
it is meaningless to report capacitance values from distinctly electrode combined with an oxide or conducting polymer cathode,
non-rectangular-shaped CVs. Electrochemical impedance spec- extending the voltage window above 1.5 V. Both approaches can be
troscopy plots in the Nyquist representation shows a sharp, almost combined for maximum performance improvement.
vertical, increase of the imaginary part of the impedance in the In 2013, Dunn’s group reported on the synthesis of nanocrys-
low-frequency region as the result of the absence of solid-state dif- talline orthorhombic-phase Nb2O5 particles (T-Nb2O5)90, with con-
fusion5,76–78. Several electrochemical methods and techniques have trolled structure and texture, capable of working in a non-aqueous
been developed to help in the characterization of pseudocapacitive electrolyte. The Li-ion intercalation capacity was mostly inde-
materials, such as the identification of the surface versus bulk con- pendent of rate, and redox peaks with small voltage offsets were
tributions to the total current79, as well as the use of electrochemi- observed even at high rates. Moreover, the Li ion intercalation
cal impedance spectroscopy at various potentials80 or multiple step reaction was found to occur in the bulk of the particles, thanks
potential techniques81,82. These tools must be used by the scientific to facile 2D lithium-ion diffusion pathways existing in the ortho-
community to correctly characterize materials and report their rhombic crystal structure74,90,91. Together with the absence of any
performance, avoiding misleading reports, such as capacitance val- phase transformation, these agreed well with a Li-ion intercala-
ues exceeding 1,000 F g–1 for faradic nickel oxide or oxy-hydroxide tion pseudocapacitive reaction mechanism74 such as defined by
in alkaline electrolyte5,19,76–78. Interestingly, using cobalt oxide as a Conway92 (Fig. 4c), which needs a network that offers 2D transport
model material, a recent report explains its pseudocapacitive elec- pathways and little structural change on intercalation. Mirror-like
trochemical signature (similar to Fig. 4a and to the one observed CVs in non-aqueous electrolytes, with broad redox peaks on top
for MnO2 in neutral electrolyte) by a transition from an insulating of a capacitive envelope, are characteristics of such materials.
to an electrically conducting phase initiated by redox reaction83. In Here, the change of redox state of Nb during Li ion intercalation,
this material, the creation of a potential gradient would result in a together with high capacity (150 mAh g–1) values achieved, also
double-layer capacitance-like storage mechanism, but more experi- excludes the possibility of a pure double-layer mechanism con-
mental work (using in operando structural analysis, for instance) is tribution. Interestingly, this material combines the key features of
required to support such a claim. faradic (battery) and pseudocapacitive materials, namely (i) high
Although pseudocapacitive materials hold promise for improv- capacity thanks to Li ion intercalation into the bulk of the particles
ing the energy density of EC, they have so far not seen widespread and (ii) high power because of the high rate, non-diffusion-limited
use in commercial products because of (i) the high costs of RuO2, reaction in the facile 2D Li-ion diffusion pathways of the
(ii) the limited capacitance and low electrical conductivity of MnO2 structure (Fig. 4d).
and other metal oxides, (iii) the short lifetime of conducting poly- The intercalation pseudocapacitive behaviour of Ti3C2 MXene
mers and (iv) the restriction of the pseudocapacitive behaviour to electrodes in Li-ion-containing non-aqueous electrolyte has recently
aqueous electrolytes with a narrow voltage window. However, the been reported in the literature (Fig. 4e)86. Capacity up to 130 mAh g–1
discoveries and advances of the past few years on metal oxides and was achieved in 1M Li-TFSI in propylene carbonate (PC) electro-
carbides could change the game. In 2011, a family of 2D early tran- lyte, within a 2.4-V voltage window (from ∼2.8 V down to 0.4 V
sition metal carbides and carbonitrides (MXenes) was produced by versus Li). The high capacitance in PC-based electrolytes was
selective etching of the A element from MAX phases84. MXenes are ascribed to the full desolvation of Li ions when intercalated between
usually referred to as Mn+1XnTx, where M is an early transition metal the MXene layers, resulting in improved Li ion intercalation com-
and X is carbon and/or nitrogen, T represents surface termination pared with other solvents86. Similar to Nb2O5, the intercalation is
(=O, –OH, –Cl and/or –F), n = 1, 2, 3 or 4, and x is the number not limited by solid-state diffusion in MXene, thanks to the com-
of terminating groups per formula unit85. Ti3C2Tx MXene has been bination of the 2D structure offering 2D paths for Li ion interca-
intensively studied for energy-storage applications85,86. CV of this lation and the metallic conductivity of Ti3C2 which results in high
material in sulfuric acid electrolyte shows broadened peaks with power performance. Interestingly, Li-ion desolvation was found to
little separation in peak position on charge/discharge, on top of a be responsible for the high performance of MXene in non-aqueous
capacitive rectangular box (Fig. 4a). In this material, the change of electrolyte (Fig. 4f), similar to what was reported earlier for porous
the valence state of the surface layers of Ti atoms (Ti+II/Ti+III) is bal- carbon1,29. Finally, in very recent work93, F-free MXene materials
anced by the reversible intercalation reaction of protons85. Needless prepared by a molten salt synthesis route achieved unique elec-
to say, the metal-like conductivity of Ti3C2Tx MXene excludes the trochemical performance in terms of energy and power density
possibility of a pure double-layer-like mechanism coming from the in Li-ion battery electrolyte. The control of the surface termina-
change in the electrical conductivity during polarization83. High tion groups and Li-ion desolvation allowed the use of this MXene
gravimetric capacitance (>500 F g–1)85,87 and volumetric capacitance as high-rate negative electrode material for electrochemical energy
(1,500 F cm–3), with an areal capacitance of up to 4 F cm–2 (ref. 88), storage, with capacities reaching 200 mAh g–1.
were obtained in aqueous sulfuric acid electrolyte. Moreover, 50% There are currently few commercial devices operating with pseu-
of the capacitance could be delivered (or stored) during 1 s dis- docapacitive materials. These materials still have a limited cycle life
charge (charge). These impressive performance is the result of the and power performance compared with EDLCs, without offering
combination of 2D structure that offers redox-active surface highly a cost advantage. Although the recently developed MXene mate-
accessible to electrolyte ions, and short transport paths, together rials appear promising for high rate pseudocapacitive electrodes,
with the high electrical conductivity of the MXene backbone, which improvements are needed in the irreversible capacity loss due to
makes these materials unique85. As MXene electrodes usually do not SEI formation, the control of the nature and content of surface ter-
require the use of metallic current collectors (at least in prismatic mination groups, and the synthesis routes. However, research over
and pouch cells), the energy density of the devices can be given an the past few years has resulted in great advances in both the basic
extra boost85,89. science and practical performance. Further efforts will be needed

Nature Materials | VOL 19 | November 2020 | 1151–1163 | [Link]/naturematerials 1157

Review Article NaTuRE MaTERIalS

to move pseudocapacitive oxides, carbides, nitrides and other mate- their higher discharge potential. As positive electrodes, high-rate,
rials to the market. Li-containing materials are also good candidates since they can sup-
Beyond offering promising power and energy performance, ply Li ions to the negative electrode, in case SEI is formed at the first
these examples illustrate changes of paradigm in the design of cycle. Defective LFP is a possible candidate, but such an approach
redox-based capacitive materials for high-energy ECs. The fron- could be extended to lamellar transition metal oxides operating at
tier between high-discharge-rate battery materials and pseudoca- higher potentials. Those redox-based capacitive materials could
pacitive materials, which combine the best of the battery and EDLC also benefit from the development of solvent-in-salt electrolytes,
worlds, becomes increasingly blurred10,12. Moreover, it is possible to which can achieve improved operating voltage window in combi-
transform the electrochemical signature of a faradic battery material nation with Li-ion battery electrodes, thanks to the decrease of the
into a pseudocapacitive one by decreasing the particle size and/or content of free solvent molecules100.
introducing defects in the structure, as has been achieved with
LiFePO4 (refs. 94,95) LiCoO2 (ref. 96) and other cathode materials74. Hybrid capacitor design
These materials are called extrinsic pseudocapacitive materials5,74 There is plenty of basic science to share and develop between the
to distinguish them from conventional intrinsic pseudocapacitive battery and the capacitive storage communities, and a common
materials (Fig. 4). In these materials, high defect concentrations playground is the field of hybrid electrochemical capacitors (HECs).
(breaking the crystal structure) and large surface area, which can be Unlike EDLCs or pseudocapacitive ECs, HECs combine a faradic
achieved by adapting the synthesis route or nanosizing the particles, battery electrode with a capacitive or pseudocapacitive electrode.
accelerate the Li solid-state diffusion74,95,96. A striking example is In these systems, the battery electrode brings the energy while the
LiFePO4 (LFP), a well-known positive electrode material for Li-ion power comes from the capacitive one. Most of the popular HECs
batteries. Li-ion intercalation into LFP is achieved via a two-phase currently commercialized combine a Li-ion-battery graphite anode
reaction mechanism, resulting in a constant potential plateau at with a porous-carbon capacitive electrode. These are known as
3.5 V versus Li during constant-current charge/discharge tests. The Li-ion capacitors (LIC)101, such as shown in Fig. 5a,b. The high
modification of the LFP structure with Fe3+ defects, for instance95, operating cell voltage (3.8–2.2 V) of LICs comes from the use of the
yields a different electrochemical behaviour. Constant-current graphite anode and the SEI layer formation.
charge/discharge tests of amorphous LFP containing Fe3+ defects High energy density (beyond 20 Wh kg–1) can be achieved using
show sloping potential profile change with time, like observed for LIC technology, and the growing number of companies develop-
intercalation pseudocapacitance mechanism. The change in elec- ing LICs demonstrate the interest in this technology, applications of
trochemical behaviour of the defective LFP was ascribed to faster which are similar to those of ECs. However, LICs are not expected
Li-ion diffusion in the presence of defects94,95. The past few years to replace ECs, since they face several challenges101. The negative
have seen a growing number of papers dealing with the synthesis Li-ion graphite anode needs to be highly over-capacitive to accept
and characterization of high-rate materials with pseudocapacitive high recharge rates (beyond the capacity of the limiting positive
or pseudocapacitive-like signature for both negative and positive electrode), limiting the power capability to a few kilowatts per kilo-
electrodes7,10,97. This looks like a promising way to improve the gram, which is a major difference from EDLCs. Also, the negative
energy density of ECs and exceed the power density of batteries. electrode must be prelithiated, which is achieved by introducing
In summary, pseudocapacitive materials have the potential to a sacrificial Li-metal electrode in the cell, short-circuited with the
improve the energy density of ECs considerably. However, conven- graphite anode before use. Although some smart solutions have
tional pseudocapacitive materials (such as RuO2, MnO2 or Fe3O4) been proposed that use sacrificial Li+ extraction at the positive elec-
are only active in aqueous electrolytes, limiting their energy density trode during the first cycle to balance the electrode capacities102, this
and practical applications. The past decade has seen rapid devel- limits to some extent the gravimetric and volumetric performance
opment of research on these materials. Designing 2D materials or of the cells. Then, the use of carbonate-based electrolytes restrains
metal oxides with crystal structure offering 2D Li-ion diffusion the operating temperature range to that of Li-ion batteries, in con-
pathways for fast Li intercalation in non-aqueous electrolyte has trast to EDLCs, which can use carbonate-free electrolytes101. Finally,
led to evidence of a fast intercalation mechanism in the bulk of the the cycling life — and specially the impact of high charge/discharge
material (‘intercalation pseudocapacitance’). Also, the modifica- rates on the negative electrode stability upon cycling — needs to be
tion of faradic battery materials by creating defects in the structure improved, despite hundreds of thousands of cycles being claimed
and incorporating structural water, as proposed by Augustyn and by manufacturers101. However, thanks to the recent development
co-workers98,99, yields pseudocapacitive signatures and offers oppor- of high-rate carbon-based materials for Li-ion-battery anodes3 and
tunities for designing high-energy and high-power-density pseudo- discovery or design of better pseudocapacitive materials74, there
capacitive ECs. is a chance of improving the LICs’ energy density beyond 20 Wh
At the same time, as we have said, the boundary between kg–1 using those nanostructured high-rate electrodes, while keeping
high-rate battery materials and high-energy pseudocapacitive high power capability close to that of EDLC (Fig. 5c,d). This would
materials is becoming increasingly blurred. This holds promise for result in a major change in possible applications of ECs. In such
great improvement in the energy density while maintaining a high designs, the presence of a Li-containing positive electrode, such
power capability and a long cycle life (>1,000,000 cycles), moving to as defective LFP, minimizes the need for an extra lithium source
the area marked with the star in Fig. 1. Some general guidelines can to form the SEI at the negative electrode. One concern might still
be given for extending the energy density of high-rate pseudoca- be the cycling life of pseudocapacitive ECs compared with EDLCs,
pacitive materials, most of them being shared with the battery com- since the introduction of a redox-based charge-storage mechanism
munity. First, a negative (positive) electrode must deliver/store its may result in chemical modification or volume change of the mate-
charge at the lowest (highest) average potential, to end up with the rials. This should be a target of future studies.
highest cell voltage and energy density. Next, the presence of a SEI The HECs (Li-ion capacitors, for instance) that have recently
layer and the risk of Li plating at high recharge rate are important appeared in the market currently offer the highest performance
drawbacks when using negative electrodes operating at low poten- in terms of energy density (>20 Wh kg–1). However, the use of a
tials in Li-containing electrolyte. This is why new materials based on Li-ion-battery negative electrode in LIC systems limits their power
intercalation capacitance (metal oxides, such as Nb2O5, LiTiNb2O7 capability compared with EDLCs and makes the cycle life dur-
or Li3VO4, and MXenes) operating at high rate with no — or lim- ing high-rate cycles uncertain. Also, HECs will face competition
ited — risk of Li-metal plating are attractive candidates, despite from high-power batteries such as the high-rate batteries, based on

1158 Nature Materials | VOL 19 | November 2020 | 1151–1163 | [Link]/naturematerials

NaTuRE MaTERIalS Review Article
a c

Porous carbon Defective LiFePO4

4.5 4
Graphite (Li-ion) T-Nb2O5
T3C2Tx MXene
3.5
Potential (V vs Li+/Li)

Potential (V vs Li+/Li)
3

~3 V
2.5 ~3 V ~4.2 V –3.7 V
2

1.5
1

0.5

0 0
0 50 370 0 50 100 400
–1
Electrode capacity (mAh g ) Electrode capacity (mAh g–1)

b d Defective LFP
Graphite Porous carbon Li0.94Fe1.05P0.92O4
T-Nb2O5
Charge
PF6– Charge
PF6–
PF6– PF6–
PF6– Li+ Li+ Li+
Li+ PF –

Current collector
Current collector

6
PF6– PF6– Li+
Li+ Li+
Li+
PF6–
Li+ Li+
Li+
PF6– PF6– Li +
Li +
Li+
Discharge
PF6– a
PF6–
PF6– PF6– b
PF6– Li Fe P c

Fig. 5 | Lithium-ion capacitors. a,b, Concept of a Li-ion capacitor (LIC), which combines a negative graphite electrode, as used in a Li-ion battery,
with a positive porous carbon EDLC electrode. The cell voltage is increased relative to an EDLC capacitor using symmetrical porous carbon electrodes.
c,d, Concepts to show that high-power batteries and high-energy ECs using pseudocapacitive materials share similar features. Here, a combination of
pseudocapacitive Nb2O5 or MXene negative electrode with a defective LFP positive electrode with sloping profile suggests the possibility of building
a pseudocapacitive device.

lithium titanium oxide, developed by Toshiba103, or emerging on flexible substrates, printed on various surfaces or incorporated
Na-ion batteries2, which show similar power performance with a into textiles has been growing quickly to address these needs105.
higher energy density. Development of micro-devices is an emerging topic, but the field is
quickly expanding, and there are high expectations in the design of
Micro-devices for IoT and wearable electronics high-energy and high-power micro-devices to support the develop-
The development of connected and smart environments, more com- ment of the IoT106.
monly termed IoT104, is a major emerging application for microscale MSCs can be prepared on various substrates using many tech-
energy-storage devices (Fig. 6a). IoT relies on the use of connected niques; the process chosen usually depends on the substrate107. MSC
sensor networks that collect and transfer the data to a central gate- devices are generally prepared in one of two designs, with stacked
way. It covers a broad range of applications: health (individual man- or planar interdigitated electrodes (Fig. 6b). In planar interdigi-
agement and continuous monitoring of patients), environmental tated electrodes, the areal capacitance per electrode polarity is 4
(fire prevention, air and water quality measurements) or industrial times the total areal capacitance of the device (footprint area)107.
and infrastructure monitoring (buildings, bridges, roads and even Although MSCs share with their EC counterparts the challenge of
museums), drug delivery (in vivo application), transportation, high energy density, gravimetric performance has little meaning in
wearable personal electronics and various radiofrequency identifi- this case. Reliable performance metrics are areal energy (Wh cm–2)
cation systems104. Those applications require a very large number of and power (W cm–2) densities, or volumetric energy (Wh cm–3) and
compact, efficient, miniaturized energy-storage devices for energy power (W cm–3) densities — calculated from areal (F cm–2) and vol-
delivery or harvesting with high power capabilities. Although umetric (F cm–3) capacitance — to take into account the electrode
micro-batteries can ensure a low continuous energy delivery, they thickness (Fig. 6c)105. In addition to these metrics, the thickness
fail to deliver the peak power needed for data acquisition and trans- of the substrate should be indicated as well as the electrode thick-
fer. Moreover, their charge-storage capability per weight or volume ness or/and the total thickness of the device. A further challenge
of the device decreases with decreasing size, because of the increas- in MSCs is the preparation of mechanically stable electrodes with
ing ratio of passive (packaging and so on) to active components. strong adhesion to the substrate and ability to withstand multiple
Starting from about 2010, research on the development of deformation cycles, which are required in the case of flexible and
micro-supercapacitors (MSCs) integrated on Si wafers, produced wearable electronics.

Nature Materials | VOL 19 | November 2020 | 1151–1163 | [Link]/naturematerials 1159

Review Article NaTuRE MaTERIalS

a b
L

Vol. = LWe (cm3)

Area = LW (cm2)
e W

L Electrodes

Collectors
W

e Substrate (Si,
polymer, textile…)

c
Area 100

100

10
Energy density (μWh cm–2)

Energy density (mWh cm–3)

10

Volume 1
1
Graphene / MnO2, aqueous
CDC onto Si wafer, in AN
0.1
0.1 Activated carbon, NEt4BF4, in AN
Graphene / ionogel
rGO in KCl
0.01 Graphene CNT in Na2SO4 0.01
0.01 0.1 1 10 100 1,000 Carbon onions, in PC 0.01 0.1 1 10 100 1,000
Power density (mW cm–2) Power density (W cm–3)

Fig. 6 | Micro-supercapacitors for current and future technologies. a, Sketch of a connected network of sensors used for the IoT. b, Calculations of areal
and volumetric capacitance of micro-devices with planar (top) or interdigitated (bottom) electrode configuration. c, Example of Ragone plots showing the
performance of MSC devices normalized per surface area (left) and volume (right). AN, acetonitrile; CNT, carbon nanotube; rGO, reduced graphene oxide.
Adapted with permission from ref. 106, AAAS.

The direct integration of MSCs on Si wafers is the most challeng- deposition of pseudocapacitive materials is more complex and lim-
ing concept, since fabrication techniques compatible with the semi- ited to a few compounds, resulting in areal capacitance and energy
conductor industry have to be used. Dry processing (atomic layer density values below the expectations for practical applications105.
deposition or magnetron sputtering) is preferred to wet routes (such Electrochemical deposition techniques can also be used to prepare
as electrophoretic or electrochemical deposition, screen printing or on-chip pseudocapacitive MSCs, with the same limitations in terms
serigraphy), but the limited electrode thickness achieved by these of the variety of materials and energy density.
techniques results in low areal capacitance107. An important step A massive trend currently observed in the field of MSCs is
forward was achieved by preparing on-chip MSCs by chlorination the preparation of flexible devices aiming to address the need for
of TiC films (∼5 μm thick) deposited on Si wafer by magnetron smart textiles and for paper-based and flexible power electronics110.
sputtering106. The partial chlorination of TiC led to the formation of A major step has been the development of the laser-writing process
mechanically stable porous carbon electrodes strongly adhering to technique, thanks to pioneering work from Kaner and co-workers,
the Si surface. Electrode areal capacitance beyond 400 mF cm–2 was who reported the fabrication of MSCs by direct reduction of
achieved in sulfuric acid electrolyte, making these micro-devices graphite oxide film under a laser beam in 2012, drawing porous
among the best reported for carbon-based MSCs (Fig. 6c). A limi- ‘laser-scribed graphene’ interdigitated devices111. Despite limited
tation of the MSC technology based on nanoporous carbon elec- capacitance achieved (about 5 mF cm–2) because of the thin elec-
trodes is that it is barely compatible with the deposition of inorganic trodes, this was important in showing that a simple, easy to imple-
solid-state electrolytes owing to the electrode porosity, although ment and scalable technique was efficient in preparation of flexible
solid-state-like electrolytes (gels or ionogels) can be used108,109. electrodes and micro-devices. Tour’s group also showed the possibil-
Further improvement of the areal capacitance can be achieved ity of direct conversion of polymer substrates into carbon under the
by developing a 3D Si surface and preparing structures with high laser beam112. Moving further, Kaner and co-workers functionalized
area enhancement factor (that is, the projected area per footprint laser-scribed graphene electrodes by electrochemical deposition of
area)105. This offers interesting routes for preparing on-chip MSCs pseudocapacitive MnO2 (ref. 113), resulting in a notable improve-
with a high areal energy density, using carbide-derived carbon elec- ment in areal capacitance and energy density. The next challenges
trodes prepared by the chlorination of TiC or other carbide films for this promising laser-scribing technology will be to implement
deposited by magnetron sputtering. Unfortunately, the vacuum newly developed pseudocapacitive materials, such as Nb2O5, MoO3

1160 Nature Materials | VOL 19 | November 2020 | 1151–1163 | [Link]/naturematerials

NaTuRE MaTERIalS Review Article
or MXene74, operating in non-aqueous electrolytes to boost the biosafety, when in direct contact with skin. This puts certain restric-
energy performance. The technique will need to be adapted to those tions on choice of active materials and electrolytes, which should
materials, since oxides require metal current collectors and MXenes be gels or polymers. Also, the capacitance and other characteristics
may oxidize under the laser beam. should be normalized by length of fibres and yarns (F cm–1) or area
Besides laser writing, electrophoretic114, electrochemical115, for textiles (F cm–2), as gravimetric data for the active material itself
stamping, spray coating and printing116 techniques have been used do not really help to understand how much charge can be stored in
to prepare electrodes made of carbon and pseudocapacitive materi- the garment117. This is a quickly maturing research area, and rapid
als on flexible substrates. Conventional and 3D printing on a variety development is expected in the next few years.
of surfaces are experiencing fast expansion116. The areal capacitance
achieved ranges from <1 mF cm–2 to hundreds of mF cm–2, depend- Summary and outlook
ing on the material used and the electrode thickness116. Although This Review aims at providing general key messages. The first is
the field is still young, some trends can be extracted from the litera- that EDLCs have certain features (no redox processes, long cycle
ture survey. First, the mechanical properties and the quality of the life, high power and a wide temperature range of operation) that
electrode/current-collector interface (adherence of the film) have a make them an important technology for energy-storage applica-
strong impact on the performance and the cycle life of the devices. tions. The energy density of EDLCs can be increased by increasing
Thus, deposition techniques leading to improved impedance of the the cell voltage via interface modification or development of chemi-
contact — electrochemical plating, for instance — are preferred. cally stable, high-surface-area carbon structures free from defects
The preparation of carbon nanotube, graphene or reduced gra- and functional groups.
phene oxide electrodes leads to a moderate areal capacitance (less Next, the recent discoveries of new pseudocapacitive materials
than about 20 mF cm–2) owing to limited electrode thickness; elec- and of the bulk intercalation pseudocapacitive reaction mecha-
trodeposition of pseudocapacitive materials (such as MnO2, RuO2 nism occurring in non-aqueous electrolytes are examples of key
or conducting polymers) in an additional synthesis step allows for advances in the field of capacitive storage, since high-energy and
great improvement of the areal capacitance to 200–300 mF cm–2. high-power electrodes and devices can now be produced. Here, the
However, the use of aqueous electrolytes comes with a narrow cell high-power challenge for batteries meets the high-energy quest for
voltage (less than ∼1.6 V) and energy density limitation, as dis- ECs, and the combination can make a huge impact on the entire
cussed earlier. The situation is different for carbon-based MSCs energy-storage field.
which can operate in non-aqueous electrolytes, with the highest The IoT is currently pushing the development of MSC technol-
capacitance being reported so far for porous carbons with a high ogy, and several techniques and processes have been developed
surface area. The use of highly conductive graphene and MXene to prepare MSCs with a focus on flexible devices. The laser-based
eliminates the need for metal current collectors, greatly simplifying techniques, ink-jet printing, stamping, 3D printing and other pro-
the manufacturing process. Finally, the use of increasingly popular cesses used are very different from conventional tape electrode
solid-state-like (gel) electrolytes may resolve the issue of electrolyte manufacturing. To meet the needs of flexible and wearable technol-
leakage and eliminate the need for hermetically sealed packaging116. ogy, new materials and new device designs should be implemented.
However, MSCs will have to address several challenges in the near Active materials with very high electronic conductivity are required
future to meet industry expectations. Solid-state or gel high-voltage to eliminate current collectors. The use of conductive 1D and 2D
electrolytes are needed to replace the aqueous polyvinyl-alcohol-based materials could eliminate the need for resistive polymer binders.
gel electrolytes with a narrow voltage window (about 1 V in acids). Non-leaking gel and polymer electrolytes are needed to eliminate
Developing inorganic ceramic solid-state electrolyte will be chal- special packaging and enable textile supercapacitors. A coating
lenging owing to the high surface area of carbon materials and the technology should be developed for separation of electrodes (gel
general issues of mastering the electrolyte/active material interfaces. and polymer electrolytes can play the role of separators) and encap-
Alternatively, ionogel electrolytes (that is, ionic liquid phase entrapped sulation of devices (for example, polymer spray-coating or vapour
inside an inorganic (SiO2) matrix) have been successfully used in ECs deposition). Conventional sandwich device designs will be replaced
in combination with high-surface-area materials, offering voltage by interdigitated electrodes and fibres/yarns.
windows up to 3 V (refs. 108,109). Also, transferring the recent advances
in the pseudocapacitive field, such as the use of bulk intercalation Received: 27 December 2019; Accepted: 18 June 2020;
pseudocapacitance mechanism (Nb2O5, MoO3 or MXene), operating Published online: 3 August 2020
in non-aqueous electrolyte into real micro-devices would result in
a major enhancement of both the areal capacitance and the energy References
1. Simon, P. & Gogotsi, Y. Materials for electrochemical capacitors. Nat. Mater.
density86,90,97. The design of a suitable 3D electrode structure would 7, 845–854 (2008).
further boost the areal power and energy performance113,115. Finally, 2. Delmas, C. Sodium and sodium-ion batteries: 50 years of research.
the combination of the pseudocapacitive materials with suitable elec- Adv. Energy Mater. 8, 1703137 (2018).
trodes using 1D, 2D and/or 3D carbon structures or even high-rate 3. Li, M., Lu, J., Chen, Z. & Amine, K. 30 years of lithium-ion batteries.
battery materials would result in major improvements, making MSC Adv. Mater. 30, 1800561 (2018).
4. Palacin, M. R. & de Guibert, A. Why do batteries fail? Science 351,
devices competitive with micro-batteries for applications in embed- 1253292 (2016).
ded sensors for battery monitoring, truly wearable systems integrated 5. Augustyn, V., Simon, P. & Dunn, B. Pseudocapacitive oxide materials
into clothes, or energy harvesting. for high-rate electrochemical energy storage. Energy Environ. Sci. 7,
As a separate class of emerging flexible devices, we need to 1597–1614 (2014).
consider textile supercapacitors. This area emerged less than ten 6. Liu, C. F., Liu, Y. C., Yi, T. Y. & Hu, C. C. Carbon materials for high-voltage
supercapacitors. Carbon 145, 529–548 (2019).
years ago117 and keeps growing because of the interest in ‘wearable 7. Wang, F. X. et al. Latest advances in supercapacitors: from new electrode
Internet’ and smart textiles110. Initially, carbon materials were used, materials to novel device designs. Chem. Soc. Rev. 46, 6816–6854 (2017).
but conductive polymers, metal oxides118 and MXenes119 recently 8. Wang, Y. G., Song, Y. F. & Xia, Y. Y. Electrochemical capacitors: mechanism,
entered the field. Storing the amount of energy required for power- materials, systems, characterization and applications. Chem. Soc. Rev. 45,
5925–5950 (2016).
ing devices, from LEDs to sensors and communication (antennas)
9. Wu, H. et al. Graphene based architectures for electrochemical capacitors.
on the limited surface of a garment (less than 1 m2) is challenging, Energy Storage Mater. 5, 8–32 (2016).
if only EDLC is used. Energy-storing textiles have some additional 10. Lin, Z. et al. Materials for supercapacitors: when Li-ion battery power is not
requirements to materials, such as stretchability, washability and enough. Mater. Today 21, 419–436 (2018).

Nature Materials | VOL 19 | November 2020 | 1151–1163 | [Link]/naturematerials 1161

Review Article NaTuRE MaTERIalS
11. Noori, A., El-Kady, M. F., Rahmanifar, M. S., Kaner, R. B. & Mousavi, M. F. crystal microbalance with dissipation monitoring. Acc. Chem. Res. 51,
Towards establishing standard performance metrics for batteries, 69–79 (2018).
supercapacitors and beyond. Chem. Soc. Rev. 48, 1272–1341 (2019). 44. Shpigel, N. et al. In situ hydrodynamic spectroscopy for structure
12. Salanne, M. et al. Efficient storage mechanisms for building better characterization of porous energy storage electrodes. Nat. Mater. 15,
supercapacitors. Nat. Energy 1, 16070 (2016). 570–575 (2016).
13. Miller, J. R., Outlaw, R. A. & Holloway, B. C. Graphene double-layer 45. Le, T. et al. Unveiling the ionic exchange mechanisms in vertically-oriented
capacitor with ac line-filtering performance. Science 329, 1637–1639 (2010). graphene nanosheet supercapacitor electrodes with electrochemical quartz
14. Horn, M., MacLeod, J., Liu, M., Webb, J. & Motta, N. Supercapacitors: a crystal microbalance and ac-electrogravimetry. Electrochem. Commun. 93,
new source of power for electric cars? Econ. Anal. Policy 61, 93–103 (2019). 5–9 (2018).
15. Ultracapacitor modules. Maxwell Technologies [Link] 46. Umeda, K., Kobayashi, K., Minato, T. & Yamada, H. Atomic-scale
products/ultracapacitors/modules (accessed May 2020). three-dimensional local solvation structures of ionic liquids. J. Phys. Chem.
16. LS Ultracapacitor [Link] (accessed May 2020). Lett. 11, 1343–1348 (2020).
17. Skeleton Technologies [Link] (accessed May 2020). 47. Tsai, W.-Y. et al. Hysteretic order–disorder transitions of ionic liquid double
18. Gogotsi, Y. & Simon, P. True performance metrics in electrochemical layer structure on graphite. Nano Energy 60, 886–893 (2019).
energy storage. Science 334, 917–918 (2011). 48. Ye, J. L. et al. Charge storage mechanisms of single-layer graphene in ionic
19. Mathis, T. S. et al. Energy storage data reporting in perspective — liquid. J. Am. Chem. Soc. 141, 16559–16563 (2019).
guidelines for interpreting the performance of electrochemical energy 49. Mao, X. et al. Self-assembled nanostructures in ionic liquids facilitate
storage systems. Adv. Energy Mater. 9, 1902007 (2019). charge storage at electrified interfaces. Nat. Mater. 18, 1350–1357 (2019).
20. Conway, B. Electrochemical Supercapacitors; Scientific Fundamentals and 50. Jackel, N., Simon, P., Gogotsi, Y. & Presser, V. Increase in capacitance by
Technological Applications (Springer, 1999). subnanometer pores in carbon. ACS Energy Lett. 1, 1262–1265 (2016).
21. Shao, H., Wu, Y.-C., Lin, Z., Taberna, P.-L. & Simon, P. Nanoporous carbon 51. Yan, R. Y., Antonietti, M. & Oschatz, M. Toward the experimental
for electrochemical capacitive energy storage. Chem. Soc. Rev. 49, understanding of the energy storage mechanism and ion dynamics in ionic
3005–3039 (2020). liquid based supercapacitors. Adv. Energy Mater. 8, 1800026 (2018).
22. Lin, T. Q. et al. Nitrogen-doped mesoporous carbon of extraordinary 52. Antonietti, M., Chen, X. D., Yan, R. Y. & Oschatz, M. Storing electricity as
capacitance for electrochemical energy storage. Science 350, chemical energy: beyond traditional electrochemistry and double-layer
1508–1513 (2015). compression. Energy Environ. Sci. 11, 3069–3074 (2018).
23. Beguin, F., Presser, V., Balducci, A. & Frackowiak, E. Carbons and 53. Futamura, R. et al. Partial breaking of the Coulombic ordering of ionic
electrolytes for advanced supercapacitors. Adv. Mater. 26, 2219–2251 (2014). liquids confined in carbon nanopores. Nat. Mater. 16, 1225–1232 (2017).
24. Zhong, C. et al. A review of electrolyte materials and compositions for 54. Bazant, M. Z., Storey, B. D. & Kornyshev, A. A. Double layer in ionic
electrochemical supercapacitors. Chem. Soc. Rev. 44, 7484–7539 (2015). liquids: overscreening versus crowding. Phys. Rev. Lett. 106, 046102 (2011).
25. Miller, J. R. & Butler, S. M. Electrical characteristics of large state-of-the-art 55. Liu, Y. M., Merlet, C. & Smit, B. Carbons with regular pore geometry yield
electrochemical capacitors. Electrochim. Acta 307, 564–572 (2019). fundamental insights into supercapacitor charge storage. ACS Cent. Sci. 5,
26. Xiong, G., Kundu, A. & Fisher, S. T. Thermal Effects in Supercapacitors 1813–1823 (2019).
(Springer, 2015). 56. Redondo, E. et al. Outstanding room-temperature capacitance of
27. Prehal, C. et al. Quantification of ion confinement and desolvation in biomass-derived microporous carbons in ionic liquid electrolyte.
nanoporous carbon supercapacitors with modelling and in situ X-ray Electrochem. Commun. 79, 5–8 (2017).
scattering. Nat. Energy 2, 16215 (2017). 57. Miller, J. R. & Burke, A. F. Electrochemical capacitors: challenges and
28. Forse, A. C. et al. Direct observation of ion dynamics in supercapacitor opportunities for real-world applications. Electrochem. Soc. Interface 8,
electrodes using in situ diffusion NMR spectroscopy. Nat. Energy 2, 53–57 (2008).
16216 (2017). 58. Pognon, G., Brousse, T., Demarconnay, L. & Belanger, D. Performance and
29. Chmiola, J. et al. Anomalous increase in carbon capacitance at pore sizes stability of electrochemical capacitor based on anthraquinone modified
less than 1 nanometer. Science 313, 1760–1763 (2006). activated carbon. J. Power Sources 196, 4117–4122 (2011).
30. Forse, A. C., Merlet, C., Griffin, J. M. & Grey, C. P. New perspectives on 59. Assresahegn, B. D., Brousse, T. & Belanger, D. Advances on the use of
the charging mechanisms of supercapacitors. J. Am. Chem. Soc. 138, diazonium chemistry for functionalization of materials used in energy
5731–5744 (2016). storage systems. Carbon 92, 362–381 (2015).
31. Griffin, J. M. et al. In situ NMR and electrochemical quartz crystal 60. Nomura, K., Nishihara, H., Kobayashi, N., Asada, T. & Kyotani, T. 4.4 V
microbalance techniques reveal the structure of the electrical double layer supercapacitors based on super-stable mesoporous carbon sheet made of
in supercapacitors. Nat. Mater. 14, 812–819 (2015). edge-free graphene walls. Energy Environ. Sci. 12, 1542–1549 (2019).
32. Deschamps, M. et al. Exploring electrolyte organization in supercapacitor 61. Antonietti, M. & Oschatz, M. The concept of ‘noble, heteroatom-doped
electrodes with solid-state NMR. Nat. Mater. 12, 351–358 (2013). carbons,’ their directed synthesis by electronic band control of
33. Boukhalfa, S. et al. In situ small angle neutron scattering revealing ion carbonization, and applications in catalysis and energy materials. Adv.
sorption in microporous carbon electrical double layer capacitors. ACS Mater. 30, 1706836 (2018).
Nano 8, 2495–2503 (2014). 62. Schutter, C., Passerini, S., Korth, M. & Balducci, A. Cyano ester as solvent
34. Batisse, N. & Raymundo-Pinero, E. Pulsed electrochemical mass spectrometry for high voltage electrochemical double layer capacitors. Electrochim. Acta
for operando tracking of interfacial processes in small-time-constant 224, 278–284 (2017).
electrochemical devices such as supercapacitors. ACS Appl. Mater. Interfaces 9, 63. Krummacher, J., Schütter, C., Hess, L. H. & Balducci, A. Non-aqueous
41224–41232 (2017). electrolytes for electrochemical capacitors. Curr. Opin. Electrochem. 9,
35. Kim, J. et al. Nondisruptive in situ Raman analysis for gas evolution in 64–69 (2018).
commercial supercapacitor cells. Electrochim. Acta 219, 447–452 (2016). 64. Xie, H. J., Gelinas, B. & Rochefort, D. Redox-active electrolyte
36. Richey, F. W., Dyatkin, B., Gogotsi, Y. & Elabd, Y. A. Ion dynamics in supercapacitors using electroactive ionic liquids. Electrochem. Commun. 66,
porous carbon electrodes in supercapacitors using in situ infrared 42–45 (2016).
spectroelectrochemistry. J. Am. Chem. Soc. 135, 12818–12826 (2013). 65. Mourad, E. et al. Biredox ionic liquids with solid-like redox density in the
37. Kondrat, S., Wu, P., Qiao, R. & Kornyshev, A. A. Accelerating charging liquid state for high-energy supercapacitors. Nat. Mater. 16, 446–453 (2017).
dynamics in subnanometre pores. Nat. Mater. 13, 387–393 (2014). 66. Birkl, C. R., Roberts, M. R., McTurk, E., Bruce, P. G. & Howey, D. A.
38. Pean, C. et al. On the dynamics of charging in nanoporous carbon-based Degradation diagnostics for lithium ion cells. J. Power Sources 341,
supercapacitors. ACS Nano 8, 1576–1583 (2014). 373–386 (2017).
39. Merlet, C. et al. On the molecular origin of supercapacitance in nanoporous 67. Singh, N. et al. Artificial SEI transplantation: a pathway to enabling lithium
carbon electrodes. Nat. Mater. 11, 306–310 (2012). metal cycling in water-containing electrolyte. ACS Appl. Energy Mater. 2,
40. Kondrat, S. & Kornyshev, A. Superionic state in double-layer capacitors 8912–8918 (2019).
with nanoporous electrodes. J. Phys. Condens. Matter 23, 022201 (2011). 68. Son, S. B. et al. An artificial interphase enables reversible magnesium
41. Bi, S. et al. Molecular understanding of charge storage and charging chemistry in carbonate electrolytes. Nat. Chem. 10, 532–539 (2018).
dynamics in supercapacitors with MOF electrodes and ionic liquid 69. Brandon, E. J., West, W. C., Smart, M. C., Whitcanack, L. D. & Plett, G. A.
electrolytes. Nat. Mater. 19, 552–558 (2020). Extending the low temperature operational limit of double-layer capacitors.
42. Tsai, W. Y., Taberna, P. L. & Simon, P. Electrochemical quartz crystal J. Power Sources 170, 225–232 (2007).
microbalance (EQCM) study of ion dynamics in nanoporous carbons. 70. Kunze, M. et al. Mixtures of ionic liquids for low temperature electrolytes.
J. Am. Chem. Soc. 136, 8722–8728 (2014). Electrochim. Acta 82, 69–74 (2012).
43. Shpigel, N., Levi, M. D., Sigalov, S., Daikhin, L. & Aurbach, D. In situ 71. Tsai, W. Y. et al. Outstanding performance of activated graphene based
real-time mechanical and morphological characterization of electrodes for supercapacitors in ionic liquid electrolyte from −50 to 80 °C. Nano Energy
electrochemical energy storage and conversion by electrochemical quartz 2, 403–411 (2013).

1162 Nature Materials | VOL 19 | November 2020 | 1151–1163 | [Link]/naturematerials

NaTuRE MaTERIalS Review Article
72. Premathilake, D. et al. Fast response, carbon-black-coated, vertically- 101. Han, P. X. et al. Lithium ion capacitors in organic electrolyte system:
oriented graphene electric double layer capacitors. J. Electrochem. Soc. 165, scientific problems, material development, and key technologies.
A924–A931 (2018). Adv. Energy Mater. 8, 1801243 (2018).
73. Miller, J. R. & Outlaw, R. A. Vertically-oriented graphene electric double 102. Jezowski, P. et al. Safe and recyclable lithium-ion capacitors using sacrificial
layer capacitor designs. J. Electrochem. Soc. 162, A5077–A5082 (2015). organic lithium salt. Nat. Mater. 17, 167–173 (2018).
74. Choi, C. et al. Achieving high energy density and high power density with 103. Takami, N. et al. High-energy, fast-charging, long-life lithium-ion batteries
pseudocapacitive materials. Nat. Rev. Mater. 5, 5–19 (2020). using TiNb2O7 anodes for automotive applications. J. Power Sources 396,
75. Lee, H. Y. & Goodenough, J. B. Supercapacitor behavior with KCl 429–436 (2018).
electrolyte. J. Solid State Chem. 144, 220–223 (1999). 104. Whitmore, A., Agarwal, A. & Xu, L. D. The Internet of Things — a survey
76. Brousse, T., Belanger, D. & Long, J. W. To be or not to be pseudocapacitive? of topics and trends. Inform. Syst. Front. 17, 261–274 (2015).
J. Electrochem. Soc. 162, A5185–A5189 (2015). 105. Lethien, C., Le Bideau, J. & Brousse, T. Challenges and prospects of 3D
77. Simon, P., Gogotsi, Y. & Dunn, B. Where do batteries end and micro-supercapacitors for powering the internet of things. Energy Environ.
supercapacitors begin? Science 343, 1210–1211 (2014). Sci. 12, 96–115 (2019).
78. Lukatskaya, M. R., Dunn, B. & Gogotsi, Y. Multidimensional materials and 106. Huang, P. et al. On-chip and freestanding elastic carbon films for
device architectures for future hybrid energy storage. Nat. Commun. 7, micro-supercapacitors. Science 351, 691–695 (2016).
12647 (2016). 107. Kyeremateng, N. A., Brousse, T. & Pech, D. Microsupercapacitors
79. Wang, J., Polleux, J., Lim, J. & Dunn, B. Pseudocapacitive contributions to as miniaturized energy-storage components for on-chip electronics.
electrochemical energy storage in TiO2 (anatase) nanoparticles. J. Phys. Nat. Nanotechnol. 12, 7–15 (2017).
Chem. C 111, 14925–14931 (2007). 108. Negre, L., Daffos, B., Turq, V., Taberna, P. L. & Simon, P. Ionogel-based
80. Ko, J. S., Sassin, M. B., Rolison, D. R. & Long, J. W. Deconvolving solid-state supercapacitor operating over a wide range of temperature.
double-layer, pseudocapacitance, and battery-like charge-storage Electrochim. Acta 206, 490–495 (2016).
mechanisms in nanoscale LiMn2O4 at 3D carbon architectures. Electrochim. 109. Brachet, M., Brousse, T. & Le Bideau, J. All solid-state symmetrical
Acta 275, 225–235 (2018). activated carbon electrochemical double layer capacitors designed with
81. Gibson, A. J. & Donne, S. W. A step potential electrochemical spectroscopy ionogel electrolyte. ECS Electrochem. Lett. 3, A112–A115 (2014).
(SPECS) investigation of anodically electrodeposited thin films of 110. Sumboja, A. et al. Electrochemical energy storage devices for wearable
manganese dioxide. J. Power Sources 359, 520–528 (2017). technology: a rationale for materials selection and cell design. Chem. Soc.
82. Shao, H., Lin, Z., Xu, K., Taberna, P.-L. & Simon, P. Electrochemical study Rev. 47, 5919–5945 (2018).
of pseudocapacitive behavior of Ti3C2Tx MXene material in aqueous 111. El-Kady, M. F., Strong, V., Dubin, S. & Kaner, R. B. Laser scribing of
electrolytes. Energy Storage Mater. 18, 456–461 (2019). high-performance and flexible graphene-based electrochemical capacitors.
83. Costentin, C. & Saveant, J. M. Energy storage: pseudocapacitance in Science 335, 1326–1330 (2012).
prospect. Chem. Sci. 10, 5656–5666 (2019). 112. Lin, J. et al. Laser-induced porous graphene films from commercial
84. Naguib, M. et al. Two-dimensional nanocrystals produced by exfoliation of polymers. Nat. Commun. 5, 5714 (2014).
Ti3AlC2. Adv. Mater. 23, 4248–4253 (2011). 113. El-Kady, M. F. et al. Engineering three-dimensional hybrid supercapacitors
85. Anasori, B., Lukatskaya, M. R. & Gogotsi, Y. 2D metal carbides and nitrides and microsupercapacitors for high-performance integrated energy storage.
(MXenes) for energy storage. Nat. Rev. Mater. 2, 16098 (2017). Proc. Natl Acad. Sci. USA 112, 4233–4238 (2015).
86. Wang, X. et al. Influences from solvents on charge storage in titanium 114. Ye, L. H. et al. Highly efficient materials assembly via electrophoretic
carbide MXenes. Nat. Energy 4, 241–248 (2019). deposition for electrochemical energy conversion and storage devices.
87. Lukatskaya, M. R. et al. Cation intercalation and high volumetric capacitance Adv. Energy Mater. 6, 1502018 (2016).
of two-dimensional titanium carbide. Science 341, 1502–1505 (2013). 115. Ferris, A., Garbarino, S., Guay, D. & Pech, D. 3D RuO2
88. Lukatskaya, M. R. et al. Ultra-high-rate pseudocapacitive energy microsupercapacitors with remarkable areal energy. Adv. Mater. 27,
storage in two-dimensional transition metal carbides. Nat. Energy 2, 6625–6629 (2015).
17105 (2017). 116. Zhang, Y. Z. et al. Printed supercapacitors: materials, printing and
89. Ghidiu, M., Lukatskaya, M. R., Zhao, M. Q., Gogotsi, Y. & Barsoum, M. W. applications. Chem. Soc. Rev. 48, 3229–3264 (2019).
Conductive two-dimensional titanium carbide ‘clay’ with high volumetric 117. Jost, K., Dion, G. & Gogotsi, Y. Textile energy storage in perspective.
capacitance. Nature 516, 78–81 (2014). J. Mater. Chem. A 2, 10776–10787 (2014).
90. Augustyn, V. et al. High-rate electrochemical energy storage through Li+ 118. Garcia-Torres, J., Roberts, A. J., Slade, R. C. T. & Crean, C. One-step
intercalation pseudocapacitance. Nat. Mater. 12, 518–522 (2013). wet-spinning process of CB/CNT/MnO2 nanotubes hybrid flexible fibres
91. Girard, H. L., Dunn, B. & Pilon, L. Simulations and interpretation of as electrodes for wearable supercapacitors. Electrochim. Acta 296,
three-electrode cyclic voltammograms of pseudocapacitive electrodes. 481–490 (2019).
Electrochim. Acta 211, 420–429 (2016). 119. Quain, E. et al. Direct writing of additive-free MXene-in-water ink for
92. Conway, B. E. Two-dimensional and quasi two-dimensional isotherms for electronics and energy storage. Adv. Mater. Technol. 4, 1800256 (2019).
Li intercalation and UPD processes at surfaces. Electrochim. Acta 38,
1249–1258 (1993). Acknowledgements
93. Li, Y. et al. A general Lewis acidic etching route for preparing MXenes P.S. acknowledges H. Shao, P. Rozier and C. Merlet for help with the figures, as well
with enhanced electrochemical performance in non-aqueous electrolyte. as the Agence Nationale de la Recherche (Labex Store-Ex) and Institut Universitaire de
Nat. Mater. 19, 894–899 (2020). France for support. Y.G.’s research on capacitive energy storage was primarily supported
94. Naoi, K. et al. Ultrafast charge–discharge characteristics of a nanosized through the Fluid Interface Reactions, Structures, and Transport (FIRST) Center,
core–shell structured LiFePO4 material for hybrid supercapacitor an Energy Frontier Research Center funded by the US Department of Energy, Office
applications. Energy Environ. Sci. 9, 2143–2151 (2016). of Science, and Office of Basic Energy Sciences.
95. Amisse, R. et al. Singular structural and electrochemical properties in
highly defective LiFePO4 powders. Chem. Mater. 27, 4261–4273 (2015). Author contributions
96. Okubo, M. et al. Nanosize effect on high-rate Li-ion intercalation in LiCoO2 Both authors contributed equally to the planning and writing of this article.
electrode. J. Am. Chem. Soc. 129, 7444–7452 (2007).
97. Kim, H. S. et al. Oxygen vacancies enhance pseudocapacitive charge storage
properties of MoO3−x. Nat. Mater. 16, 454–460 (2017).
Competing interests
The authors declare no competing interests.
98. Wang, R. C. et al. Operando atomic force microscopy reveals mechanics of
structural water driven battery-to-pseudocapacitor transition. ACS Nano 12,
6032–6039 (2018). Additional information
99. Mitchell, J. B. et al. Confined interlayer water promotes structural stability Correspondence should be addressed to P.S. or Y.G.
for high-rate electrochemical proton intercalation in tungsten oxide Reprints and permissions information is available at [Link]/reprints.
hydrates. ACS Energy Lett. 4, 2805–2812 (2019).
100. Yamada, Y., Wang, J., Ko, S., Watanabe, E. & Yamada, A. Advances and Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in
issues in developing salt-concentrated battery electrolytes. Nat. Energy 4, published maps and institutional affiliations.
269–280 (2019). © Springer Nature Limited 2020

Nature Materials | VOL 19 | November 2020 | 1151–1163 | [Link]/naturematerials 1163