Article

pubs.acs.org/est

Microscopic Characterization of FO/PRO Membranes − A

Comparative Study of CLSM, TEM and SEM
Yi-Ning Wang,†,‡ Jing Wei,†,‡ Qianhong She,†,‡ Federico Pacheco,§ and Chuyang Y. Tang*,†,‡
†
School of Civil and Environmental Engineering, Nanyang Technological University, Singapore 639798
‡
Singapore Membrane Technology Centre, Nanyang Technological University, Singapore 639798
§
Environmental Engineering and Science, Department of Civil and Environmental Engineering, Stanford University, Stanford,
California 94305-4020, United States
*
S Supporting Information

ABSTRACT: Osmotically driven membrane processes (including forward osmosis

(FO) and pressure retarded osmosis (PRO)) have received increasing attention in
recent decades. The performance of an FO/PRO membrane is significantly limited by
internal concentration polarization, which is a strong function of the membrane
support layer pore structure. The objective of the current study was to develop
microscopic characterization methods for quantitative/semiquantitative analysis of
membrane pore structure (both pore diameter and porosity distribution across
membrane thickness). The use of confocal laser scanning microscopy (CLSM) for
noninvasive characterization of the internal pore structure of FO/PRO membranes is
reported for the first time. By performing optical sectioning, information on pore
diameter, porosity depth profile and pore connectivity can be obtained. The CLSM
porosity results are further compared to those obtained using scanning electron
microscopy (SEM) and transmission electron microscopy (TEM), and reasonably
good agreement was observed. A comparison of these characterization methods
reveals their complementary nature, and a combination of these techniques may allow a more comprehensive understanding of
membrane structure. The current study also provided a comprehensive insight into the pore structure of commercially available
FO/PRO membranes.

■ INTRODUCTION
Osmotically driven membrane (OM) processes, including
forward osmosis (FO) and pressure retarded osmosis (PRO),
have received increasing attention in the past decades.1,2 FO
involves the water transport across a semipermeable membrane
from a low-osmotic-pressure feed solution (FS) to a high-
osmotic-pressure draw solution (DS) in the absence of an
externally applied hydraulic pressure. In PRO, a hydraulic
pressure is applied to the DS, which retards the water flux.
Nevertheless, as long as the applied pressure is smaller than the
osmotic pressure difference across the membrane, water will
still permeate through the membrane from the FS to the DS. In
recent years, FO has gained significant interests for applications
in wastewater treatment3−7 and seawater desalination,8,9
whereas PRO is explored for osmotic power production.10−15
The performance of OM processes can be severely limited by
concentration polarization (CP). In addition to the CP that
occurs at the solution−membrane interface (i.e., the external
CP or ECP), an FO/PRO membrane also experiences internal
CP (ICP) due to the accumulation of feed solutes or the
dilution of DS inside its porous membrane support layer.16−18
ECP is generally moderate19,20 and can be effectively controlled
by increasing crossflow velocity or by the use of spacers,21
whereas ICP tends to severely limit the flux efficiency of FO/
PRO membranes (in many cases by an order of magni-
tude).11,16,17,19,21,22 Prior studies have shown that the intensity
of ICP is strongly affected by the pore structure of the support
layer. Thinner, more porous and less tortuous support layers
are generally preferred to reduce ICP.16,17,21−25 Therefore, pore
structure characterization of FO/PRO membranes is critical for
a better understanding of their flux performance.
Only a handful of characterization methods have been
reported in the literature on studying the porosity/pore
structure of FO/PRO membranes. Dry-wet weighing method
determines the average porosity of a membrane, but this
method does not provide any information on membrane pore
structure.22,26 In existing OM literature, scanning electron
microscopy (SEM) is the most commonly employed method
for studying the internal structure of FO/PRO mem-
branes.22,26−28 It is generally applied to examine the cross-
section of a membrane to provide information on pore
diameter, pore shape (e.g., spongy-like vs finger-like pore
structures), and pore connectivity (how membrane pores are
connected to each other).22 Compared to SEM, transmission
electron microscopy (TEM) can provide significantly higher
Received: May 10, 2012
Revised: August 16, 2012
Accepted: August 16, 2012
Published: August 16, 2012

© 2012 American Chemical Society 9995 dx.doi.org/10.1021/es301885m | Environ. Sci. Technol. 2012, 46, 9995−10003
Environmental Science & Technology
Table 1. Membrane Properties of Commercial FO/PRO Membranes Used in the Current Study
thickness (μm)a
membrane TEM SEM contact angle (deg)b
CTA-HW 103 ± 4 106 ± 6 63 ± 3
CTA-W 91 ± 2 86 ± 5 73 ± 2
Notes:
a
From current study. bFrom ref 22. cTested in crossflow reverse osmosis filtration mode with 10 mM NaCl as feed solution.23
resolution. Although it has been widely used for characterizing
pressure-driven membranes (such as reverse osmosis and
nanofiltration membranes),29−31 its use in FO/PRO membrane
characterization has not been documented. Both SEM and
TEM are destructive methods. In this regard, light microscopy
may serve as a useful alternative.21 In particular, confocal laser
scanning microscopy (CLSM) can be potentially valuable in
providing depth-dependent porosity information in a non-
invasive way. CLSM can acquire in-focus images at desired
depth by performing optical sectioning of a sample.32 This
method has been used to study membrane interior structures in
the context of microfiltration (MF) and ultrafiltration (UF)
membranes33−35 in addition to its use in membrane biofouling
characterization.36 For FO/PRO membranes, it is known that
the porosity distribution across the membrane thickness plays a
critical role on the mass transport in the porous structure.26
Nevertheless, to the best knowledge of the authors, there has
not been any published work on the quantitative analysis of
such structural information. The ability of CLSM in providing a
porosity-depth profile and a three-dimensional pore imaging
can thus be extremely valuable for enhancing our understanding
of the concentration polarization phenomenon during OM
processes.37
The objectives of current study were (1) to compare the use
of different microscopic methods (SEM, TEM, and CLSM) for
the characterization of FO/PRO membranes, and (2) to
provide a quantitative (or semiquantitative in some cases)
analysis of membrane pore structure for commercially available
FO/PRO membranes. To the best knowledge of the authors,
this is the first comprehensive study of using CLSM to
characterize the internal pore structure of FO/PRO mem-
branes. Improved knowledge of FO/PRO membrane pore
structure can help process engineers to better understand mass
transport inside these membranes and thus to better predict
their performance in OM processes for environmental
applications.
■ MATERIALS AND METHODS
Membranes and Chemicals. Unless otherwise specified,
all the chemicals used in the study were of analytical grade and
were used without further purification. Ultrapure water with a
resistivity of 18.2 MΩ.cm (Millipore Integral 10 Water
Purification System) was used to prepare all working solutions.
The two commercially available FO/PRO membranes were
obtained from Hydration Technology Innovations (HTI,
Albany, OR). Both membranes are made of cellulose triacetate
(CTA) and are supported with embedded woven mesh. The
membrane cut from a HydroWell module was denoted as CTA-
HW. The other membrane was received as a flat coupon, and
was denoted as CTA-W in accordance with a previous study.22
The surface properties and separation characteristics (water
permeability and salt rejection) have been well characterized in
our works22,23 and are summarized in Table 1. LR White resin
Article
pure water permeability, A ( × 10−12 m/s.Pa)c NaCl permeability, B ( × 10‑8 m/s)c
2.55 ± 0.11 27.2 ± 3.1
1.07 ± 0.12 6.45 ± 0.53
used for membrane embedding during TEM sample prepara-
tion was supplied by Polysciences (Warrington, PA).
Fluorescein isothiocyanate (FITC, Fluka 46950) was used for
staining membrane samples prior to CLSM imaging.
Characterization Methods. SEM. Membrane surfaces and
cross sections were characterized with a Zeiss EVO 50 scanning
electron microscope (Carl Zeiss Pte Ltd.). To obtain cross
sections, membrane samples were frozen in liquid nitrogen and
fractured manually. All samples were dried in vacuum at room
temperature (∼24 °C) for 24 h, and sputter coated with a thin
layer of gold using EMITECH SC7620 sputter coater (Quorum
Technologies Ltd., UK). The cross sections were imaged at an
accelerating voltage of 10 keV. The membrane active layer was
further scanned at accelerating voltages of 2, 5, 10, 20, and 25
keV to obtain morphological information at different sampling
depths.38
TEM. The cross sections of FO/PRO membranes were also
characterized with a JEM 1230 Electron Microscopy at an
accelerating voltage of 80 keV. Aromatic acrylic LR white resin
was used to embed dehydrated samples in capsules according to
our previously reported procedure.29 Briefly, membrane
samples were dehydrated by several changes of ethanol (3
times 100% ethanol, 15 min for each step). Samples were then
infiltrated in 50, 67, and 100% LR White resin (volumetric %,
prepared in ethanol) sequentially, followed by embedding them
in fresh LR White resin that was subsequently polymerized at
48 °C for 3 days. After complete polymerization of the resin,
thin TEM sections (<100 nm thickness) were cut with a
diamond knife using Leica Ultracut S ultramircotome (Leica,
Wetzlar, Germany) and transferred onto copper TEM grids for
imaging. Similar procedures are commonly adopted to prepare
TEM biological samples.39 Although TEM sample preparation
procedures might result in some slight changes in feature
dimensions, the overall dimensional change is generally less
than 10% from its hydrated state.39
CLSM. An Olympus Fluoview FV300 confocal laser scanning
microscopy (CLSM, Olympus Optical, Tokyo, Japan) was used
in both fluorescent mode and reflective mode to visualize the
membranes. Membrane samples were stained by a fluorescent
dye FITC that has excitation/emission wavelengths of 488/
500−550 nm. A membrane sample of ∼1 × 1 cm was cut and
immersed in 1 g/L FITC solution for ∼0.5 h and was then
rinsed with ultrapure water. The sample, while wet, was
mounted onto a glass slide and covered by a thin coverslip prior
to imaging. Lateral x−y scan was performed with a ×40
objective, and images were stored with a resolution of 512 ×
512 pixels (representing an area of 360 × 360 μm). Scan was
performed either from the active layer (active layer facing the
objective lens) or from back support surface (support layer
facing the objective lens). Images were collected at a series of
sampling depth at a scanning depth increment of 1 μm in z-
direction to obtain the internal structure of membrane samples.
For the ease of reporting, we set z = 0 at the membrane active
9996 dx.doi.org/10.1021/es301885m | Environ. Sci. Technol. 2012, 46, 9995−10003
Environmental Science & Technology
Figure 1. Cross sections of membrane CTA-HW. (a) TEM image; and (b) SEM image.
layer surface, with the positive z-direction pointing toward the
supporting layer.
Image Analysis. TEM, SEM, and CLSM images were
analyzed using an image analysis software (MediaCybernetics,
Inc.).21 The CLSM fluorescent images and SEM images (active
surface) were converted to gray scale images that contained 8
bits of data per pixel. The gray scale images were then
transformed to binary images (i.e., black and white images) by
adjusting the threshold value (see Supporting Information S1).
The porosity was determined as the ratio of the total void area
to the gross area analyzed using the software. The sampled area
for each analysis was about 100 × 100 μm for CLSM images,
and was about 38 × 57 μm for SEM images. TEM cross-
sectional images were analyzed using the same software to
determine the porosity distribution across the depth. The
analysis was performed over a sample width of 100 μm. The
cross-section was divided into segments that were parallel to
the active layer surface for determining porosity at different
sample depth. In all the microscopic methods (CLSM, SEM,
and TEM), multiple independent sample locations (≥3) were
taken to determine the mean porosity for each depth. Unless
specified otherwise, membrane area adjacent to the woven
mesh was excluded for the pore diameter and porosity
determination.
9997
Article
■ RESULTS AND DISCUSSIONS
TEM and SEM Observation of Membrane Cross
Sections. Figure 1 presents the TEM and SEM images of
the cross-section of membrane CTA-HW. Additional TEM and
SEM images are also presented in the Supporting Information
S2. Both the TEM (Figure 1(a)) and SEM (Figure 1(b))
images show a membrane thickness of ∼100 μm (Table 1).
The TEM images were able to provide clear structural details.
For example, the woven polyester mesh fibers (represented by
the dark circles) were clearly visible in the lower-magnification
TEM image, with a fiber-to-fiber distance of about 160 μm. In
the higher-magnification TEM image, the active layer can be
differentiated from the support layer since pores can only be
seen in the support layer. The active layer had a thickness of
∼2−3 μm, beneath which some small pores (5−10 μm) can be
observed. In addition, larger elongated finger-like pores were
also present, and these pores are believed to greatly facilitate
the mass transport in the porous support.12,22 The pore
diameter distribution and porosity profile of CTA-HW based
on the TEM image analysis are presented in the section
“Membrane Pore Diameter and Porosity Distribution”.
Compared to the highly porous structure of CTA-HW, the
TEM image of CTA-W membrane does not show pores across
its thickness (Supporting Information S3).
dx.doi.org/10.1021/es301885m | Environ. Sci. Technol. 2012, 46, 9995−10003
Environmental Science & Technology Article

Figure 1(b) shows the SEM cross-sectional image of CTA-

no special preparation required

overestimation of porosity due

to water absorption by poly-
dry−wet weighing method
HW. A difference in cross-section thickness can be noticed by

low cost; simple and fast.

comparing Figure 1(a) and (b): the TEM image shows a cross-
section with nearly homogeneous thickness (∼103 μm),

meric material
average porosity
whereas the SEM image shows one with varied thickness

not applicable
(greater thickness of ∼106 μm for where a mesh fiber is located
and thinner thickness of ∼93 μm for locations away from the
mesh). This difference is more pronounced for membrane

yes

no
CTA-W (Supporting Information S3). The observed undulat-

3-D pore structure and pore connectivity; quantitative depth

depth; not able to sample the entire membrane thickness

no special preparation required; imaging can be performed
ing structure in the SEM images may be explained by the fact

possible interference of fluorescent signal at large sample

microscale (∼ 0.5 μm in z direction and ∼0.2 μm in xy
that SEM samples were scanned in a dry state. Indeed, direct
optical microscopic observation showed significant dimensional
differences between dry and wet samples (Supporting
Information S4). For TEM observations, some slight feature

under wet conditions (hydrated state)

profiles of pore diameter and porosity

changes may also be introduced during sample preparation as a
result of alcohol dehydration and resin embedding. In this

CLSM
regard, CLSM (see section “CLSM Images”), which allows

for thick samples (>80 μm)

microscopic imaging of wet membrane samples, is preferred to
avoid artifacts arising from sample preparation, despite its lower
resolution (Table 2).

moderate cost; fast

Both TEM and SEM cross sections of CTA-HW (Figure 1)
revealed the presence of a support skin. The pore structure of

direction)
this skin may significantly influence FO/PRO membrane
fouling behaviors.40,41 The high-magnification SEM image

yes

yes
(Figure S5(b), Supporting Information S2) further shows
that the support skin have small pores (diameter ∼10−40 nm),

embedding) may potentially cause some mild

extensive sample preparation (e.g., dehydration,

nanoscale (atomic scale for crystalline samples)

2-D (cross-sectional) pore structure and pore

connectivity; quantitative depth profiles of
which agrees reasonably well with a prior study.42
resin embedding, and sample sectioning)

sample preparation (dehydration and resin

SEM Observation of the Membrane Active Layer. SEM
was further used to characterize the pores near the active layer
of membrane CTA-HW. The accelerating voltage was varied

expensive and time-consuming

pore diameter and porosity
over 2 − 25 keV to obtain different sample depth penetration.
TEM

By using a higher SEM accelerating voltage, the electrons can

Table 2. Comparison of Methods for FO/PRO Membrane Porosity Characterization

dimensional change
penetrate deeper into a sample to interact with materials
beneath the surface. The resulting SEM image thus shows
information that is convoluted over the sampling depth
(including that from the sample surface). According to the
Kanaya−Okayama depth penetration formula38 (see Support- yes
no

ing Information S5), the maximum penetration depth of

electrons is approximately 12 μm at an accelerating voltage of
improper sample drying may lead to sample damage; difficult to prepare
cross sections without sample damage; sputter coating may reduce
quantitative porosity depth profile by varying accelerating voltage;

25 keV, while it is less than 3 μm at accelerating voltages of ≤10

relatively simple sample preparation (e.g., sputter coating for non-

2-D (cross-sectional) pore structure and pore connectivity; Semi-

KeV.
Figure 2 shows a series of SEM images over the same sample
area scanned under different accelerating voltages. At 2 keV, the
SEM image shows relatively smooth active layer with no
nanoscale (up to 1 nm for field emission SEM)

observable pores. When the voltage was increased to 10 keV,

many tiny pores can be observed. The membrane appeared to
limited sampling depth (∼ or <10 μm)

be more porous at 25 keV. The more porous structure observed

SEM

can be rationalized by the greater electron penetration depth/

interaction volume at higher accelerating voltage (Supporting
yes; difficult to avoid artifacts

Information S6). While the SEM images obtained at low

membrane pore diameter

accelerating voltage show mainly the surface features of the

conductive samples)

sample, those at high beam energy reflect more features from

the bulk sample43 (pores in the current study). Based in Figure
moderate cost

2, the pore diameter close to the active layer was a few μm to

∼10 μm, which is in accordance with the TEM (Figure 1(a))
and CLSM observations (refer to Section “Membrane Pore
no

Diameter and Porosity Distribution”). We did not observe any

suitable for fouling
characterization?
noninvasive meth-
pore structure in-

differences in the images of CTA-W at different penetration

potential artifacts
and limitations
sample prepara-

cost and time

methods

depths. The current study demonstrates that SEM character-

formation
resolution

ization using a range of accelerating voltages can be a valuable

tion

tool for detecting membrane pores underneath a dense thin

od

surface layer.
9998 dx.doi.org/10.1021/es301885m | Environ. Sci. Technol. 2012, 46, 9995−10003
Environmental Science & Technology
Figure 2. SEM images showing the active surface of membrane CTA-
HW. A range of accelerating voltages were applied: (a) 2 keV, (b) 10
keV, (c) 25 keV.
CLSM Images. Figure 3 and 4 show the CLSM images of
membrane CTA-HW scanned under fluorescent mode (see
more information in Supporting Information S7 and S8).
Membrane material absorbs the fluorescein isothiocyanate
(FITC) and emits a fluorescent signal of yellow color. By
comparison, pores and woven mesh appear in dark color due to
no or little staining. In the current study, clear CLSM images
9999
Article
Figure 3. CLSM images of membrane CTA-HW viewed from the
active layer side. z is the distance to the membrane active surface. (a) z
= 0 μm, (b) z = 4 μm, (c) z = 8 μm and (d) z = 12 μm.
Figure 4. CLSM images of membrane CTA-HW viewed from support
layer side. z is the distance to the membrane active surface. (a) z = 100
μm, (b) z = 92 μm, (c) z = 83 μm, and (d) z = 74 μm.
can be obtained for a maximum scan depth of ∼23 μm from the
active layer surface (data not shown) or ∼54 μm from the
support skin, which are limited by the attenuation of
fluorescent signal by the membrane material.44 The lower
penetration depth from the active layer side was likely due to
light scattering by the closely packed small pores under the
layer.
Figures 3(a-d) show representative CLSM images of CTA-
HW obtained at scan depth of 0, 4, 8, and 12 μm from the
dx.doi.org/10.1021/es301885m | Environ. Sci. Technol. 2012, 46, 9995−10003
Environmental Science & Technology Article

active layer surface (i.e., z = 0, 4, 8, and 12 μm). The active

layer surface (z = 0) was generally free of pores; numerous
small pores (diameter ∼ or <10 μm) were visible at z ≥ 4 μm.
In addition, the pore diameter and porosity increases with
depth (from 4 to 12 μm) gradually. These features are in
qualitative agreement with the TEM and SEM observations
(Figure 1).
Figures 4(a−d) show representative CLSM images of CTA-
HW obtained at scan depth of 0, 8, 17, and 26 μm from the
support skin (corresponding to z values of ∼100, 92, 83, and 74
μm, respectively). Membrane pores were clearly visible at a
scan depth of 8 μm. More pores were observed at scan depths
of 17 and 26 μm, and the pore diameter (a few tens of
micrometers) was in reasonable agreement with TEM
characterization (Figure 1(a)). A three-dimensional (3-D)
reconstruction of the pore structures can be obtained by
stacking the individual images obtained over a range of scan
depth. As shown in Supporting Information S9, elongated 3-D
pores were obtained, confirming the TEM and SEM cross-
sectional observations (Figure 1). In addition to the pore
structure, the woven mesh fibers can also be clearly visualized in
Figure 4. Finally, membrane CTA-W was also characterized
using CLSM (Supporting Information S10). There was no
observable pore, in good agreement with TEM characterization
(Supporting Information S3). Change in scan depth only shows
the presence of woven mesh at different sample depth.
Membrane Pore Diameter and Porosity Distribution.
Quantitative information on membrane pore structure (pore
diameter and porosity distribution) can be obtained from TEM
and CLSM micrographs. Figure 5(a) presents the depth profile
of pore diameter (i.e., average pore diameter vs z) for
membrane CTA-HW. The profile obtained from TEM agrees
well with that based on CLSM - smaller pores (∼ or <10 μm)
were present close to the active layer and larger pores (10−40
μm) were observed on the support skin side. Pore diameter
measured from CLSM was slightly larger than that from TEM,
which may be attributed to the different sample preparation:
CLSM was performed for wet samples while TEM was
performed for alcohol-dehydrated and resin-embedded sam-
ples. Figure 5(b) presents the depth profile of porosity for
membrane CTA-HW. Over the active layer side (z from 0 to 6
μm), the membrane porosity increased sharply with scan depth,
and it remained at ∼50% over the bulk of the sample depth (z
from 10 to 80 μm). Porosity reduced toward the support skin
side (z from 80 to 100 μm) due to the termination of finger-like
pores. Compared to the CLSM porosity analysis, the TEM
analysis had significantly larger error bars, which can be Figure 5. The pore diameter and porosity distribution of membrane
attributed to its much smaller sampling volume (i.e., sampling CTA-HW in z direction. (a) pore diameter distribution obtained from
area × sampling thickness (<100 nm for TEM)). In principle, CLSM and TEM; and (b) porosity distribution obtained from CLSM
similar information on pore diameter and porosity distribution and TEM; (c) porosity adjacent to the active layer obtained from SEM
can be obtained from SEM cross-sectional images. Never- under different accelerating voltages (Kanaya-Okayama penetration
depth). The results were based on at least three different sample
theless, due to the lack of clear contrast for some pore features, locations.
quantitative analysis of these parameters was not performed in
the current study.
For comparison purpose, the porosity measured by the dry- The porosity near the active layer side can also be
wet weighing method22 is also shown in Figure 5(b). characterized by SEM using a range of accelerating voltages
Membrane CTA-HW had an average porosity of 64% based (see Figure 2). Figure 5(c) presents porosity as a function of
on the weighing method,22 which was significantly higher than the Kanaya-Okayama penetration depth (dKO). The porosity
those based on TEM and CLSM. The higher reported value for depth profile of CTA-HW measured by the SEM technique
the weighing method may be attributed to the absorption of agreed reasonably well with those obtained based on TEM and
water by the membrane material45 (in addition to water filling CLSM (Figure 5(b)). Since it does not require the
the pores), leading to an overestimation of membrane porosity. sophisticated sample preparation steps that are needed for
10000 dx.doi.org/10.1021/es301885m | Environ. Sci. Technol. 2012, 46, 9995−10003
Environmental Science & Technology
TEM and it provides better resolution compared to CLSM,
SEM can be a useful method for porosity characterization of the
active layer (and regions near the active layer). Nevertheless,
the SEM method reported in the current study is only
semiquantitative, and dKO estimates the maximum penetration
range rather than the average penetration depth.46 Further
studies are needed for more accurate interpretation of SEM
results (e.g., by using Monte Carlo simulation).46
Comparison of Microscopic Methods for Porosity
Characterization of FO/PRO Membrane. Table 2 summa-
rizes the three microscopic methods (SEM, TEM, and CLSM)
for characterizing pore structure of FO/PRO membranes. The
weighing method is also presented for comparison. Among the
four methods, the weighing method is the simplest and also the
cheapest approach. This method measures the membrane
average porosity, but provides no information on its depth
profile. It also tends to overestimate the porosity as a result of
water absorption by the polymeric substrate.
SEM and TEM are two electron microscopic methods with
nanoscale resolution. SEM is a relatively simple and
straightforward method, and it has been widely used to gain
qualitative understanding of membrane pore structures. By
varying the accelerating voltage (thus sample penetration
depth), one may even obtain a semiquantitative measurement
of porosity depth profile near the membrane active layer.
Compared to SEM, TEM has better resolution and it can
provide quantitative analysis of the pore diameter and porosity
variation across an entire membrane cross-section. However,
TEM characterization is expensive and it also involves tedious
and time-consuming sample preparation, which limits its use in
sampling large membrane areas.
In the current study, the porosity distribution results from
CLSM image analysis showed a good consistency with those
analyzed from TEM images. Compared to the electron
microscopic methods, CLSM offers several important advan-
tages:
• CLSM requires minimal sample preparation and the
method is noninvasive. This also avoids artifacts related
to sample preparation (such as sample drying).
• CLSM is simple, fast, and relatively cheap, and it can be
used to scan relatively large sample areas to improve
statistical accuracy.
• CLSM can provide 3-D pore structural information,
including pore diameter and porosity distribution, and
pore connectivity.
• Membrane samples can be scanned under hydrated
conditions, which means that the membrane pore
characteristics during CLSM imaging match closely
with those during typical FO/PRO membrane operation.
For thick membrane samples, however, the method is not
able to image over the entire membrane thickness (CLSM
performed image analysis over the membrane thickness of ∼75
μm in the current study). Finding the exact location of surfaces
can also be difficult for membranes with large surface features.
Implications. FO and PRO have attracted increasing
interests in various environmental applications such as
wastewater treatment, seawater desalination, and osmotic
energy harvesting. The performance of these membranes can
be severely limited by ICP, which is a strong function of the
membrane support layer structure. The current study provides
a comprehensive microscopic characterization of commercially
available FO/PRO membranes using SEM, TEM, and CLSM.
10001
Article
As discussed above, these techniques generally provide
complementary information on porosity and pore diameter
characteristics. When combined together, these methods can
lead to improved understanding of membrane pore structures.
For example, CLSM can provide detailed information on the 3-
D pore structure of wet membrane samples at a microscale
resolution, while TEM and SEM are able to resolve finer
structures near the membrane skins for dehydrated samples.
For the first time, CLSM is proven to be a versatile method for
characterizing internal pore structure of FO/PRO membranes.
Since the dimension of CTA membranes may change
significantly upon drying (Supporting Information S4),
CLSM is the preferred method that allows characterization of
membranes in wet state. Future studies may further explore its
use as a noninvasive method for (1) characterizing the adhesion
between membrane matrix and supporting mesh (Supporting
Information S7), and (2) real-time monitoring of FO/PRO
membrane fouling (particularly internal fouling/scaling inside
the support layer18 that is difficult to characterize using
conventional methods).
■
*
ASSOCIATED CONTENT
S Supporting Information
S1. Conversion of color image to binary black-white image; S2.
Additional TEM and SEM images of membrane CTA-HW; S3.
TEM and SEM cross-sectional images of membrane CTA-W;
S4. Optical microscopic images of dry and wet membrane
samples (CTA-HW); S5. Kanaya-Okayama depth penetration
formula; S6. Evaluation of the effect of high accelerating voltage
to membrane samples; S7. Additional CLSM images of
membrane CTA-HW taken at different locations; S8. A
comparison of CLSM images under fluorescent mode and
reflective mode; S9. 3-D pore structure of CTA-HW based on
CLSM characterization; S10. CLSM images of membrane
CTA-W. This material is available free of charge via the
Internet at http://pubs.acs.org.
■
AUTHOR INFORMATION
Corresponding Author
*Phone: +65 6790 5267; fax: +65 6791 0676; e-mail: cytang@
ntu.edu.sg.
Notes
The authors declare no competing financial interest.
■
ACKNOWLEDGMENTS
This research was supported by Tier 1 Research Grant No.
RG6/07, Ministry of Education, Singapore. We also thank the
support by the Singapore Membrane Technology Centre.
Hydration Technology Innovations is thanked for providing us
membrane materials. We thank John Perrino and the Cell
Sciences Imaging Facility at Stanford University for the help on
preparing TEM sections. The comments and feedbacks from
the editor and the four anonymous reviewers have greatly
improved the quality of the manuscript.
■
REFERENCES
(1) Cath, T. Y.; Childress, A. E.; Elimelech, M. Forward osmosis:
Principles, applications, and recent developments. J. Membr. Sci. 2006,
281 (1−2), 70−87.
(2) Zhao, S.; Zou, L.; Tang, C. Y.; Mulcahy, D. Recent developments
in forward osmosis: Opportunities and challenges. J. Membr. Sci. 2012,
396, 1−21.
dx.doi.org/10.1021/es301885m | Environ. Sci. Technol. 2012, 46, 9995−10003
Environmental Science & Technology
(3) Holloway, R. W.; Childress, A. E.; Dennett, K. E.; Cath, T. Y.
Forward osmosis for concentration of anaerobic digester centrate.
Water Res. 2007, 41 (17), 4005−4014.
(4) Cornelissen, E. R.; Harmsen, D.; de Korte, K. F.; Ruiken, C. J.;
Qin, J. J.; Oo, H.; Wessels, L. P. Membrane fouling and process
performance of forward osmosis membranes on activated sludge. J.
Membr. Sci. 2008, 319 (1−2), 158−168.
(5) Achilli, A.; Cath, T. Y.; Marchand, E. A.; Childress, A. E. The
forward osmosis membrane bioreactor: A low fouling alternative to
MBR processes. Desalination 2009, 238 (1−3), 10−21.
(6) Xiao, D.; Tang, C. Y.; Zhang, J.; Lay, W. C. L.; Wang, R.; Fane, A.
G. Modeling salt accumulation in osmotic membrane bioreactors:
Implications for FO membrane selection and system operation. J.
Membr. Sci. 2011, 366 (1−2), 314−324.
(7) Lay, W. C. L.; Zhang, Q.; Zhang, J.; McDougald, D.; Tang, C.;
Wang, R.; Liu, Y.; Fane, A. G. Study of integration of forward osmosis
and biological process: Membrane performance under elevated salt
environment. Desalination 2011, 283, 123−130.
(8) Kessler, J. O.; Moody, C. D. Drinking water from sea water by
forward osmosis. Desalination 1976, 18 (3), 297−306.
(9) Elimelech, M. Yale constructs forward osmosis desalination pilot
plant. Membrane Technology 2007, 1, 7−8.
(10) McGinnis, R. L.; McCutcheon, J. R.; Elimelech, M. A novel
ammonia-carbon dioxide osmotic heat engine for power generation. J.
Membr. Sci. 2007, 305 (1−2), 13−19.
(11) Xu, Y.; Peng, X.; Tang, C. Y.; Fu, Q. S.; Nie, S. Effect of draw
solution concentration and operating conditions on forward osmosis
and pressure retarded osmosis performance in a spiral wound module.
J. Membr. Sci. 2010, 348 (1−2), 298−309.
(12) Yip, N. Y.; Tiraferri, A.; Phillip, W. A.; Schiffman, J. D.; Hoover,
L. A.; Kim, Y. C.; Elimelech, M. Thin-film composite pressure retarded
osmosis membranes for sustainable power generation from salinity
gradients. Environ. Sci. Technol. 2011, 45 (10), 4360−4369.
(13) Ramon, G. Z.; Feinberg, B. J.; Hoek, E. M. V. Membrane-based
production of salinity-gradient power. Energy Environ. Sci. 2011, 4
(11), 4423−4434.
(14) Chou, S.; Wang, R.; Shi, L.; She, Q.; Tang, C.; Fane, A. G. Thin-
film composite hollow fiber membranes for pressure retarded osmosis
(PRO) process with high power density. J. Membr. Sci. 2012, 389, 25−
33.
(15) She, Q.; Jin, X.; Tang, C. Y. Osmotic power production from
salinity gradient resource by pressure retarded osmosis: Effects of
operating conditions and reverse solute diffusion. J. Membr. Sci. 2012,
401−402, 262−273.
(16) Loeb, S.; Titelman, L.; Korngold, E.; Freiman, J. Effect of porous
support fabric on osmosis through a Loeb-Sourirajan type asymmetric
membrane. J. Membr. Sci. 1997, 129 (2), 243−249.
(17) McCutcheon, J. R.; Elimelech, M. Influence of concentrative
and dilutive internal concentration polarization on flux behavior in
forward osmosis. J. Membr. Sci. 2006, 284 (1−2), 237−247.
(18) Tang, C. Y.; She, Q.; Lay, W. C. L.; Wang, R.; Fane, A. G.
Coupled effects of internal concentration polarization and fouling on
flux behavior of forward osmosis membranes during humic acid
filtration. J. Membr. Sci. 2010, 354, 123−133.
(19) Xiao, D.; Li, W.; Chou, S.; Wang, R.; Tang, C. Y. A modeling
investigation on optimizing the design of forward osmosis hollow fiber
modules. J. Membr. Sci. 2012, 392−393, 76−87.
(20) Qin, J. J.; Chen, S.; Oo, M. H.; Kekre, K. A.; Cornelissen, E. R.;
Ruiken, C. J. Experimental studies and modeling on concentration
polarization in forward osmosis. Water Sci. Technol. 2010, 61 (11),
2897−2904.
(21) Wang, Y.; Wicaksana, F.; Tang, C. Y.; Fane, A. G. Direct
microscopic observation of forward osmosis membrane fouling.
Environ. Sci. Technol. 2010, 44 (18), 7102−7109.
(22) Wei, J.; Qiu, C.; Tang, C. Y.; Wang, R.; Fane, A. G. Synthesis
and characterization of flat-sheet thin film composite forward osmosis
membranes. J. Membr. Sci. 2011, 372 (1−2), 292−302.
10002
Article
(23) Jin, X.; Tang, C. Y.; Gu, Y.; She, Q.; Qi, S. Boric acid permeation
in forward osmosis membrane processes: Modeling, experiments, and
implications. Environ. Sci. Technol. 2011, 45 (6), 2323−2330.
(24) Chou, S.; Shi, L.; Wang, R.; Tang, C. Y.; Qiu, C.; Fane, A. G.
Characteristics and potential applications of a novel forward osmosis
hollow fiber membrane. Desalination 2010, 261, 365−372.
(25) Qiu, C.; Qi, S.; Tang, C. Y. Synthesis of high flux forward
osmosis membranes by chemically crosslinked layer-by-layer poly-
electrolytes. J. Membr. Sci. 2011, 381 (1−2), 74−80.
(26) Wang, R.; Shi, L.; Tang, C. Y.; Chou, S.; Qiu, C.; Fane, A. G.
Characterization of novel forward osmosis hollow fiber membranes. J.
Membr. Sci. 2010, 355 (1−2), 158−167.
(27) McCutcheon, J. R.; McGinnis, R. L.; Elimelech, M. A novel
ammonia-carbon dioxide forward (direct) osmosis desalination
process. Desalination 2005, 174 (1), 1−11.
(28) Tang, C. Y.; She, Q.; Lay, W. C. L.; Wang, R.; Fane, A. G.
Coupled effects of internal concentration polarization and fouling on
flux behavior of forward osmosis membranes during humic acid
filtration. J. Membr. Sci. 2010, 354 (1−2), 123−133.
(29) Tang, C. Y.; Fu, Q. S.; Robertson, A. P.; Criddle, C. S.; Leckie, J.
O. Use of reverse osmosis membranes to remove perfluorooctane
sulfonate (PFOS) from semiconductor wastewater. Environ. Sci.
Technol. 2006, 40 (23), 7343−7349.
(30) Pacheco, F. A.; Pinnau, I.; Reinhard, M.; Leckie, J. O.
Characterization of isolated polyamide thin films of RO and NF
membranes using novel TEM techniques. J. Membr. Sci. 2010, 358 (1−
2), 51−59.
(31) She, F. H.; Nihara, K.; Gao, W. M.; Hodgson, P. D.; Jinnai, H.;
Kong, L. X. 3-Dimensional characterization of membrane with
nanoporous structure using TEM tomography and image analysis.
Desalination 2010, 250 (2), 757−761.
(32) Conchello, J.-A.; Lichtman, J. W. Optical sectioning microscopy.
Nat. Methods 2005, 2 (12), 920−931.
(33) Charcosset, C.; Bernengo, J. C. Comparison of microporous
membrane morphologies using confocal scanning laser microscopy. J.
Membr. Sci. 2000, 168 (1−2), 53−62.
(34) Wang, J.; Dismer, F.; Hubbuch, J.; Ulbricht, M. Detailed analysis
of membrane adsorber pore structure and protein binding by advanced
microscopy. J. Membr. Sci. 2008, 320 (1−2), 456−467.
(35) Wickramasinghe, S. R.; Carlson, J. O.; Teske, C.; Hubbuch, J.;
Ulbricht, M. Characterizing solute binding to macroporous ion
exchange membrane adsorbers using confocal laser scanning
microscopy. J. Membr. Sci. 2006, 281 (1−2), 609−618.
(36) Yun, M. A.; Yeon, K. M.; Park, J. S.; Lee, C. H.; Chun, J.; Lim,
D. J. Characterization of biofilm structure and its effect on membrane
permeability in MBR for dye wastewater treatment. Water Res. 2006,
40 (1), 45−52.
(37) Li, W.; Gao, Y.; Tang, C. Y. Network modeling for studying the
effect of support structure on internal concentration polarization
during forward osmosis: Model development and theoretical analysis
with FEM. J. Membr. Sci. 2011, 379 (1−2), 307−321.
(38) Kanaya, K.; Okayama, S. Penetration and energy-loss theory of
electrons in solid targets. J. Phys. D: Appl. Phys. 1972, 5 (1), 43−58.
(39) Glauert, A. M.; Lewis, P. R., Biological Specimen Preparation for
Transmission Electron Microscopy; Princeton University Press: Prince-
ton, NJ: 1998.
(40) Qi, S.; Qiu, C. Q.; Zhao, Y.; Tang, C. Y. Double-skinned
forward osmosis membranes based on layer-by-layer assembly-FO
performance and fouling behavior. J. Membr. Sci. 2012, 405−406, 20−
29.
(41) Shi, L.; Chou, S. R.; Wang, R.; Fang, W. X.; Tang, C. Y.; Fane,
A. G. Effect of substrate structure on the performance of thin-film
composite forward osmosis hollow fiber membranes. J. Membr. Sci.
2011, 382 (1−2), 116−123.
(42) McCutcheon, J. R.; McGinnis, R. L.; Elimelech, M. Desalination
by ammonia-carbon dioxide forward osmosis: Influence of draw and
feed solution concentrations on process performance. J. Membr. Sci.
2006, 278 (1−2), 114−123.
dx.doi.org/10.1021/es301885m | Environ. Sci. Technol. 2012, 46, 9995−10003
Environmental Science & Technology Article

(43) A Guide to Scanning Microscope Observation; Japan Electron

Optics Laboratory Company (JEOL), 2012, http://www.jeolusa.com/
tabid/320/DMXModule/692/EntryId/1/Default.aspx.
(44) Marroquin, M.; Bruce, T.; Pellegrino, J.; Wickramasinghe, S. R.;
Husson, S. M. Characterization of asymmetry in microporous
membranes by cross-sectional confocal laser scanning microscopy. J.
Membr. Sci. 2011, 379 (1−2), 504−515.
(45) Freger, V. Swelling and morphology of the skin layer of
polyamide composite membranes: An atomic force microscopy study.
Environ. Sci. Technol. 2004, 38 (11), 3168.
(46) Ng, F. L.; Wei, J.; Lai, F. K.; Goh, K. L. Metallic thin film depth
measurements by X-ray microanalysis. Appl. Surf. Sci. 2006, 252 (11),
3972−3976.

10003 dx.doi.org/10.1021/es301885m | Environ. Sci. Technol. 2012, 46, 9995−10003