Author’s Accepted Manuscript

A high-energy, long cycle-life hybrid

Bote Zhao, Dongchang Chen, Xunhui Xiong, Bo

Ching-Ping Wong, Meilin Liu

PII: S2405-8297(16)30268-9
DOI: [Link]
Reference: ENSM107
To appear in: Energy Storage Materials
Received date: 25 September 2016
Revised date: 29 November 2016
Accepted date: 30 November 2016
Cite this article as: Bote Zhao, Dongchang Chen, Xunhui Xiong, Bo Song,
Renzong Hu, Qiaobao Zhang, Benjamin H. Rainwater, Gordon H. Waller,
Dongxing Zhen, Yong Ding, Yu Chen, Chong Qu, Dai Dang, Ching-Ping Wong
and Meilin Liu, A high-energy, long cycle-life hybrid supercapacitor based on
graphene composite electrodes, Energy Storage Materials,
[Link]
This is a PDF file of an unedited manuscript that has been accepted for
publication. As a service to our customers we are providing this early version of
the manuscript. The manuscript will undergo copyediting, typesetting, and
review of the resulting galley proof before it is published in its final citable form.
Please note that during the production process errors may be discovered which
could affect the content, and all legal disclaimers that apply to the journal pertain.
A high-energy, long cycle-life hybrid supercapacitor

based on graphene composite electrodes

Bote Zhao, Dongchang Chen, Xunhui Xiong, Bo Song, Renzong Hu, Qiaobao Zhang, Benjamin

H. Rainwater, Gordon H. Waller, Dongxing Zhen, Yong Ding, Yu Chen, Chong Qu, Dai Dang,

Ching-Ping Wong, Meilin Liu*

School of Materials Science and Engineering, Georgia Institute of Technology, Atlanta, Georgia

30332-0245, United States.

*Corresponding Author. [Link]@[Link]

ABSTRACT

The energy density of a hybrid supercapacitor consisting of a battery-type electrode and a

capacitive electrode could be significantly higher than that of an electrical double-layer capacitor

(EDLC) due to the broadened voltage window and the high capacity of the battery-type

1
electrode. However, the commercialization of hybrid supercapacitors is still hampered by lacking

of proper electrode materials of desired nanostructures. Here we report a class of nanocomposite

electrodes composed of CoxNi1-x(OH)2 and reduced graphene oxide (rGO), derived from a facile

process at room temperature using hydrous hydrazine and proper ratios of Co(II) to Ni(II) to tune

the composition and morphology. In particular, an architectural composite electrode consisting

of porous CoxNi1-x(OH)2 disks wrapped by rGO achieves large capacity, high rate capability (743

and 545 C g-1 at 1 and 20 A g-1, respectively), and long cycling life. When coupled with a p-

phenylenediamine (PPD)-modified rGO, the resulting hybrid supercapacitor exhibits superior

energy densities of 72 and 44 W h Kg-1 at a power density of 797 W Kg-1 and 16.7 kW Kg-1,

respectively, and excellent cycling stability for 20,000 cycles at 20 A g-1, implying that it is a

very promising device for portable power and next-generation energy storage.

KEYWORDS

graphene, composites, hybrid supercapacitors, hydroxides, organic molecular

1. Introduction

Supercapacitors (or electrochemical capacitors) have attracted intense attention due to

their higher power density and longer cycle-life than existing rechargeable batteries [1-4]. They

are well suited for applications requiring rapid power delivery and recharging, such as

regenerative braking, short-term energy storage, hybrid electric vehicles, large industrial

2
equipment, and portable devices. However, commercially available supercapacitors are still

based on activated carbon materials, known as electrical double-layer capacitors (EDLCs), and

have much less energy density than rechargeable batteries, which severely limit their potential

for many applications [1].

During the past few years, asymmetric/hybrid supercapacitors with much higher energy

density than that of EDLCs, due to the extended voltage window and the significantly enhanced

capacitance, are considered a promising energy storage device to meet the demands for many

emerging applications [5-18]. The term “asymmetric supercapacitors” is suggested by Brousse et

al. to be used for the devices with only capacitive/pseudocapacitive electrodes (e.g.,

MnO2/graphene//carbon nanotubes [5]) to avoid confusion with “hybrid” devices in-between a

supercapacitor and a battery. [19]. Compared to asymmetric supercapacitors, hybrid

supercapacitors consisting of a battery-type faradaic electrode (for high energy density) and a

capacitive electrode (for high power density and long cycle life) could achieve even higher

energy density because the hybridization of these two electrodes can further broaden the

operating voltage and increase the capacitance of the hybrid capacitor [20, 21]. The

electrochemical performance of hybrid supercapacitors is sensitive to the nature of the electrode

materials, which are the key to developing a long cycle-life hybrid supercapacitor with high

energy and power density. In a typical hybrid supercapacitor, the capacitive electrode is usually

activated carbon[11-14, 22] or graphene[15, 18, 23, 24], and the battery-type electrode is lithium

intercalation compounds [7, 8] or transition metal compounds[16, 17, 22, 24, 25].

Recently, Ni and/or Co-based oxides/hydroxides have been reported as one of the most

promising candidates for battery-type electrodes in hybrid supercapacitors due to their attractive

theoretical specific capacity and potentially high rate capability in aqueous electrolyte (e.g.,

3
KOH). These oxides/hydroxides that were reported as advanced “pseudocapacitive”

materials[26-28], have been recently considered as “extrinsic pseudocapacitors” [29] (with other

examples including V2O5[29] and metal sulfides[30, 31]), which do not exhibit

pseudocapacitance but battery-type behavior in the bulk state due to phase transformations

during ion storage, but an improved capacitive behavior in nanosize with increased surface area

especially at high current density. Now these Ni and/or Co-based oxides/hydroxides have been

termed as battery-type faradaic electrode materials based on their electrochemical behavior in

aqueous electrolytes [2, 19, 32], which have gained significant credence. However, these

oxides/hydroxides suffer from relatively poor electronic conductivity, leading to limited rate

capability and cycling stability. Growth of these Ni and/or Co-based oxides/hydroxides on

conductive substrates such as carbon fiber paper[27], or nickel foam[33-35] have been

demonstrated as an effective approach to improve the electronic conductivity, further resulting in

an improved rate capability and cycling performance. Compared to commercial carbon fiber

papers and nickel foams, the formation of composites with graphene is preferred. Graphene-

based materials have been widely reported with attractive electrical/electrochemical properties

[36-41]. The much higher surface area and ultrathin thickness of graphene results in significantly

enhanced mass loading of active materials (i.e., Ni and/or Co-based oxides/hydroxides), and

further improved capacity based on the total mass of active materials/conductive substrates. To

date, numerous efforts have been devoted to engineering the microstructures of Ni and/or Co-

based oxides/hydroxides for improved electrochemical performance[24-26, 42-44]. However,

advanced architecture with stable cycling performance as well as high rate capability derived

from a facile method is still rarely reported. Porous microstructures in micrometer-size derived

from a facile method are significantly preferred for commercial applications.

4
Figure 1. Schematic illustration of the hybrid supercapacitor device with a CoxNi1-x(OH)2@

reduced graphene oxide (rGO) composite as a battery-type faradaic electrode and a p-

phenylenediamine (PPD)-modified rGO composite as a capacitive electrode.

In this work, we report our findings on the synthesis and characterization of a series of

CoxNi1-x(OH)2-reduced graphene oxide (rGO) nanocomposites prepared by a facile process at

room temperature where hydrous hydrazine was used as the OH- source and reducing agent.

Different ratios of initial Co(II) to Ni(II) and different amounts of hydrazine have been used to

tune the composition and morphology. Specifically, a porous disk-like CoxNi1-x(OH)2 structure

wrapped by rGO was prepared, in which the unique porous structures facilitate rapid access of

liquid electrolyte to electrode materials and rGO offers effective electron paths for fast energy

storage. The resulting composite electrode exhibited a high specific capacity, rate capability, and

cycling stability. More importantly, when coupled with a high-performance p-phenylenediamine-

modified rGO (PPD/rGO) capacitive electrode, the hybrid supercapacitor (Figure 1) delivered

greatly enhanced gravimetric/volumetric energy density and cycle-life compared to those of

recently reported advanced hybrid/asymmetric supercapacitors[22-25, 45-49]. The simple

5
preparation process and high performance make the hybrid supercapacitor attractive for a wide

variety of applications.

2. Experimental Section

Materials synthesis: Graphite oxide was synthesized by a modified Hummers’ method [50]. The

as-prepared aqueous suspension of graphite oxide was diluted with deionized water under

ultrasonication for 5 h to obtain an exfoliated graphene oxide (GO) suspension (~1 mg mL-1). 50

mL of mixed Ni(CH3COO)2·4H2O and Co(CH3COO)2·4H2O solution with 30 mmol of total

metal ions was poured into a 50 mL of GO suspension (1 mg mL-1) and stirred for 10 min, and

then 1.5 ~ 4.5 mL of hydrazine monohydrate (N2H4 64-65 %, reagent grade, 98%, Sigma-

Aldrich) was dropped into the solution, followed by stirring for 36 h at room temperature. The

as-obtained composite was washed repeatedly with deionized water by vacuum filtration until

the filtrate had a neutral pH and the filtrate became colorless. The collected black solids were

dried at room temperature in air under fume hood.

By adding 1.5 mL of hydrazine monohydrate, a series of CoxNi1-x(OH)2-rGO composites

with different Co2+:Ni2+ ratios but fixed total metal ion amount (30 mmol) were prepared and

named CNOHa:b (a:b means that the mole ratio of Co(II):Ni(II) in the raw materials). By fixing

the initial Ni(II) to Co(II) ratio as 1:1, the increased amounts of hydrous hydrazine (3.0 and 4.5

mL) were further used to synthesize the CoxNi1-x(OH)2-rGO nanocomposites, which were named

CNOH1:1-II, and CNOH1:1-III, respectively. A pure rGO was also prepared in a similar fashion

by using 4.5 mL of hydrous hydrazine as the reducing agent only without adding

Ni(CH3COO)2·4H2O or Co(CH3COO)2·4H2O.

6
 To prepare the PPD/rGO material, 50 mg of GO was dispersed in 35 mL of deionized water.

Then, 54 mg of PPD was dissolved in 15 mL of deionized water and added into the GO

suspension. The mixture was stirred for 30 min and reacted in a Teflon-lined stainless steel

autoclave at 180 °C for 12 h. After the reaction, the composite was washed repeatedly with

deionized water and dried at 55 °C for 24 h.

Basic characterization: The X-ray diffraction (XRD) patterns were recorded with an X’Pert PRO

Alpha-1 X-ray diffractometer. A scanning electron microscope (SEM, LEO 1530) coupled with

an energy dispersive X-ray spectrometer (EDX) was used to analyze the morphologies and

element distribution/ratio of samples. A high-resolution transmission electron microscope (TEM,

FEI G2 Tecnai F30) was used to analyze the microstructures and morphologies of samples. Prior

to TEM analysis, the samples were ultrasonically dispersed in ethanol for several hours. The

nitrogen adsorption–desorption isotherm of the sample was measured at the boiling point of

liquid nitrogen (77 K) using a Micromeritics ASAP 2020 analyzer. Thermogravimetric analysis

(TGA) was carried out with an SDTQ-600 (TA Instruments Co.) apparatus in air with a heating

rate of 10 °C min−1. X-ray photoelectron spectroscopy (XPS) was measured on a Thermo K-

Alpha XPS spectrometer (Thermo Fisher Scientific) equipped with a monochromatic Al-Kα X-

ray source (hv = 1468.6 eV). Raman spectroscopic measurement was performed using a

Renishaw RM1000 microspectroscopic system. A He-Ne laser with a wavelength of 633 nm

(Thorlab HRP-170) was applied as excitation laser through a 20x/0.40 objective.

7
Electrochemical measurement: The electrochemical performances of the as-obtained CoxNi1-

x(OH)2-rGO materials were investigated under a three-electrode configuration in a beaker with a

2 M KOH aqueous solution as electrolyte at room temperature. A Ag/AgCl electrode prefilled

with 4 M KCl aqueous solution saturated with AgCl and a Pt mesh electrode were used as the

reference and the counter electrodes, respectively. The working electrode was prepared as

follows: the CoxNi1-x(OH)2-rGO materials and Super P Li powders (TIMCAL Graphite &

Carbon) were ground and sieved through a 170-mesh sieve before using; a free-standing film

(5/16 in. in diameter) was made from a mixture of CoxNi1-x(OH)2-rGO materials, Super P Li and

polytetrafluoroethylene (PTFE) at a mass ratio of [Link] with assistance of ethanol; the working

electrode was obtained by pressing the free-standing film between two nickel foams (3/8 in. in

diameter). Besides acting as a conductive additive, the Super P Li powders was also used as

additive to assist the preparation of free-standing electrode films; it appears that we were unable

cast the CoxNi1-x(OH)2-rGO into a free-standing electrode film without adding Super P Li

powders. The typical mass loading of the active material (i.e., CoxNi1-x(OH)2-rGO) in the

electrode is 2.7 ～ 3.3 mg cm-2. The working electrode was soaked in a 2 M KOH aqueous

solution under vacuum for > 8 h before the electrochemical measurement.

The electrochemical performances of the hybrid supercapacitor were investigated under a

two-electrode configuration in a split test cell (MTI Corporation). The free-standing film mixture

of CoxNi1-x(OH)2-rGO materials, Super P Li and PTFE at a mass ratio of [Link] was used as the

positive electrode. The negative electrode was a free-standing film of mixture of PPD/rGO

materials and PTFE at a mass ratio of 95:5. Carbon fiber papers (7/16 in. in diameter) were used

as current collectors by only physical contact with the working electrode. A 2 M KOH aqueous

solution was used as electrolyte, and a NKK separator (MPF30AC-100, Nippon Kodoshi

8
Corporation, Kochi, Japan) was used as the separator. The electrodes were soaked in the

electrolyte solution for approximately 1 h under vacuum and were assembled in a split test cell

under ambient atmosphere. The cell was tightened by 4 nuts and bolts, and was standing for

another 9 h under atmospheric pressure before the electrochemical measurement.

Cyclic voltammograms were recorded on a Solartron electrochemical workstation (Solartron

SI 1287 electrochemical interface). The galvanostatic charge/discharge measurements were

carried out using an Arbin testing system (BT-2143) at room temperature. At the beginning of

the electrochemical measurement, the working electrode or cell was always activated via cyclic

voltammetry test at a scan rate of 30 mV s-1 for 5-10 cycles until cyclic voltammograms became

relatively stable. The cycling performance of three-electrode system or two-electrode device was

carried out after the initial activating process and rate capability testing.

The specific capacity (C g-1) of CoxNi1-x(OH)2-rGO electrode based on the galvanostatic

discharge curve was calculated using the following equation:

Q = IΔt/m (1)

where I (mA) is the discharge current, Δt (s) is the discharge time, and m (mg) is the mass of the

active material (i.e., CoxNi1-x(OH)2-rGO). The specific capacity is based on the total mass of the

CoxNi1-x(OH)2-rGO composite.

The mass ratio of positive electrode to negative electrode in the hybrid supercapacitor device

was determined to better balance the obtainable charge of the two electrodes (q+ = q−). The

charge of the positive electrode (q+) was calculated with that of the negative electrode (q−) based

on the cyclic voltammograms acquired from a cell with a three-electrode configuration.

9
 q = ∫imdV/v (2)

where q (C) is the charge, i (A g-1) is the current density, m (mg) is the mass of the active

material, V (V) is the voltage, v (mV s-1) is the scan rate, and ∫idV is the integral area of the cyclic

voltammogram.

To achieve q+ = q−, the mass ratio of positive electrode active material to negative electrode

active material is determined from the following equation:

m+:m− = (∫idV/v)−: (∫idV/v)+ (3)

The mass ratio of the positive electrode active material (CoxNi1-x(OH)2-rGO) to the negative

electrode active material (PPD/rGO) was calculated to be 1:2.28 for the hybrid supercapacitor, as

determined from the cyclic voltammograms acquired at 10 mV s−1 (Figure 4a) for both

electrodes.

The energy density (W h kg-1) of the hybrid supercapacitor device was calculated based on

the galvanostatic discharge curve using the following equation:

E = I ∫Vdt/3.6(mCoxNi1-x(OH)2-rGO+ mPPD/rGO) (4)

The constant 3.6 is adopted when the energy density (E) is in W h kg-1, the current (I) is in mA,

the voltage (V) is in V, the time (t) is in seconds, and the mass (m) is in mg. The mass used in the

calculation for hybrid supercapacitor device was based on the total mass of CoxNi1-x(OH)2-rGO

composite and PPD/rGO composite.

The power density (W kg-1) is calculated using the following equation:

P = E/Δt (5)

10
3. Results and Discussion

The preparation processes of CoxNi1-x(OH)2-rGO composites are schematically illustrated

in Figure 2a. Co(II)/Ni(II)-containing aqueous solution was mixed with GO suspension, followed

by adding 1.5 mL of hydrous hydrazine, stirring for 36 h at room temperature. A series of

CoxNi1-x(OH)2-rGO composites with different Co(II):Ni(II) ratios were prepared and named

CNOHa:b (a:b means that the mole ratio of Co(II):Ni(II) in the raw materials; the actual ratios in

the as-prepared samples were also tested by EDX analysis with results shown in Table S1, which

were similar to initial ratios by adding 1.5 mL of hydrous hydrazine). After washing, the

morphologies of the as-obtained CoxNi1-x(OH)2-rGO composites (Figure S1) were very similar,

although the degree of aggregation may appear somewhat different. The morphological features

are dominated by wrinkled graphene sheets, suggesting that the as-prepared CoxNi1-x(OH)2 are

small nanoparticles, as reflected by the relatively low intensity of X-ray diffraction (XRD) peaks

(Figure S2). The XRD patterns of CNOH3:0, CNOH2:1, CNOH1:1, and CNOH1:2 samples

were similar with four broad peaks corresponding to α-Ni(OH)2- and α-Co(OH)2-like structures

[51]. No diffraction peak corresponding to metallic Ni or Co was observed. The CNOH0:3

sample without Co(II) was almost amorphous instead of the α-Ni(OH)2-like structure, which may

be attributed to some special reaction with hydrous hydrazine in absence of Co(II). The XRD

peak intensity of the CNOH3:0 sample without Ni(II) was much lower than those of the

CNOH2:1, CNOH1:1, and CNOH1:2 samples. These results suggest that the presence of both

Co(II) and Ni(II) ions in the alkaline solution (hydrous hydrazine) improves the crystallinity of

the hydroxides. The absence of the obvious diffraction peak corresponding to rGO could be

attributed to the distinctly reduced layer-to-layer interaction between adjacent graphene sheets

11
Figure 2. (a) Schematic illustration of the preparation of CoxNi1-x(OH)2-rGO composites via a

facile hydrazine-induced process at room temperature. (b, c, d) BF-TEM images of (b) CNOH1:1,

(c) CNOH1:1-II, and (d) CNOH1:1-III (i.e., CoxNi1-x(OH)2-rGO composite; 1:1 means that the

mole ratio of Co(II):Ni(II) in the raw materials is 1:1; II and III represent the sample prepared

with different amounts of hydrous hydrazine); (e) DF-TEM, and (f) HRTEM images of

CNOH1:1-III obtained from the selected rectangle region with orange border (dotted line), and

red border (solid line) in (d), respectively; (g) HAADF STEM image of CNOH1:1-III; the

arrows in (f) indicate layered graphene structure; the arrows in (g) indicate the graphene sheet

structures, the circles (dotted line) indicate the typical pores in the unique architecture. (h) XRD

12
patterns, (i) XPS spectra of C 1s, and (j) XPS spectra of Ni 2p of as-prepared samples using

different amounts of hydrous hydrazine: CNOH1:1, CNOH1:1-II, and CNOH1:1-III; XPS

spectrum of C 1s in GO sample was also given in (i); “Sat.” in (j) means satellite.

which could be an indication that the surface of rGO was mostly covered by as-formed CoxNi1-

x(OH)2.

The hydrous hydrazine in the synthesis process plays two key roles: offering OH- for the

formation of CoxNi1-x(OH)2 and acting as the reducing agent for reducing GO. We systematically

varied the used amount of hydrous hydrazine to explore the effect on the material properties

(Figure 2a). With increasing amount of hydrous hydrazine from 1.5 to 3.0 mL, the morphology

of as-obtained sample (named CNOH1:1-II) (Figure S3a) still remained similar, but somewhat

looser compared to that of the CNOH1:1 sample (Figure S3b). When the amount of hydrous

hydrazine was increased to 4.5 mL, many CoxNi1-x(OH)2 disk-like structures of micrometer-size

appeared in the as-prepared sample (named CNOH1:1-III), as shown in Figure S3c. More

importantly, it seems that the CoxNi1-x(OH)2 disks were wrapped by graphene sheets. The Co:Ni

ratios in the samples measured by EDX analysis increased from 44:56 for CNOH1:1 sample to

65:35 for CNOH1:1-II sample, and 69:31 for CNOH1:1-III sample, respectively, with increasing

amount of hydrous hydrazine (Table S1). As seen from the transmission electron microscopy

(TEM)/scanning TEM (STEM) images shown in Figure 2b-g, a significant amount of folding

graphene sheets were observed for the samples. The bright-field (BF)-TEM image of CNOH1:1-

II sample (Figure 2c) confirmed that it had relatively loose microstructure compared to that of

CNOH1:1 sample (Figure 2b). Figure 2d presents a typical CoxNi1-x(OH)2@rGO disk, in which

the CoxNi1-x(OH)2 disk appeared to be wrapped by graphene sheets. Dark-field (DF)-TEM

images demonstrate that the as-formed CoxNi1-x(OH)2 in CNOH1:1-III (Figure 2e), CNOH1:1-II

13
(Figure S4) and CNOH1:1 (Figure S5) samples were actually ultrafine nanocrystals. These

nanocrystals were relatively uniformly distributed in the composites especially for CNOH1:1-II

and CNOH1:1. Figure 2f and S6 show two typical high-resolution TEM (HRTEM) images of the

border of CoxNi1-x(OH)2@rGO disk, in which the periodic lattice fringe corresponding to CoxNi1-

x(OH)2 phase were wrapped by clear graphene sheets with 3-5 layers. High-angle anular dark-

field (HAADF) STEM image (Figure 2g) confirmed that the CoxNi1-x(OH)2 disk was wrapped by

graphene sheets. Importantly, the as-obtained CoxNi1-x(OH)2 disks were porous with a significant

quantity of mesopores from the observation of HAADF STEM (Figure 2g and S7). The porosity

of the samples was further analyzed by nitrogen adsorption–desorption measurements,

confirming the mesoporosity in CNOH1:1-III sample (Figure S8). Such porous structure would

facilitate the access of liquid electrolyte to electrode materials for fast ionic transport while the

wrapped graphene sheets promote effective electron transport and current collection.

The phase structures of these as-prepared nanocomposites were almost the same, as

reflected by the XRD patterns shown in Figure 2h. No diffraction peaks corresponding to

metallic Ni or Co were observed even when the amount of hydrous hydrazine was increased to

4.5 mL. The character of broad peaks in XRD patterns was consistent with the ultrafine

nanocrystal nature as observed by DF-TEM (Figure 2e, S4, and S5). As determined by

thermogravimetric analysis (TGA) (Figure S9), the content of CoxNi1-x(OH)2 in the composite

increased with increasing amount of hydrous hydrazine from CNOH1:1 to CNOH1:1-II sample

and almost remained unchanged for the CNOH1:1-III sample.

Figure 2i shows X-ray photoelectron spectroscopy (XPS) spectra of C1s in different

samples with fitted results. After the synthesis process, the intensity of C-O-type carbon (286.8

eV, including C-OH and C-O-C (epoxide groups)) significantly decreased, confirming that the

14
GO was successfully reduced by hydrous hydrazine. The peak shift of C-O-type carbon was

attributed to the sharp reduction of C-O-C groups in C-O-type carbon. With increasing amount

of hydrous hydrazine, the content of C=O- and O-C=O-type carbon somewhat increased. This

result is anomalous since hydrous hydrazine is a well-known reducing agent. The result may be

due to the competition between the formation of metal hydroxide and the reduction of graphene

oxide. In the Ni 2p spectra (Figure 2j), expected Ni 2p1/2 (873.3 eV) and Ni 2p3/2 (855.7 eV)

signals were observed with two corresponding satellites for all the samples, consistent with that

of Ni(OH)2 [52, 53] and Ni-Co hydroxide [28]. No Ni 2p3/2 peak centered at ~852.6 eV was

observed, confirming that no metallic Ni formed under this reducing environment. The Co 2p3/2

was centered at 780.8 eV (Figure S10), suggesting the formation of cobalt hydroxide [54]. The

spin-orbit splitting value of Co 2p1/2 and Co 2p3/2 was 15.6 eV, and the intensity of the Co 2p3/2

satellite line was low, indicating the presence of both Co3+ and Co2+ in these CoxNi1-x(OH)2-rGO

samples[28, 52]. The formation of CoxNi1-x(OH)2-rGO composite was further supported by

Raman analysis (Figure S11).

To characterize the electrochemical performance, the as-prepared CoxNi1-x(OH)2-rGO

composites were tested using a three-electrode configuration in a 2 M KOH aqueous solution.

Shown in Figure 3a are some typical cyclic voltammetric (CV) curves of CoxNi1-x(OH)2-rGO

composite electrodes derived from different Co(II) to Ni(II) ratios and 1.5 mL of hydrous

hydrazine. While one pair of redox peaks were observed for all samples, the oxidation peak of

the as-prepared composite electrodes shifted in the positive potential direction while increasing

the initial Ni(II) content. The CNOH2:1 and CNOH1:1 samples exhibited the largest integral

area, indicating the highest specific capacities. Figure 3b shows the galvanostatic charge-

discharge curves of the CoxNi1-x(OH)2-rGO composite electrodes at 1 A g-1; the electrochemical

15
Figure 3. (a) CV curves at a scan rate of 2 mV s-1, (b) 5th galvanostatic charge-discharge curves

at 1 A g-1, (c) specific capacities of CoxNi1-x(OH)2-rGO composite electrodes with different

initial Co(II) to Ni(II) ratios using 1.5 mL of hydrous hydrazine. (d) 5th galvanostatic charge-

discharge curves at 1 A g-1, and (e) rate performance of as-prepared samples using different

amounts of hydrous hydrazine: CNOH1:1, CNOH1:1-II, and CNOH1:1-III, the inset in (e) is

discharge curves of the CNOH1:1-III electrode at different current densities (1-20 A g-1), the

rectangular triangles shown in the figure are used for clearly understanding the different slopes

of the curves at 1 (red) and 20 A g-1 (violet). (f) Cycling performance of the CNOH1:1-III

electrode at 20 A g-1.

behaviors are consistent with the CV results. The corresponding discharge capacities were

summarized in Figure 3c. The capacities of the CNOH2:1 and CNOH1:1 electrodes were similar,

demonstrating 692 and 707 C g-1 (based on the total mass of CoxNi1-x(OH)2-rGO composite),

respectively, much better than those of the CNOH1:2, CNOH3:0 and CNOH0:3 electrodes. The

16
substitution of Ni with an appropriate amount of cobalt definitely increased the specific capacity,

consistent with the reported results [25, 51, 55]. Based on the slightly higher capacity and higher

redox potentials (better for hybrid device) of CNOH1:1 electrode compared to those of

CNOH2:1 electrode, the initial Co(II) to Ni(II) ratio of 1:1 was selected to study the effect of the

amounts of hydrous hydrazine.

The galvanostatic charge-discharge curves of the CNOH1:1, CNOH1:1-II, and

CNOH1:1-III electrodes at 1 A g-1 are shown in Figure 3d. The different potential plateaus

suggest the difference of Co to Ni ratio in these samples, which was consistent with EDX

analysis (Table S1). The corresponding typical rate capability results are summarized in Figure

3d. With increasing amount of hydrous hydrazine, the capacities at low current density were

somewhat increased due to the increased content of CoxNi1-x(OH)2 in the composite. The rate

capability was also enhanced with increasing amount of hydrous hydrazine. This effect is likely

due to the improved microstructures of CNOH1:1-II and CNOH1:1-III samples, where more

open channels promote easy ion access whereas the graphene allows fast current collection. In

particular, specific capacities of 743 and 545 C g-1 were achieved, respectively, at 1 A g-1 and 20

A g-1 for the CNOH1:1-III electrode. For comparison, a CNOH1:1-III sample without graphene

was also prepared in a similar fashion with typical electrochemical performance shown in Figure

S13. The CNOH1:1-III electrode exhibited significantly improved specific capacity and rate

performance compared to that of the CNOH1:1-III without graphene electrode, suggesting the

presence of a synergistic effect of CoxNi1-x(OH)2 and graphene in the composite.

Figure 3e shows the typical discharge curves of the CNOH1:1-III electrode at different

current densities (1 to 20 A g-1). A potential plateau between 0.15 and 0.05 V with relatively

small slope prevails over a wide capacity range, although the slope increased with increasing

17
current density, suggesting an increase in the contribution of capacitive behavior to capacity. A

high capacitive contribution is more beneficial to cycling stability [56, 57]. When evaluated at 20

A g-1, the CNOH1:1-III electrode demonstrated an attractive capacity retention of ~91% after

5,000 cycles (Figure 3f). The morphology stability of CoxNi1-x(OH)2 disks in the CNOH1:1-III

electrode was verified by SEM observation of CoxNi1-x(OH)2 disks grown on nickel foam before

and after 5,000 cycles (Figure S14). Additionally, the Co:Ni ratio in the sample was found to be

a key factor for cycling stability; the composite sample with a Co:Ni ratio of ~2:1 had a better

electrochemical stability compared to that of the composite sample with a Co:Ni ratio of ~1:1

(Figure S15). The high specific capacity, rate capability, and superior cycling stability could be

attributed to the proper Co:Ni ratio, unique porous disk-like architecture, the wrapped rGO, and

the high contribution of capacitive behavior to capacity at high current densities.

To further enhance the energy density, the operation voltage of the capacitor should be

broadened using a hybrid cell configuration. We built a hybrid device using a high-performance

PPD/rGO composite as the negative electrode and the CNOH1:1-III as the positive electrode, as

schematically illustrated in Figure 1. Some typical CV curves of the two electrodes are presented

in Figure 4a. Clearly, the CNOH1:1-III electrode exhibited a pair of redox peaks within a

potential window of -0.15~0.45 V whereas the PPD/rGO electrode showed a typical capacitive

behavior between -1.15 and -0.15 V (Figure 4a and S16) with electrical double-layer capacitive

contribution from rGO and amine groups-derived pseudocapacitive contribution from PPD. The

PPD/rGO composite achieved a much higher specific capacitance than that of the pure rGO

sample (Figure S16). Since the specific surface area of PPD/rGO is 7.7 m2 g-1 and that of rGO is

188.7 m2 g-1 (Figure S17), the PPD in the PPD/rGO composite contributes the most to the

pseudocapacitance of the sample. However, a comparison of the CV curve of pure rGO with that

18
of the PPD/rGO electrode at the same scan rate (Figure S16) indicates that the double-layer

contribution from PPD/rGO electrode was still higher than that from pure rGO electrode,

suggesting that the PPD also functions as a molecular spacer to prevent the graphene sheets from

restacking. An operation voltage of 1.6 V could be achieved for the hybrid

Figure 4. (a) CV curves of the CNOH1:1-III electrode and PPD/rGO electrode at a scan rate of

10 mV s-1 obtained in a three-electrode system; (b) CV curves at different scan rates (5-100 mV

s-1), (c) galvanostatic discharge curves at different current densities (1-20 A g-1) of hybrid

supercapacitor device with CNOH1:1-III//PPD/rGO couple; (d, e) Ragone plot of the CNOH1:1-

III//PPD/rGO hybrid supercapacitor device compared to that of traditional EDLCs and LIBs

based on the mass of active materials [34], and recently reported advanced hybrid/asymmetric

supercpacitors [22-25, 45-49] with electrode materials shown in (e); all the energy/power density

was calculated based on the mass of active materials; The inset time is obtained by dividing the

19
energy density by the power density, which is a time constant of the devices; (e) is an enlarged

image of (d); and (f) cycling stability at current density of 20 A g-1 between 0 and 1.6 V of the

CNOH1:1-III//PPD/rGO hybrid supercapacitor device, the inset shows the corresponding

galvanostatic charge/discharge curves of the 9990-10000th and last 10 cycles; the inset digital

photo is a GT logo consisting of 32 green LEDs in parallel powered by two hybrid

supercapacitor devices in series.

supercapacitor. The corresponding CV curves and galvanostatic discharge curves of the hybrid

supercapacitor are shown in Figure 4b and 4c. The shapes of CV curves were well maintained as

the scan rate was increased from 5 to 100 mV s-1, suggesting reasonably high rate capability of

the hybrid supercapacitor, attributed to the electrochemical properties of both the positive and

negative electrodes. The galvanostatic discharge curves of the hybrid supercapacitor (Figure 4c)

appear to be battery-type with a marked sloping plateau at low current densities, but seems to be

capacitor-like behavior at high current densities (inset in Figure 4f). The rate performance was

summarized in Figure S18.

Shown in Figure 4d and 4e (the Ragone plot) are the gravimetric energy and power

densities of the hybrid supercapacitor calculated from the discharge curves (Figure 4c),

demonstrating a high energy density of 72 W h kg-1 at a power density of 797 W kg-1. Even at a

power density of 16.7 kW kg-1, energy density of 44 W h kg-1 still remains. The energy and

power densities of the traditional EDLCs and lithium-ion batteries (LIBs) based on the mass of

active materials [34] were also presented in Figure 4d for comparison. The hybrid supercapacitor

delivered much higher energy density compared to that of EDLCs at the same power density and

much higher power density compared to that of LIBs at the same energy density. The energy

20
densities of the hybrid supercapacitor are comparable to those of the LIBs at the same power

density. Furthermore, the gravimetric energy and power densities of the hybrid supercapacitor

were superior to that of recently reported advanced hybrid/asymmetric supercapacitors, including

NiCo2S4 ball-in-ball hollow spheres//graphene/carbon spheres [45], NiCo2S4/N-doped carbon (C)

foam//ordered mesoporous C/N-doped C foam[46],

Co0.5Ni0.5(OH)2/graphene(G)/CNTs//activated carbon (AC)[25], Ni-Co-Fe hydroxide//AC[22],

MnO2/GO//porous C[47], Co-Al hydroxide/dodecyl sulfate anions/G//sandwiched G/porous

C[48], Fe2O3/functionalized G//MnO2/functionalized G[49], NiO-porous C hollow sphere//N-

doped G[23], Ni(OH)2/G//porous G[24]. Additionally, the volumetric energy and power densities

of the presented hybrid supercapacitor were also evaluated with results shown in Figure S19,

which were comparable or even superior to that of recently reported advanced

hybrid/asymmetric supercapacitors [6, 33]. Moreover, a simple application example of the hybrid

supercapacitor is presented. A GT (Georgia Institute of Technology) logo consisting of 32 green

LEDs in parallel can be powered by two hybrid supercapacitor devices in series after being

charged to 3.2 V (inset in Figure 4f). Cycling performance, which is an important property for a

supercapacitor, of the hybrid supercapacitor at 20 A g-1 between 0 and 1.6 V is shown in Figure

4f. An excellent cycling stability without undergoing decay compared to the initial Q m was

achieved even after 20,000 cycles at a high current density of 20 A g-1, which is much better than

that of LIBs and superior to those of recently reported advanced hybrid/asymmetric

supercapacitors (Table S2)[22-24, 45-49]. The corresponding Coulombic efficiency was around

99% for the initial 2,000 cycles, and was above 99% (close to 100%) for the subsequent 18,000

cycles (Figure S20). Furthermore, the corresponding galvanostatic charge/discharge curves of the

21
9990-10000th and the last 10 cycles are given in Figure 4f (inset). There is little observable

change in these curves, confirming the excellent cycling stability of the hybrid supercapacitor.

4. Conclusion

In summary, a high-energy, long cycle-life hybrid supercapacitor was constructed using

graphene-based hybrid materials: CoxNi1-x(OH)2-rGO and PPD/rGO. We have successfully

prepared a series of CoxNi1-x(OH)2-rGO nanocomposites via a facile hydrazine-induced process

and studied the effect of different Co(II) to Ni(II) ratios and hydrous hydrazine amounts on the

phase structure, microstructure, and the electrochemical performance. In particular, a porous

CoxNi1-x(OH)2 disks@rGO nanocomposite delivered outstanding specific capacities, rate

capability, and cycling stability. When coupled with a capacitive-type PPD/rGO electrode, the

hybrid supercapacitor achieved superior energy and power density, and a stable cycling

performance for 20,000 cycles at 20 A g-1. The attractive electrochemical performance and

simple preparation process make the hybrid supercapacitor a promising candidate for

commercial applications. Furthermore, the unique architecture with porous micrometer-sized

disks wrapped by rGO provides new insight into the design of better electrode materials. The

facile preparation strategy further opens a general avenue to the development of cost-effective

fabrication of high-performance supercapacitors based on transition metal hydroxides for many

applications.

Supporting Information.

SEM images and XRD patterns of CNOHa:b; SEM, TEM images, N2 adsorption–desorption

results, TGA curves, XPS spectra and Raman results of CNOH1:1, -II, -III samples; CV curves

22
of blank Ni foam; electrochemical performance of CNOH1:1-III without graphene;

microstructure stability of CNOH1:1-III; cycling performance of the CNOH2:1, CNOH1:1,

CNOH1:1-II; CV curves and rate performances of PPD/rGO and pure rGO; volumetric

energy/power density of the CNOH1:1-III//PPD/rGO device; EDX analysis of CoxNi1-x(OH)2-

rGO; N2 adsorption–desorption results of rGO and PPD/rGO; Rate performance and Coulombic

efficiency of the hybrid device; comparison of the cycling stability of the supercapacitors.

Author Contributions

B.T.Z. and M.L.L. conceived the project. B.T.Z. prepared the samples, electrodes, performed the

electrochemical experiments, and analyzed the data. D.C.C. carried out measurements and data

analyses of N2 adsorption-desorption and Raman. X.H.X., R.Z.H., and Y.C. collected and

analyzed the SEM data. B.S. contributed to the synthesis of GO and PPD/rGO. B.H.R., and

G.H.W. collected and analyzed the XPS and XRD data, respectively. D.X.Z. assisted B.T.Z. with

the synthesis of CoxNi1-x(OH)2-rGO. C.Q. and D.D. assisted B.T.Z. with the electrochemical

experiments. Y.D. carried out measurements and data analyses of TEM. B.T.Z., M.L.L., D.C.C.,

Q.B.Z., and C.P.W. contributed to the discussion of the results. B.T.Z. drafted the manuscript,

and all authors contributed to the final version of the manuscript.

23
ACKNOWLEDGMENT

This work was supported by the US Department of Energy ARPA-E Program under award

number DE-AR0000303 and by the US National Science Foundation under award number

DMR-1410320.

REFERENCES

[1] P. Simon, Y. Gogotsi, Nat. Mater., 7 (2008) 845-854.

[2] P. Simon, Y. Gogotsi, B. Dunn, Science, 343 (2014) 1210-1211.

[3] C. Zhong, Y.D. Deng, W.B. Hu, J.L. Qiao, L. Zhang, J.J. Zhang, Chem. Soc. Rev., 44 (2015)

7484-7539.

[4] Y.M. Sun, R.B. Sills, X.L. Hu, Z.W. Seh, X. Xiao, H.H. Xui, W. Luo, H.Y. Jin, Y. Xin, T.Q.

Li, Z.L. Zhang, J. Zhou, W. Cai, Y.H. Huang, Y. Cui, Nano Lett., 15 (2015) 3899-3906.

[5] G. Yu, L. Hu, M. Vosgueritchian, H. Wang, X. Xie, J.R. McDonough, X. Cui, Y. Cui, Z.

Bao, Nano Lett., 11 (2011) 2905-2911.

[6] Y. Zeng, Y. Han, Y. Zhao, Y. Zeng, M. Yu, Y. Liu, H. Tang, Y. Tong, X. Lu, Adv. Energy

Mater., 5 (2015) 1402176.

[7] Y.F. Ma, H.C. Chang, M. Zhang, Y.S. Chen, Adv. Mater., 27 (2015) 5296-5308.

[8] B.T. Zhao, R. Ran, M.L. Liu, Z.P. Shao, Mat. Sci. Eng., R., 98 (2015) 1–71.

[9] F. Bonaccorso, L. Colombo, G.H. Yu, M. Stoller, V. Tozzini, A.C. Ferrari, R.S. Ruoff, V.

Pellegrini, Science, 347 (2015) 1246501.

24
[10] M. Yang, Y.R. Zhong, J.J. Ren, X.L. Zhou, J.P. Wei, Z. Zhou, Adv. Energy Mater., 5

(2015) 1500550.

[11] E. Lim, C. Jo, H. Kim, M.H. Kim, Y. Mun, J. Chun, Y. Ye, J. Hwang, K.S. Ha, K.C. Roh,

K. Kang, S. Yoon, J. Lee, ACS Nano, 9 (2015) 7497-7505.

[12] Y. Cai, B.T. Zhao, J. Wang, Z.P. Shao, J. Power Sources, 253 (2014) 80-89.

[13] H. Kim, M.Y. Cho, M.H. Kim, K.Y. Park, H. Gwon, Y. Lee, K.C. Roh, K. Kang, Adv.

Energy Mater., 3 (2013) 1500-1506.

[14] A. Le Comte, T. Brousse, D. Belanger, J. Power Sources, 326 (2016) 702-710.

[15] F. Zhang, T.F. Zhang, X. Yang, L. Zhang, K. Leng, Y. Huang, Y.S. Chen, Energy Environ.

Sci., 6 (2013) 1623-1632.

[16] Y.G. Wang, D.D. Zhou, D. Zhao, M.Y. Hou, C.X. Wang, Y.Y. Xia, J. Electrochem. Soc.,

160 (2013) A98-A104.

[17] C.D. Gu, X. Ge, X.L. Wang, J.P. Tu, J. Mater. Chem. A, 3 (2015) 14228-14238.

[18] T.W. Lin, M.C. Hsiao, S.W. Chou, H.H. Shen, J.Y. Lin, J. Electrochem. Soc., 162 (2015)

A1493-A1499.

[19] T. Brousse, D. Belanger, J.W. Long, J. Electrochem. Soc., 162 (2015) A5185-A5189.

[20] W.G. Pell, B.E. Conway, J. Power Sources, 136 (2004) 334-345.

[21] Q. Abbas, P. Babuchowska, E. Frckowiak, F. Beguin, J. Power Sources, 326 (2016) 652-

659.

[22] H.B. Li, Y.Q. Gao, C.X. Wang, G.W. Yang, Adv. Energy Mater., 5 (2015) 1401767.

[23] S.Y. Kim, H.M. Jeong, J.H. Kwon, I.W. Ock, W.H. Suh, G.D. Stucky, J.K. Kang, Energy

Environ. Sci., 8 (2015) 188-194.

25
[24] J. Yan, Z.J. Fan, W. Sun, G.Q. Ning, T. Wei, Q. Zhang, R.F. Zhang, L.J. Zhi, F. Wei, Adv.

Funct. Mater., 22 (2012) 2632-2641.

[25] Y.W. Cheng, H.B. Zhang, C.V. Varanasi, J. Liu, Energy Environ. Sci., 6 (2013) 3314-3321.

[26] H.L. Wang, H.S. Casalongue, Y.Y. Liang, H.J. Dai, J. Am. Chem. Soc., 132 (2010) 7472-

7477.

[27] L. Huang, D.C. Chen, Y. Ding, S. Feng, Z.L. Wang, M.L. Liu, Nano Lett., 13 (2013) 3135-

3139.

[28] H. Chen, L.F. Hu, M. Chen, Y. Yan, L.M. Wu, Adv. Funct. Mater., 24 (2014) 934-942.

[29] V. Augustyn, P. Simon, B. Dunn, Energy Environ. Sci., 7 (2014) 1597-1614.

[30] Q. Mahmood, S.K. Park, K.D. Kwon, S.J. Chang, J.Y. Hong, G.Z. Shen, Y.M. Jung, T.J.

Park, S.W. Khang, W.S. Kim, J. Kong, H.S. Park, Adv. Energy Mater., 6 (2016) 1501115.

[31] Q. Mahmood, M.G. Kim, S. Yun, S.M. Bak, X.Q. Yang, H.S. Shin, W.S. Kim, P.V. Braun,

H.S. Park, Nano Lett., 15 (2015) 2269-2277.

[32] Y. Gogotsi, ACS Nano, 8 (2014) 5369-5371.

[33] C. Zhou, Y.W. Zhang, Y.Y. Li, J.P. Liu, Nano Lett., 13 (2013) 2078-2085.

[34] L.B. Qu, Y.L. Zhao, A.M. Khan, C.H. Han, K.M. Hercule, M.Y. Yan, X.Y. Liu, W. Chen,

D.D. Wang, Z.Y. Cai, W.W. Xu, K.N. Zhao, X.L. Zheng, L.Q. Mai, Nano Lett., 15 (2015) 2037-

2044.

[35] X.H. Xiong, D. Ding, D.C. Chen, G. Waller, Y.F. Bu, Z.X. Wang, M.L. Liu, Nano Energy,

11 (2015) 154-161.

[36] P.G. Campbell, M.D. Merrill, B.C. Wood, E. Montalvo, M.A. Worsley, T.F. Baumann, J.

Biener, J. Mater. Chem. A, 2 (2014) 17764-17770.

26
[37] Y. Zhao, J. Liu, Y. Hu, H.H. Cheng, C.G. Hu, C.C. Jiang, L. Jiang, A.Y. Cao, L.T. Qu, Adv.

Mater., 25 (2013) 591-595.

[38] X.T. Zhang, Z.Y. Sui, B. Xu, S.F. Yue, Y.J. Luo, W.C. Zhan, B. Liu, J. Mater. Chem., 21

(2011) 6494-6497.

[39] M.A. Worsley, P.J. Pauzauskie, T.Y. Olson, J. Biener, J.H. Satcher, T.F. Baumann, J. Am.

Chem. Soc., 132 (2010) 14067-14069.

[40] B. Song, L.Y. Li, Z.Y. Lin, Z.K. Wu, K.S. Moon, C.P. Wong, Nano Energy, 16 (2015) 470-

478.

[41] B. Song, C. Sizemore, L.Y. Li, X.G. Huang, Z.Y. Lin, K.S. Moon, C.P. Wong, J. Mater.

Chem. A, 3 (2015) 21789-21796.

[42] H.L. Wang, Y.Y. Liang, T. Mirfakhrai, Z. Chen, H.S. Casalongue, H.J. Dai, Nano Res., 4

(2011) 729-736.

[43] W.C. Jiang, D.S. Yu, Q. Zhang, K.L. Goh, L. Wei, Y.L. Yong, R.R. Jiang, J. Wei, Y. Chen,

Adv. Funct. Mater., 25 (2015) 1063-1073.

[44] L.F. Shen, L. Yu, X.Y. Yu, X.G. Zhang, X.W. Lou, Angew. Chem. Int. Ed., 54 (2015)

1868-1872.

[45] L. Shen, L. Yu, H.B. Wu, X.-Y. Yu, X. Zhang, X.W. Lou, Nat. Commun., 6 (2015) 6694.

[46] L.F. Shen, J. Wang, G.Y. Xu, H.S. Li, H. Dou, X.G. Zhang, Adv. Energy Mater., 5 (2015)

1400977.

[47] Y.F. Zhao, W. Ran, J. He, Y.Z. Huang, Z.F. Liu, W. Liu, Y.F. Tang, L. Zhang, D.W. Gao,

F.M. Gao, Small, 11 (2015) 1310-1319.

[48] X.L. Wu, L.L. Jiang, C.L. Long, T. Wei, Z.J. Fan, Adv. Funct. Mater., 25 (2015) 1648-

1655.

27
[49] H. Xia, C.Y. Hong, B. Li, B. Zhao, Z.X. Lin, M.B. Zheng, S.V. Savilov, S.M. Aldoshin,

Adv. Funct. Mater., 25 (2015) 627-635.

[50] X.N. Lu, L.Y. Li, B. Song, K. Moon, N. Hu, G.L. Liao, T.L. Shi, C.P. Wong, Nano Energy,

17 (2015) 160-170.

[51] Z.A. Hu, Y.L. Xie, Y.M. Wang, H.Y. Wu, Y.Y. Yang, Z.Y. Zhang, Electrochim. Acta, 54

(2009) 2737-2741.

[52] J.B. Liang, R.Z. Ma, N.B.O. Iyi, Y. Ebina, K. Takada, T. Sasaki, Chem. Mater., 22 (2010)

371-378.

[53] J.W. Lee, T. Ahn, D. Soundararajan, J.M. Ko, J.D. Kim, Chem. Commun., 47 (2011) 6305-

6307.

[54] J.F. Moulder, W.F. Sticlke, P.E. Sobol, K.D. Bomben, Handbook of X-ray Photoelectron

Spectroscopy, Perkin-Elmer Corporation, Eden Prairie, Minnesota, 1992.

[55] L.J. Xie, Z.A. Hu, C.X. Lv, G.H. Sun, J.L. Wang, Y.Q. Li, H.W. He, J. Wang, K.X. Li,

Electrochim. Acta, 78 (2012) 205-211.

[56] B.T. Zhao, Z.P. Shao, J. Phys. Chem. C, 116 (2012) 17440-17447.

[57] B.T. Zhao, S.M. Jiang, C. Su, R. Cai, R. Ran, M.O. Tade, Z.P. Shao, J. Mater. Chem. A, 1

(2013) 12310-12320.

28