AngewandteInternational Edition

A Journal of the Gesellschaft Deutscher Chemiker

www.angewandte.orgChemie

Accepted Article

Title: Molecular bridges stabilize graphene oxide membranes in water

Authors: Mengchen Zhang, Yangyang Mao, Guozhen Liu, GongpingLiu, Yiqun Fan, and Wanqin Jin

This manuscript has been accepted after peer review and appears as anAccepted Article online prior to editing, proofing, and formal publicationof the final Version of Record (VoR). This work is currently citable byusing the Digital Object Identifier (DOI) given below. The VoR will bepublished online in Early View as soon as possible and may be differentto this Accepted Article as a result of editing. Readers should obtainthe VoR from the journal website shown below when it is publishedto ensure accuracy of information. The authors are responsible for thecontent of this Accepted Article.

To be cited as: Angew. Chem. Int. Ed. 10.1002/anie.201913010Angew. Chem. 10.1002/ange.201913010

Link to VoR: http://dx.doi.org/10.1002/anie.201913010http://dx.doi.org/10.1002/ange.201913010

RESEARCH ARTICLE

Molecular bridges stabilize graphene oxide membranes in water

Mengchen Zhang, Yangyang Mao, Guozhen Liu, Gongping Liu,* Yiqun Fan and Wanqin Jin*

Abstract: Recent innovations highlight the great potential of two-

dimensional materials, such as graphene oxide (GO) films, in water-

related applications. However, undesirable water-induced effects,

such as the redispersion and peeling of stacked GO laminates, greatly

degrade their established performance and impact their practical

application. It remains a great challenge to stabilize GO membranes

in water. Here, we report on a molecular bridge strategy, in which an

interlaminar short-chain molecular bridge generates a robust GO

laminate that resists the tendency to swell. Furthermore, an interfacial

long-chain molecular bridge adheres the GO laminate onto a porous

substrate to increase the mechanical strength to the membrane. By

rationally creating and tuning the molecular bridges, the stabilized GO

membranes can exhibit outstanding durability in harsh operating

conditions, such as cross-flow, high-pressure, and long-term filtration.

The proposed general and scalable stabilizing approach for GO

membranes opens new opportunities for reliable two-dimensional

laminar films used in aqueous environments.

Introduction

Graphene oxide (GO) has attracted extraordinary interest for

its distinct two-dimensional structure and tunable

physicochemical properties[1]. This material displays fascinating

characteristics as a selective barrier and precise sieve in aqueous

solutions, making GO-based films widely applicable in numerous

important fields, such as water purification[2], desalination[3], anti-

corrosion and anti-fouling coatings[4], and energy storage and

conversion[5]. After nearly a decade of effort, GO film technologies

have matured and are realizing their potential. However, stability,

which is a prerequisite for broad applications, has become the

bottleneck that seriously impedes the use of GO films in water-

based implementations.

In particular, unfavorable swelling, redispersion, and peeling

of GO membranes in water have been reported often[6], and these

unstable characteristics greatly deteriorate the membranes’

established performance[7]. For example, the plentiful oxygenated

functional groups on GO nanosheets can become strongly

hydrated with water molecules and yield extra electrostatic

repulsion against π-π attraction, causing GO membranes to swell

or even disintegrate in water[6,7]. Several attempts, such as cross-

linking by molecules or ions[3a,8], insertion of polyelectrolytes[9], or

blending of graphene-based materials[10], have been proposed to

introduce multiple interactions between the GO laminate layers.

Unfortunately, the weak secondary bonding (e.g., hydrogen

bonding or intermolecular interaction) is insufficient to rigidly

connect neighboring GO nanosheets, while excessive cross-

linking is likely to disturb and block water transport within GO

nanochannels. In addition, GO membranes easily peel off the

underlying substrate due to the inadequate interfacial adhesion

between the GO layer and substrate, and this severely degrades

the mechanical strength and operational durability of GO

membranes. Huang and co-workers discovered that Al3+ released

from an anodized aluminum oxide (AAO) filter can immobilize the

GO layer on the AAO substrate[6a]. Also, chemically modifying the

substrate surface can enhance the interfacial compatibility of

composite membranes[11]. However, these methods are limited to

the specific chemistry of the substrate and fail to control substrate

nanostructures that also significantly affect the adhesion of the

GO layer.

Scheme 1. Schematic illustration for stabilizing a GO membrane through hierarchical interlaminar short-chain and interfacial long-chain molecular bridges. An amine

monomer acted as the interlaminar short-chain molecular bridge (short wavy line in red) to interlock the GO laminate through condensation and nucleophilic addition

reactions. Aldehyde-modified chitosan served as an interfacial long-chain molecular bridge (long wavy line in blue) to covalently bond with the amine-interlocked

GO laminate through a Schiff base reaction while firmly adhering to a porous substrate through a sufficient physical van der Waals force (dotted line).

To address the challenging instability issues of GO

membranes in water, we introduced hierarchical molecular

bridges to stabilize the GO membrane. As illustrated in Scheme

1, interlaminar and interfacial interactions of a GO membrane

were respectively reinforced by two types of molecular bridges.

To covalently interlock the GO laminate without sacrificing fast

water permeation, interlaminar short-chain molecular bridges

Interlaminar short-chain

molecular bridge

Interfacial long-chain

molecular bridge

GO nanosheet

Porous substrate

M. Zhang, Y. Mao, G. Liu, G.Liu, Y. Fan and W. Jin

State Key Laboratory of Materials-Oriented Chemical Engineering, College

of Chemical Engineering

Nanjing Tech University

30 Puzhu South Road, Nanjing 210009, P.R. China

E-mail: wqjin@njtech.edu.cn (W.J.); gpliu@njtech.edu.cn (G.L.)

Supporting information for this article is given via a link at the end of the

document.

10.1002/anie.201913010

Acc

epte

d M

anus

crip

t

Angewandte Chemie International Edition

This article is protected by copyright. All rights reserved.

RESEARCH ARTICLE

were designed to ensure a robust assembly structure of GO

laminate via condensation and nucleophilic addition reactions that

resist the tendency toward swelling. Meanwhile, to firmly adhere

the GO laminate onto the substrate through sufficient interactions,

interfacial long-chain molecular bridges were introduced that

connect the GO layer and substrate by synergistically coupling

the chemical Schiff base reaction and physical van der Waals

interactions. With rational design of the molecular bridges, the GO

membranes were stabilized and exhibited outstanding durability

that can survive harsh conditions and maintain performance in

cross-flow, high-pressure, and long-term filtrations.

Results and Discussion

Interlaminar short-chain molecular bridge. When a GO

membrane is applied in an aqueous solution, excessive water

molecules are easily captured inside the GO interlayer channels

due to the strong hydration enabled by the plentiful oxygenated

functional groups on GO nanosheets[6.7]. The extra absorbed

water molecules in the GO nanochannels can lead to undesirable

swelling that enlarges the interlayer spacing and disrupts the well-

stacked GO laminates designed for selective water transport.

Hence, we expect that applying an interlaminar short-chain

molecular bridge can interlock the GO laminate against this

swelling tendency (Figure S1). Ethylenediamine (EDA) is

commonly used to covalently bond adjacent GO nanosheets[12].

As confirmed by the emergence of amide (C-N) absorption peaks

in IR spectra (Figure S2), the amine monomer generated

chemical linking with GO nanosheets through condensation

reactions with carboxyl groups and nucleophilic addition reactions

with epoxy groups[8,12]. The d-spacing measured by XRD

demonstrates the size of the GO interlayer channels (Figure 1a).

Compared with a pristine GO membrane that underwent

substantial d-spacing enlargement when wetted, from ~0.80 to

~1.45 nm, the EDA-bridged GO membrane exhibited only mild d-

spacing stretching after soaking in water, from ~0.84 to ~0.99 nm.

This indicated the anti-swelling effect of covalent bonding within

the GO laminate. Analyzing the molecular structure and

configuration aspects, we find that EDA with aliphatic amine

groups that donate electrons forms a strong nucleophile with high

electron density. However, we also speculate that it bears the high

electrostatic repulsion of negative charges from upper and lower

GO nanosheets at a very close distance, due to its short chain

length with only two carbon atoms. Alternatively,

paraphenylenediamine (PPD) featuring a six-carbon aromatic ring

is expected to exhibit high steric hindrance to afford much less

electrostatic repulsion. Interestingly, although a benzene ring in

PPD occupied a large interlayer spacing, the PPD-bridged GO

membrane displayed a d-spacing expansion of ~0.11 nm (from

~0.91 to ~1.02 nm), which was less than that of the EDA-bridged

GO membrane (~0.15 nm) and showed a reduced swelling

tendency. It is noteworthy that the distinct d-spacings of

interlaminar molecular-bridged GO membranes in a dry state are

mainly dependent on the spatial configuration of intercalated

amine monomers, whereas their variable d-spacings in a wet

state are likely determined by the bonding strength of amine

monomers within the GO laminate. On the basis of our

observations, we find that the latter may play a more vital role in

restraining GO swelling, as the swelling is closely related to the

binding energy between the amine monomer and GO nanosheet.

Unfortunately, the reactivity of amine groups in the aromatic PPD

is relatively low because of the high resonance stability of the

amine lone pair and benzene ring[12]. Considering integration

issues, we were inspired to use polydopamine (PDA) as the

interlaminar short-chain molecular bridge based on its fascinating

properties: it has both aliphatic amine groups with strong binding

reactivity and an appropriate molecular configuration with high

steric hindrance. Compared with its other counterparts (i.e., GO-

EDA, GO-PPD), the PDA-bridged GO membrane exhibited only a

slight 2θ shift in water, corresponding to a minimal d-spacing

expansion from ~0.90 to ~0.96 nm; thus, it has the most attractive

anti-swelling capability in an aqueous environment. Moreover, the

unique self-polymerization characteristics of PDA further offers a

degree of controllable linking within GO laminates[13].

The suppression effect of an interlaminar molecular bridge on

the swelling of the GO laminate was further demonstrated by the

variation in membrane separation performance over operation

time (Figure 1b). As expected, typical swelling-induced

permeation behavior was observed in the pristine GO membrane:

a progressive decline of rejection along with a rise in permeance.

Remarkably, the interlaminar molecular-bridged GO membranes

well maintain the rejection and permeance once reaching steady

state. Here, the variation ratio of the water permeance or rejection

(i.e., the initial value of the normalized difference between the final

and initial values) was calculated to quantify the swelling-caused

variation in transport properties. As displayed in Figure 1c, the

GO-PDA membrane exhibited a minimal swelling tendency with

only a rejection variation of 0.4% and a water permeance variation

of 6.0%, which are significantly less than those of a pristine GO

membrane. The permeation result confirms that strong

interlaminar interactions are able to hinder excessive intercalation

of water molecules into the GO laminate, thereby preserving the

ordered interlayer channels at an appropriate size for selective

water transport.

To further explore the formation and critical role of an

interlaminar molecular bridge within a GO laminate, we

investigated and regulated the linking degree of GO-PDA. We

observed that the yellow-brown GO dispersion turned into a dark-

brown solution exhibiting a visible change in color with increasing

linking time. On basis of observations in UV-vis (Figure S3) and

XRD (Figure S4) meaturements, we envisioned that DA

monomers self-polymerized to form an entwining network of

covalent interlocked GO nanosheets[14], offering significantly

enhanced interlaminar interactions within the GO laminate. The

separation performance of GO-PDA membranes (Figure S5)

showed that the bulge in the water transport channels was

remarkably relieved as the PDA linking time was prolonged.

These results further demonstrate the favorable suppressed GO

10.1002/anie.201913010

Acc

epte

d M

anus

crip

t

Angewandte Chemie International Edition

This article is protected by copyright. All rights reserved.

RESEARCH ARTICLE

swelling resulting from the interlaminar molecular bridge that

closely interconnected adjacent GO nanosheets. However,

excessive PDA chains may also cause unfavorable blockage in

transport channels and thus lower the membrane permeance.

Figure 1. Anti-swelling capabilities of interlaminar short-chain molecular-bridged GO membranes. a, XRD patterns (left) of pristine GO and EDA-, PPD-, and PDA-

bridged GO membranes (color coded). Solid line: dry state; dotted line: wet state. Schematic illustrations (right) of GO-EDA, GO-PPD, and GO-PDA laminates.

Amine monomers covalently interlock GO nanosheets through condensation reactions with carboxyl groups and nucleophilic addition reactions with epoxy groups.

Amine monomers with aliphatic amine groups can form a strong nucleophile to generate high binding strength with GO nanosheets. Meanwhile, amine monomers

with large molecular configuration can exhibit high steric hindrance to afford much less electrostatic repulsion from negatively charged GO nanosheets. b, Operation

time dependence of water permeance and rejection of pristine GO and EDA-, PPD-, and PDA-bridged GO membranes (color coded). Solid line: water permeance;

dotted line: rejection rate. c, Relative variation ratios (initial value normalized difference between final value and initial value) of water permeance and rejection of

pristine GO and EDA-, PPD-, and PDA-bridged GO membranes. Blue: permeance variation; Gray: rejection variation.

Interfacial long-chain molecular bridge. Although the

interlaminar short-chain molecular bridge efficiently averts the

undesired redispersion of GO nanosheets in water, a GO laminate

could easily peel away from the substrate during the separation

process due to inadequate adhesion[12,14]. Intrinsically, flexible GO

nanosheets with a unique 2D structure tend to attach only to the

surface of the substrate, and the internal elastic strain energy from

bending further induces GO laminate peeling and freestanding[15].

Moreover, it is difficult to create a highly adhesive platform for a

porous substrate with a micrometer-scale rough and inert surface

using conventional surface modifications with short-chain

agents[16]. Here, we proposed to establish an interfacial long-

chain molecular bridge, which not only provided large contact

areas with the porous substrate layer but also generated chemical

binding with the GO laminate layer. Aldehyde (glutaraldehyde or

maleic anhydride)-modified chitosan (O=CS), with a moderate

molecular weight and abundant functional groups, was designed

as the interfacial long-chain molecular bridge. We expect that a

properly controlled molecular weight would enable O=CS to form

an ultrathin and uniform covering on a porous substrate with

appropriate pore penetration (Figure S6), thus offering sufficient

physical interaction with the underlying substrate[17]. Furthermore,

plentiful aldehyde groups of O=CS would be beneficial for

generating chemical bindings with the amine groups of the upper

GO-PDA laminate[18].

The AFM images (Figure 2a and b) show typical surface

topographies for a porous ceramic tube substrate before and after

O=CS modification, and its roughness remarkably dropped from

282.4 nm to 67.7 nm. This is due to the decoration of O=CS

molecular bridges on the substrate filling in the low-lying regions.

Meanwhile, the skeletons of the underlying substrate remained

clearly identified due to the ultrathin O=CS decoration. These

morphologies agreed well with SEM results (Figure 2c and d, left),

demonstrating an ultrathin decoration of O=CS molecular bridges

uniformly covering the surface of a ceramic substrate. In the

cross-sectional EDX mapping (Figure 2c and d, right),

overlapping of the N elements with the CS and Zr elements from

the ceramic suggested a transition layer formed by partial

penetration of the O=CS molecular bridge into the ceramic

substrate, which significantly enlarged the contact areas to yield

substantial physical van der Waals interactions between them.

We further assembled the GO-PDA laminate on the O=CS-

decorated ceramic substrate and performed XPS depth analysis

(Figure 2e) to distinguish the individual layers of GO-PDA/O=CS

/ceramic according to the contents of N elements (from the GO-

0

5

10

15

20

25 Permeance variation

Rela

tive v

ariation r

atio (

%)

GO-EDA GO-PPD GO-PDAGO

0.0

0.5

1.0

1.5

2.0

Rejection variation

Rela

tive v

ariation r

atio (

%)

0 1 2 3 4 5 63

6

9

12

GO

Perm

eance (

Lm

-2h

-1bar-1

)

Operation time (h)

GO-PDAGO-PPDGO-EDA

GO

GO-EDAGO-PDAGO-PPD

90

92

94

96

98

100

Re

jection

(%

)

5 6 7 8 9 10 11 12

Inte

nsity (

a.u

.)

2 (degree)

GO-PDA

GO-PPD

GO-EDA

Wet state

GO

Dry state d-spacing (nm)

0.99 0.84

0.96 0.90

1.02 0.91

1.45 0.80

GO-EDA

GO-PPD

GO-PDA

b c

a

10.1002/anie.201913010

Acc

epte

d M

anus

crip

t

Angewandte Chemie International Edition

This article is protected by copyright. All rights reserved.

RESEARCH ARTICLE

PDA laminate and O=CS chains) and Zr elements (from the

ceramic substrate) varying with cross-sectional depth[19]. The

transition region with crisscrossing N and Zr content at 140–160

nm depths further demonstrated the partial pore penetration of CS

molecular bridges into the porous ceramic substrate with

sufficient interaction contact areas. In addition to physically

interacting with the underlying ceramic substrate, the interfacial

O=CS molecular bridges formed chemical bindings with the upper

GO-PDA laminate as confirmed in the C1s XPS spectra (Figure

S7). A new peak arose at ~288.0 eV that can be assigned to C=N

bonding, suggesting the formation of a covalent connection

between PDA and O=CS[20]. In a further IR analysis (Figure S8),

the C=N stretching vibration at ~1690 cm-1 again indicated the

reaction of extra free aldehyde groups of O=CS with amine

groups from GO-PDA through a Schiff base reaction[21].

Figure 2. Adhesion characterizations of interfacial long-chain molecular-bridged GO membranes. AFM images with roughness data for a, the ceramic substrate

surface and b, the O=CS/ceramic substrate surface. c, Cross-sectional SEM images of O=CS-decorated ceramic substrate (left) with EDX mappings of selected

region (right). d, Surface view SEM images of O=CS-decorated ceramic substrate (left) with EDX mappings of selected region (right). Gray points: Zr signal; red

points: N signal. e, XPS depth analysis results of GO-PDA/O=CS/ceramic membrane. Red triangle: N element content from GO-PDA laminate and O=CS chains;

blue square: Zr element content from ceramic substrate. Depth 0–80 nm: GO-PDA laminate; depth 80–100 nm: transition region; depth 100–140 nm: O=CS

decoration; depth 140–160 nm: transition region; depth 160–200 nm: ceramic substrate. f, Cross-sectional SEM images of GO/ceramic tube with GO layer peeling

off the substrate. g, Cross-sectional SEM images of GO-PDA/O=CS/ceramic membrane with GO layer firmly adhering to the substrate. Insert: local enlarged image

of GO membrane layer with thickness of ~150 nm. h, Schematic illustration of nanoscratch technique. i, Results of nanoscratch measurements of GO/ceramic tube

and GO-PDA/O=CS/ceramic membrane. Red line: scratching depth signal of GO-PDA/O=CS/ceramic membrane with critical load of 46.3 mN; blue line: scratching

depth signal of GO/ceramic membrane with critical load of 20.9 mN; black line: scratching load.

The resulting GO-PDA/O=CS/ceramic membrane exhibited

excellent integrity without incurring any visible damage under

finger touch, whether in a dry or wet state, whereas the pristine

GO laminate easily peeled off the ceramic substrate. Further SEM

characterizations showed that the pristine GO/ceramic membrane

yielded an evident boundary with a huge crack between the GO

laminate and ceramic substrate (Figure 2f); this was caused by

only a slight vibration during sample preparation. Such severe

detachment of a GO laminate is due to its weak combination with

the substrate. In contrast, with the aid of O=CS as an interfacial

molecular bridge, the GO-PDA laminate firmly adhered to the

ceramic substrate without any gaps (Figure 2g).

We further employed the nanoscratch technique to in situ

quantify the membrane interfacial binding force in order to monitor

and record the scratch distance relationships with the dynamic

load (Figure 2h)[22]. The membrane layer with strong interfacial

adhesion with the substrate is expected to display greater

mechanical strength and endure a higher scratch load. The critical

load at failure is considered to be the interfacial binding force of

the membrane onto the substrate. As shown in the results of

nanoscratch measurements (Figure 2i), the GO-

PDA/O=CS/ceramic membrane had a critical load (46.3 mN)

more than twice that of the pristine GO/ceramic membrane (20.9

mN). Such excellent mechanical properties of the GO-

PDA/O=CS/ceramic membrane again proved that the created and

tuned interlaminar short-chain (PDA) and interfacial long-chain

(O=CS) molecular bridges significantly consolidated the

integrated GO membrane.

Membrane stability and strategy generality. Cross-flow, high-

pressure, and continuous filtration separation processes were

0 40 80 120 160 200

0

2

20

25

30 N element

Zr element

Ato

mic

(%

)

Etch depth (nm)

0 100 200 300 400

0

20

40

60

80

100

120

Scra

tch

ing

lo

ad

(m

N)

Displacement (m)

20.9 mN

46.3 mN

GO/Ceramic tube

Scra

tch

ing

de

pth

sig

na

l GO-PDA/O=CS/Ceramic tube

1 µm

Ra=282.4 nm

Rq=313.6 nm

Ceramic tube

1 µm

Ra=67.7 nm

Rq=89.5 nm

O=CS/Ceramic tube

500 nm

O=CS/Ceramic tube

1 µm

O=CS/Ceramic tube

c

d

ea

b

10 µm

10 µm

GO/Ceramic tube

~150 nm

hf

g

i

Stylus

GO laminate on substrate

Progressive scratch

Applied load

Displacement

GO-PDA laminate O=CS

decoration

Ceramic

substrate

GO-PDA/O=CS/Ceramic tube

10.1002/anie.201913010

Acc

epte

d M

anus

crip

t

Angewandte Chemie International Edition

This article is protected by copyright. All rights reserved.

RESEARCH ARTICLE

performed on GO membranes to validate their stability and

durability in water. As shown in Figure 3a, the GO-

PDA/O=CS/ceramic membrane exhibited remarkable integrity

while maintaining excellent rejection and stable permeance when

subjected to a cross-flow feed at 50 mL/min for 300 min. Figure

3b demonstrated that the water flux showed an approximately

linear increase with no slowdown, and the rejection rate displayed

a steady high value as the feed pressure increased to 10 bar. This

suggested that the PDA interlaminar molecular bridge not only

suppressed the swelling of the GO laminate, but also enhanced

the rigidity of GO nanochannels so they could withstand high

transmembrane pressure. The results of the long-term continuous

filtration operation in Figure 3c demonstrated excellent stability of

the GO-PDA/O=CS/ceramic membrane over a duration of 25

days. In comparison, the pristine GO/ceramic membrane was

incapable of withstanding shear stress[10], and fragmented GO

pieces were easily carried away when applying the above cross-

flow condition.

Figure 3. Stability and durability of GO-PDA/O=CS/ceramic membrane. Nanofiltration performance of GO-PDA/O=CS/ceramic membrane under a, cross-flow up

to 300 min, b, transmembrane pressure up to 10 bar, and c, continuous operation up to 25 days. Blue: water permeance; gray: rejection rate. d, Photos of GO/ceramic,

GO-PDA/ceramic, and GO-PDA/O=CS/ceramic membranes after high-power sonication for 30 min in water. e, Comparison of long-term operation times for state-

of-the-art two-dimensional material membranes applied in water separation[10,11,24-27]. More details about the membrane and operation conditions are provided in

Table S1.

Cleaning processes involving vibration and sonication are

often adopted when reusing membranes in practical separation

processes[23]. To further evaluate the structural stability under

cleaning treatment, the pristine GO/ceramic, GO-PDA/ceramic,

and GO-PDA/O=CS/ceramic membranes were subjected to high-

power (40 kHz frequency) sonication for 30 min in water. As

revealed in Figure 3d, the violent vibration from sonication

crumbled the pristine GO membrane and caused exfoliation of

GO stacks into small pieces. The GO-PDA membrane maintained

the integrity of the GO laminate, implying that the PDA

interlaminar molecular bridge greatly enhanced the interlaminar

interactions of the GO laminate to prevent its disassembly.

Unfortunately, the layer entirely peeled off and disclosed the

underlying white ceramic substrate because of insufficient

interfacial interaction between the laminate and substrate. In

contrast, by simultaneously stabilizing the GO layer and

GO/substrate interface, the integrated GO-PDA/O=CS/ceramic

membrane remained intact without any visible gaps, cracks, or

damages.

We summarized representative results in the recent literature

involving stability evaluations of two-dimensional material

membranes, including membranes based on GO[6a, 7, 9b, 10, 11, 15, 20,

24], covalent-organic-frameworks (COF)[25], g-C3N4[26], and

MXene[27] (Table S1), and illustrated their long-term operation

time in water-based separations, such as desalination,

nanofiltration, and pervaporation (Figure 3e). Promisingly, the

superior stability of our GO-PDA/O=CS/ceramic membrane offers

a greater competitive advantage over other counterparts,

especially in long-term operations that demand much more

stringent requirements for membrane stability.

Additionally, it is well known that membrane stability is

correlated with the physicochemical properties of GO materials[28].

Thus, we also tried to stabilize GO membranes by tuning the

physical and chemical characteristics of GO nanosheets (Figure

S9-S12). Due to the inevitable water swelling, it is not surprising

that the d-spacings of wet GO membranes were remarkably larger

than those of dry GO membranes, regardless of the lateral size

and oxidation degree of GO (Figure 4a and Figure S13). The GO

0 200 400 600

Long-term operation time (h)

GO-PDA/O=CS/ceramic

COF@CNF/PAN

GO/PDA-ceramic

GO-PDA/PES

PDI-GOF/ceramic

Cation-GO/PAN

COF-LZU1/ceramic

Polycation-GO/PAN

Fe(OH)3/g-C

3N

4/AAO

MXene/AAO

MXene/nylon

rGO-MWCNT/PC

This work

60 90 120 150 180 210 240 270 3000

2

4

6

8

Pe

rme

an

ce

(L

m-2h

-1b

ar-1

)

Operation time (min)

90

92

94

96

98

100R

eje

ctio

n (

%)

0 5 10 15 20 250

2

4

6

8

Perm

eance (Lm

-2h

-1bar-1

)

Operation time (day)

90

92

94

96

98

100

Reje

ction (

%)

0 2 4 6 8 100

10

20

30

40

50

Flu

x (Lm

-2h

-1)

Operation pressure (bar)

90

92

94

96

98

100

Re

jectio

n (

%)

a cb

e

GO/Ceramic GO-PDA/Ceramic GO-PDA/O=CS/Ceramic

d After sonication in water

10.1002/anie.201913010

Acc

epte

d M

anus

crip

t

Angewandte Chemie International Edition

This article is protected by copyright. All rights reserved.

RESEARCH ARTICLE

membrane derived from a small lateral size (SGO) suffered the

most severe swelling, with d-spacing stretching from 0.80 nm to

1.52 nm, and the highly reduced, large-lateral-sized GO

membrane (rLGO-3) yielded the slightest d-spacing stretching

from 0.69 nm to 0.97 nm. The lowest d-spacing expansion (0.28

nm) was achieved by increasing the lateral size and reducing the

oxidation degree of GO; however, this was more than 400% larger

than that of the GO-PDA membrane (merely 0.06 nm). These

results clearly demonstrated that the nonlinked GO membranes,

even those with the fewest hydrophilic groups and narrowed

interlayer spacing (rLGO-3) after careful tuning, endured a greater

swelling tendency than the GO-PDA membrane. The above

control experiments highlighted the indispensable role of creating

and tuning molecular bridges in order to stabilize GO membranes

with desirable anti-swelling capability in an aqueous environment.

Figure 4. Generality and scalability of the molecular-bridge stabilization strategy

for GO membranes. a, D-spacing expansion of GO membranes with different

lateral sizes and oxygenated functionality in dry and wet states as calculated

from their XRD results. Gray: dry state; blue: wet state. Cross-sectional SEM

images of b, GO-PDA/O=CS/nylon and c, GO-PDA/O=CS/MCE membranes.

Inserts are photographs of each membrane. d, Photograph of scaled inner

surface GO-PDA/O=CS/ceramic membrane with 40 cm length.

The generality of the proposed molecular-bridge stabilization

strategy for GO membranes was demonstrated earlier by utilizing

various amines (e.g., EDA, PPD, and PDA) for interlaminar short-

chain molecular bridges, as well as alterable O=CS derived from

glutaraldehyde or maleic anhydride for interfacial long-chain

molecular bridges. More importantly, this methodology is

applicable to other porous substrates beyond the ceramics used

above. We took commonly used nylon and mixed- cellulose

acetate (MCE) substrates with a nominal pore size of ~200 nm as

examples (Figure S14 and S15). As shown in Figure 4b and c,

both photographs and SEM images revealed a GO-PDA laminate

that was closely immobilized upon the O=CS-bridged substrate

with outstanding integrity. The separation performance and

stability in water of the as-prepared GO-PDA/O=CS/nylon and

GO-PDA/O=CS/MCE membranes were tested under a cross-flow

feed condition, and they exhibited comparable water permeance

of 7.6 and 8.1 L m-2 h-1 bar-1 with remarkable dye rejection of

98.9% and 98.3%, respectively.

Scale-up is another key step for implementation of GO

membranes[29]. Compared with polymers, ceramics show

advantages in structural, chemical, and thermal stability[17]. The

scalability of a ceramic-supported membrane can be further

enhanced if overcoming an inner-surface membrane technology

that offers better protection of the separation layer as well as

higher packing density if using multi-channels[24d]. In this work, we

successfully scaled up our stabilized GO membrane supported on

the inner surface of a ceramic tube with length from 7 cm to 40

cm (Figure 4d), using PDA as an interlaminar molecular bridge

and O=CS as an interfacial molecular bridge. The large-area

(~100 cm2) stabilized GO membrane exhibited very promising

nanofiltration performance, with water permeance of 3.6 L m-2 h-1

bar-1 and dye rejection of 97.2%.

Conclusion

In conclusion, we presented a facile and efficient

methodology for stabilizing GO membranes by creating and

tuning molecular bridges. A properly designed amine acted as an

interlaminar short-chain molecular bridge to interlock the GO

laminate through condensation and nucleophilic addition

reactions, contributing to a mechanically robust GO assembly

structure with anti-swelling capability. Meanwhile, ultrathin O=CS

served as an interfacial long-chain molecular bridge to covalently

bond with the amine-interlocked GO laminate through the Schiff

base reaction. It firmly adhered to a porous substrate through

sufficient physical attachment areas. The molecular-bridged GO

membrane featured superior stability in a harsh water separation

process, showing excellent durability in cross-flow, high-pressure,

and long-term filtration and the ability to withstand vibration and

sonication treatment. This facile molecular-bridge stabilization

strategy, with additional generality and scalability, paves a new

avenue toward the implementation of new two-dimensional

laminar films for reliable water-related applications, such as water

purification, desalination, anti-corrosion and anti-fouling coatings,

and energy storage and conversion.

Experimental Section

Creating interlaminar molecular bridge. GO powder was dissolved in

deionized water followed by ultrasonication to obtain a 0.01 mg/mL GO

dispersion. 0.4 g tris (hydroxymethyl) aminomethane was dissolved in 200

0.0

0.3

0.6

0.9

1.2

1.5

1.8 Dry state

Wet state

d-s

pacin

g (

nm

)

SGO MGO LGO rLGO-1 rLGO-2 rLGO-3 GO-PDA

0.72 0.70.65

0.61

0.38

0.28 0.06

1 µm

GO-PDA/O=CS/Nylon

1 µm

GO-PDA/O=CS/MCE

GO-based membrane on inner surface

of ceramic tube

40 cm

a

b c

d

10.1002/anie.201913010

Acc

epte

d M

anus

crip

t

Angewandte Chemie International Edition

This article is protected by copyright. All rights reserved.

RESEARCH ARTICLE

mL dilute GO dispersion followed by adjusting the pH to 8.5 through adding

1 M HCl dropwise. Then, 0.1 g DA was added in the aforementioned GO

dispersion under stirring at room temperature to obtain PDA linked GO

dispersion. Similarly, 0.1 g diamine monomer (EDA or PPD) was added in

the GO dispersion under stirring to prepare EDA and PPD linked GO

dispersion, respectively.

Creating interfacial molecular bridge. CS solution was prepared by

dissolving 0.05 g CS in a 50 mL 0.5 wt% acetic acid aqueous solution.

Excess GA (25% aqueous solution) or MA (0.1 M) with a high molar ratio

of crosslinkers to CS units (2:1) was added in the CS solution and reacted

under stirring for 2 min. The final O=CS solution was dip-coated on the

surface of ceramic tube and kept still for 5 min before being poured out.

The resultant ceramic tube was rinsed with deionized water and dried at

room temperature.

Preparation of GO membrane on inner surface of ceramic tube.

Ceramic tube was mounted into tube component and sealed by rubber O-

rings on both ends of the substrate. A solution container was connected to

the tube component, and the dilute GO dispersion was drove onto the inner

side of the tube by pressured nitrogen under 2 bar. The flat substrate was

mounted in a self-designed filtration device and sealed by a rubber gasket,

and the dilute GO dispersion was deposited on the substrates driven by

pressured nitrogen under 2 bar as well. It worth noticing that all the GO

solution should be filtrated in order to avoid the redispersion of GO

membrane when relieving the driving pressure. All the as-prepared GO

membranes were dried and solidified overnight at 60 ºC in vacuum before

use.

Acknowledgements

This work was financially supported by the National Natural

Science Foundation of China (Nos. 21490585, 51861135203,

21922805, 21776125), the Innovative Research Team Program

by the Ministry of Education of China (No. IRT_17R54) and the

Topnotch Academic Programs Project of Jiangsu Higher

Education Institutions (TAPP). We thank Prof. Xiaoming Ren for

helpful discussion.

Keywords: graphene oxide • molecular bridge • stability •

membrane • nanofiltration

[1] a) R. R. Nair, H. A. Wu, P. N. Jayaram, I. V. Grigorieva, A. K. Geim,

Science 2012, 6067, 442-444; b) G. Liu, W. Jin, N. Xu, Chem. Soc. Rev.

2015, 44, 5016-5030; c) B. Mi, Science 2014, 343, 740-742.

[2] a) R. K. Joshi, P. Carbone, F. Wang, V. G. Kravets, Y. Su, I. V. Grigorieva,

H. Wu, A. K. Geim, R. R. Nair, Science 2014, 343, 752-754; b) M. Hu, B.

Mi, Environ. Sci. Technol. 2013, 47, 3715-3723.

[3] a) L. Chen, G. Shi, J. Shen, B. Peng, B. Zhang, Y. Wang, F. Bian, J.

Wang, D. Li, Z. Qian, G. Xu, G. Liu, J. Zeng, L. Zhang, Y. Yang, G. Zhou,

M. Wu, W. Jin, J. Li, H. Fang, Nature 2017, 550, 415-418; b) A. Morelos-

Gomez, R. Cruz-Silva, H. Muramatsu, J. Ortiz-Medina, T. Araki, T.

Fukuyo, S. Tejima, K. Takeuchi, T. Hayashi, M. Terrones, M. Endo, Nat.

Nanotechnol. 2017, 12, 1083–1088; c) M. Zhang, K. Guan, Y. Ji, G. Liu,

W. Jin, N. Xu, Nat. Commun. 2019, 10, 1253.

[4] C. Cui, A. Lim, J. Huang, Nat. Nanotechnol. 2017, 12, 834.

[5] a) J. Huang, J. Huang, T. Zhuang, Q. Zhang, H. Peng, C. Chen, F. Wei,

ACS Nano 2015, 9, 3002-3011; b) E. S. Cho, A. M. Ruminski, S. Aloni,

Y. Liu, J. Guo, J. J. Urban, Nat. Commun. 2016, 7, 10804.

[6] a) C. Yeh, K. Raidongia, J. Shao, Q. Yang, J. Huang, Nat. Chem. 2015,

7, 166-170; b) S. Zheng, Q. Tu, J. J. Urban, S. Li, ACS Nano 2017, 11,

6440-6450.

[7] Y. Xi, J. Hu, Z. Liu, R. Xie, X. Ju, W. Wang, L. Chu, ACS Appl. Mat.

Interfaces 2016, 8, 15557-15566.

[8] J. Yang, D. Gong, G. Li, G. Zeng, Q. Wang, Y. Zhang, G. Liu, P. Wu, E.

Vovk, Z. Peng, X. Zhou, Y. Yang, Z. Liu, Y. Sun, Adv. Mater. 2018, 30,

1705775.

[9] a) J. W. Burress, S. Gadipelli, J. Ford, J. M. Simmons, W. Zhou, T.

Yildirim, Angew. Chem. Int. Ed. 2010, 49, 8902-8904; b) Y. Zhang, S.

Zhang, T. Chung, Environ. Sci. Technol. 2015, 49, 10235-10242.

[10] K. Goh, W. Jiang, H. E. Karahan, S. Zhai, L. Wei, D. Yu, A. G. Fane, R.

Wang, Y. Chen, Adv. Funct. Mater. 2015, 25, 7348-7359.

[11] C. Wang, Z. Li, J. Chen, Y. Yin, H. Wu, Appl. Surf. Sci. 2018, 427, 1092-

1098.

[12] W. Hung, C. Tsou, M. De Guzman, Q. An, Y. Liu, Y. Zhang, C. Hu, K.

Lee, J. Lai, Chem. Mater. 2014, 26, 2983-2990.

[13] H. Lee, S. M. Dellatore, W. M. Miller, Science 2007, 318, 426-430.

[14] S. Wan, F. Xu, L. Jiang, Q. Cheng, Adv. Funct. Mater. 2017, 27, 1605636.

[15] R. Hu, Y. He, M. Huang, G. Zhao, H. Zhu, Langmuir 2018, 34, 10569-

10579.

[16] M. Zhang, J. Sun, Y. Mao, G. Liu, W. Jin, J. Membr. Sci. 2019, 574, 196-

204.

[17] K. Huang, G. Liu, J. Shen, Z. Chu, H. Zhou, X. Gu, W. Jin, N. Xu, Adv.

Funct. Mater. 2015, 25, 5809-5815.

[18] W. Zhang, Q. Zhao, J. Yuan, Angew. Chem. Int. Ed. 2018, 57, 6754-6773.

[19] Y. Ji, M. Zhang, K. Guan, J. Zhao, G. Liu, W. Jin, Adv. Funct. Mater. 2019,

29, 1900735.

[20] Y. T. Nam, J. Choi, K. M. Kang, D. W. Kim, H. T. Jung, ACS Appl. Mat.

Interfaces 2016, 8, 27376-27382.

[21] Y. Xin, J. Yuan, Poly. Chem. 2012, 3, 3045-3055.

[22] J. Shen, G. Liu, K. Huang, W. Jin, K. R. Lee, N. Xu, Angew. Chem. Int.

Ed. 2015, 54, 578-582.

[23] B. V. Bruggen, M. Mänttäri, M. Nyström, Sep. Purif. Technol. 2008, 63,

251-263.

[24] a) B. Feng, K. Xu, A. Huang, Desalination 2016, 394, 123-130; b) K. Xu,

B. Feng, C. Zhou, A. Huang, Chem. Eng. Sci. 2016, 146, 159-165; c) X.

Xu, F. Lin, Y. Du, X. Zhang, J. Wu, Z. Xu, ACS Appl. Mat. Interfaces 2016,

8, 12588-12593; d) M. Zhang, K. Guan, J. Shen, G. Liu, Y. Fan, W. Jin,

AIChE J. 2017, 63, 5054-5063; e) N. F. D. Aba, J. Y. Chong, B. Wang,

C. Mattevi, K. Lia, J. Membr. Sci. 2015, 484, 87-94; f) K. Guan, Q. Liu, G.

Zhou, G. Liu, Y. Ji, G. Liu, W. Jin, Sci. China Mater. 2019, 62, 925-935;

g) Y. Xi, Z. Liu, Q. Liao, R. Xie, X. Ju, W. Wang, Y. Faraj, L. Chu, Ind.

Eng. Chem. Res. 2018, 57, 9439-9447; h) L. Wang, N. Wang, J. Li, J. Li,

W. Bian, S. Ji, Sep. Purif. Technol. 2016, 160, 123-131; i) C. M. Kim, S.

Hong, R. Li, I. S. Kim, P. Wang, ACS Sustain. Chem. Eng. 2019, 7, 7252-

7259; j) A. L. Ghaffar, L. Zhang, X. Zhu, B. Chen, Environ. Sci. Nano

2019, 6, 904-915.

[25] a) H. Fan, J. Gu, H. Meng, A. Knebel, J. Caro, Angew. Chem. Int. Edit

2018, 57. 4083-4087; b) H. Yang, L. Yang, H. Wang, Z. Xu, Y. Zhao, Y.

Luo, N. Nasir, Y. Song, H. Wu, F. Pan, Z. Jiang, Nat. Commun. 2019, 10,

2101.

[26] Y. Wang, L. Liu, J. Hong, J. Cao, C. Deng, J. Membr. Sci. 2018, 564,

372-381.

[27] a) L. Ding, Y. Wei, Y. Wang, H. Chen, J. Caro, H. Wang, Angew. Chem.

Int. Edit 2017, 56, 1825-1829; b) J. Wang, P. Chen, B. Shi, W. Guo, M.

Jaroniec, S. Qiao, Angew. Chem. Int. Edit 2018, 57, 6814-6818.

[28] Y. Cho, H. W. Kim, H. D. Lee, J. E. Shin, B. M.Yoo, H. B. Park, J. Membr.

Sci. 2017, 544, 425-435.

[29] a) A. Akbari, P. Sheath, S. T. Martin, D. B. Shinde, M. Shaibani, P. C.

Banerjee, R. Tkacz, D. Bhattacharyya, M. Majumder, Nat. Commun.

2016, 7, 10891; b) J. Zhong, W. Sun, Q. Wei, X. Qian, H. Cheng, W. Ren,

Nat. Commun. 2018, 9, 3484.

10.1002/anie.201913010

Acc

epte

d M

anus

crip

t

Angewandte Chemie International Edition

This article is protected by copyright. All rights reserved.

RESEARCH ARTICLE

Entry for the Table of Contents

A facile stabilization strategy of GO membranes was proposed by creating and tuning

interlaminar short-chain and interfacial long-chain molecular bridges.

Mengchen Zhang, Yangyang Mao,

Guozhen Liu, Gongping Liu,* Yiqun

Fan and Wanqin Jin*

Page No. – Page No.

Molecular bridges stabilize

graphene oxide membranes in

water

Interlaminar short-chain

molecular bridge

Interfacial long-chain

molecular bridge

GO nanosheet

Porous substrate

10.1002/anie.201913010

Acc

epte

d M

anus

crip

t

Angewandte Chemie International Edition

This article is protected by copyright. All rights reserved.