Article

pubs.acs.org/IECR

3D Membrane Imaging and Porosity Visualization

Ganesh Sundaramoorthi,† Markus Hadwiger,† Mohamed Ben-Romdhane,† Ali R. Behzad,‡
Poornima Madhavan,§ and Suzana P. Nunes*,†
†
Visual Computing Center (VCC), ‡Imaging and Characterization Core Lab, §Biological and Environmental Science and Engineering
Division (BESE), King Abdullah University of Science and Technology (KAUST), 23955-6900 Thuwal, Saudi Arabia
*
S Supporting Information

ABSTRACT: Ultrafiltration asymmetric porous membranes were imaged by two

microscopy methods, which allow 3D reconstruction: focused ion beam and serial
block face scanning electron microscopy. A new algorithm was proposed to evaluate
porosity and average pore size in different layers orthogonal and parallel to the
membrane surface. The 3D reconstruction enabled additionally the visualization of
pore interconnectivity in different parts of the membrane. The method was
demonstrated for a block copolymer porous membrane and can be extended to
other membranes with application in ultrafiltration, supports for forward osmosis,
etc., offering a complete view of the transport paths in the membrane.

■ INTRODUCTION
Synthetic membranes are integrated in a variety of processes,
gives us quantified information on total surface area or the
overall porosity, accessible to the probing gas. SEM and AFM
which go from hemodialysis to water desalination, beverages are restricted in most cases to image the membrane surface
and dairy product manufacture, gas separation in petrochemical morphology. 2D imaging of membrane cross sections is
and pharmaceutical industry. They have different requirements possible by SEM. Porosimetry gives information on average
of stability, flux and selectivity. Integral asymmetric or pore size and distribution. However, in none of these methods
multilayer asymmetric porous membranes are used in almost it is possible to have a detailed characterization of porosity in all
all membrane-based processes.1−4 Although the top layer different layers of the membrane.
determines the membrane selectivity and the separation As far as morphological characterization is concerned,
capability, the porous sublayer provides the needed mechanical advanced microscopy and scattering methods have been very
stability and should not negatively interfere with the transport. useful for membrane characterization,11−13 including field
Porosity control is relevant for many other reasons. The main emission electron microscopy (FESEM) and transmission
drawback of forward osmosis is concentration polarization, electron microscopy (TEM) tomography, associated with
which is highly affected by tortuosity and porosity of the synchrotron scattering methods. Cryo-TEM and cryo-SEM
substrate layer.3,5−7 Porosity influences how fast water transfer helped in investigating copolymer assembly in solution,13,17−19
through membrane and condensation takes place.8 The before the membrane manufacture. We used for the first time
porosity of porous electrodes and gas diffusion layers in cryo-SEM and environmental SEM (ESEM) to image the effect
membrane−electrode-assemblies for fuel cell contributes to the of pH on stimuli-responsive porous membranes.12 Mottern et
efficiency of the reactive conversion.9,10 al.20 used focus ion beam (FIB) to prepare ceramic membrane
The preparation of asymmetric porous membranes based on samples as single slices for 2D TEM imaging. Kyotani et al.21
block copolymer self-assembly combined to phase inversion, reported a similar method combined to grazing incidence X-ray
block sacrifice by selective etching or reactive phase diffraction analysis of tubular zeolites.
separation,11−16 leads to the formation of a selective top layer More recently, new methods of tomography or 3D
with isoporous structure on the top of the membrane and a microscopy image reconstruction have become available with
sublayer with gradient porosity. Morphology control in solution potential use in membrane technology, offering then the
and in the final asymmetric membrane is decisive for the final perspective of having a complete overview of porosity in the
membrane application in ultrafiltration. whole volume.
Well-known methods for pore size characterization are
extensively used for membranes.17 Examples are scanning Received: January 27, 2016
electron microscopy (SEM), atomic force microscopy (AFM), Revised: February 27, 2016
bubble pressure and gas transport, mercury porosimetry, Accepted: March 2, 2016
permporometry, BET adsorption and thermoporometry. BET Published: March 3, 2016

© 2016 American Chemical Society 3689 DOI: 10.1021/acs.iecr.6b00387

Ind. Eng. Chem. Res. 2016, 55, 3689−3695
Industrial & Engineering Chemistry Research Article

Figure 1 summarizes the resolution of different 3D imaging Opposite to TEM tomography or AFM, X-ray computed
methods and the sampling volume. tomography29 or synchrotron radiation computed micro-
tomography30,31 image relatively large volumes, being consid-
ered a multilength scale methodology, which can be non-
destructive and used for in situ experiments to follow process
evolution (4D investigation). 3D reconstruction of polymeric
membranes by X-ray tomography is being used for 3D
reconstruction of polymeric30 and inorganic32 membranes.
The image resolution is commonly in the micrometer range.
This technology is constantly improving with the implementa-
tion of better X-ray optics in synchrotron facilities, allowing the
resolution to go far bellow 1 μm. As far as membrane is
concerned the methodology has been successful for micro-
filtration and solid oxide fuel cell. Although the absorption
Figure 1. 3D imaging of membranes. contrast of X-ray tomography is more suitable for imaging
ceramic and metallic structures, phase contrast X-ray imaging is
TEM tomography provides the best resolution (<1 nm) and becoming widespread and more successfully for low atomic
is therefore the best choice if fine structures need to be number materials. A recent example is the characterization of
observed. We have used TEM tomography before to image graphite electrodes in Li-ion batteries.33
block copolymer membranes,22 which offered precise and Confocal microscopy gives information on porosity in
detailed information on the membrane selective layer. TEM different layer depths for a porous structure. Snyder et al.34
tomography was recently used by Pacheco et al.23 to visualize reported a quantitative analysis of zeolite membranes imaged
the structure of the top selective layer of thin film composite by laser scanning confocal microscopy. Confocal microscopy is
membranes. Maijoinen et al.24 and Kaushik et al.25 used 3D a method of much more widespread and of better access to
cryo-TEM tomography for investigating cellulose nanocrystals most laboratories than X-ray tomography. But the resolution of
close to native conditions. The disadvantage of TEM confocal is also not high, with restrictions known for optical
tomography is the small sampling volume (ca. 2−5 μm in the microscopy and requires fluorescent samples. The resolution is
three dimensions), which does not give an overview of the not enough for membranes with pore size in the range of nano-
whole membrane structure. Slices thinner than 100 nm, or ultrafiltration. 3D reconstruction associated with the electron
obtained by ultramicrotomy or focused ion beam, are normally microscopy resolution would be more suitable.
required. In addition to the morphology normally revealed by Focused ion beam (FIB/SEM) and serial block face SEM
TEM using elastic electrons, more information relative to (SBF/SEM) fill the gap between TEM and X-ray tomography
contrast promoted by differences in crystallinity or composition (or confocal microscopy) in terms of resolution and imaged
can be obtained in combination with spectroscopy techniques.
volume. FIB removes thin slices of a sample by incidence of ion
Examples useful technologies for this purpose are dark-field
beam, consecutively exposing a fresh cross section, which is
TEM, electron energy loss spectroscopy (EELS) and energy
dispersive spectroscopy (EDS).26 imaged by the electron beam of a SEM. FIB has been much
Atomic force microscopy (AFM) provides images with high more applied to inorganic porous materials. Holzer et al.35,36
resolution with 3D visualization with relatively easy sample applied focused ion beam (FIB/SEM) for 3D reconstruction of
preparation and the advantage of observing samples at porous ceramics. FIB application to polymeric membranes has
environmental conditions without the need for a high vacuum. been much less explored, because additional care and control of
Valuable achievements on quantified morphological character- experimental parameters are required to avoid sample damage
ization of membranes have been recently reported using AFM. for instance when exposed to Gallium ions and charging. We
ElHadidy et al.27,28 and AlMarzooqi et al.28 worked on pore were the first to report the used of FIB coupled to FESEM for a
analysis for ultrafiltration membranes based on AFM images qualitative 3D reconstruction of polymeric porous mem-
correlating with porometry techniques. The main disadvantage branes.11 The 3D visualization and semiautomated segmenta-
of AFM for 3D imaging is that it just samples the superficial tion were carried out using software tools implemented in the
layers of a membrane. The information is closer to that program Amira (TGS, San Diego, CA). Zils et al.37 applied FIB
obtained by 2D imaging methods than to that supplied by other nanotomography in a similar way for characterization of
3D characterization methods discussed here. polymer electrolyte fuel cell. The alignment of the image

Table 1. Comparison of Different 3D Imaging Methods

method image staining slicing

TEM tomography elastic electrons (inelastic electrons for element contrast) needed ultramicrotomy or ion
beam
AFM probe-sample force or interaction not required not required
FIB/SEM secondary electrons (backscattered electrons for composition not needed ion beam
contrast)
SBF/SEM backscattered electrons needed ultramicrotomy
X-ray computed X-rays heavy elements have better not required
tomography contrast
confocal microscopy light/fluorescence needed if sample not fluorescent not required

3690 DOI: 10.1021/acs.iecr.6b00387

Ind. Eng. Chem. Res. 2016, 55, 3689−3695
Industrial & Engineering Chemistry Research Article

stacks in this case was done with the software ImageJ and been explored by our group in the past few years. The
StackReg Plugin and further analysis by using Otsu algorithm. membranes characterized by FIB/SEM and by SBF/SEM were
Reingruber et al.38 recently published the quantitative prepared by slightly different procedures:
analysis of a 3D reconstruction of microfiltration membranes Procedure 1. The membrane imaged by FIB/SEM was
using SBF/SEM. The method consists of serial sectioning prepared by casting a PS-b-P4VP solution in a mixture of
provided by an ultramicrotome in the specimen chamber of an dimethylformamide and tetrahydrofuran with copper acetate, as
ESEM, which is a variable pressure FESEM. The ESEM previously reported in ref.11
resolution is normally lower than that of a high vacuum Procedure 2. The membrane analyzed by SBF/SEM was
FESEM, but the method was appropriate for microfiltration prepared by casting a PS-b-P4VP solution in a 1:1:1 (wt %)
membranes. mixture of DMF, THF and dioxane, followed by immersion in
In addition to Figure 1, Table 1 summarizes characteristics of water, analogous to the description reported in ref.19 The
different 3D imaging methods, which could be applied for prepared PS-b-P4VP membrane was then used to filter a 1 wt %
membranes. gold(III) chloride solution. Gold ions complex with pyridine
More than an optical visualization method, a quantitative blocks.40 The complexed gold was then reduced to gold
analysis is required to correlate fully morphology to transport, nanoparticles by treating the membrane with a 0.05 M sodium
selectivity and additional parameters such as concentration citrate solution.
polarization. Although 2D microscopy images do not allow Imaging. Two methods were used for imaging: (1) FIB/
measurements of volume specific surface area with pore length SEM and (2) SBF/SEM.
distribution and connectivity across the membrane, a 3D 1. Focused Ion Beam/Field Emission Scanning Electron
quantitative analysis is ideal for this purpose. We recently Microscopy (FIB/SEM). Membranes prepared by procedure 1
demonstrated how 3D image simulation can help in predicting were sectioned, by using focused ion beam. To minimize the
transport characteristics in membranes.39 Quantitative analysis polymer sample damage by the ion beam, the beam currents
can be incorporated as starting point for flux simulations to were lower than usually applied for semiconductors analysis,
design better membranes. By providing experimental 3D and a protective Pt layer was deposited, prior to milling the
quantitative visualization of membrane morphology, transport surface at the region of interest. The layer was deposited using
simulations would be even more effective. There is a need for ion beam induced CVD, inside the FIB/SEM chamber. To
more complete approaches, offering quantitative character- dissipate effectively surface charge from the surface, the
ization associated with 3D reconstruction with high resolution. membrane surface was also sputter-coated with Au prior to
Because most membranes have a gradient porosity, it would imaging/milling. The series of 150 slices (each with 50 nm
be very useful to have a quantitative analysis of pore size thickness) was obtained by using a FEI Helios 400S small dual
distribution layer by layer for a porous membrane, obtaining a beam system and Slice & View software.
complete 3D mapping of the structure. This can be hardly done 2. Serial Block Face Scanning Electron Microscopy (SBF/
experimentally. Confocal microscopy reveals the pore structure SEM). The morphology of membranes prepared by procedure 2
in different depths, but as mentioned above the resolution is was analyzed by serial block-face scanning electron microcopy
not enough for most ultra- and nanofiltration membranes. (SBF/SEM), performed with a GATAN 3View sectioning
In this work, we explore and compare two electron system. SBF/SEM combines environmental scanning electron
microscopy methods for 3D imaging, FIB/SEM and SBF/ microscopy and an automated ultramicrotome placed inside the
SEM, for the characterization of block copolymer membranes microscope chamber. Backscattered electrons detector is
with pore size of 20−50 nm, a range, which requires higher normally used for imaging. To increase the conductivity and
resolution than for microfiltration and can not be achieved by coefficient of backscattered electron signals, the membrane was
X-ray computed tomography. We propose a new segmentation stained with vapors of methyl iodide for 4 h before embedding
method for quantitative analysis of porosity of membranes slice it in resin blocks. The resin block with embedded membrane
sets applied to the two methods. Information on porosity can was cut off with jewelry saw and glued with epoxy resin to
be obtained in different layers or depths. We demonstrate the aluminum rivet in such a way that a profile of the membrane
feasibility, by applying the methods to membranes prepared by cross section would be observed once the block face is trimmed
block copolymer self-assembly and nonsolvent induced phase with razor blade and ultramicrotome. To minimize charging,
separation. The membranes are asymmetric with a top layer the lateral sides of the block face were coated with colloidal
containing pores in the range of 20−50 nm. As the depth silver paint and the block face was coated with 5 nm gold inside
increases, the pore size increases and the pores become less a sputter coater. A Quanta 200 FEG SEM (FEI) equipped with
regular. The same method can be extended in principle to any the GATAN 3View door and backscattered electron detector
other polymeric porous membrane.

■
was used for serial sectioning and imaging of the block face.
The image stack of 50 nm slices was acquired in an automated
EXPERIMENTAL SECTION fashion in low vacuum mode under the following conditions:
Materials. Polystyrene-b-poly(4-vinylpyridine) block co- 2.0 kV voltage, current of 43 pico Amp, resolution 1024 ×
polymer (PS-b-P4VP 175000-b-65000)) (P9828-S4VP was 1024, pixel size 8 nm, pixel time 32 μs, chamber pressure of
purchased from Polymer Source, Inc., Canada. 1,4-Dioxane, 0.2239 Torr, magnification of 16477×. A set of 1000 slices was
dimethylformamide (DMF), gold(III) chloride and sodium obtained.
citrate were supplied by Sigma-Aldrich. Tetrahydrofuran was Segmentation Method. 3D reconstruction with detailed
supplied by Fischer Scientific. Methyl iodide used for staining and quantitative information on pore size and porosity in
was supplied by Acros-organics. different layers of the membrane requires first the segmentation
Membrane Preparation. The membranes analyzed in this of the microscopy slices images. Details of the segmentation
work were prepared by block copolymer self-assembly and method using the Mumford−Shah Model41−44 and an
nonsolvent induced phase separation, a method, which has algorithm developed at KAUST for this purpose are described
3691 DOI: 10.1021/acs.iecr.6b00387
Ind. Eng. Chem. Res. 2016, 55, 3689−3695
Industrial & Engineering Chemistry Research Article

in the Supporting Information. For segmentation, the image of Figure 3 shows the 3D reconstruction of the membrane from
the membrane consists of empty spaces (pores) and membrane slices obtained by FIB/SEM, similar to those depicted in Figure
matrix, which have to be automatically identified and separated 2. Figure 3b represents the first step of segmentation.
from each other. The image intensity in different regions is
spatially homogeneous. The membrane can be considered as
consisting of several regions whose image intensity is described
by smooth function plus additive noise. The goal of the
segmentation algorithm is to determine the regions, in which
the image intensities are smooth functions. Because both the
functions and regions are initially unknown, the algorithm
below estimates both simultaneously. The problem described is
a classical problem in computer vision, which is normally
referred as Mumford−Shah Model.41−44 Several numerical
optimization methods have been proposed to solve this
problem,45,46 but the algorithms tend to be computationally
Figure 3. (a) 3D visualization of the membrane reconstructed from
expensive, prone to local optima, and extremely sensitive to
the set of slices obtained by FIB/SEM and respective (b)
initialization. Therefore, we employ a recent technique,42 which segmentation.
is computationally efficient (compared to those in refs 45 and
46), less susceptible to local minima, insensitive to initialization, FIB/SEM allows the imaging of slices obtained orthogonal to
and therefore may be employed in a fully automated scheme. the membrane surface. For practical reasons, slices parallel to

■ RESULTS AND DISCUSSION

The first method explored in this work was FIB/SEM. FIB has
the membrane surface can be hardly experimentally obtained.
By 3D reconstruction and application of the developed
algorithm, the porosity and pore average size in different layers
been recognized as a successful method for morphology parallel to the membrane surface can be now quantified, giving
investigation of different kinds of materials, but has been a complete picture of the membrane.
particularly well applied for nanofabrication in the semi Figure 4 shows the results of the porosity calculation using
conductor industry47 or preparation of ceramic samples for the algorithm, as well as the visualization of different cuts
transmission microscopy,48 which could not be easily cut by corresponding to layers of different porosity.
ultramicrotomy. More recently, FIB/SEM has been increasingly
used for the characterization of biological samples.49 The use of
FIB for polymeric membranes has been much more restrict and
a reason for that is the potential damage of the sample by
exposure to the beam, if the conditions are not properly
controlled. Boussu et al.50 were one of the first to report the use
of FIB for polymeric membrane characterization. However,
they used it only to expose cross sections of different polymeric
membranes for SEM investigation in an alternative way to the
classical sample fracture. Our group was probably the pioneer
in using FIB sectioning for 3D reconstruction of polymeric
porous membranes, as reported 6 years ago.11 Low beam
currents and a protective Pt layer were applied to minimize
damages. Our previous report used FIB for qualitative
visualization purpose. Here we propose a method to quantify
porosity in different membrane layers, demonstrating it for
asymmetric porous membranes. Typical membrane slices 2D
images obtained by FIB/SEM are shown in Figure 2. Figure 4. Porosity layer by layer, calculated for each virtual slice
parallel to the membrane surface, each slice with thickness of 50 nm.
Images obtained by FIB/SEM.

As a result of the segmentation it is possible also to

distinguish the pore regions, which are highly interconnected.
This is visualized in Figure 5, with different colors representing
regions with preferential interpore connectivity.
The second imaging method used in this work was SBF/
SEM. We used for that a GATAN 3View system. A typical slice
image can be seen in Figure 6. The sample for FIB/SEM was
not embedded in resin and the contrast for each slice was
topographic, using secondary electrons, analogously to the
images obtained by regular FESEM. For SBF/SEM the sample
was embedded in resin and sliced by ultramicroty inside the
Figure 2. FIB/SEM membrane slices images to be segmented. microscope chamber. Because the imaged surface is smooth,
3692 DOI: 10.1021/acs.iecr.6b00387
Ind. Eng. Chem. Res. 2016, 55, 3689−3695
Industrial & Engineering Chemistry Research Article

As in Figure 5, different colors emphasize pore regions with

high interconnectivity.

Figure 5. 3D reconstruction of the membrane, after segmentation and

algorithm application. Each color represents pore regions with
interconnectivity. Figure 7. Porosity layer by layer, calculated for each virtual slice
parallel to the membrane surface, each slice with thickness of 50 nm.
The images were obtained by SBT/SEM, by using a GATAN 3View.

Both imaging methodologies used in this work gave similar

quantitative information on porosity. The convenience of using
one or other depends on the characteristic of membrane being
characterized, contrast for backscattering electrons and stability
in the ion beam. Figure 6 for SBF/SEM is closer to images
obtained by TEM and takes advantage of analogous staining or
intrinsic contrast provided by elements of higher electron
density or different capacity of scattering electrons. A diamond
knife is used as for TEM. The samples must be sliceable by the
knife and properly prepared, including adequate embedding in
Figure 6. Membrane slice, obtained by SBF/SEM, after staining with resin. SBF/SEM can in principle sample larger volumes in
CH3I and gold. shorter times than FIB/SEM, as claimed also for other
materials.49,51 Although for the membranes investigated in
this work, SBF/SEM was a very convenient method, the same
the contrast is provided by differences in local composition. staining might not work for membranes based on other
The image is formed by backscattered electrons. The contrast materials, which do not complex with gold or do not strongly
given by atoms constituting each blocks of a plain membrane interact with CH3I. Other staining methods need then to be
and the resin is not enough for a satisfactory image. SBF/SEM evaluated. Membranes containing acrylic acid blocks can easily
is convenient to image structures already with heavy elements complex with other metal ions. For membranes without
such as samples decorated with inorganic nanoparticles. If the charged segments, long staining with osmium tetroxide before
image of plain polymeric membranes is the target, additional embedding might provide enough contrast for SBF/SEM.
staining is necessary. For biological or synthetic polymers, According to previous literature,49 for instance for biological
staining with osmium tetroxide is a common procedure. The samples, whose characteristics are not much different than that
block copolymer we used contains styrene and pyridine blocks. of polymeric membranes, FIB/SEM is slightly superior in
Pyridine can be selectively stained by CH3I. Pyridine complexes resolution and lower in terms of coverable volumes. By using
with gold ions, which provide even stronger contrast. The backscattered electrons instead of secondary electrons for FIB/
membrane is preformed and then stained by exposure to the SEM, it would be possible to obtain additional composition
vapor phase (CH3I) or by filtration of an aqueous gold solution, contrast, however with lower image resolution. In general, both
followed by reduction to elementary gold. Under these SBF/SEM and FIB/SEM technologies are developing fast with
conditions, the membrane morphology is not expected to improvement of resolution. FIB/SEM would be the method of
change, but the it must be clear that the dark regions in Figure choice in the case of ceramic membranes or other hard
6 corresponds to regions occupied mainly by pyridine blocks. materials, which should not be cut by diamond knifes. Both
For the membranes investigated here by SBF/SEM, gold was SBF/SEM and FIB/SEM cover much larger volumes than
not only added for pure staining purposes but also has been TEM tomography. We have used TEM tomography before52
previously explored for catalytic purpose.40 for membrane characterization, which provides excellent
As for the FIB/SEM slices, segmentation and 3D resolution, but the imaged area is restricted. The possibility
reconstruction were conducted for SBF/SEM images and the of imaging a large part of the membrane, typically with
porosity was quantitatively analyzed and presented in Figure 7. thickness of 20−50 μm or more, can be very useful to have a
3693 DOI: 10.1021/acs.iecr.6b00387
Ind. Eng. Chem. Res. 2016, 55, 3689−3695
Industrial & Engineering Chemistry Research Article

complete understanding of transport. Both SBF/SEM and FIB/

SEM cover this range. The membranes investigated here have
■ AUTHOR INFORMATION
Corresponding Author
pores larger than 20 nm and for that the resolution of both
*S. P. Nunes. E-mail: suzana.nunes@kaust.edu.sa.
methods is sufficient. This would be the case for most
ultrafiltration membranes. When imaging smaller pores, the Notes
resolution of FIB/SEM could be advantageous. The authors declare no competing financial interest.

■
The results obtained by applying segmentation, 3D
reconstruction and the developed algorithm allowed the ACKNOWLEDGMENTS
quantification of porosity in any layer parallel to the surface
This work was supported by the King Abdullah University of
going from one side to the other of the membrane. Also
Science and Technology (KAUST), KAUST Project “Visual-
information on the pore interconnectivity in adjacent areas is ization and Pore Tuning of Asymmetric Membranes”.
provided. This is very useful in the field of synthetic
membranes. Most commercial membranes for ultra-nano-
filtration and reverse osmosis as well as gas separation are
asymmetric membranes with complex morphology. Morpho-
■ REFERENCES
(1) Nunes, S. P.; Peinemann, K.-V. Membrane technology: in the
logical characterization is essential as tool for membrane chemical industry; John Wiley & Sons: Hoboken, NJ, 2006.
development and to take decisions on best choices for each (2) Shannon, M. A.; Bohn, P. W.; Elimelech, M.; Georgiadis, J. G.;
Mariñas, B. J.; Mayes, A. M. Science and technology for water
application and process optimization. For fully understanding purification in the coming decades. Nature 2008, 452 (7185), 301−
or simulating transport in membranes the 3D reconstruction 310.
and analysis described in this paper can be much more useful (3) Loeb, S.; Titelman, L.; Korngold, E.; Freiman, J. Effect of porous
than regular 2D images obtained by SEM or TEM, giving the support fabric on osmosis through a Loeb-Sourirajan type asymmetric
complete view and quantitative information. Here we provide membrane. J. Membr. Sci. 1997, 129 (2), 243−249.
experimental 3D images and we anticipate that this information (4) Lee, A.; Elam, J. W.; Darling, S. B. Membrane materials for water
could be used in combination with transport simulation tools purification: design, development, and application. Environmental
Science: Water Research & Technology 2016, 2 (1), 17−42.
similar to those used in ref 39 and additional methods under
(5) Gruber, M.; Johnson, C.; Tang, C.; Jensen, M. H.; Yde, L.; Hélix-
development for a much better transport and structure Nielsen, C. Computational fluid dynamics simulations of flow and
optimization in the near future. concentration polarization in forward osmosis membrane systems. J.

■
Membr. Sci. 2011, 379 (1), 488−495.
CONCLUSION (6) Li, W.; Gao, Y.; Tang, C. Y. Network modeling for studying the
effect of support structure on internal concentration polarization
Two microscopy methods for 3D imaging visualization, FIB/ during forward osmosis: model development and theoretical analysis
SEM and SBF/SEM, were applied to porous block copolymer with FEM. J. Membr. Sci. 2011, 379 (1), 307−321.
membranes. Both methods offered high resolution for adequate (7) Park, M.; Lee, J. J.; Lee, S.; Kim, J. H. Determination of a
pore analysis and covered sample volumes higher than normally constant membrane structure parameter in forward osmosis processes.
possible by transmission electron microscopy tomography. J. Membr. Sci. 2011, 375 (1), 241−248.
With help of a new algorithm, the set of images obtained by (8) Maab, H.; Al Saadi, A.; Francis, L.; Livazovic, S.; Ghafour, N.;
Amy, G. L.; Nunes, S. P. Polyazole hollow fiber membranes for direct
both methods could be analyzed to provide a complete contact membrane distillation. Ind. Eng. Chem. Res. 2013, 52 (31),
mapping of porosity and average pore size in different 10425−10429.
membrane layers. The 3D-visualization combining information (9) Prehn, K.; Adelung, R.; Heinen, M.; Nunes, S. P.; Schulte, K.
gained by the image segmentation showed the pore Catalytically active CNT−polymer-membrane assemblies: From
connectivity in different membrane regions. We demonstrate synthesis to application. J. Membr. Sci. 2008, 321 (1), 123−130.
here how effective the method is for asymmetric block (10) Tripathi, B. P.; Schieda, M.; Shahi, V. K.; Nunes, S. P.
copolymer membranes. The same analysis using the proposed Nanostructured membranes and electrodes with sulfonic acid
functionalized carbon nanotubes. J. Power Sources 2011, 196 (3),
algorithm can be applied for any kind of membranes or porous
911−919.
materials. The imaging method to be chosen depends however (11) Nunes, S. P.; Sougrat, R.; Hooghan, B.; Anjum, D. H.; Behzad,
on the mechanical properties, feasibility of slicing with diamond A. R.; Zhao, L.; Pradeep, N.; Pinnau, I.; Vainio, U.; Peinemann, K.-V.
knife or ion beam, pore size, volume to be sampled and Ultraporous films with uniform nanochannels by block copolymer
feasibility of staining. The method is particularly valuable for micelles assembly. Macromolecules 2010, 43 (19), 8079−8085.
asymmetric membranes, which are used in many applications. It (12) Nunes, S. P.; Behzad, A. R.; Hooghan, B.; Sougrat, R.;
can help the development of new systems for forward osmosis Karunakaran, M.; Pradeep, N.; Vainio, U.; Peinemann, K.-V.
and other applications for which not only the porosity of the Switchable pH-responsive polymeric membranes prepared via block
copolymer micelle assembly. ACS Nano 2011, 5 (5), 3516−3522.
top selective layer is relevant,but also that of the whole porous (13) Nunes, S. P.; Karunakaran, M.; Pradeep, N.; Behzad, A. R.;
substrates. Hooghan, B.; Sougrat, R.; He, H.; Peinemann, K.-V. From micelle

■
*
ASSOCIATED CONTENT
S Supporting Information
supramolecular assemblies in selective solvents to isoporous
membranes. Langmuir 2011, 27 (16), 10184−10190.
(14) Jackson, E. A.; Hillmyer, M. A. Nanoporous membranes derived
from block copolymers: From drug delivery to water filtration. ACS
The Supporting Information is available free of charge on the Nano 2010, 4 (7), 3548−3553.
ACS Publications website at DOI: 10.1021/acs.iecr.6b00387. (15) Seo, M.; Moll, D.; Silvis, C.; Roy, A.; Querelle, S.; Hillmyer, M.
A. Interfacial Polymerization of Reactive Block Polymers for the
Proposed algorithm used for segmentation and quanti- Preparation of Composite Ultrafiltration Membranes. Ind. Eng. Chem.
fication of porosity in different layers (PDF). Res. 2014, 53 (48), 18575−18579.

3694 DOI: 10.1021/acs.iecr.6b00387

Ind. Eng. Chem. Res. 2016, 55, 3689−3695
Industrial & Engineering Chemistry Research Article

(16) Nunes, S. P.; Car, A. From charge-mosaic to micelle self- (34) Snyder, M.; Vlachos, D.; Nikolakis, V. Quantitative analysis of
assembly: block copolymer membranes in the last 40 years. Ind. Eng. membrane morphology, microstructure, and polycrystallinity via laser
Chem. Res. 2013, 52 (3), 993−1003. scanning confocal microscopy: application to NaX zeolite membranes.
(17) Nakao, S.-i. Determination of pore size and pore size J. Membr. Sci. 2007, 290 (1), 1−18.
distribution: 3. Filtration membranes. J. Membr. Sci. 1994, 96 (1), (35) Holzer, L.; Indutnyi, F.; Gasser, P.; Münch, B.; Wegmann, M.
131−165. Three-dimensional analysis of porous BaTiO3 ceramics using FIB
(18) Marques, D. S.; Dorin, R. M.; Wiesner, U.; Smilgies, D.-M.; nanotomography. J. Microsc. 2004, 216 (1), 84−95.
Behzad, A. R.; Vainio, U.; Peinemann, K.-V.; Nunes, S. P. Time- (36) Mangipudi, K.; Radisch, V.; Holzer, L.; Volkert, C. A FIB-
resolved GISAXS and cryo-microscopy characterization of block nanotomography method for accurate 3D reconstruction of open
copolymer membrane formation. Polymer 2014, 55 (6), 1327−1332. nanoporous structures. Ultramicroscopy 2016, 163, 38−47.
(19) Marques, D. S.; Vainio, U.; Chaparro, N. M.; Calo, V. M.; (37) Zils, S.; Timpel, M.; Arlt, T.; Wolz, A.; Manke, I.; Roth, C. 3D
Bezahd, A. R.; Pitera, J. W.; Peinemann, K.-V.; Nunes, S. P. Self- visualization of PEMFC electrode structures using FIB tomography.
assembly in casting solutions of block copolymer membranes. Soft Fuel Cells 2010, 10 (6), 966.
Matter 2013, 9 (23), 5557−5564. (38) Reingruber, H.; Zankel, A.; Mayrhofer, C.; Poelt, P. Quantitative
(20) Mottern, M. L.; Shqau, K.; Shi, J.; Yu, D.; Verweij, H. Thin characterization of microfiltration membranes by 3D reconstruction. J.
supported inorganic membranes for energy-related gas and water Membr. Sci. 2011, 372 (1), 66−74.
purification. Int. J. Hydrogen Energy 2007, 32 (16), 3713−3723. (39) Shi, M.; Printsypar, G.; Iliev, O.; Calo, V. M.; Amy, G. L.;
(21) Kyotani, T.; Mizuno, T.; Katakura, Y.; Kakui, S.; Shimotsuma, Nunes, S. P. Water flow prediction for membranes using 3D
N.; Saito, J.; Nakane, T. Characterization of tubular zeolite NaA simulations with detailed morphology. J. Membr. Sci. 2015, 487, 19−
membranes prepared from clear solutions by FTIR-ATR, GIXRD and 31.
FIB-TEM-SEM. J. Membr. Sci. 2007, 296 (1), 162−170. (40) Hilke, R.; Pradeep, N.; Madhavan, P.; Vainio, U.; Behzad, A. R.;
(22) Madhavan, P.; Sougrat, R.; Behzad, A. R.; Peinemann, K.-V.; Sougrat, R.; Nunes, S. P.; Peinemann, K.-V. Block copolymer hollow
Nunes, S. P. Ionic liquids as self-assembly guide for the formation of fiber membranes with catalytic activity and pH-response. ACS Appl.
nanostructured block copolymer membranes. J. Membr. Sci. 2015, 492, Mater. Interfaces 2013, 5 (15), 7001−7006.
568−577. (41) Blake, A.; Zisserman, A. Visual reconstruction; MIT Press:
(23) Pacheco, F.; Sougrat, R.; Reinhard, M.; Leckie, J. O.; Pinnau, I. Cambridge, 1987; Vol. 2.
3D visualization of the internal nanostructure of polyamide thin films (42) Hong, B.-W.; Lu, Z.; Sundaramoorthi, G. A new model and
in RO membranes. J. Membr. Sci. 2016, 501, 33−44. simple algorithms for multi-label mumford-shah problems. In
(24) Majoinen, J.; Haataja, J. S.; Appelhans, D.; Lederer, A.; Computer Vision and Pattern Recognition (CVPR), 2013 IEEE
Olszewska, A.; Seitsonen, J.; Aseyev, V.; Kontturi, E.; Rosilo, H.; Conference on; IEEE: New York, 2013; pp 1219−1226.
Ö sterberg, M. Supracolloidal multivalent interactions and wrapping of (43) Merriman, B.; Bence, J. K.; Osher, S. Diffusion generated motion
dendronized glycopolymers on native cellulose nanocrystals. J. Am. by mean curvature; Department of Mathematics, University of
Chem. Soc. 2014, 136 (3), 866−869. California, Los Angeles, 1992.
(25) Kaushik, M.; Basu, K.; Benoit, C.; Cirtiu, C. M.; Vali, H.; (44) Mumford, D.; Shah, J. Optimal approximations by piecewise
Moores, A. Cellulose Nanocrystals as Chiral Inducers: Enantioselective smooth functions and associated variational problems. Communications
Catalysis and Transmission Electron Microscopy 3D Characterization. on pure and applied mathematics 1989, 42 (5), 577−685.
J. Am. Chem. Soc. 2015, 137 (19), 6124−6127. (45) Tsai, A.; Yezzi, A., Jr; Willsky, A. S. Curve evolution
(26) Schryvers, D.; Cao, S.; Tirry, W.; Idrissi, H.; Van Aert, S. implementation of the Mumford-Shah functional for image
Advanced three-dimensional electron microscopy techniques in the segmentation, denoising, interpolation, and magnification. Image
quest for better structural and functional materials. Sci. Technol. Adv. Processing, IEEE Transactions on 2001, 10 (8), 1169−1186.
Mater. 2013, 14, 014206. (46) Vese, L. A.; Chan, T. F. A multiphase level set framework for
(27) ElHadidy, A. M.; Peldszus, S.; Van Dyke, M. I. Development of image segmentation using the Mumford and Shah model. International
a pore construction data analysis technique for investigating pore size journal of computer vision 2002, 50 (3), 271−293.
distribution of ultrafiltration membranes by atomic force microscopy. (47) Cooper, D.; Rivallin, P.; Guegan, G.; Plantier, C.; Robin, E.;
J. Membr. Sci. 2013, 429, 373−383. Guyot, F.; Constant, I. Field mapping of focused ion beam prepared
(28) AlMarzooqi, F. A.; Bilad, M.; Mansoor, B.; Arafat, H. A. A semiconductor devices by off-axis and dark field electron holography.
comparative study of image analysis and porometry techniques for Semicond. Sci. Technol. 2013, 28 (12), 125013.
characterization of porous membranes. J. Mater. Sci. 2016, 51 (4), (48) Volkert, C. A.; Minor, A. M. Focused ion beam microscopy and
2017−2032. micromachining. MRS Bull. 2007, 32 (05), 389−399.
(29) Dai, S.; Seol, Y.; Wickramanayake, S.; Hopkinson, D. (49) Kizilyaprak, C.; Daraspe, J.; Humbel, B. Focused ion beam
Characterization of hollow fiber supported Ionic liquid membranes scanning electron microscopy in biology. J. Microsc. 2014, 254 (3),
using microfocus X-ray computed tomography. J. Membr. Sci. 2015, 109−114.
(50) Boussu, K.; De Baerdemaeker, J.; Dauwe, C.; Weber, M.; Lynn,
492, 497−504.
K. G.; Depla, D.; Aldea, S.; Vankelecom, I. F.; Vandecasteele, C.; Van
(30) Remigy, J.-C.; Meireles, M.; Thibault, X. Morphological
der Bruggen, B. Physico-Chemical Characterization of Nanofiltration
characterization of a polymeric microfiltration membrane by
Membranes. ChemPhysChem 2007, 8 (3), 370−379.
synchrotron radiation computed microtomography. J. Membr. Sci.
(51) Müllner, T.; Zankel, A.; Svec, F.; Tallarek, U. Finite-size effects
2007, 305 (1), 27−35.
in the 3D reconstruction and morphological analysis of porous
(31) Tung, K.-L.; Chang, K.-S.; Wu, T.-T.; Lin, N.-J.; Lee, K.-R.; Lai,
polymers. Mater. Today 2014, 17 (8), 404−411.
J.-Y. Recent advances in the characterization of membrane
(52) Madhavan, P.; Sougrat, R.; Behzad, A. R.; Peinemann, K.-V.;
morphology. Curr. Opin. Chem. Eng. 2014, 4, 121−127.
Nunes, S. P. Ionic Liquids As Self-Assembly Guide for the Formation
(32) Shearing, P. R.; Eastwood, D. S.; Bradley, R. S.; Gelb, J.; Cooper,
of Nanostructured Block Copolymer Membranes. J. Membr. Sci. 2015,
S. J.; Tariq, F.; Lee, P. Exploring electrochemical devices using X-ray
492, 568−577.
microscopy: 3D micro-structure of batteries and fuel cells. Microsc.
Anal 2013, 27 (2), 19−22.
(33) Eastwood, D.; Bradley, R.; Tariq, F.; Cooper, S.; Taiwo, O.;
Gelb, J.; Merkle, A.; Brett, D.; Brandon, N.; Withers, P.; Lee, P. D.;
Shearing, P. R. The application of phase contrast X-ray techniques for
imaging Li-ion battery electrodes. Nucl. Instrum. Methods Phys. Res.,
Sect. B 2014, 324, 118−123.

3695 DOI: 10.1021/acs.iecr.6b00387

Ind. Eng. Chem. Res. 2016, 55, 3689−3695