Download as pdf or txt
Download as pdf or txt
You are on page 1of 8

This is an open access article published under a Creative Commons Attribution (CC-BY)

License, which permits unrestricted use, distribution and reproduction in any medium,
provided the author and source are cited.

Research Article

Cite This: ACS Appl. Mater. Interfaces 2018, 10, 16140−16147 www.acsami.org

Robust Covalently Cross-linked Polybenzimidazole/Graphene Oxide


Membranes for High-Flux Organic Solvent Nanofiltration
Fan Fei,†,‡ Levente Cseri,§ Gyorgy Szekely,*,§ and Christopher F. Blanford*,†,‡

School of Materials, University of Manchester, Oxford Road, Manchester, M13 9PL, United Kingdom

Manchester Institute of Biotechnology, University of Manchester, 131 Princess Street, Manchester, M1 7DN, United Kingdom
§
School of Chemical Engineering and Analytical Science, University of Manchester, The Mill, Sackville Street, Manchester, M1 3BB,
United Kingdom
*
S Supporting Information

ABSTRACT: Robust, readily scalable, high-flux graphene


oxide (GO) mixed matrix composite membranes were
developed for organic solvent nanofiltration. Hydroxylated
polybenzimidazole was synthesized by N-benzylation of
Downloaded by 86.134.154.244 at 13:24:46:855 on June 20, 2019

polybenzimidazole with 4-(chloromethyl)benzyl alcohol,


from https://pubs.acs.org/doi/10.1021/acsami.8b03591.

which was confirmed by FTIR and NMR spectroscopy. Flat-


sheet composite membranes comprising of polybenzimida-
zoles and 1 or 2 wt % GO were fabricated via conventional
blade coating and phase inversion. Subsequently, GO was
covalently anchored to the hydroxyl groups of the polymer
using a diisocyanate cross-linking agent. The even distribution
of GO in the membranes was mapped by visible-light microscopy. Hydroxylation and incorporation of GO in the polymer matrix
increased the permeance up to 45.2 ± 1.6 L m−2 h−1 bar−1 in acetone, nearly 5 times higher than the unmodified benchmark
membrane. The enhancement in permeance from the addition of GO did not compromise the solute rejection. The composite
membranes were found to be tight in seven organic solvents, having molecular weight cut-offs (MWCO) as low as 140 g mol−1.
Permeance increased with increasing solvent polarity, while rejection of a 420 g mol−1 pharmaceutical remained over 93%. The
covalent anchoring resulted in robust composite membranes that maintained constant performance over 14 days in a continuous
cross-flow configuration.
KEYWORDS: composite membrane, graphene sieve, mixed matrix membrane, surface modification, cross-linking

1. INTRODUCTION research suggests the potential of advanced materials, such as


Conventional separation processes account for up to 70% of the metal−organic frameworks,8 carbon nanosheets,9 and gra-
total capital and operational cost of chemical plants, and phene,10 for the fabrication of more efficient membranes.
consequently remain an obstacle to sustainable manufacturing.1 Graphene oxide (GO), a chemical derivative of graphite in the
Membrane-based technologies provide more economic and family of graphene nanomaterials, comprises of carbon sheets
sustainable alternatives to conventional separation processes.2 decorated with oxygen-containing functionalities on the edges
Organic solvent nanofiltration (OSN) is a pressure-driven (hydroxyl, carbonyl, and carboxyl groups) and on the surface
separation technique that uses semipermeable membranes that (hydroxyl and epoxide groups). GO is a versatile starting material
can selectively distinguish solutes between 50 and 2000 g mol−1.3 because of its ready dispersibility and reactivity in both organic
OSN has found applications mainly in petrochemical and and aqueous media. GO can be used as a membrane either on its
pharmaceutical industries, being used for solvent exchange and own as a laminate,11 or mixed with polymers to fabricate mixed
recycling, product purification and concentration, and catalyst matrix membranes.12 A series of different techniques, such as
recovery. vacuum filtration,11 drop casting,13 dip-coating,14 spin-coating,15
Membrane materials are core parts of any membrane-based spray-evaporation,16 layer-by-layer assembly,17 and shear align-
technologies, and consequently the prime focus of the current ment18 have been used to prepare GO membranes.
membrane research lies in developing new materials to improve GO membranes have been mainly developed for desalina-
stability, permeance, rejection, and robustness in general. State- tion19 and gas separation.20 However, efforts aiming at molecular
of-the-art OSN membranes provide long-term stability in organic separation are emerging with promising results in the field of
media, low molecular weight cutoff (MWCO) valuesthe
molecular weight of the compound that is 90% rejectedand a Received: March 2, 2018
trade-off between selectivity and ultrahigh permeance to Accepted: April 19, 2018
minimize the processing time and membrane area.4−7 Recent Published: April 19, 2018

© 2018 American Chemical Society 16140 DOI: 10.1021/acsami.8b03591


ACS Appl. Mater. Interfaces 2018, 10, 16140−16147
ACS Applied Materials & Interfaces Research Article

Figure 1. Reaction scheme showing the N-benzylation of polybenzimidazole and the diisocyanate-based (A) cross-linking of hydroxylated
polybenzimidazole chains, (B) anchoring of GO sheets, and (C) cross-linking of GO sheets.

OSN.21−23 Incorporation of GO into a polypyrrole matrix and solvents were used as received without further purification. All
significantly enhanced solvent permeance without compromis- solutions were prepared using water with a resistivity of 18.2 MΩ cm at
ing solvent rejection.21 In contrast, GO-embedded polyethyle- 25 °C (Milli-Q).
nimine membranes displayed enhanced solute rejection and 2.2. Synthesis of Hydroxylated Polybenzimidazole. Hydroxy-
lated polybenzimidazole (PBI−OH) was synthesized through N-
adequate solvent flux.22 These studies were carried out in a
benzylation of polybenzimidazole (Figure 1). In a typical synthesis, 26
limited number of mild solvents (acetone and alcohols), in a wt % PBI dope solution (4.46 g) was diluted with DMAc (45 mL) stirred
dead-end configuration with limited volume of test solutions, at room temperature for 0.5 h, followed by the addition of tBuOK (0.886
hindering long-term performance studies. g, 7.91 mmol, 2.1 eq. per repeating unit). The color of the reaction
The cohesion between the components of a mixed matrix mixture turned from brown to red indicating the deprotonation of the
membrane plays an important role in the performance and long- amine groups of the polymer. The mixture was stirred at room
term stability.24 Therefore, not only the individual characteristics temperature for 3 h to ensure completion of deprotonation. 4-
but also the interaction of the phases should be considered when (Chloromethyl)benzyl alcohol (1.182 g, 7.53 mmol, 2 equiv per
designing mixed matrix membranes. The distribution of GO repeating unit) was added in one portion and the reaction was stirred at
flakes in polymer matrices is an important aspect for composite room temperature for 3 days. The product was precipitated in 1 L H2O/
ethanol (1:1 v/v) mixture, and then filtered. The filter cake was dialyzed
filtration membranes; the decline in membrane permeance in
in water and freeze-dried to obtain the final product as a yellow powder.
previous reports was attributed to aggregation of GO.25 2.3. Membrane Fabrication. Dope solutions containing PBI, PBI−
Herein, we report the fabrication, characterization, perform- OH, and GO were prepared. The total solid content in all dope solutions
ance, and long-term stability of mixed matrix membranes was 20 wt % by adding suitable amount of DMAc as solvent. The
comprised of polybenzimidazole (PBI) and GO. PBI was percentage of PBI, PBI−OH, and GO in the solid content for each
selected as polymer matrix due to its outstanding ability to membrane is shown in Table 1. Homogeneous dope solutions were
withstand extreme thermal, chemical and mechanical con- realized by mechanical stirring of the components at 50 rpm and room
ditions.26 PBI-based OSN membranes have been recently temperature for 6 h, followed by degassing under argon in an incubator
developed by the Livingston group,27,28 and they have been shaker at 400 rpm and 30 °C for 12 h. Membranes were cast onto
successfully used for solvent exchange29 and solvent recovery.30 nonwoven polypropylene support using an Elcometer 4340 film

2. EXPERIMENTAL SECTION Table 1. Abbreviations and Compositions of Membranesa


2.1. Materials and Reagents. Polybenzimidazole dope solution membrane PBI PBI−OH GO TDI cross-linking
(PBI, 26 wt % in DMAc) was purchased from PBI Performance designation (wt %) (wt %) (wt %) reaction
Products, Inc. (Charlotte, USA). Graphene oxide (flake) was purchased M1 100 0 0 no
from William Blythe Ltd. (Accrington, UK). Potassium tert-butoxide M2 90 10 0 no
(tBuOK) and analytical grade organic solvents were purchased from
M3 90 10 0 yes
Fisher Scientific. 4-(Chloromethyl)benzyl alcohol, toluene 2,4-
M4 89 10 1 no
diisocyanate (TDI), methylstyrene dimer, and mepenzolate bromide
(N,N-dimethyl-3-piperidiniobenzylate bromide) were purchased from M5 89 10 1 yes
Sigma-Aldrich. Novatexx 2471 polypropylene nonwoven backing was M6 88 10 2 no
purchased from Freudenberg Filtration Technologies (Crewe, UK). M7 88 10 2 yes
Polystyrene markers for solute rejection evaluation were purchased from
a
Agilent Technologies. Microscope slides and dialysis tubing (SnakeSkin The total solid content in all dope solutions were 20 wt %. The table
10 kDa MWCO 35 mm diameter) were purchased from Thermo shows the percentage of PBI, PBI−OH, and GO in the solid content.
Scientific. Si/SiOx (silicon surface with a silica layer of approximately Composite membranes having 3 wt % GO were prepared but were
290 nm) wafer was purchased from IDB Technologies Ltd. All materials found to be unsuitable for testing because of their brittleness.

16141 DOI: 10.1021/acsami.8b03591


ACS Appl. Mater. Interfaces 2018, 10, 16140−16147
ACS Applied Materials & Interfaces Research Article

applicator with a casting knife set to a thickness of 250 μm and casting ⎛ Cp ⎞


speed of 5 cm s−1. The membranes were formed by the precipitation of R[%] = ⎜1 − ⎟ × 100
⎝ Cf ⎠ (2)
the film in a coagulation bath containing deionized water (15.0 MΩ cm)
at 20 °C. The A4-size membrane sheets were cut into 9 cm diameter
discs. The membranes were washed and soaked in acetonitrile prior to 3. RESULTS AND DISCUSSION
cross-linking to remove the water. The GO sheets were anchored to the
polymer matrix and the polymer chains were simultaneously cross-
3.1. Polymer and Membrane Characterization. The N-
linked using TDI (Figure 1), which was previously reported for cross- benzylation of PBI was confirmed by IR and 1H NMR
linking Pebax membranes.31 Cross-linking was performed by immersing spectroscopy. In the IR spectrum (Figure 2A), PBI−OH exhibits
the membrane discs in 2 wt % TDI in acetonitrile for 30 min followed by
washing with 3 × 40 mL acetonitrile. Covalent cross-linking is often used
to lower the MWCO and improve the stability of membranes. In this
particular work, the GO is also cross-linked to the polymer matrix to
prevent it leaching out of the membrane over time.
2.4. Characterization. NMR analyses were performed on a B500
Bruker Avance II+ 500 MHz instrument using DMSO-d6 as a solvent.
The solvent peak was used as an internal chemical shift reference. Solid
state NMR spectra were recorded on a Bruker Avance III 400 MHz solid
state NMR spectrometer with cross-polarization (CP) pulse and magic
angle spinning (MAS) using adamantane as an external 13C chemical
shift reference.
Attenuated total reflection Fourier-transform infrared spectroscopy
(ATR-FTIR) spectra of powder and membrane samples were acquired
using a Bruker Alpha-P run in air, mounting on a zinc−selenium/
diamond plate. The spectra were recorded at a resolution of 4 cm−1 as an
average of 16 scans. Raman analysis of the PBI-containing materials was
not possible because of the intrinsic fluorescence of the polymer.
Optical microscope images were obtained by Keyence VHX-5000
digital microscope. Samples were prepared by sticking the dried
membranes on glass slides using double-sided tape.
Membrane surface and cross-sectional microstructure was deter-
mined by scanning electron microscopy (SEM) using a Hitachi S-
3000N with a tungsten hairpin filament emission gun at an accelerating
voltage of 5 kV. Samples were prepared by sticking the membranes on a
conductive carbon tab and sputter coated with gold/palladium under an
argon atmosphere using a Quorum Q150TES in order to make the
samples conductive.
Atomic force microscopy (AFM) images were acquired in tapping
Figure 2. (A) ATR-FTIR spectra of PBI and PBI−OH powders. The
mode in air using a Digital Instruments Dimension 3100 with Bruker additional C−H and C−O peaks in the PBI−OH spectrum arise from
TESPA-V2 probe with a nominal spring constant of 37 N m−1 and a the presence of hydroxymethyl groups. (B) 1H NMR spectra of PBI and
nominal tip apex radius of 7 nm. Samples were prepared by sticking the PBI−OH. In the spectrum of PBI−OH, the five new peaks between 7.5
membranes on a glass slide using double-sided tape. For roughness and 4.0 ppm, and the change of the signal structure in the aromatic
calculations, three membranes of each type with an area of 25 μm2 were region confirm the N-benzylation.
scanned and analyzed with the NanoScope Analysis software.
Thermogravimetric analysis (TGA) was determined using a TA
Instruments Q500 in N2 atmosphere with the ramp rate of 10 °C min−1.
two additional peaks compared to PBI, which can be assigned to
2.5. Membrane Performance Tests. Membrane performance was
tested in a typical stainless steel cross-flow nanofiltration apparatus at 10
alkane C−H stretch bonds of methylene (2867 cm−1) and
bar (Figure S9). Two independently prepared membrane discs of each alcohol C−O stretch bonds (1015 cm−1). The N-benzylation
type were tested, and the reported results are the mean values of these causes two red shifts in the spectrum: CN stretching vibrations
measurements. The effective area (A) of each membrane was 52.8 cm2. shifted from 1612 to 1619 cm−1 due to the electronic effects of N-
The permeance was calculated as given in eq 1. substitution, and the out-of-plane bending of C−H groups in the
imidazole ring shifted from 687 to 700 cm−1.34
permeance [L m−2 h−1 bar −1] =
J
=
V The 1H NMR spectrum (Figure 2B) is also in good agreement
ΔP ΔPAt (1) with the expectations. In the case of PBI−OH, five additional
signals were observed compared with PBI. Two significant peaks
The permeance of each membrane was calculated by dividing the in the upfield aromatic region, at 7.21 and 7.02 ppm, are
solvent flux through the membrane (J) by the transmembrane pressure attributed to the protons of the para-substituted benzene ring.
(ΔP). The flux was obtained by measuring the volume of solvent (V) Two peaks of methylene protons were identified at 5.69 ppm
that permeates through the membrane per membrane area (A) per time
(N−CH2−φ) and 4.40 ppm (φ−CH2−O) while the peak at 5.13
(t). The model system for the MWCO curve determination composed
of a mixture of 1 g L−1 PS580 and PS1300 polystyrene markers and 0.1 g ppm belongs to OH protons. The relative peak areas of these five
L−1 of divinylbenzene (130 g mol−1) and 0.1 g L−1 methylstyrene dimer signals are 2:2:2:2:1 in agreement with the expected structure. A
(236 g mol−1) solutions.3,32,33 Because of its versatile solubility, peak at 13.24 ppm in the spectrum of PBI−OH, identified as NH
mepenzolate bromide (420 g mol−1) at 0.1 g L−1 was used for solute protons, shows that the N-benzylation was not complete. The
rejection determinations in various solvents. The rejection (R) of solutes degree of benzylation was calculated to be 70% based on the ratio
was determined as the ratio of its measured concentration in the between aromatic backbone protons and the methylene protons
permeate (Cp) and the feed (Cf) as defined in eq 2. found at 4.40 ppm. The incomplete N-benzylation can be
16142 DOI: 10.1021/acsami.8b03591
ACS Appl. Mater. Interfaces 2018, 10, 16140−16147
ACS Applied Materials & Interfaces Research Article

attributed to the steric crowding around the polymeric backbone.


Furthermore, a significant change can be observed in the
structure of aromatic protons compared with PBI. The
phenylene protons at 9.19 and 8.36 ppm significantly down-
shifted due to the extra shielding provided by the introduced
benzyl groups. Moreover, their signals are split up as a result of
the incomplete modification.
Membranes were characterized by FTIR and TGA analysis.
The cross-linking was verified by two characteristic signals of
urethane bond in the IR spectrum (Figure 3). In the cross-linked

Figure 4. 13C NMR analysis of polymers PBI and PBI−OH, and


membranes M6 and M7. Characteristic signals are indicated by Greek
letters, and they are assigned to carbons showed in Figure 1. An asterisk
(*) is used to mark the solvent (DMSO-d6) in liquid state spectra. ls:
Liquid-state NMR (homogeneous). ls-s: Liquid-state NMR (suspen-
sion). ss: CP-MAS solid-state NMR.

Membrane morphology was characterized by optical micros-


copy, SEM, and AFM. None of the techniques revealed any
difference between non-cross-linked membranes and their cross-
linked counterparts. The images of cross-linked membranes
(M3, M5, and M7) can be found in the Supporting Information.
Figure 3. FTIR spectra of GO flake and membranes of different Visible-light images of the membranes are shown in Figure 5A.
compositions. The cross-linked membranes exhibit two characteristic Images of GO containing membranes (M4 and M6) revealed
peaks which confirm the presence of the urethane groups formed during shiny flakes which are similar in shape and size to the pristine GO
the cross-linking. GO peaks could not be identified in the membrane (Figure S1). The flakes were evenly distributed on the membrane
spectra due to the low GO content.
surfaces. ImageJ analysis (Figure S8) revealed the mean apparent
area of GO flakes in the top layers of M4 and M6 to be 0.39 and
0.37 μm2, and their coverage to be 1.5% and 2.0%, respectively.
membranes, the peaks of the CO stretch (1665 cm−1) and C− SEM surface and cross-sectional images of the membranes are
N stretch (1225 cm−1) were observed. Owing to the small GO shown in Figure 5B and 5C, respectively. All the membranes
content (1−2 wt %), it has no characteristic peak in the FTIR show typical structure of integrally skinned asymmetric polymer
spectra. membranes formed via phase inversion, with dense top layers,
The chemical composition of membranes was characterized by spongy support structures in the middle of the membrane and
suspension and solid-state 13C NMR spectroscopy (Figure 4). macroscopic voids at the bottom of the membranes. AFM height
Both techniques provided similar results; however the solid-state images (Figure 5D) revealed similar surface morphology with
NMR showed higher sensitivity, while the suspension NMR had cavities in a few micrometers for all membranes. The
higher resolution. The PBI signals were the most prominent in incorporation of PBI−OH reduces the root-mean-square
the spectra with aromatic signals between 110−140 ppm, and the (RMS) roughness. However, there is no trend for the influence
imidazole carbon at 151 ppm (δ). Weak signals of benzyl carbons of the incorporation of GO and cross-linking toward surface
originating from PBI−OH (β and γ) were identified in the solid roughness (Table S1).
state spectra of M6 and M7. The spectra of M7 contain the 3.2. Membrane Performance. As shown in Figure 6A, the
characteristic methyl carbon of the cross-linker at 17 ppm (α). 10 wt % addition of PBI−OH (M2) increased the initial
Furthermore, the two additional aromatic signals of the cross- permeance by almost two and half times compared to the pristine
linker can be observed in the suspension NMR spectra at 130 and PBI membrane (M1), which could be attributed to the
138 ppm. The appearance of urethane carbons at 153 ppm (ε) hydrophilicity (Figure S11),35 and lower packing density of
indicated successful diisocyanate cross-linking, which was also PBI−OH.36 Nonetheless, the M2 membrane showed consid-
confirmed with TGA analysis (Figure S12). erable compaction over time, which resulted in a 33% permeance
16143 DOI: 10.1021/acsami.8b03591
ACS Appl. Mater. Interfaces 2018, 10, 16140−16147
ACS Applied Materials & Interfaces Research Article

Figure 5. Images of membranes cast from pure PBI (M1), PBI + 10 wt % PBI−OH (M2), PBI + 10 wt % PBI−OH and 1 wt % GO (M4), and PBI + 10
wt % PBI−OH and 2 wt % GO. (A) Visible-light microscope reflection images (frame size = 100 μm × 100 μm), (B) surface and (C) cross-sectional
SEM images (Scale bar = 10 μm; see Figure S6 for active layer thickness), and (D) AFM height images (scan size = 25 μm × 25 μm) of membranes.

decline in 14 days. Having cross-linked the membrane (M3), the The reported OSN performances of GO-based membranes are
decline in permeance was found to be only 11%, which is less compared with this work in Table 2. The MWCO values for
prominent than for the non-cross-linked membrane (M2). The M4−M7 show comparable, or even tighter membranes than
incorporation of GO increased both the permeance and the long- previously reported GO/polymer mixed matrix membranes.21,22
term stability of the membranes (M4−M7). The decrease of Moreover, the permeance obtained for M6 is 1 order of
membrane top layer thickness (Figure S6) can contribute to the magnitude higher than the others. Our work outperforms the
increase of the permeance. Moreover, it is speculated that the GO laminate membrane38 in permeance, although the latter has
presence of GO inhibits compaction of the polymer matrix, an unprecedented ultrasharp sieving cutoff. Such improvement
subsequently the permeance is maintained over 14 days of comes with the requirement for excessive GO quantity as 100%
continuous operation. The highest flux of 45.2 ± 1.6 L m−2 h−1 GO is used for the membrane fabrication, which hinders scale-up
bar−1 was demonstrated by M6, which is around 18 times higher due to currently limited GO availability and high cost.
M6 and M7 were tested in a series of common solvents with
than commercial OSN membranes (2.5 L m−2 h−1 bar−1).3,37
different polarity such as toluene, tetrahydrofuran, dichloro-
The increase from 1 wt % to 2 wt % GO content improved the
methane, acetonitrile, methanol, and water, through continuous
permeance, but further increase to 3 wt % GO resulted in a fragile
operation over 14 days in cross-flow configuration (Figure 6C
membrane not suitable for filtration. and 6D). The cross-linked M7 showed lower permeance but
The benchmark M1 membrane showed the highest MWCO of higher rejection in all the solvents. The permeance increased with
485 g mol−1 (Figure 6B). On the basis of the cutoff curves, the the dielectric constant of the solvent, and reached a plateau in
rest of the membranes formed two distinguished clusters. Cluster highly polar solvents (Figure 6C), which can be explained by the
1 consists of the non-cross-linked membranes (M2, M4, M6) higher affinity of polar solvents for the hydrophilic membrane
with MWCO of about 277 ± 7 g mol−1, and cluster 2 consists of surface.39,40 This result is in agreement with the reported trend
the cross-linked membranes (M3, M5, M7) with MWCO of for GO laminates.11 Rejection values were obtained for
about 137 ± 15 g mol−1. The results show that the TDI cross- mepenzolate (420 g mol−1), an active pharmaceutical ingredient,
linking tightened the membranes via decreasing the permeance which has a versatile solubility covering both polar and nonpolar
(Figure 6A) and increasing the solute rejection and the MWCO solvents. Both membranes showed high rejection (93−100%) in
values (Figure 6B). Interestingly, the GO-induced increase in all solvents, but no correlations between rejection and
permeance does not compromise the cutoff profile. permeance or rejection and solvent polarity were observed
16144 DOI: 10.1021/acsami.8b03591
ACS Appl. Mater. Interfaces 2018, 10, 16140−16147
ACS Applied Materials & Interfaces Research Article

Figure 6. Comparison of membrane performance in acetone: (A) permeance of acetone over time and (B) solute rejection. Error bars represent
standard deviations derived from two independently prepared membranes cast from separately prepared dope solutions. The effect of solvent on
membrane performance: (C) permeance of various solvents, and (D) rejection values for mepenzolate (420 g mol−1). Error bars represent standard
deviations derived from two experiments using independently prepared membranes.

Table 2. Graphene Oxide Membranes for Molecular Separations in Organic Solventsa


maximum continuous
GO permeance MWCO stability test process
polymer matrix (wt %) (L m−2 h−1 bar−1) (g mol−1) solvents (days) configuration
Ding et al.22 polyethylenimine 1−4 5 (acetone) 200−400 acetone, ethanol dead-end cell
(batch)
21 b
Shao et al. polypyrrole 1 3 (2-propanol) <973 methanol, ethanol, 2-propanol dead-end cell
(batch)
Yang et al.38 N/A 100 18 (hexane) <249c butanol, isopropanol, ethanol, water, butyl dead-end cell
acetate, methanol, tetrahydrofuran, (batch)
acetonitrile, acetone, hexane
this work polybenzimidazole 1−2 45 (acetone) 140−490 acetone, toluene, tetrahydrofuran, 14 cross-flow
dichloromethane, acetonitrile, methanol, (continuous)
water, dimethylformamide
a
Only this work includes GO crosslinking. bReported rejection for rose bengal (973 g mol−1) is 98.5%. cReported rejection for chrysoidine G (249 g
mol−1) is >99.9%.

(Figure 6D). The non-cross-linked M6 was found to be stable in 4. CONCLUSIONS


DMF, which is speculated to be the result of GO stabilizing the Mixed matrix membranes comprising of graphene oxide and
polymer matrix through π−π interactions with the polymer, as polybenzimidazoles were developed for nanofiltration in organic
observed in the literature.41,42 The high rejection and water media. The hydrophilic modification of PBI (PBI−OH) was
permeance of M6 suggests potential applications also in aqueous confirmed by ATR-FTIR and 1H NMR analysis. Membrane
separations including desalination. cross-linking and GO anchoring was achieved by diisocyanate
16145 DOI: 10.1021/acsami.8b03591
ACS Appl. Mater. Interfaces 2018, 10, 16140−16147
ACS Applied Materials & Interfaces Research Article

treatment of the membranes. The cross-linking was followed by


13
C NMR, ATR-FTIR, and TGA analysis. Since GO flakes are
■ ABBREVIATIONS
GO, graphene oxide; MWCO, molecular weight cutoff; OSN,
widely regarded as difficult to be identified by electron organic solvent nanofiltration; PBI, polybenzimidazole; PBI−
microscopic analyses and AFM, we proposed visible-light OH, hydroxylated polybenzimidazole; TDI, toluene 2,4-
microscopy as a method to map the distribution of GO flakes diisocyanate; DMAc, N,N-dimethylacetamide; DMF, N,N-
in polymer matrices. Incorporation of GO into the polymer dimethylformamide; THF, tetrahydrofuran; DCM, dichloro-
matrix improved the membrane performance. First, the presence methane; MeOH, methanol; MeCN, acetonitrile


of 2 wt % GO increased the permeance by 2.9 and 2.5 times for
the cross-linked and non-cross-linked membranes, respectively. REFERENCES
The membranes showed superior permeance to OSN mem- (1) Adler, S.; Beaver, E.; Bryan, P.; Robinson, S.; Watson, J. Vision
branes composed entirely from GO laminates.38 Second, the 2020:2000 Separations Roadmap; EERE Publication and Product
stability of the membranes improved, which was demonstrated Library, 2000.
via continuous filtration over 14 days in a cross-flow (2) Szekely, G.; Jimenez-Solomon, M. F.; Marchetti, P.; Kim, J. F.;
configuration. The higher polarity solvents had higher Livingston, A. G. Sustainability assessment of organic solvent nano-
permeance values, which plateaued at 54 L m−2 h−1 bar−1. The filtration: from fabrication to application. Green Chem. 2014, 16 (10),
membranes showed high rejection (>93% for 420 g mol−1) in 4440−4473.
both polar and nonpolar solvents. Owing to the high permeance (3) Marchetti, P.; Jimenez Solomon, M. F.; Szekely, G.; Livingston, A.
and rejection, the desalination potential of these membranes will G. Molecular separation with organic solvent nanofiltration: a critical
be further studied. review. Chem. Rev. 2014, 114 (21), 10735−10806.


(4) Shi, B.; Marchetti, P.; Peshev, D.; Zhang, S.; Livingston, A. G. Will
ultra-high permeance membranes lead to ultra-efficient processes?
ASSOCIATED CONTENT Challenges for molecular separations in liquid systems. J. Membr. Sci.
*
S Supporting Information
2017, 525, 35−47.
(5) Hennessy, J. Membranes from academia to industry. Nat. Mater.
The Supporting Information is available free of charge on the 2017, 16 (3), 280−282.
ACS Publications website at DOI: 10.1021/acsami.8b03591. (6) Park, H. B.; Kamcev, J.; Robeson, L. M.; Elimelech, M.; Freeman, B.
Characterization of commercial GO flakes by visible-light D. Maximizing the right stuff: The trade-off between membrane
microscopy, SEM, AFM, and Raman spectroscopy; permeability and selectivity. Science 2017, 356 (6343), No. eaab0530.
(7) Marchetti, P.; Peeva, L.; Livingston, A. The Selectivity Challenge in
viscosity values for the dope solutions; AFM, SEM, and
Organic Solvent Nanofiltration: Membrane and Process Solutions.
visible-light microscope images of membranes; thermog- Annu. Rev. Chem. Biomol. Eng. 2017, 8 (0), 473−497.
ravimetric and nanoindentation analyses of membranes; (8) Campbell, J.; Davies, R. P.; Braddock, D. C.; Livingston, A. G.
nanofiltration process scheme; permeance and rejection Improving the permeance of hybrid polymer/metal−organic framework
data in eight solvents; NMR and IR spectra of membranes (MOF) membranes for organic solvent nanofiltration (OSN)−
(PDF) development of MOF thin films via interfacial synthesis. J. Mater.


Chem. A 2015, 3 (18), 9668−9674.
(9) Karan, S.; Samitsu, S.; Peng, X.; Kurashima, K.; Ichinose, I. Ultrafast
AUTHOR INFORMATION viscous permeation of organic solvents through diamond-like carbon
nanosheets. Science 2012, 335 (6067), 444−447.
Corresponding Authors (10) Joshi, R. K.; Carbone, P.; Wang, F.-C.; Kravets, V. G.; Su, Y.;
*Tel.: +44 161 306 4366. E-mail: gyorgy.szekely@manchester. Grigorieva, I. V.; Wu, H. A.; Geim, A. K.; Nair, R. R. Precise and ultrafast
ac.uk. molecular sieving through graphene oxide membranes. Science 2014,
*Tel.: +44 161 306 8915. E-mail: christopher.blanford@ 343 (6172), 752−754.
manchester.ac.uk. (11) Huang, L.; Li, Y.; Zhou, Q.; Yuan, W.; Shi, G. Graphene oxide
membranes with tunable semipermeability in organic solvents. Adv.
ORCID
Mater. 2015, 27 (25), 3797−3802.
Fan Fei: 0000-0003-0548-5775 (12) Zinadini, S.; Zinatizadeh, A. A.; Rahimi, M.; Vatanpour, V.;
Gyorgy Szekely: 0000-0001-9658-2452 Zangeneh, H. Preparation of a novel antifouling mixed matrix PES
Christopher F. Blanford: 0000-0002-0112-7818 membrane by embedding graphene oxide nanoplates. J. Membr. Sci.
2014, 453, 292−301.
Notes (13) Sun, P.; Zhu, M.; Wang, K.; Zhong, M.; Wei, J.; Wu, D.; Xu, Z.;
The authors declare no competing financial interest. Zhu, H. Selective ion penetration of graphene oxide membranes. ACS
In accordance with the University of Manchester’s guidelines, the Nano 2013, 7 (1), 428−437.
data are openly available from Mendeley Data (DOI: 10.17632/ (14) Lou, Y.; Liu, G.; Liu, S.; Shen, J.; Jin, W. A facile way to prepare
tnh4sgrspy.1). ceramic-supported graphene oxide composite membrane via silane-graft


modification. Appl. Surf. Sci. 2014, 307, 631−637.
(15) Nair, R.; Wu, H.; Jayaram, P.; Grigorieva, I.; Geim, A. Unimpeded
ACKNOWLEDGMENTS permeation of water through helium-leak−tight graphene-based
The authors thank the experimental support from Dr. Ben membranes. Science 2012, 335 (6067), 442−444.
Spencer (AFM), Mr. Andrew Forrest (nanoindentation), Mr. (16) Guan, K.; Shen, J.; Liu, G.; Zhao, J.; Zhou, H.; Jin, W. Spray-
Michael Faulkner (SEM), and Mr. Martin Jennings (TGA). This evaporation assembled graphene oxide membranes for selective
hydrogen transport. Sep. Purif. Technol. 2017, 174, 126−135.
work was supported by the UK’s Biotechnology and Biological (17) Nan, Q.; Li, P.; Cao, B. Fabrication of positively charged
Sciences Research Council (BBSRC) and Engineering and nanofiltration membrane via the layer-by-layer assembly of graphene
Physical Sciences Research Council (EPSRC) through an award oxide and polyethylenimine for desalination. Appl. Surf. Sci. 2016, 387,
from the BioProNET Network in Industrial Biotechnology and 521−528.
Bioenergy (BB/L013770/1). F.F. acknowledges the financial (18) Akbari, A.; Sheath, P.; Martin, S. T.; Shinde, D. B.; Shaibani, M.;
support from his family for his doctoral studies. Banerjee, P. C.; Tkacz, R.; Bhattacharyya, D.; Majumder, M. Large-area

16146 DOI: 10.1021/acsami.8b03591


ACS Appl. Mater. Interfaces 2018, 10, 16140−16147
ACS Applied Materials & Interfaces Research Article

graphene-based nanofiltration membranes by shear alignment of polyimide nanofiltration membranes. J. Membr. Sci. 2013, 447, 107−
discotic nematic liquid crystals of graphene oxide. Nat. Commun. 118.
2016, 7, 10891. (37) Vandezande, P.; Gevers, L. E.; Vankelecom, I. F. Solvent resistant
(19) Abraham, J.; Vasu, K. S.; Williams, C. D.; Gopinadhan, K.; Su, Y.; nanofiltration: separating on a molecular level. Chem. Soc. Rev. 2008, 37
Cherian, C. T.; Dix, J.; Prestat, E.; Haigh, S. J.; Grigorieva, I. V.; Carbone, (2), 365−405.
P.; Geim, A. K.; Nair, R. R. Tunable sieving of ions using graphene oxide (38) Yang, Q.; Su, Y.; Chi, C.; Cherian, C.; Huang, K.; Kravets, V.;
membranes. Nat. Nanotechnol. 2017, 12 (6), 546−550. Wang, F.; Zhang, J.; Pratt, A.; Grigorenko, A.; et al. Ultrathin graphene-
(20) Kim, H. W.; Yoon, H. W.; Yoon, S.-M.; Yoo, B. M.; Ahn, B. K.; based membrane with precise molecular sieving and ultrafast solvent
Cho, Y. H.; Shin, H. J.; Yang, H.; Paik, U.; Kwon, S.; Choi, J.-Y.; Park, H. permeation. Nat. Mater. 2017, 16, 1198−1202.
B. Selective gas transport through few-layered graphene and graphene (39) Bhanushali, D.; Kloos, S.; Kurth, C.; Bhattacharyya, D.
oxide membranes. Science 2013, 342 (6154), 91−95. Performance of solvent-resistant membranes for non-aqueous systems:
(21) Shao, L.; Cheng, X.; Wang, Z.; Ma, J.; Guo, Z. Tuning the solvent permeation results and modeling. J. Membr. Sci. 2001, 189 (1),
performance of polypyrrole-based solvent-resistant composite nano- 1−21.
filtration membranes by optimizing polymerization conditions and (40) Geens, J.; Van der Bruggen, B.; Vandecasteele, C. Transport
incorporating graphene oxide. J. Membr. Sci. 2014, 452, 82−89. model for solvent permeation through nanofiltration membranes. Sep.
(22) Ding, R.; Zhang, H.; Li, Y.; Wang, J.; Shi, B.; Mao, H.; Dang, J.; Purif. Technol. 2006, 48 (3), 255−263.
Liu, J. Graphene oxide-embedded nanocomposite membrane for (41) Ramanathan, T.; Abdala, A.; Stankovich, S.; Dikin, D.; Herrera-
solvent resistant nanofiltration with enhanced rejection ability. Chem. Alonso, M.; Piner, R.; Adamson, D.; Schniepp, H.; Chen, X.; Ruoff, R.;
Eng. Sci. 2015, 138, 227−238. et al. Functionalized graphene sheets for polymer nanocomposites. Nat.
(23) Huang, L.; Chen, J.; Gao, T.; Zhang, M.; Li, Y.; Dai, L.; Qu, L.; Shi, Nanotechnol. 2008, 3 (6), 327.
G. Reduced graphene oxide membranes for ultrafast organic solvent (42) Potts, J. R.; Dreyer, D. R.; Bielawski, C. W.; Ruoff, R. S. Graphene-
nanofiltration. Adv. Mater. 2016, 28 (39), 8669−8674. based polymer nanocomposites. Polymer 2011, 52 (1), 5−25.
(24) Wang, Z.; Ren, H.; Zhang, S.; Zhang, F.; Jin, J. Polymers of
intrinsic microporosity/metal−organic framework hybrid membranes
with improved interfacial interaction for high-performance CO2
separation. J. Mater. Chem. A 2017, 5, 10968−10977.
(25) Bano, S.; Mahmood, A.; Kim, S.-J.; Lee, K.-H. Graphene oxide
modified polyamide nanofiltration membrane with improved flux and
antifouling properties. J. Mater. Chem. A 2015, 3 (5), 2065−2071.
(26) Chung, T.-S. A critical review of polybenzimidazoles: historical
development and future R&D. Journal of Macromolecular Science, Part C:
Polymer Reviews 1997, 37 (2), 277−301.
(27) Valtcheva, I. B.; Kumbharkar, S. C.; Kim, J. F.; Bhole, Y.;
Livingston, A. G. Beyond polyimide: crosslinked polybenzimidazole
membranes for organic solvent nanofiltration (OSN) in harsh
environments. J. Membr. Sci. 2014, 457, 62−72.
(28) Valtcheva, I. B.; Marchetti, P.; Livingston, A. G. Crosslinked
polybenzimidazole membranes for organic solvent nanofiltration
(OSN): Analysis of crosslinking reaction mechanism and effects of
reaction parameters. J. Membr. Sci. 2015, 493, 568−579.
(29) Peeva, L.; Da Silva Burgal, J.; Heckenast, Z.; Brazy, F.; Cazenave,
F.; Livingston, A. Continuous Consecutive Reactions with Inter-
Reaction Solvent Exchange by Membrane Separation. Angew. Chem.
2016, 128 (43), 13774−13777.
(30) Didaskalou, C.; Buyuktiryaki, S.; Kecili, R.; Fonte, C. P.; Szekely,
G. Valorisation of agricultural waste with an adsorption/nanofiltration
hybrid process: from materials to sustainable process design. Green
Chem. 2017, 19, 3116−3125.
(31) Aburabie, J.; Peinemann, K.-V. Crosslinked poly (ether block
amide) composite membranes for organic solvent nanofiltration
applications. J. Membr. Sci. 2017, 523, 264−272.
(32) See Toh, Y. H.; Loh, X.; Li, K.; Bismarck, A.; Livingston, A. In
search of a standard method for the characterisation of organic solvent
nanofiltration membranes. J. Membr. Sci. 2007, 291 (1-2), 120−125.
(33) Campbell, J.; Székely, G.; Davies, R.; Braddock, D. C.; Livingston,
A. G. Fabrication of hybrid polymer/metal organic framework
membranes: mixed matrix membranes versus in situ growth. J. Mater.
Chem. A 2014, 2 (24), 9260−9271.
(34) Wang, K. Y.; Xiao, Y.; Chung, T.-S. Chemically modified
polybenzimidazole nanofiltration membrane for the separation of
electrolytes and cephalexin. Chem. Eng. Sci. 2006, 61 (17), 5807−5817.
(35) Eren, E.; Sarihan, A.; Eren, B.; Gumus, H.; Kocak, F. O.
Preparation, characterization and performance enhancement of
polysulfone ultrafiltration membrane using PBI as hydrophilic modifier.
J. Membr. Sci. 2015, 475, 1−8.
(36) Jansen, J. C.; Darvishmanesh, S.; Tasselli, F.; Bazzarelli, F.;
Bernardo, P.; Tocci, E.; Friess, K.; Randova, A.; Drioli, E.; Van der
Bruggen, B. Influence of the blend composition on the properties and
separation performance of novel solvent resistant polyphenylsulfone/

16147 DOI: 10.1021/acsami.8b03591


ACS Appl. Mater. Interfaces 2018, 10, 16140−16147

You might also like