You are on page 1of 10

www.afm-journal.

de
www.MaterialsViews.com

A New Approach to Tuning Carbon Ultramicropore Size at

full paper
Sub-Angstrom Level for Maximizing Specific Capacitance
and CO2 Uptake
Jin Zhou, Zhaohui Li, Wei Xing,* Honglong Shen, Xu Bi, Tingting Zhu, Zhipeng Qiu,
and Shuping Zhuo*

was obtained when the pore size of carbon


Ultramicroporous carbon materials with uniform pore size accurately materials matched the dimensions of
adjusted to the dimension of electrolyte ions or CO2 molecule are highly the electrolyte ion.[4] With regard to CO2
desirable for maximizing specific capacitance and CO2 uptake. However, capture applications, it has been widely
efficient ways to fine-tuning ultramicropore size at angstrom level are scarce. accepted that CO2 uptake at ambient
pressure is determined by the volume
A completely new approach to precisely tuning carbon ultramicropore size at
of ultramicropores rather than the total
sub-angstrom level is proposed herein. Due to the varying activating strength pore volume.[5,6] Presser et al. reported
and size of the alkali ions, the ultramicropore size can be finely tuned in the that most of the CO2 uptake capacity of
range of 0.60–0.76 nm as the activation ion varies from Li+ to Cs+. The car- carbide derived carbons at ambient pres-
bons prepared by direct pyrolysis of alkali salts of carboxylic phenolic resins sure to be originated by pores with a
diameter smaller than 0.8 nm.[7] Further
yield ultrahigh capacitances of up to 223 F g-1 (205 F cm-3) in ionic liquid
studies showed that CO2 uptake is lim-
electrolyte, and superior CO2 uptake of 5.20 mmol g-1 at 1.0 bar and 25 °C. ited by ultramicropores smaller than a
Such outstanding performance of the finely tuned carbons lies in its adjust- certain diameter at different pressures or
able pore size perfectly adapted to the electrolyte ions and CO2 molecule. This temperatures.[7,8] On the other hand, the
work paves the way for a new route to finely tuning ultramicropore size at the volumetric performance is of great impor-
sub-angstrom level in carbon materials. tance for practical application of carbon
materials in supercapacitors and CO2 cap-
ture. In this sense, high packing density of
the carbon materials is preferred. As large
1. Introduction pore size and broad pore size distribution are usually in contra-
diction with high packing density, uniform ultramicroporosity
Carbon materials have received considerable attention in super- are highly desirable for maximizing volumetric capacitance and
capacitors and CO2 capture.[1] Recent studies have shown that CO2 uptake for carbon materials.
ultramicropores (diameter <0.80 nm) play a key role in deter- As pointed out above, carbons with uniform and fine tai-
mining specific capacitance and CO2 uptake of carbon mate- lored ultramicropore size matching the dimensions of the
rials.[2–8] Chmiola et al. pointed out that micropores with a electrolyte ions or the CO2 molecule may result in materials
diameter smaller than 1.0 nm to be responsible for an anom- exhibiting optimal performance. These impressively show the
alous increase in specific capacitance in organic electrolyte importance of uniformly adjusting the ultramicropore size at
media.[3] It is suggested that desolvated ions rather than larger- sub-angstrom levels. Well-established methods such as chem-
sized solvated ionic species are stored in these ultramicropores. ical activation,[9,10] physical activation,[11] template methods,[12]
In the case of ionic liquid (IL) media, maximum capacitance and polymer carbonization[13] allow production of porous
carbon materials with developed porosity. These methods, how-
ever, typically lead to poor fine tuning of ultramicropore size
Dr. J. Zhou, Z. Li, H. Shen, X. Bi, T. Zhu, Z. Qiu, at sub-angstrom levels. The most successful synthesis strate-
Prof. S. Zhuo gies to control micropore size are based on the selective deal-
School of Chemical Engineering loying of metal carbides by chlorination, although chlorine gas
Shandong University of Technology is very toxic.[14] Hou et al. reported a hot-pressing method to
Zibo 255049, P. R. China
E-mail: zhuosp_academic@yahoo.com adjust the average pore size of zeolite-templated carbons from
Prof. W. Xing 0.7 to 1.0 nm at the expense of using ultrahigh pressures of up
School of Science to 147 MPa.[15] Cyclic oxidation/thermal desorption treatment
State Key Laboratory of Heavy Oil Processing allows gradual adjustment of average pore size, although the
China University of Petroleum time- and energy-consuming features of this process seriously
Qingdao 266580, P. R. China
E-mail: xingwei@upc.edu.cn
hinder large-scale carbon manufacturing.[16] In a word, efficient
ways to precise fine-tuning the ultramicropore size of carbons
DOI: 10.1002/adfm.201601904 are scarce. It is still very challenging to develop facile approach

Adv. Funct. Mater. 2016, 26, 7955–7964 © 2016 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim wileyonlinelibrary.com 7955
www.afm-journal.de
www.MaterialsViews.com

the nucleophilic addition reaction between phenols and for-


full paper

to tuning carbon ultramicropore size at sub-angstrom level


when keeping the pore uniformity. maldehyde leads to the formation of hydroxymethyl derivatives
Herein, we proposed a completely new strategy to finely that subsequently undergo intermolecular condensation (i.e.,
tune the pore size of ultramicroporous carbons using different dehydration) to yield methylene and methylene ether bridges.
alkali metal ions (Li+, Na+, K+, Rb+, Cs+) as activating agents. The formation of hydroxymethyl derivatives is very fast at
Ultramicroporous carbons with highly uniform pores and high the basic conditions used herein (pH ≈10), this resulting in a
packing density were prepared by direct carbonization of alkali large number of primary particles and thus small-sized resin
salts of carboxylic phenolic resins. Due to the varying activating gels (about 200 nm, Figure S1 (Supporting Information)).[17]
strength and size of alkali ions, the ultramicropore size can be The high surface tension generated during the hydrothermal
finely tuned in the range of 0.60–0.76 nm as the activation ion process causes the packing texture of gel nanoparticles to col-
varies from Li+ to Cs+. The resulting carbons present ultrahigh lapse, thereby resulting in compacting of gel particles.[18] The
capacitances of up to 223 F g−1 or 205 F cm−3 in ionic liquid resulting red hydrogels were dried to form dark red xerogels
electrolyte, and superior CO2 uptake of 5.20 mmol g−1 at 1.0 bar (Figure S2, Supporting Information), followed by carboniza-
and 25 °C. To the best of knowledge, these values are much tion at 900 °C (3 °C min−1) for 2 h in argon. The alkali metal
higher than most of the reported carbon materials. The out- ion-activated carbon materials were finally obtained by washing
standing performance of the finely tuned carbons was proved with diluted HCl and deionized water to neutrality. For conven-
to lie in its pore size perfectly adjusted to dimensions of the ience, the alkali salts of carboxylic phenolic resin and its corre-
ions composing the ionic liquids and the CO2 molecule. sponding carbon material are denoted as PR-COOM and MAC,
where PR and M stand for phenolic resin and alkali metal ion,
respectively.
2. Results and Discussion As illustrated by X-ray energy dispersive spectroscopy (EDS)
mapping (Figure S3, Supporting Information), the distribu-
2.1. Formation of the Ultramicroporous Carbon Materials tions of alkali ions are homogeneous. These monodispersed
ions as a form of COOM could produce a homogeneous “in
Figure 1 illustrates the preparation of the ultramicroporous situ activation” effect. To study the carbonization process and
carbon materials. In a typical synthesis, 2,4-dihydroxybenzoic activation mechanism, thermogravimetry-mass spectrometry
acid, alkali hydroxide (MOH), and formaldehyde were dis- (TG-MS) analysis and in situ X-ray diffraction (XRD) were per-
solved in deionized water to form a pale yellow homogeneous formed. Figure 2a–d shows the TG curves and MS responses
solution. Immediately, a complete neutral reaction will occur of potassium salt of carboxylic phenolic resin (PR-COOK)
between strongly basic MOH and carboxyl groups (COOH) and phenolic resin carboxylic acid (PR-COOH). Three gases,
of 2,4-dihydroxybenzoic acid to form carboxylates (COOM). including H2O (m/z 18), CO (m/z 28), and CO2 (m/z 44), were
This solution was then hydrothermally treated at 120 °C for detected by mass spectrometry. The little weight loss (<10 wt%)
24 h to promote polymerization of phenols with formaldehyde at T < 200 °C could be attributed to the evaporation of mois-
resulting in alkali salts of carboxylic phenolic resins. Initially, ture and decarboxylation confirmed by the release of H2O

Figure 1.  Schematic diagram of the synthesis of MAC.

7956 wileyonlinelibrary.com © 2016 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim Adv. Funct. Mater. 2016, 26, 7955–7964
www.afm-journal.de
www.MaterialsViews.com

full paper
Figure 2.  a) TG curves, MS responses of evolved gases in TG-MS analysis b) H2O, c) CO2, and d) CO, e) in situ XRD patterns (Cu Ka) during the
carbonization of PR-COOK, and f) possible activation mechanism.

and CO2 (Figure 2a–c). In the case of PR-COOH, a significant which are attributed to K (100) (PDF card 01-0500) and KC8
weight loss of 38.3 wt% in 200–800 °C is due to the continual (100) (PDF card 04-0221), and large amount of CO were
thermal decomposition of PR-COOH, and a negligible weight detected (Figure 2d). These facts indicate that K2O is reduced
loss <0.5 wt% at the last step between 800 and 1000 °C were by carbon to metallic potassium via the reaction of K2O + C →
observed, confirming stable carbon structure was formed 2K + CO, and partial carbon atoms were etched into CO to give
above 800 °C. By contrast, the pyrolysis process of PR-COOK rise to the porosity.[19] Meanwhile, potassium vapors may inter-
is much more complicated at the temperatures above 200 °C. calate between the carbon layers to form graphite intercalation-
The weight loss of about 9.6 wt% in 200–400 °C may be due like compounds (e.g., KC8 here) and cause swelling and the
to further cross-linking and initial thermal degradation of the disruption of the carbon microstructure, which thereby gener-
phenolic resin with the formation of K2CO3 and evolution of ates even more ultramicroporosity. Similar results of TG-MS
CO2 and H2O at about 350 °C. The weight loss in 400–600 °C analyses were obtained for the other alkali salts of carboxylic
could be partially attributed to the transformation of K2CO3 phenolic resins (Figure S4, Supporting Information), indi-
into K2O. The sharp weight loss (>20 wt%) above 800 °C indi- cating the carbonization process is essentially same for all the
cates that the carbon framework is severely etched to form the samples.
microporosity. Overall, based on the above observations, a homogeneous
Figure 2e presents the in situ XRD patterns at 200, 400, “in situ activation” process was presented in Figure 2f. That is:
600, and 800 °C during the pyrolysis of PR-COOK. It can be First, alkali salts of phenolic resins decompose into alkali car-
observed that the PR-COOK began to decompose into K2CO3 bonate (M2CO3), H2O, and CO2 below 400 °C. Second, alkali
(PDF card 71-1466) at around 200 °C. At 600 °C, the diffrac- oxides (M2O) are generated via the thermal decomposition of
tion peaks of K2CO3 are no longer visible and K2O (PDF card M2CO3 or the redox reaction between M2CO3 and C. Third,
26-1327) can be detected, indicating that most of K2CO3 has framework carbon atoms are etched by M2O via the vigorous
been transformed into K2O. We further designed a simple redox reaction of M2O + C → 2M + CO, leading to a significant
experiment to prove the decomposition of K2CO3. As shown weight loss, then the produced metallic alkali intercalates into
in Figure S5 (Supporting Information), no precipitates the lattices of the carbon matrix, which is responsible for both
were observed when the extract of carbonization residual of stabilization and widening of the interlayer spacing.[20] The
PR-COOK at 600 °C was dropped into BaCl2 solution, indi- interlayer spacing will be primarily determined by the size of
cating the absence of CO32− in the residual. This further proves metal intercalate, and this spacing (i.e., pore size) will be sys-
that most K2CO3 have been transformed into K2O at 600 °C. tematically widened with increasing metal ion size from Li+ to
Considering the evolution of CO2 and CO at about 520 °C, the Cs+ (Figure S6, Supporting Information).[21] After the removal
generation of K2O could be explained by the possible reactions of the intercalated metallic alkali and other alkali compounds
of K2CO3 → K2O + CO2 or K2CO3 + C → K2O + 2CO. When the by washing, the expanded carbon lattices cannot return to their
carbonization temperature was increased up to 800 °C, besides previous nonporous structure and thus achieved to finely tun-
the peaks of K2O, new peaks at 10.4° and 16.5° appeared, able ultramicropore size.

Adv. Funct. Mater. 2016, 26, 7955–7964 © 2016 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim wileyonlinelibrary.com 7957
www.afm-journal.de
www.MaterialsViews.com
full paper

Figure 3.  SEM images of: a,b) KAC and e,f) CsAC. TEM images of: c,d) KAC and g,h) CsAC.

2.2. Structural Characteristics and Chemical Properties of MAC <900 °C for other alkali metals) may be a reason. At 900 °C,
Materials the alkali metals except Li become metallic vapor with high
diffusivity, and could easily intercalate into the deep of carbon
Microscopic morphology and pore texture of MAC materials lattices, thus resulting in much more developed porosity. Even
were imaged by scanning electron microscopy (SEM) and though the dosage of activation agents used herein was much
transmission electron microscopy (TEM) techniques (Figure 3). lower compared to traditional chemical activation processes
As revealed by SEM images, the MAC materials were shown (30 wt% of resins as counted in KOH vs 100–400 wt%), the sur-
as compact blocks with a smooth surface and no clear pores. face area of KAC material (897 m2 g−1) was comparable to that
However, when observed by TEM, the as-prepared carbons of KOH-activated carbons.[10,22]
were found to possess abundant micropores without apparent Dubinin−Radushkevich (D–R) plots from CO2 adsorption
mesopore or macropore signs. High resolution TEM images data provide instructive information about the size of ultrami-
(Figure 3c,d,g,h and Figure S7 (Supporting Information)) cropores (diameter <0.8 nm). The slope from the linear D–R
clearly evidenced highly connected and worm-like micropores plots can be used to estimate micropore size and uniformity
within the MAC structures. These pore morphology resulted (see the Supporting Information for further details). All the
from the homogeneous “in situ activation” produced by highly MAC materials except LiAC exhibited well-defined linear D–R
dispersed (at atomic level) alkali ions. plots (correlation coefficient R2 > 0.995, Figure 4c–f). In the
The porosity properties of the MAC materials were further case of excellent linear fittings, Dubinin postulated the exist-
analyzed by N2 and CO2 adsorption at −196 and 0 °C, respec- ence of highly uniform ultramicropores.[6,23] LiAC showed a
tively. As shown in Figure 4a, all the MAC samples except D–R plot with two well-defined linear ranges, thereby indi-
RbAC presented a standard type I N2 adsorption isotherm with cating the narrow microporosity to be composed of two
a narrow knee at very low relative pressures (P/P0 < 0.02) and micropore systems (Figure S8, Supporting Information). Due
nearly unchanged adsorption amount at higher relative pres- to the weak basicity of LiOH, the neutral reaction between
sures. This shape is characteristic of porous materials showing 2,4-dihydroxybenzoic acid and LiOH is incomplete. The exist-
a narrow micropore size distribution and a minimal presence ence of unreacted LiOH leads to an uneven dispersion of Li+
of mesopores. As shown in Table 1, the SBET versus Smicro and in the phenolic resins, thus resulting in a wide pore size dis-
VT versus Vmicro values were very similar in all cases, further tribution of LiAC. Apparently, as the activation ions vary from
demonstrating the microporous nature of MAC materials. Li+ to Cs+, a systematic widening of micropores takes place,
Remarkably, the porosity of LiAC is much undeveloped, and the whose size gradually increases from 0.60 up to 0.76 nm. Thus,
pore volume and the apparent surface areas sharply increased N2 and CO2 adsorption data, along with SEM and TEM obser-
(from 0.07 to 0.70 cm3 g−1 and from 111 to 1312 m2 g−1, respec- vations revealed a purely uniform ultramicropore texture for
tively) (Table 1) while varying the activating ion from Li+ to Cs+, MAC samples. Remarkably, the ultramicropore size can be
thereby revealing activation strength significantly increasing finely and simply adjusted at sub-angstrom level by varying
from Li+ to Cs+. The observed trend in activating strengths of the alkali metal ion exchanged in the parent carboxylic phe-
alkali ions can be explained by the reactivity difference of M2O nolic resins. The pore size tuning resulted from the varying
and carbon. Gibbs free energies change (ΔG) calculated by Van’t activating strength and ion sizes of the alkali metals from Li+
Hoff equation (Table S1, Supporting Information), revealed to Cs+.
the reduction process to be more thermodynamically feasible The crystalline structure of MAC samples was characterized
at 900 °C with increasingly larger negative values (i.e., higher by XRD (Figure S9a, Supporting Information). XRD profiles
reactivity of M2O and carbon) from Li to Cs. Furthermore, the typical of activated carbon materials were obtained with two
difference of boiling points for alkali metals (1347 °C for Li, broad bands centered at 23.4° and 43° corresponding to the

7958 wileyonlinelibrary.com © 2016 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim Adv. Funct. Mater. 2016, 26, 7955–7964
www.afm-journal.de
www.MaterialsViews.com

full paper
Figure 4.  a) Nitrogen adsorption isotherms. b) CO2 adsorption isotherms. Analysis of the narrow microporosity by Dubinin−Radushkevich equation
of: c) NaAC, d) KAC, e) RbAC, and f) CsAC.

Table 1.  Textural properties of MAC samples. reflections of the (002) and (101) graphitic planes, respectively.
The intensities of these bands decreased from Li+ to Cs+, in
N2 sorption CO2 sorption agreement with the higher activating strength of Cs+ that leads
SBETa) Smicrob) VTc) Vmicrod) V0e) Lf) to a carbonaceous structure with more defects. Raman spectra
Sample [m2 g−1] [m2 g−1] [cm3 g−1] [cm3 g−1] [cm3 g−1] [nm] (Figure S9b, Supporting Information) showed well-resolved G
LiAC 111 93 0.07 0.05 0.05 0.60/1.17 and D bands centered at 1589 and 1340 cm−1, corresponding
NaAC 668 621 0.35 0.32 0.36 0.63 to the reflections of the ideal and the disordered graphitic lat-
tice, respectively. The low value of IG/ID (about 0.8) further
KAC 897 816 0.49 0.42 0.42 0.66
revealed the amorphous nature of MAC materials. X-ray pho-
RbAC 1243 1073 0.70 0.55 0.51 0.75 toelectron spectroscopy (XPS) measurements were conducted
CsAC 1312 1239 0.67 0.63 0.53 0.76 to determine the surface chemical properties of MAC mate-
a)Brunauer–Emmett–Teller surface area; b)Micropore surface area calculated by
rials (Figure S10, Supporting Information). Only C and O
the t-plot method; c)Total pore volume; d)Micropore volume calculated by the
species were detected, thereby revealing complete removal of
t-plot method; e)Micropore volume given by CO2 adsorption; f)Average micropore surface metal species by thorough washing with diluted HCl
size. and deionized water. The oxygen contents of MAC materials

Adv. Funct. Mater. 2016, 26, 7955–7964 © 2016 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim wileyonlinelibrary.com 7959
www.afm-journal.de
www.MaterialsViews.com

(M = Li, Na, K, Rb, Cs) were determined to be 12.3, 5.5, 5.2,


full paper

2.3. Capacitive Performance of MAC Materials


6.1, and 8.2 at%, respectively. It could be seen that LiAC pos-
sesses much higher oxygen content than the others. The cause Capacitive performance of MAC materials was studied both
is not clear exactly. One possible reason is the difference of in 6 m KOH and pure IL of 1-ethyl-3-methylimidazolium
alkali metals in boiling point (1347 °C for Li, <900 °C for other tetrafluoroborate (EMImBF4) electrolytes (Figure 5). Nearly
alkali metals). At high temperatures (e.g., 900 °C), the alkali rectangular-like cyclic voltammetry (CV) curves, characteristic
metals except Li become metallic vapor with high diffusivity, of double-layer capacitance, were obtained in all cases. To fur-
and could easily contact with carbon surface and efficiently ther investigate the electrochemical performance of the MAC
remove surface oxygen. This explains why surface oxygen con- materials, galvanostatic charge/discharge measurements were
tent of LiAC is apparently higher than the others. High reso- carried out (Figure S12, Supporting Information). At 0.2 A g−1
lution XPS was used to investigate the atom binding states of and in KOH, the specific capacitances were measured to be 43,
the prepared carbons (Figure S11, Supporting Information). In 140, 205, 200, and 221 F g−1 for LiAC, NaAC, KAC, RbAC, and
the case of C-species, the peaks at 284.7, 285.4, and 286.4 eV CsAC, respectively. At 0.2 A g−1 and in EMImBF4, those were
are assigned to sp2 hybridized carbon, carbon atoms single or 48, 97, 128, 189, and 223 F g−1 for LiAC, NaAC, KAC, RbAC,
double bonded to oxygen, respectively. In the case of O-spe- and CsAC, respectively. The specific capacitance was plotted
cies, three types of O-containing groups could be verified on versus discharge current density (Figures 5c,d). Clearly, CsAC
the surface of the prepared carbons, including CO, OCO, showed the highest specific capacitance among the MAC sam-
and OCO, corresponding to the peaks at 531.2, 532.3, and ples in the entire range of current densities. CsAC showed
533.5 eV, respectively.[24] ultrahigh specific capacitance (223 F g−1) at 0.2 A g−1 in ionic

Figure 5.  Capacitive performance: a) CV curves at 10 mV s−1 in KOH electrolyte; b) CV curves at 5 mV s−1 in EMImBF4; c) specific capacitances in KOH
electrolyte; and d) specific capacitance in EMImBF4; e) areal capacitance at 0.2 A g−1; f) capacitance ratio.

7960 wileyonlinelibrary.com © 2016 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim Adv. Funct. Mater. 2016, 26, 7955–7964
www.afm-journal.de
www.MaterialsViews.com

liquid electrolyte, and this value is one of the highest among 32.5 and 6.2 Wh kg−1 for CsAC in IL and KOH electrolytes,

full paper
carbon electrode materials (Table S2, Supporting Information). respectively. The cycling stability of CsAC against charge–dis-
When evaluating the potential of a porous carbon for superca- charge cycles was investigated at 1 A g−1 in KOH (Figure S13d,
pacitor applications (especially in portable electronic devices), Supporting Information). Remarkably, the specific capacitance
the specific volumetric capacitance is a more reliable parameter of the electrode was maintained at ≈191 F g−1 (nearly 100%
as compared to the gravimetric one. CsAC (packing density of capacitance retention), even after 10 000 cycles.
ca 0.92 cm3 g−1) showed very high volumetric capacitances of
203 and 205 F cm−3 in KOH and EMImBF4, respectively.
The ultrahigh capacitance properties can be better under- 2.4. CO2 Sorption Performance of MAC Materials
stood by studying the relationship between pore and ion size.
Results of the areal capacitance and retention ratio as a func- CO2 capture has attracted considerable attention in recent
tion of MAC material clearly pointed out a pore size effect years as CO2 is the main anthropogenic contributor to climate
(Figure 5e, please note that LiAC was not included because of change. In consideration of the microporous characteristics of
its broad pore size distribution[25]). A maximum in specific areal MAC, we carried out CO2 adsorption measurements at 1.0 and
capacitance is reached by KAC in KOH electrolyte, with signifi- 20 bar (Figure 6). As reported in Table S3 (Supporting Infor-
cantly lower values obtained for the rest of carbon materials. In mation), CsAC showed very high CO2 uptake values at 1.0 bar
contrast, the specific areal capacitance in EMImBF4 was found (6.77 and 5.20 mmol g−1 at 0 and 25 °C, respectively) well above
to monotonously increase with pore size, with CsAC showing than those reported for carbon materials under identical condi-
the maximum normalized capacitance. Even though KAC and tions (Table S4, Supporting Information) and higher than those
NaAC have very fine difference in pore sizes (≈0.66 vs 0.63 nm), of metal-organic frameworks (MOFs)[28] and covalent organic
the specific areal capacitance of the latter in KOH electro- frameworks (COFs)[29] that are also considered as good CO2
lyte is much larger than that of NaAC. These results can be adsorbents. Figure 6c shows the CO2 adsorption isotherms of
explained in terms of pore size and electrolyte ion size (EMIm+: the MAC materials at 25 °C and 0–20 bar. RbAC exhibits the
0.76 nm, BF4−: 0.45 nm, solvated K+: 0.31 nm, and solvated highest CO2 capacity, i.e., 9.63 mmol g−1 (42.4 wt%) at 25 °C
OH−: 0.35 nm). According to the equation of C = εA/d, the and 20 bar. The trend of CO2 uptakes at 20 bar is in good agree-
capacitance (C) of a supercapacitor electrode mainly depends on ment with that of total pore volume. As shown in Table S5 (Sup-
the electrode specific surface area (A), which must be as high as porting Information), the CO2 uptakes of CsAC are much less
possible, and the distance (d) between the adsorbed ions and at high pressure, but much higher at low pressure than those of
the electrode surface which should be minimized. In this sense, the carbon materials previously reported. This can be explained
most efficient ion adsorption is achieved in the pores perfectly by the highly exclusive ultramicroporosity but low total pore
adapted to the ion size. Theoretical studies have revealed that volumes (<1.0 cm3 g−1) of MAC materials because the CO2
the capacitance shows an oscillatory behavior as a function uptake at low pressure is determined by the volume of ultra-
of nanopore size in a theoretical slit pore.[26] The capacitance micropores,[7] and the high-pressure CO2 capacity largely cor-
versus pore size was found to exhibit two peaks located at the relates to the volumes of micropores and narrow mesopores.[30]
diameters equal to ion size or twice of that, respectively. CsAC Therefore, the prepared MAC materials are more suitable to
possesses a uniform pore size (0.76 nm) that perfectly matches CO2 sorption at low pressure.
with that of EMIm+, thus leading to efficient ion adsorption For large-scale CO2 capture applications, an adsorbent should
and maximum specific areal capacitance.[4] The pore size of ideally have high CO2/N2 selectivity, good stability during repet-
KAC is approximately twice that of the KOH ions (especially itive adsorption–desorption cycles, and low regeneration energy.
solvated OH− in distortion). Since ions can be adsorbed on both CsAC showed promising CO2 uptakes (1.52 mmol g−1) at typical
pore walls in KOH electrolyte, there may be a contribution to flue gas conditions (25 °C and P(CO2): 0.15 bar) while main-
the capacitance of KAC from both compact layers of ions. This taining low N2 uptakes (0.61 mmol g−1, <1/8 of the CO2 uptake)
result is also in good agreement with the experimental results at 25 °C and 1.0 bar (Figure 6c). Additionally, CsAC showed a
by Béguin and co-workers.[27] CO2/N2 selectivity (calculated from the ratio of the initial slopes
CsAC showed standard rectangular-like CV curves in KOH of N2 and CO2 adsorption isotherms) of 18 (Figure S14a, Sup-
electrolyte at a very high scan rate of 500 mV s−1 (Figure S13a, porting Information),[31,32] this value being comparable (or even
Supporting Information), and preserved 73% of its specific superior) than that of some nitrogen-doped carbons.[32,33] With
capacitance (162 F g−1) after increasing current density to the aim to test the recyclability of CsAC, four consecutive CO2
20 A g−1. These results reveal excellent power performance, adsorption cycles were carried out at 25 °C, and nearly identical
being indicative of the good characteristics of the carbon pore isotherms were obtained (Figure S14b, Supporting Informa-
system (i.e., high uniformity and short pore length) designed tion). With the aim to investigate the energy needed for regen-
herein. The as-prepared carbons showed much higher capaci- eration, we calculated the isosteric heat of adsorption (Qst) by
tance retention ratio in KOH than in IL (Figure 5f), and this applying the Clausius–Clapeyron equation to the CO2 adsorp-
may be ascribed to the high conductivity, low viscosity, and tion isotherm data at 0 and 25 °C. As shown in Figure 6e, CsAC
small ion size of the KOH electrolyte. The retention ratio also showed a relative low Qst (17–21 kJ mol−1) that envisages low
showed a pore size effect. The capacitance retention of MAC energy consumption during regeneration. We believe that the
increases with pore size being attributed to the fast diffusion of presence of a well-developed ultramicropore network with
ions in larger pores. Ragone plots (Figure S13b,c, Supporting narrow pore size distribution (diameter ≈0.76 nm) in CsAC
Information) revealed remarkable energy densities values of results in a high utilization degree of the overall porosity and

Adv. Funct. Mater. 2016, 26, 7955–7964 © 2016 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim wileyonlinelibrary.com 7961
www.afm-journal.de
www.MaterialsViews.com
full paper

Figure 6.  a) CO2 adsorption isotherms at 0 °C and 1.0 bar. b) CO2 adsorption isotherms at 25 °C and 1.0 bar. c) CO2 adsorption isotherms at 25 °C
and 20 bar. d) CO2 and N2 adsorption isotherms on CsAC at 25 °C. e) CO2 heats of adsorption on MACs. f) Water vapor adsorption isotherms at 25 °C.

surface area of the carbon material. These characteristics are of water vapor at very low relative pressures (7.7 mmol g−1
responsible for excellent CO2 adsorption performance at room at 10 mbar).[34] As long as the partial pressure of H2O is kept
temperature and ambient pressure. below 10 mbar, H2O adsorption may not be expected to inter-
In practice, flue gas is a mixture of mostly N2, CO2, and fere with CO2 adsorption. Over 10 mbar, a sharp increase in the
H2O, while its exact composition depends on the design of amount of H2O adsorbed was observed, and then the presence
the power plant and the source of natural gas or coal. H2O is of water in the pores will lead to a decrease of the accessible
strongly adsorbed on many adsorbents (e.g., zeolites), and thus, volume causing a decrease of CO2 adsorption capacity. It is also
its presence can cause a reduction in the adsorption capacity important to note that the partial pressure of H2O in flue gases
of CO2. Figure 6f represents the water vapor adsorption iso- is always over 10 mbar, especially in those after wet desulphuri-
therms over MAC materials at 25 °C. Although the capacity of zation. So the flue gases would likely need to be dried to below
water vapor (20.54 mmol g−1 for CsAC) is very high compared 10 mbar for H2O adsorption to remain negligible. Another con-
to that of CO2 (5.20 mmol g−1), it has to be pointed that there is siderable method is appropriately controlling the adsorption
negligible H2O adsorption on hydrophobic carbon pore surface time for CO2 sorption to maximize the concentration of CO2
until pressures above 10 mbar (corresponding to a relative pres- in the adsorbents due to the difference of diffusivity between
sure of about 0.33). In contrast, zeolite 13X, intensively studied CO2 and H2O.[35] The higher diffusivity of CO2 will facilitate the
for CO2 capture, presents much higher adsorption capacity kinetic separation of CO2 from H2O.

7962 wileyonlinelibrary.com © 2016 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim Adv. Funct. Mater. 2016, 26, 7955–7964
www.afm-journal.de
www.MaterialsViews.com

3. Conclusion mass of the active materials was 2.0 mg for each electrode. The model

full paper
capacitors were assembled in a glove box filled with argon (Mikrouna
A simple ultramicropore size tuning method for carbon mate- Universal 2440, H2O and O2 <1 ppm) by facing two electrodes
sandwiched with a fiber separator for the ionic liquid electrolyte of
rials was proposed herein. The pore size can be finely tuned in EMImBF4. The electrochemical measurements were carried on a
the range of 0.60–0.76 nm. The pore size tuning resulted from CHI660D electrochemical testing station (Chenhua Instruments Co.
the varying activating strength and ion sizes of the alkali metals Ltd., Shanghai). The specific capacitance was calculated by the following
from Li+ to Cs+. The carbon materials prepared by direct pyrol- equation
ysis of alkali salts of carboxylic phenolic resins showed highly
developed and uniform short length ultramicroporosity. The 4I × t
Cm = (1)
prepared carbon materials showed ultrahigh specific capaci- ∆V × m
tances of 205 F cm−3 in ionic liquid electrolyte as a result of a where Cm (F g−1) was the gravimetric specific capacitance of the carbon
perfect match in pore and ion sizes. The CsAC material showed samples, I (A) was the discharge current, t (s) was the time spent in
superior CO2 adsorption capacity of 5.20 mmol g−1 at 25 °C discharge, ΔV (V) was the potential window of the cell, and m (g) was
and 1.0 bar, high CO2-to-N2 selectivities and full regenerability the total mass of active materials in the cell.
Energy density (E) and powder density (P) could be calculated from
for four consecutive cycles. This work paves the way for a new
the galvanostatic charge/discharge test using the equations of
route to finely tuning ultramicropores size at the sub-angstrom
level in carbon materials. It is anticipated that these intriguing 1
carbon materials might also find applications as molecular E= × Cm × V 2 (2)
8
sieves, catalysts, battery electrodes, and water/air filters.
E
P= (3)
t
4. Experimental Section where the Cm, V, and t were the gravimetric specific capacitance of the
Preparation of MAC Materials: In a typical synthesis, carbon materials, the voltage of a supercapacitor cell before discharge,
2,4-dihydroxybenzoic acid (0.5 g, 3.2 mmol), alkali hydroxide (3.5 mmol), and the time spent in discharge, respectively.
and formaldehyde (0.53 g, 37 wt%, 0.64 mmol) were dissolved into
6 mL H2O, and stirred for 1 h to form a pale yellow clear solution. The
solution was then solidified at 120 °C for 24 h. The resulting solid was
dried at 100 °C overnight and carbonized at 900 °C for 2 h with a heating
Supporting Information
ramp of 3 °C min−1 in argon. Finally, the alkali metal ion-activated carbon Supporting Information is available from the Wiley Online Library or
materials were liberated by washing with diluted HCl and deionized from the author.
water to neutrality.
Characterization of MAC Materials: Microscopic morphology of the
prepared carbon materials were observed with a scanning electron
microscope (Sirion 200 FEI Netherlands) and a transmission electron Acknowledgements
microscope (JEM2100, JEOL, Japan). Surface chemical properties were
determined by X-ray energy dispersive spectroscopy (EDS, BRUKER The authors are grateful for the financial supports by National Natural
AXS) and X-ray photoelectron spectroscopy (XPS, Escalab 250, USA). Science Foundation of China (NSFC51302156, 21476264, 21576158 and
XRD patterns of the carbon materials were conducted on a Bruker D8 21576159) and Distinguished Young Scientist Foundation of Shandong
Advance diffractometer with Cu Kα radiation. In situ XRD patterns Province (JQ201215).
during the pyrolysis of PR-COOK were collected on a PANalytical X’Pert
Pro system operated at 40 kV and 40 mA current with Cu Kα source Received: April 16, 2016
under N2. TG-MS studies were performed on a thermogravimetric Revised: July 11, 2016
analyzer (NETZSCH STA 409 thermobalance) coupled to Balzers MID Published online: August 29, 2016
mass spectrometer (heating rate 20 °C min−1, helium flow). Raman
spectra were collected by a LabRAM HR800 from JY Horiba. N2
adsorption–desorption isotherms were measured at liquid nitrogen
temperature (−196 °C) and CO2 adsorption was performed at 0 °C [1] a) Y. Zhu, S. Murali, M. D. Stoller, K. J. Ganesh, W. Cai, P. J. Ferreira,
using a surface area and porosity analyzer (ASAP2020M, Micromeritics, A. Pirkle, R. M. Wallace, K. A. Cychosz, M. Thommes, D. Su,
USA). N2 and CO2 gases with super high purity (99.999%) were used for E. A. Stach, R. S. Ruoff, Science 2011, 332, 1537; b) F. Beguin,
the physisorption measurements. The carbon samples were degassed V. Presser, A. Balducci, E. Frackowiak, Adv. Mater. 2014, 26, 2219;
at 350 °C for 6 h under turbomolecular vacuum before sorption c) L. Borchardt, M. Oschatz, S. Kaskel, Mater. Horiz. 2014, 1, 157;
measurements. Brunauer–Emmett–Teller (BET) surface area was d) A.-H. Lu, G.-P. Hao, Annu. Rep. Prog. Chem., Sect. A: Inorg.
calculated using the N2 adsorption isotherm data within the relative
Chem. 2013, 109, 484; e) Y. Xia, R. Mokaya, G. S. Walker, Y. Zhu,
pressure ranging from 0.05 to 0.25. Total pore volume was obtained at a
Adv. Energy Mater. 2011, 1, 678.
relative pressure of 0.995. Micropore surface area and micropore volume
[2] P. Simon, Y. Gogotsi, Acc. Chem. Res. 2012, 46, 1094.
were calculated by the t-plot method. CO2 sorption of MAC materials at
low pressure (0–1.0 bar) were determined by a static adsorption method [3] J. Chmiola, G. Yushin, Y. Gogotsi, C. Portet, P. Simon, P. L. Taberna,
carried on the ASAP2020M analyzer. The measurements of high- Science 2006, 313, 1760.
pressure CO2 adsorption and water vapor adsorption were performed on [4] C. Largeot, C. Portet, J. Chmiola, P.-L. Taberna, Y. Gogotsi,
an intelligent gravimetric analyzer. P. Simon, J. Am. Chem. Soc. 2008, 130, 2730.
Fabrication of Electrodes and Electrochemical Measurements: Working [5] a) N. P. Wickramaratne, M. Jaroniec, J. Mater. Chem. A 2013, 1, 112;
electrodes were prepared by mixing 95 wt% carbon materials and b) G.-P. Hao, Z.-Y. Jin, Q. Sun, X.-Q. Zhang, J.-T. Zhang, A.-H. Lu,
5 wt% polytetrafluoroethylene (PTFE) binders, pressing the mixture Energy Environ. Sci. 2013, 6, 3740; c) H. S. Kim, M. S. Kang,
onto nickel foam at 15 MPa, and then drying at 120 °C for 10 h. The W. C. Yoo, J. Phys. Chem. C 2015, 119, 28512.

Adv. Funct. Mater. 2016, 26, 7955–7964 © 2016 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim wileyonlinelibrary.com 7963
www.afm-journal.de
www.MaterialsViews.com
full paper

[6] M. Sevilla, J. B. Parra, A. B. Fuertes, ACS Appl. Mater. Interfaces [21] Y. Takahashi, Tanso 1993, 160, 301.
2013, 5, 6360. [22] a) Y. Lv, F. Zhang, Y. Dou, Y. Zhai, J. Wang, H. Liu, Y. Xia, B. Tu,
[7] V. Presser, J. McDonough, S.-H. Yeon, Y. Gogotsi, Energy Environ. D. Zhao, J. Mater. Chem. 2012, 22, 93; b) G. A. Ferrero, A. B. Fuertes,
Sci. 2011, 4, 3059. M. Sevilla, J. Mater. Chem. A 2015, 3, 2914.
[8] Z. Zhang, J. Zhou, W. Xing, Q. Xue, Z. Yan, S. Zhuo, S. Z. Qiao, [23] M. Dubinin, Carbon 1989, 27, 457.
Phys. Chem. Chem. Phys. 2013, 15, 2523. [24] J. Figueiredo, M. Pereira, M. Freitas, J. Orfao, Carbon 1999, 37,
[9] H. Wang, Q. Gao, J. Hu, J. Am. Chem. Soc. 2009, 131, 7016. 1379.
[10] a) P. Gong, Z. Wang, J. Wang, H. Wang, Z. Li, Z. Fan, Y. Xu, X. Han, [25] a) T. A. Centeno, O. Sereda, F. Stoeckli, Phys. Chem. Chem. Phys.
S. Yang, J. Mater. Chem. 2012, 22, 16950; b) P. Cheng, S. Gao, 2011, 13, 12403; b) S. Kondrat, C. R. Perez, V. Presser, Y. Gogotsi,
P. Zang, X. Yang, Y. Bai, H. Xu, Z. Liu, Z. Lei, Carbon 2015, 93, 315. A. A. Kornyshev, Energy Environ. Sci. 2012, 5, 6474.
[11] a) C. H. Kim, J.-H. Wee, Y. A. Kim, K. Seung Yang, C.-M. Yang, [26] G. Feng, P. T. Cummings, J. Phys. Chem. Lett. 2011, 2, 2859.
J. Mater. Chem. A 2016, 4, 4763; b) T. Tooming, T. Thomberg, [27] E. Raymundo-Piñero, K. Kierzek, J. Machnikowski, F. Béguin,
H. Kurig, A. Jänes, E. Lust, J. Power Sources 2015, 280, 667. Carbon 2006, 44, 2498.
[12] a) Z. Yang, Y. Xia, X. Sun, R. Mokaya, J. Phys. Chem. B 2006, 110, [28] S. Keskin, T. M. van Heest, D. S. Sholl, ChemSusChem 2010, 3, 879.
18424; b) H. Itoi, H. Nishihara, T. Kogure, T. Kyotani, J. Am. Soc. [29] L. Stegbauer, M. W. Hahn, A. Jentys, G. Savasci, C. Ochsenfeld,
Chem. 2011, 133, 1165. J. A. Lercher, B. V. Lotsch, Chem. Mater. 2015, 27, 7874.
[13] B. Xu, S. Hou, H. Duan, G. Cao, M. Chu, Y. Yang, J. Power Sources [30] a) G. Srinivas, V. Krungleviciute, Z.-X. Guo, T. Yildirim, Energy
2013, 228, 193. Environ. Sci. 2014, 7, 335; b) B. Ashourirad, A. K. Sekizkardes,
[14] Y. Gogotsi, A. Nikitin, H. Ye, W. Zhou, J. E. Fischer, B. Yi, H. C. Foley, S. Altarawneh, H. M. El-Kaderi, Chem. Mater. 2015, 27, 1349.
M. W. Barsoum, Nat. Mater. 2003, 2, 591. [31] R. Banerjee, H. Furukawa, D. Britt, C. Knobler, M. O’Keeffe,
[15] P.-X. Hou, H. Orikasa, H. Itoi, H. Nishihara, T. Kyotani, Carbon O. M. Yaghi, J. Am. Chem. Soc. 2009, 131, 3875.
2007, 45, 2011. [32] G.-P. Hao, W.-C. Li, D. Qian, G.-H. Wang, W.-P. Zhang, T. Zhang,
[16] a) X. Py, A. Guillot, B. Cagnon, Carbon 2003, 41, 1533; b) R. Mysyk, A.-Q. Wang, F. Schüth, H.-J. Bongard, A.-H. Lu, J. Am. Chem. Soc.
Q. Gao, E. Raymundo-Pinero, F. Béguin, Carbon 2012, 50, 3367. 2011, 133, 11378.
[17] a) L. Zubizarreta, A. Arenillas, A. Domínguez, J. Menéndez, J. Pis, [33] a) J. Wang, I. Senkovska, M. Oschatz, M. R. Lohe, L. Borchardt,
J. Non-Cryst. Solids 2008, 354, 817; b) G. Zhou, H. Tian, H. Sun, A. Heerwig, Q. Liu, S. Kaskel, J. Mater. Chem. A 2013, 1, 10951;
S. Wang, C. E. Buckley, Chem. Eng. J. 2011, 171, 1399. b) J. Chen, J. Yang, G. Hu, X. Hu, Z. Li, S. Shen, M. Radosz, M. Fan,
[18] J. Zhou, Z. Zhang, Z. Li, T. Zhu, S. Zhuo, RSC Adv. 2015, 5, 46947. ACS Sustainable Chem. Eng. 2016, 4, 1439.
[19] J. Dıaz-Terán, D. Nevskaia, J. Fierro, A. Lopez-Peinado, A. Jerez, [34] J.-H. Kim, C.-H. Lee, W.-S. Kim, J.-S. Lee, J.-T. Kim, J.-K. Suh,
Microporous Mesoporous Mater. 2003, 60, 173. J.-M. Lee, J. Chem. Eng. Data 2003, 48, 137.
[20] J. Romanos, M. Beckner, T. Rash, L. Firlej, B. Kuchta, P. Yu, [35] M. G Plaza, A. S. González, F. Rubiera, C. Pevida, J. Chem. Technol.
G. Suppes, C. Wexler, P. Pfeifer, Nanotechnology 2012, 23, 015401. Biotechnol. 2015, 90, 1592.

7964 wileyonlinelibrary.com © 2016 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim Adv. Funct. Mater. 2016, 26, 7955–7964

You might also like