Understanding the desulphurization process in an ionic porous aromatic framework

Article abstract of DOI:10.1039/c8sc03727b

An ionic porous aromatic framework, iPAF-1, was successfully synthesized from a designed monomer with imidazolium functional groups. The iPAF-1 exhibits the highest dibenzothiophene uptake among all reported adsorptive desulphurization adsorbents. The so-

Full text of DOI:10.1039/c8sc03727b

Chemical
Science
View Article Online
View Journal | View Issue
EDGE ARTICLE
Understanding the desulphurization process in an
ionic porous aromatic framework†
Cite this: Chem. Sci., 2019, 10, 606
Yuyang Tian, ‡a Jian Song,‡a Youliang Zhu,b Huanyu Zhao,c Faheem Muhammad,a
All publication charges for this article
have been paid for by the Royal Society
of Chemistry
a
Tingting Ma,a Mo Chena and Guangshan Zhu
*
An ionic porous aromatic framework, iPAF-1, was successfully synthesized from a designed monomer with
imidazolium functional groups. The iPAF-1 exhibits the highest dibenzothiophene uptake among all
reported adsorptive desulphurization adsorbents. The so-called precursor designed synthetic route
provides the stoichiometric and homogeneous introduction of desired functional groups into the
framework. Molecular dynamics simulation was performed to understand the structure and the
desulphurization process within the amorphous iPAF-1. The insight into the key role of the moderate
bonding interaction between the adsorbate and the functional groups of iPAF-1 for improved uptake is
highlighted in this work.
Received 21st August 2018
Accepted 19th October 2018
DOI: 10.1039/c8sc03727b
rsc.li/chemical-science
Activated carbons have been extensively reported for
adsorptive desulphurization of fuel oils. Furthermore, zeolites
Introduction
and MOFs as functional porous adsorbent materials have been
explored for use in desulphurization, with reported adsorption
capacities of about 8–50 mg gꢀ1 in the form of S uptake.10
However, they suffer from either relatively low sulphur
compound uptake or low chemical stability. Therefore, the
design of new and effective adsorbents is highly demanded.
Porous organic frameworks (POFs)11 with sub-class materials,
including porous aromatic frameworks (PAFs),12–16 conjugated
microporous polymers (CMPs),17,18 covalent organic frameworks
(COFs),19–21 covalent triazine frameworks22 and hyper-
crosslinked polymers,23 possess large specic surface areas
and nanosized pores. Therefore, they are intensively studied as
new adsorbents for various molecules.24,25 A representative
material is the rst porous aromatic framework named PAF-1,
which was synthesized via cross-coupling of tetrakis(4-
bromophenyl)methane through the Yamamoto-type Ullmann
reaction.12 Due to the rigidity of the tetrahedral building blocks
and the high stability of the C–C bonds, PAF-1 possesses
a specic BET surface area of 5600 m2 gꢀ1 and exhibits high
uptake capacities for hydrogen, carbon dioxide and benzene.
Besides their practical application as effective adsorbents, the
fundamental investigations of adsorption in PAFs are particularly
important for further rational design of new functional adsor-
bents. Compared to activated carbons and zeolites, certain
chemical functionalities can be introduced into POFs via func-
tional group modication or structural engineering.26,27 For
example, imidazolium salts are exclusively used for desulphuri-
zation of aromatic sulphur compounds from transport oils
because of the affinity between electron-rich thiophene-based
compounds and electrophilic imidazolium compounds.28 On the
other hand, PAFs are entirely composed of organic building units,
Porous materials are extensively applied for removal of
contaminants in water and air as primary adsorbents.1–4 Classic
porous materials including activated carbons, zeolites, and
metal organic frameworks (MOFs) have attracted wide attention
as valuable adsorbents for various molecules that are of envi-
ronmental concern, such as carbon dioxide, ammonia, volatile
organic compounds and heavy metals. Sulphur emission during
fossil fuel combustion has led to environmental pollution, fog
and haze around the world. Thus, it is necessary to control the
growing environmental issues, particularly the automobile
exhaust emission pollutants in the air.5 Due to their harmful
effects, many countries mandate the reduction of sulphur
compounds in fuel oils below 10 ppm and anticipate sulphur-
free gasoline and diesel in the future.6 Therefore, desulphuri-
zation from fuel oils has become a vital research topic.
Numerous processes have been developed to remove sulphur
compounds from fuel oils.7,8 Among them, adsorptive desul-
phurization is promising, due to its mild operating conditions,
low energy consumption and effective removal of aromatic
sulphur pollutants.9
aKey Laboratory of Polyoxometalate Science of the Ministry of Education, Faculty of
Chemistry, Northeast Normal University, Changchun 130024, P. R. China. E-mail:
zhugs100@nenu.edu.cn
bState Key Laboratory of Polymer Physics and Chemistry, Changchun Institute of
Applied Chemistry, Chinese Academy of Sciences, Changchun 130022, P. R. China
cLaboratory of Theoretical and Computational Chemistry, Institute of Theoretical
Chemistry, Jilin University, Changchun 130023, P. R. China
† Electronic supplementary information (ESI) available: Synthetic details and
spectra of all synthesized compounds. DBT adsorption computational details.
See DOI: 10.1039/c8sc03727b
‡ These authors contributed equally to this work.
606 | Chem. Sci., 2019, 10, 606–613
This journal is © The Royal Society of Chemistry 2019

View Article Online
Edge Article
Chemical Science
and this simplies the studies on the adsorption and diffusion the two necks was tted with a rubber septum and the other one
behaviors of guest molecules in the channels. Currently, the was tted with a side arm adapter connected to the fume hood.
complicated amorphous structure of PAFs hampers the under- Neat bromine (3.5 mL, 67.3 mmol) was added dropwise through
standing of these behaviors, and employing suitable computa- the septum via a syringe for 5 min; subsequently, the resulting
tional simulations is especially an interesting research topic.
solution was stirred at room temperature for 0.5 h. Ethanol
In this work, in an effort to understand sulphur uptake from (60 mL) was then added to the reaction mixture, and the reac-
gasoline with PAF materials, we present a viable synthetic tion was stirred for an additional 1 h. The resulting precipitate
approach for an ionic porous aromatic framework material with was collected by suction ltration and washed with 150 mL of
imidazolium functional groups, referred to as iPAF-1, using ethanol. The collected crude product was then puried by
a
precursor-designed synthetic method. Specically, the column chromatography to afford compound 2 (4.9 g,
precursor-designed approach stoichiometrically introduces func- 7.6 mmol, 85%) as a white solid.
tional groups and ensures evenly distributed adsorptive active
sites in the products, thus a computational simulation based on
molecular dynamics has made application to the functionalized
PAF structures possible for the rst time. The simulation results
agree well with the experimental results of the porous structures
and the adsorptive capacity of iPAF-1 for dibenzothiophene (DBT).
In particular, iPAF-1 exhibits the highest adsorption desulphuri-
zation capacity of 769.23 mg gꢀ1 for dibenzothiophene (DBT)
compared with all other reported porous materials to date.10,29–44
Meanwhile, with the aid of computational simulation, the effect of
binding energy on the DBT uptake during desulphurization was
investigated to gain insight into the structural engineering of
porous aromatic frameworks and emerging applications for
environmentally friendly processes.
Synthesis of (4-bromo-3-(bromomethyl)phenyl)tris(4-
bromophenyl)methane (compound 3)
Compound 2 (1.0 g, 1.53 mmol) was dissolved in anhydrous
CCl4 (40 mL) in a 100 mL round-bottom ask. The mixture was
degassed for 10 min. N-Bromosuccinimide (0.356 g, 1.99 mmol)
and benzoyl peroxide (0.010 g, 0.04 mmol) were then added
under N2, and the reaction mixture was reuxed overnight. The
reaction mixture was cooled to room temperature and ltered.
The ltrate was concentrated to dryness under reduced pressure
to give a pale yellow oil, which was triturated with 25 mL of
ethanol to give an off-white powder. The collected crude
product was puried by column chromatography to afford 3
(2.6 g, 45%) as an off-white powder.
Synthesis of (4-bromo-3-(3-methylimidazolemethyl)phenyl)
tris(4-bromophenyl)methane (iTBPM)
Experimental section
Synthesis of (3-methylphenyl)triphenylmethane (compound
Compound 3 (1.0 g, 1.53 mmol) and N-methylimidazole
(0.356 g, 1.99 mmol) were dissolved with 40 mL of CH2Cl2 in
a 100 mL round-bottom ask. The reaction mixture was reuxed
overnight, cooled to room temperature and ltered. The white
powder was washed with ethyl acetate and collected by ltration
to afford iTBPM (1.10 g, 98%).
1)
Trityl chloride (9.2 g, 32.9 mmol) and o-toluidine (9.4 mL, 88.8
mmol) were added to a 250 mL round-bottom ask. The mixture
was stirred under reux for 0.5 h. The resulting purple slurry
reaction mixture was allowed to cool to room temperature when
it solidied. The solid was ground with a spatula, and the
resulting powder was combined with a mixture of 2 M HCl and
methanol (25 mL/60 mL). The mixture was then heated at 80 ꢁC
for 0.5 h. Aer cooling to room temperature, the reaction
mixture was ltered and washed with 125 mL of DI H2O to
afford a light purple solid, which was briey air-dried in an
oven. Aer that, the crude solid was combined with 65 mL of
ethanol and concentrated H2SO4 (10 mL, 96 wt%). The resulting
mixture was cooled to ꢀ15 ꢁC. Isoamyl nitrite (7.5 mL, 55.8
mmol) was added slowly, and the resulting mixture was stirred
Synthesis of iPAF-1
1,5-Cyclooctadiene (cod, 1.05 mL, 8.32 mmol) was added to
a clear solution of bis(1,5-cyclooctadiene)nickel(0) ([Ni(cod)2],
2.25 g, 8.18 mmol) and 2,20-bipyridyl (1.28 g, 8.18 mmol) in
anhydrous DMF (120 mL). The reaction mixture was heated at
ꢁ
80 C for 1 h. Dry iTBPM (1.27 g, 1.57 mmol) was added to the
resulting purple solution, and the reaction mixture was stirred
overnight to give a deep purple suspension. Aer cooling to
room temperature, the residue was washed with concentrated
HCl, DI H2O and THF, respectively. The resulting light yellow
powder was puried by Soxhlet extraction with THF for 48 h to
afford iPAF-1 (576 mg, 81%).
ꢁ
at ꢀ15 C for 1 h. Aqueous hypophosphoric acid (15 mL, 50%)
was then added to the reaction mixture at ꢀ15 ꢁC, and the
resulting mixture was brought to room temperature before
being heated at 50 ꢁC for 2 h. The resulting precipitate was
collected by suction ltration and washed with DI H2O (100 mL)
and ethanol (100 mL). The tan brown crude product of 1 (8.6 g,
78%) was collected without any further purication.
Results and discussion
Synthesis of the materials
The idea of precursor design of PAF materials was rst reported
by Nguyen et al.45 as a “de novo approach”. They synthesized
–CH3, –CH2OH, –CH2–NH2, –CH2N]CMe2 and phthalimide-
Synthesis of (4-bromo-3-methylphenyl)tris(4-bromophenyl)
methane (compound 2)
Compound 1 (3.0 g, 8.96 mmol) was added to a 150 mL 2-necked based PAF-1 derivatives for CO2 adsorption materials.
round-bottom ask equipped with a magnetic stir bar. One of However, ionic imidazolium functional groups within PAF
This journal is © The Royal Society of Chemistry 2019
Chem. Sci., 2019, 10, 606–613 | 607

View Article Online
Chemical Science
Edge Article
materials have never been reported before. Recently, the prep- 1160 cmꢀ1 for the C–N stretching band, indicating the presence
aration of some ionic COF and PAF materials has been achieved of the imidazolium group aer the cross-coupling reaction. The
by selecting ionic building blocks, and these materials have peak for aromatic C–H stretching vibration is located at
various applications for proton conduction, iodine adsorption 3037 cmꢀ1 and, the vibrations at 2980 and 2940 cmꢀ1 are
and ion-exchange for water treatment.46–48 Until now, the ionic ascribed to aliphatic C–H stretching vibrations. Moreover,
building blocks were limited to a few molecules that were either a broad O–H band is found around 3410.7 cmꢀ1 due to trace
toxic or difficult to prepare, which limited their extensive amounts of absorbed water in the iPAF-1 sample. Thus, the IR
applications. Herein, cationic imidazolium sites were intro- data conrm the successful formation of iPAF-1 as our initial
duced within a PAF material in order to improve the adsorption design. More detailed IR spectra of the precursor and PAF-1,
capacity of aromatic sulphur compounds. Compared to the with and without 3-methylimidazolium groups, are also
post-modication treatment of PAFs, inhomogeneity would be shown in Fig. S6 and Table S1.† Elemental analyses of the
avoided by precursor design. For the rst time, tetrakis(4- iTBPM precursor and iPAF-1, with respect to C, H and N, are
bromophenyl)methane-based ionic porous aromatic frame- shown in Table S2.† The total content of these elements in iPAF-
works were synthesized from designed precursors with 3-meth- 1 is 86.01%, and the remaining fraction is ascribed to the
ylimidazolium ionic functional groups. To achieve this objective, oxygen of the absorbed water, as evidenced by the IR spectra
we started from the synthesis of (3-methylphenyl)triphenyl- and Clꢀ ions as counter anions to imidazolium groups. Energy
methane (compound 1), (4-bromo-3-methylphenyl)tris(4- dispersive spectrometry (EDS) analysis (Fig. S7†) of iPAF-1
bromophenyl)methane (compound 2) and (4-bromo-3- further conrms the presence of excess Clꢀ ions, rather than
(bromomethyl)phenyl)tris(4-bromophenyl)methane (compound Brꢀ ions, due to anion exchange when the product is washed
3) to build the desired monomer of (4-bromo-3-(3- with excess HCl solution. Moreover, the X-ray photoelectron
methylimidazolemethyl)phenyl)tris(4-bromophenyl)methane spectroscopy (XPS) N 1s spectrum (Fig. S8†) suggests that the
(referred to as iTBPM), as demonstrated in the Experimental broad bands in the range of 399.8 to 400.6 eV were produced by
section and ESI.† Next, iPAF-1 was successfully synthesized the imidazolium ring. An enlarged view of Cl 1s and Br 1s XPS
via a Yamamoto-type Ullmann cross-coupling reaction from spectra (Fig. S8†) veries the presence of Clꢀ ions, rather than
the iTBPM building units.
Brꢀ, in the iPAF-1 network; this result is consistent with the EDS
data. Considering the above elemental analyses, the composi-
tion of iPAF-1 was conrmed to be C28.8H20.0N2Cl. Compared
with the calculated molecular formula of C30H23N2Cl, the
precursor design method can stoichiometrically and homoge-
neously introduce the desired functional groups into the nal
product. The thermal gravimetric (TG) prole of iPAF-1 under
air (Fig. S9†) shows that the iPAF-1 network is more thermally
stable than its parent monomer of iTBPM. The mass loss from
100–500 ꢁC, which is due to the decomposition of absorbed
water followed by the loss of chloride ions and imidazolium
moieties, is consistent with the composition calculations. The
scanning electron microscopy (SEM) images reveal the
micrometer-sized aggregation of hundreds of nanometer-sized
particles of iPAF-1 (Fig. S10†).
N2-sorption analysis revealed the porosity and textural
properties of iPAF-1 at 77 K, as shown in Fig. 2. The obtained
type I isotherm indicates a microporous network. The observed
Brunauer–Emmett–Teller (BET) surface area and total micro-
pore volume were 594 m2 gꢀ1 and 0.280 cm3 gꢀ1, respectively.
The pore-size distribution predominantly lies in the micropo-
rous range of less than 2 nm. Three kinds of pores, centered at
0.54 nm, 0.86 nm and 1.10 nm, are calculated according to
nonlocal density functional theory (NLDFT), as shown in the
inset of Fig. 2. The presence of the functional group 3-methyl-
imidazolium and heavier halide counter anions in the PAF
diamond cavity reduces the BET surface area and micropore
volume compared to the robust PAF-1 network.
Structural characterization of iTBPM and iPAF-1
The synthetic details of the monomer with designed imidazo-
lium functional groups and iPAF-1 can be found in the Experi-
mental section and ESI.† A schematic representation of iPAF-1
synthesis is shown in Fig. 1.
The successful synthesis and structural characterization of
the designed precursor iTBPM and iPAF-1 were conrmed using
spectroscopic methods and elemental analysis techniques. The
FT-IR spectra of precursor iTBPM and iPAF-1 are presented in
Fig. S6.† The intense characteristic bands found at 537 and
517 cmꢀ1 in the iTBPM spectrum do not appear in the iPAF-1
spectra, indicating that the phenyl-Br groups of the monomer
iTBPM almost completely cross-coupled in the PAF-1 derivative
network. Additionally, the spectra of iTBPM and iPAF-1 retain
the region around 1607 cmꢀ1 to 1160 cmꢀ1, which are charac-
teristic of the imidazole skeleton region: 1488 cmꢀ1 for the
C]N stretching vibration, 1393 cmꢀ1 for the deformation
vibration of the C–H bond in the imidazolium ring and at
Similar to most POF materials, iPAF-1 is amorphous, as
indicated by its powder X-ray diffraction (PXRD) pattern
(Fig. S11†). Although the local structure is assumed as
Fig. 1 Schematic of the synthesis of iPAF-1 from the designed imi-
dazolium-modified precursor. The iPAF-1 framework is schematically
represented as an adamantane-like cage for clarity.
a
preferred diamond net built from tetrahedral blocks,
according to the topological view, the overall structure obtained
608 | Chem. Sci., 2019, 10, 606–613
This journal is © The Royal Society of Chemistry 2019

View Article Online
Edge Article
Chemical Science
Fig. 2 N2-sorption isotherm of iPAF-1 measured at 77 K. Square plots
represent adsorption, and circle plots represent desorption. The inset
shows the pore-size distribution calculated from the NLDFT method.
Fig. 3 Simulated structure of iPAF-1. The inset shows the pore size
distribution calculated from the move-and-grow algorithm.
from the irreversible cross-coupling reactions possesses defects
and mismatched connections. Thus, the product is an amor-
phous and sophisticated network. In order to elucidate its
structure, we used a coarse-graining (CG) polymerization model
to simulate the fabrication of iPAF-1. The CG simulations are all
performed on graphics processing units (GPU) using molecular
dynamics (MD) soware GALAMOST.49–51 In the CG model of
iTBPM, the particle A represents the quaternary carbon center
linked to four benzene groups. The particle B represents the
reactive brominated benzene group and has the position of the
benzene group center. The particle C represents the imidazo-
lium group and has the position of its center (see the details in
Section S5.1 in the ESI†). The porous aromatic framework is
formed by the cross-coupling reaction between B and B.
However, the polymerization is a step growth process; therefore,
not all of B can take part in the reaction for obtaining a perfect
crystalline diamond structure. Instead, the tetrahedral precur-
sors connect to a defected and complex network. Also, the entire
framework can be regarded as a superposition of multiple local
diamond structural networks, as shown in Fig. 3. In order to
evaluate the credibility of the simulation, a new algorithm,
referred to as “move-and-grow algorithm”, was designed and
employed to measure the pore size distribution in the simulated
structure (detailed in the ESI, Section S5.2†). The simulation
result shows that three types of pores, with sizes of 0.54 nm,
0.90 nm and 1.26 nm, were detected in the simulated iPAF-1
framework (Fig. 3 inset). Specic patterns of each opening
and surrounding framework are detailed in the ESI, Fig. S20.†
The good agreement between the simulated pore size distribu-
tion and the experimental results from N2 adsorption indicates
good credibility of the simulation model, and further simula-
tion of the adsorption process will be conducted based on this
model. The porous aromatic framework consists of three
different pore sizes, which could facilitate the diffusion and
absorption of sulphur compounds inside the framework.
adsorbent, using dibenzothiophene (DBT) as the target analyte
in n-octane, a model fuel oil, at 298 K. All desulphurization
experiments were performed at various solution concentrations
of DBT in n-octane ranging between 0 and 21 000 mg Lꢀ1. The
results indicate that the adsorption capacity increased with
increasing concentration of DBT, as shown in Fig. 4a. A linear
relationship was observed between Ce/qe and Ce, with R2
>
0.9905, according to the Langmuir adsorption model (Fig. 4a
inset). The maximum uptake of DBT was 769.23 mg gꢀ1. The
relationship was more closely correlated with the Langmuir
adsorption model than the Freundlich model, as shown in
Fig. S12.† This implies the existence of more dominant mono-
layer adsorption than multilayer physical adsorption, indicating
strong affinity between electron-rich thiophene-based
compounds and electrophilic imidazolium compounds.
Furthermore, the kinetic desulphurization behaviors of
iPAF-1 were examined in a solution of 20 000 mg Lꢀ1 DBT in
n-octane at 298 K for 24 h, as shown in Fig. 4b. The results
revealed that the DBT uptake enhanced swily in the rst hour,
but when the adsorption time was extended to 24 h, the DBT
adsorption almost reached equilibrium. In addition, the DBT
adsorption isotherm on iPAF-1 can be described by a better
tting pseudo-second-order model, with R2 > 0.999 (Fig. 4b inset
and Fig. S13†). This result further veries the existence of nite
adsorptive sites for DBT within iPAF-1, and these sites are
attributed to the presence of the imidazolium group.
Fig. 4 (a) DBT adsorption isotherm and Langmuir adsorption model
fitting (inset) for iPAF-1 at various concentrations of DBT in n-octane.
(b) Kinetic adsorption curve of DBT and the inset shows the fitting plot
DBT adsorption performance of iPAF-1
Gas-chromatography (GC) was used to experimentally investi-
gate the desulphurization performance of iPAF-1 as an for the pseudo-second-order model.
This journal is © The Royal Society of Chemistry 2019
Chem. Sci., 2019, 10, 606–613 | 609

View Article Online
Chemical Science
Edge Article
The calculated maximum uptake of DBT at 769.23 mg gꢀ1 is parameters for the optimization were 2.7 ꢂ 10ꢀ4 eV (energy),
ꢀ1
far beyond any other adsorbent reported10,29–44 (Fig. 5a). Thus, 0.05 eV$A (gradient), and 0.005 A (displacement), as reported
iPAF-1 is the best adsorbent to date. Furthermore, Fig. 5b elsewhere.29 The structures of DBT and segments of iTBPM were
describes a brief and comprehensive comparison between well- optimized before the BE calculations. Then, the binding struc-
known reported porous and non-porous adsorbents and iPAF-1, ture of DBT and segments of iTBPM were constructed and
concerning the adsorption capacity value of DBT. It is under- optimized for BE calculation. The BE values were calculated as
stood that the uptake of an adsorbent usually scales linearly the energy difference between the products and the reactants in
with its surface area. However, the plot of DBT uptake (qm) the adsorption process, as dened by BE ¼ Etotal ꢀ Eseg ꢀ EDBT
versus specic surface area of multiple adsorbents (SBET), shown where Etotal, Eseg and EDBT represent the total energy of the
in Fig. 5b, highlights the different behaviors of iPAF-1, which monomer/adsorbates, segment and the adsorbate, respectively.
fall outside of this line. Generally speaking, compared to other The calculated BE was ꢀ36.6 kJ molꢀ1, where the negative value
reported porous adsorbents, iPAF-1 exhibits a relatively low indicates an exothermic adsorption and higher absolute value
surface area (594 m2 gꢀ1), but it offers extraordinary DBT means a higher adsorption strength (Fig. 6).
˚
˚
adsorption capacity. It is important to elucidate the key role of
Considering the large spatial and temporal scales of the
DBT adsorbents in light of targeted material design and engi- desulphurization process, we employed coarse-graining (CG)
neering. With this goal, computational simulation of the simulations to gain insight into the desulphurization
desulphurization process was conducted.
phenomenon, based on the simulated model of iPAF-1. We
First, the binding energy (BE) was calculated using the constructed a model for desulphurization simulation, as shown
Dmol3 module in Materials Studio 7.0 soware (Accelrys Inc., in Fig. 7. The porous iPAF-1, which was obtained by previous
San Diego). The density functional theory (DFT) method was polymerization simulation, was placed in the middle of a 15 nm
employed using the gradient corrected (GGA) correlation func- ꢂ 15 nm ꢂ 30 nm box. 15 000 DBT molecules were placed to the
tion of Perdew and Wang (PW91), with the double numerical le of iPAF-1 in a 15 nm ꢂ 15 nm ꢂ 35 nm box. Wall1, Wall2
plus (DNP) polarization basis set. The convergence threshold and Wall3 described by WCA interaction with 3 ¼ 1.0 kJ molꢀ1
and d ¼ 0.2 nm were employed to separate the molecules in the
equilibrium simulation. The equilibrium simulation was per-
formed for 105 time steps. Aer equilibrium, Wall2 and Wall3
allow the passage of DBT molecules. Then, the adsorption
simulation began. The adsorption simulation was performed
for approximately 2 ꢂ 109 time steps. To obtain an equilibrium
Fig. 6 Binding configuration of DBT and the imidazolium segment.
Fig. 5 (a) Comparison of DBT desulphurization capacity of iPAF-1 with
other reported materials (ref. 10 and 29–44). (b) Plot of DBT uptake
(qm) versus specific surface area of multiple adsorbents (SBET).
Fig. 7 DBT desulphurization process simulation model.
610 | Chem. Sci., 2019, 10, 606–613
This journal is © The Royal Society of Chemistry 2019

View Article Online
Edge Article
Chemical Science
state of absorption and desorption, when the DBT molecules binding interaction, the adsorbed DBT blocked the pores and
diffused through the iPAF-1 framework and passed through channels, preventing further diffusion. The maximum diffu-
Wall4, they were removed from the system.
sion length along the Z direction was 27 nm for BE of
The adsorption doses of DBT along the Z direction within ꢀ40 kJ molꢀ1 and 21 nm for BE of ꢀ50 kJ molꢀ1. Both lengths
iPAF-1 for different BEs were collected. Fig. 8 shows three typical are shorter than the diameter of iPAF-1 particles, as observed
adsorption types, with different BEs. When the BE was by SEM. In this case, the inner part of the iPAF-1 bulk material
ꢀ20 kJ molꢀ1, all DBT molecules entered the box of iPAF-1 but will be free from DBT, which results in a lower adsorption
only some of them were adsorbed. When BE was ꢀ35 kJ molꢀ1
,
capability overall. The maximum uptake achieved was at a BE
most of the DBT molecules entered the box and were retained of ꢀ35 kJ molꢀ1, and the simulated uptake of about 800 mg
inside it. When BE was ꢀ50 kJ molꢀ1, the diffusion of DBT gꢀ1 is close to the experimental result of 769.23 mg gꢀ1, which
molecules in iPAF-1 became quite difficult because of the was calculated according to the Langmuir model. This indi-
stronger binding interaction between the imidazolium group cates the reliability of the simulation. The BE by DFT calcu-
and DBT. To further examine the effect of BE, a plot of DBT lation is ꢀ36.6 kJ molꢀ1; thus, we concluded that the strength
adsorption capability at different BE values versus the diffusion of the interaction between the imidazole group of iPAF-1 and
depth is shown in Fig. 8d. In general, the adsorption capability DBT is an appropriate value for DBT adsorption, within the
increased with increasing absolute BE value (Fig. 8). However, iPAF-1 network. The high DBT uptake of iPAF-1 may be
when the BE was ꢀ40 kJ molꢀ1 and ꢀ50 kJ molꢀ1, with stronger attributed to the presence of favorable chemical interactions
for DBT. Due to the diffusion control of the desulphurization
process and the high sulphur uptake, the iPAF-1 is believed to
be a suitable adsorbent for deep desulphurization, which is of
practical signicance in industry. The deep desulphurization
of 100 ppm sulphur in model oil can be achieved via a sulphur
breakthrough experiment. As shown in Fig. S15,† DBT mole-
cules with concentrations below 100 ppm can be successfully
removed from the model oil, indicating the potential appli-
cation of iPAF-1 for deep removal of organic sulphur
compounds at low concentrations.
Conclusions
In conclusion, an imidazolium-modied building block was
utilized to successfully synthesize an ionic porous aromatic
framework with the goal of increasing the DBT uptake from
gasoline. The synthetic route employs an engineered precursor
and one-step cross coupling, so that the imidazolium functional
sites are stoichiometrically and homogeneously introduced into
the framework; this was conrmed by multiple characterization
techniques. Compared to other porous adsorbents for adsorp-
tive desulphurization, the ionic porous aromatic framework
possesses a relatively low surface area of 594 m2 gꢀ1 but the
highest DBT uptake of 769.23 mg gꢀ1. In order to understand
the adsorptive desulphurization process and the role of the
imidazolium functional group in iPAF-1, computational calcu-
lations were conducted to simulate the sophisticated structure
of iPAF-1 and the desulphurization process. The simulation
results indicate that the appropriate binding energy plays a key
role in guest molecule diffusion and adsorption. In the case of
iPAF-1, imidazolium functional groups, with a binding energy
of ꢀ36.6 kJ molꢀ1, provided appropriate active sites for desul-
phurization of DBT from gasoline. This work gives insight into
the design and engineering of porous aromatic frameworks for
diverse applications in the future.
Conflicts of interest
Fig. 8 Three types of adsorption models (a–c) and the adsorption
capacity for diffusion depths at different BEs (d).
There are no conicts to declare.
This journal is © The Royal Society of Chemistry 2019
Chem. Sci., 2019, 10, 606–613 | 611

View Article Online
Chemical Science
Edge Article
21 S. Dalapati, E. Jin, M. Addicoat, T. Heine and D. Jiang, J. Am.
Chem. Soc., 2016, 138, 5797–5800.
Acknowledgements
22 P. Kuhn, M. Antonietti and A. Thomas, Angew. Chem., Int.
Ed., 2008, 47, 3450–3453.
This work was supported by the National Basic Research
Program of China (973 Program, grant no. 2014CB931804) and
the National Natural Science Foundation of China (NSFC
Project, grant no. 91622106, 21531003 and 21601031). Y. Z. also
thanks the National Natural Science Foundation of China
(NSFC Project, grant no 21774129).
23 H. Lau Cher, X. Mulet, K. Konstas, M. Doherty Cara,
M. A. Sani, F. Separovic, R. Hill Matthew and D. Wood
Colin, Angew. Chem., Int. Ed., 2016, 55, 1998–2001.
24 B. Li, Y. Zhang, R. Krishna, K. Yao, Y. Han, Z. Wu, D. Ma,
Z. Shi, T. Pham, B. Space, J. Liu, P. K. Thallapally, J. Liu,
M. Chrzanowski and S. Ma, J. Am. Chem. Soc., 2014, 136,
8654–8660.
25 Q. Sun, B. Aguila, J. Perman, L. D. Earl, C. W. Abney,
Y. Cheng, H. Wei, N. Nguyen, L. Wojtas and S. Ma, J. Am.
Chem. Soc., 2017, 139, 2786–2793.
Notes and references
1 B. Aguila, Q. Sun, J. A. Perman, L. D. Earl, C. W. Abney,
R. Elzein, R. Schlaf and S. Ma, Adv. Mater., 2017, 29, 1700665.
2 B. P. Biswal, D. Becker, N. Chandrasekhar, J. S. Seenath, 26 A. Nagai, Z. Guo, X. Feng, S. Jin, X. Chen, X. Ding and
S. Paasch, S. Machill, F. Hennersdorf, E. Brunner, D. Jiang, Nat. Commun., 2011, 2, 536.
J. J. Weigand, R. Berger and X. Feng, Chem.–Eur. J., 2018, 27 L. H. Li, X. L. Feng, X. H. Cui, Y. X. Ma, S. Y. Ding and
24, 10868–10875.
W. Wang, J. Am. Chem. Soc., 2017, 139, 6042–6045.
3 Q. Sun, B. Aguila, J. Perman, A. S. Ivanov, V. S. Bryantsev, 28 W. Jiang, W. Zhu, Y. Chang, H. Li, Y. Chao, J. Xiong, H. Liu
L. D. Earl, C. W. Abney, L. Wojtas and S. Ma, Nat.
Commun., 2018, 9, 1644.
4 Y. Yuan, F. Sun, H. Ren, X. Jing, W. Wang, H. Ma, H. Zhao
and G. Zhu, J. Mater. Chem., 2011, 21, 13498.
and S. Yin, Energy Fuels, 2014, 28, 2754–2760.
29 L. Peng, F. Guo, C. Zhang, J. Xu, S. Xu, C. Peng, J. Hu and
H. Liu, Ind. Eng. Chem. Res., 2017, 56, 4319–4326.
30 J. Xiao, C. Song, X. Ma and Z. Li, Ind. Eng. Chem. Res., 2012,
51, 3436–3443.
5 X. Xu, S. Zhang, P. Li and Y. Shen, Fuel, 2014, 117, 499–508.
6 P. Wasserscheid and A. Jess, Green Chem., 2004, 6, 316–322. 31 T. Jin, S. An, X. Yang, J. Hu, H. Wang, H. Liu, Z. Tian,
7 H. Yang, B. Jiang, Y. Sun, L. Zhang, Z. Huang, Z. Sun and
D. e. Jiang, N. Mehio and X. Zhu, AIChE J., 2016, 62, 1740–
N. Yang, J. Hazard. Mater., 2017, 333, 63–72.
1746.
8 M. Soleimani, A. Bassi and A. Margaritis, Biotechnol. Adv., 32 J. Xiong, L. Yang, Y. Chao, J. Pang, P. Wu, M. Zhang, W. Zhu
2007, 25, 570–596.
and H. Li, Green Chem., 2016, 18, 3040–3047.
33 M. Huang, G. Chang, Y. Su, H. Xing, Z. Zhang, Y. Yang,
Q. Ren, Z. Bao and B. Chen, Chem. Commun., 2015, 51,
12205–12207.
9 X. L. Tang and L. Shi, Langmuir, 2011, 27, 11999–12007.
10 J. Xiong, W. Zhu, H. Li, L. Yang, Y. Chao, P. Wu, S. Xun,
W. Jiang, M. Zhang and H. Li, J. Mater. Chem. A, 2015, 3,
12738–12747.
34 J. Wang and J. Wei, J. Mater. Chem. A, 2017, 5, 4651–4659.
11 S. Das, P. Heasman, T. Ben and S. Qiu, Chem. Rev., 2017, 117, 35 J. Zhang, H. H. Xu, Y. Z. Lu, H. Meng, C. X. Li, B. H. Chen and
1515–1563.
Z. G. Lei, Energy Fuels, 2016, 30, 5035–5041.
12 T. Ben, H. Ren, S. Ma, D. Cao, J. Lan, X. Jing, W. Wang, J. Xu, 36 J. Zhang, C. Ma, X. Zhu, Y. Lu, H. Meng, C. Li, B. Chen and
F. Deng, M. Simmons Jason, S. Qiu and G. Zhu, Angew.
Chem., Int. Ed., 2009, 48, 9457–9460.
13 Y. Yuan, F. Sun, L. Li, P. Cui and G. Zhu, Nat. Commun., 2014,
5, 4260.
14 J. F. Van Humbeck, T. M. McDonald, X. Jing, B. M. Wiers,
G. Zhu and J. R. Long, J. Am. Chem. Soc., 2014, 136, 2432–
2440.
15 Y. Yuan, F. Sun, F. Zhang, H. Ren, M. Guo, K. Cai, X. Jing,
X. Gao and G. Zhu, Adv. Mater., 2013, 25, 6619–6624.
16 T. Ben, Y. Li, L. Zhu, D. Zhang, D. Cao, Z. Xiang, X. Yao and
S. Qiu, Energy Environ. Sci., 2012, 5, 8370.
Z. Lei, Ind. Eng. Chem. Res., 2016, 55, 8079–8086.
37 N. A. Khan, J. W. Yoon, J. S. Chang and S. H. Jhung, Chem.
Commun., 2016, 52, 8667–8670.
38 H. X. Zhang, H. L. Huang, C. X. Li, H. Meng, Y. Z. Lu,
C. L. Zhong, D. H. Liu and Q. Y. Yang, Ind. Eng. Chem.
Res., 2012, 51, 12449–12455.
39 K. A. Cychosz, A. G. Wong-Foy and A. J. Matzger, J. Am. Chem.
Soc., 2008, 130, 6938–6939.
40 T. H. Park, K. A. Cychosz, A. G. Wong-Foy, A. Dailly and
A. J. Matzger, Chem. Commun., 2011, 47, 1452–1454.
41 J. Xiong, W. Zhu, H. Li, W. Ding, Y. Chao, P. Wu, S. Xun,
M. Zhang and H. Li, Green Chem., 2015, 17, 1647–1656.
17 J. X. Jiang, F. Su, A. Trewin, D. Wood Colin, L. Campbell Neil,
H. Niu, C. Dickinson, Y. Ganin Alexey, J. Rosseinsky 42 N. Farzin Nejad, E. Shams, M. K. Amini and J. C. Bennett,
Matthew, Z. Khimyak Yaroslav and I. Cooper Andrew,
Angew. Chem., Int. Ed., 2007, 46, 8574–8578.
18 Y. Xie, T. T. Wang, X. H. Liu, K. Zou and W. Q. Deng, Nat.
Commun., 2013, 4, 1960.
Microporous Mesoporous Mater., 2013, 168, 239–246.
43 Z. Hasan and S. H. Jhung, ACS Appl. Mater. Interfaces, 2015,
7, 10429–10435.
44 L. Wang, B. Sun, F. H. Yang and R. T. Yang, Chem. Eng. Sci.,
2012, 73, 208–217.
ˆ ´
19 A. P. Cote, A. I. Benin, N. W. Ockwig, M. Keeffe, A. J. Matzger
and O. M. Yaghi, Science, 2005, 310, 1166.
20 S. Y. Ding, J. Gao, Q. Wang, Y. Zhang, W. G. Song, C. Y. Su
and W. Wang, J. Am. Chem. Soc., 2011, 133, 19816–19822.
45 S. J. Garibay, M. H. Weston, J. E. Mondloch, Y. J. Colon,
O. K. Farha, J. T. Hupp and S. T. Nguyen, CrystEngComm,
2013, 15, 1515–1519.
612 | Chem. Sci., 2019, 10, 606–613
This journal is © The Royal Society of Chemistry 2019

View Article Online
Edge Article
Chemical Science
46 H. Ma, B. Liu, B. Li, L. Zhang, Y. G. Li, H. Q. Tan, H. Y. Zang 49 Y. L. Zhu, H. Liu, Z. W. Li, H. J. Qian, G. Milano and Z. Y. Lu,
and G. Zhu, J. Am. Chem. Soc., 2016, 138, 5897–5903. J. Comput. Chem., 2013, 34, 2197–2211.
47 Y. Su, Y. Wang, X. Li, X. Li and R. Wang, ACS Appl. Mater. 50 Y. L. Zhu, D. Pan, Z. W. Li, H. Liu, H. J. Qian, Y. Zhao, Z. Y. Lu
Interfaces, 2016, 8, 18904–18911.
and Z. Y. Sun, Mol. Phys., 2018, 1–13, DOI: 10.1080/
48 Z. Li, H. Li, X. Guan, J. Tang, Y. Yusran, Z. Li, M. Xue,
00268976.2018.1434904.
¨
Q. Fang, Y. Yan, V. Valtchev and S. Qiu, J. Am. Chem. Soc., 51 H. Liu, Y. L. Zhu, Z. Y. Lu and F. Muller-Plathe, J. Comput.
2017, 139, 17771–17774.
Chem., 2016, 37, 2634–2646.
This journal is © The Royal Society of Chemistry 2019
Chem. Sci., 2019, 10, 606–613 | 613