Post-print of:
Nature Materials 11, 53–57 (2012)
doi:10.1038/nmat3179
Collective osmotic shock in ordered materials
Paul Zavala-Rivera, Kevin Channon, Vincent Nguyen, Easan Sivaniah, Dinesh Kabra,
Richard H. Friend,
S. K. Nataraj, Shaheen A. Al-Muhtaseb, Alexander Hexemer,
Mauricio E. Calvo & Hernán Míguez
Affiliations
Biological and Soft Systems, Cavendish Laboratory, Cambridge University, CB3 0HE,
UK
Paul Zavala-Rivera, Kevin Channon, Vincent Nguyen, Easan Sivaniah & S. K. Nataraj
Optoelectronics Group, Cavendish Laboratory, Cambridge University, CB3 0HE, UK
Dinesh Kabra & Richard H. Friend
Department of Chemical Engineering, Qatar University, PO Box 2713, Doha, Qatar
S. K. Nataraj & Shaheen A. Al-Muhtaseb
Lawrence Berkeley National Lab, 1 Cyclotron Road MS 2R0400, Berkeley, California
94720, USA
Alexander Hexemer
Instituto de Ciencia de Materiales de Sevilla (CSIC-US), C/Américo Vespucio 49,
41092 Seville, Spain
Mauricio E. Calvo & Hernan Miguez
Osmotic shock in a vesicle or cell is the stress build-up and subsequent rupture of the
phospholipid membrane that occurs when a relatively high concentration of salt is
unable to cross the membrane and instead an inflow of water alleviates the salt
concentration gradient. This is a well-known failure mechanism for cells and vesicles
(for example, hypotonic shock) and metal alloys (for example, hydrogen
embrittlement)1, 2, 3. We propose the concept of collective osmotic shock, whereby a
coordinated explosive fracture resulting from multiplexing the singular effects of
osmotic shock at discrete sites within an ordered material results in regular bicontinuous
1
structures. The concept is demonstrated here using self-assembled block copolymer
micelles, yet it is applicable to organized heterogeneous materials where a minority
component can be selectively degraded and solvated whilst ensconced in a matrix
capable of plastic deformation. We discuss the application of these self-supported,
perforated multilayer materials in photonics, nanofiltration and optoelectronics.
Main
Consider an ordered nanoscale assembly of solute-containing spheres that are trapped in
a semi-impermeable matrix. The matrix prevents the diffusion of solute into a
surrounding bath of solvent but permits solvent ingress. It is reasonable to expect that
large solvent-induced osmotic stresses within the spheres could cause a regular series of
small explosions within the matrix. The resulting ruptures between the spheres would
create a pathway for the complete release of solute. This collective osmotic shock of the
material would be an innovative method for the creation of bicontinuous and
nanoporous media or to obtain controlled release from a material. A first simple
demonstration of collective osmotic shock reported here is performed using selfassembled materials.
Thin films of a commercially available spherical block copolymer (BCP), polystyreneblock-polymethyl methacrylate [PS- b-PMMA], were used to create materials
susceptible to collective osmotic shock. The films consist of several layers of closepacked PMMA spherical cores, discretely spaced and surrounded by a PS matrix.
Exposure to ultraviolet light crosslinks the PS phase and breaks the PMMA down to
small oligomers. The film is immersed in acetic acid, a solvent for PMMA oligomers.
The PS matrix acts as a semi-impermeable barrier for solvent transfer such that the act
of acetic acid immersion initiates collective osmotic shock (COS).
After ultraviolet and acetic acid processing (see Methods), the transformed BCP film
architecture was characterized by scanning electron microscope (SEM). Film crosssections revealed a self-supported multilayer structure. The removal of the PMMA
component leaves behind perforated PS layers (~65 nm thick) separated from each other
by PS columns (Fig. 1a) with a period of ~120 nm for 82.5 kDa PS- b-PMMA. The film
surface itself seems to be non-porous, with a faint impression of a close packed
morphology (Fig. 1b). However, this upper layer must contain pores to remove the
film’s degraded PMMA contents. As shown later by filtration studies, these surface
pores are ~1–2 nm in diameter; this explains the lack of obvious surface pores in Fig. 1b
(noting that SEM samples are also gold coated). The internal structure of the supported
multilayer is exposed by peeling layers through the repeated application of adhesive
tape to the film (Fig. 1c) to reveal the perforations in the internal layers. Plan views of
the internal layers also indicate perforations (Fig. 1d) with larger pores than at the
surface; their sizes increase with molecular weight; 82.5 kDa and 146 kDa BCP had
interlayer pore diameters of 11.6±1.8 and 19.5±2.5 nm respectively. Grazing incidence
small angle X-ray scattering (GISAXS) studies reveal that the in-plane pore–pore and/or
column–column distances are unaltered from the original BCP nearest-neighbour
2
spacings (Supplementary Figs S1, S2). Filtration studies also implied that the surface
pore sizes (~1–2 nm) increase with BCP molecular weight. Therefore, the COSgenerated architectures resemble a self-supported perforated multilayer system, shown
schematically in Fig. 1e. These architectures were seen for a broad (38–146 kDa)
molecular weight (Mw) range of pure spherical PS- b-PMMA systems (Supplementary
Fig. S3) and produced in a broad composition range of 0.065<φPMMA<0.235 (±0.025)
(Supplementary Fig. S4). A further advantage of spherical structures is their ease of
assembly, because it has a contained architecture (unlike the intricate bicontinuous
gyroid or extended cylinder mesophase); although the reported work required 5 h of
total treatment, in practice a multilayer can be produced in just 40 min using shorter
anneal times (see Supplementary Fig. S5). Importantly, the perforated multilayers are
also observed in another spherical BCP, polystyrene-block-polybutadiene, subjected to
our ultraviolet/ozone treatment (Supplementary Fig. S3f). Besides highlighting the
broader applicability of the COS process to other systems, this is a compelling result
because styrene-butadiene BCPs are commodity materials.
The overall film thicknesses of PS- b-PMMA BCP thin films, after exposure to a range
of ultraviolet dosages, were correlated to these structures. Films, from three different
spherical BCP systems, expanded for a particular range of dosages (~5–15 J cm−2; Fig.
2a), reaching a maximal expansion at an optimal (~7.5 J cm−2) dosage. For 82.5 kDa
PS- b-PMMA, the film expansion over its initial thickness was 28 ± 3% (Fig. 2a inset)
at optimal (7.5 J cm−2) dosage. The perforated multilayers were observed in such
expanded spherical BCP films. Expansion, and the corresponding multilayer structures,
were not observed for a cylindrical BCP system, nor for samples exposed to ultraviolet
in a nitrogen atmosphere.
Ultraviolet irradiation has two effects; crosslinking the polystyrene phase and degrading
PMMA, whilst the accompanying ozone generation (in air) degrades the PS phase4. At
optimal dosage there is minimal net degradation of PS, whereas PMMA has been
degraded to low molecular weight oligomers. Initially identical films that received
minimal (2 J cm−2) and optimal dosages (Fig. 2b) were compared. Under minimal
dosage there is no film expansion or poration. Under optimal dosage, a film
corresponding to approximately eight layers of spheres expanded, with half the sphere
layers forming perforated PS layers and the other half forming PS columns. During
expansion, the films (on a reflective silicon wafer) undergo the same number of
interference colour changes during solvent dialysis as there are perforated layers. Five
characteristic spectral bands (see Supplementary Fig. S6 for 2D spectra plots) are
observed in the reflectance intensity colour map from the initial state (i) when the film
is first immersed in acetic acid via three states (1, 2, 3) to a final state (f). The
intermediate states (1–3) last ~8 s whereas the transition from one state to the next takes
~2 s. The abrupt changes are indicative of a fast expansion and fracture event in the
material.
This structure formation is a clear example of the proposed COS mechanism. Osmotic
pressure builds up owing to solvent swelling in the degraded spherical domains; having
3
no vertical constraint, the sphere layers deform in that direction, resulting in thinning of
the PS walls between the spheres. The walls subsequently rupture, releasing the
degraded polymer, and the process continues to underlying sphere layers, as shown
schematically in Fig. 2d. The smaller pores observed at the upper surface are due to the
relative thinness of the PS membrane closest to the air surface. That half the sphere
layers produce perforated layers and the other half form columns is a particular result of
alternate layers of spheres deforming to higher or lesser degrees (Supplementary Fig.
S7). Therefore surface-induced layering5, 6 of the BCP morphology plays a role in
generating the final COS multilayers. A different explanation for the pore/column
morphologies could be the initial alternation in crystalline stacking of the BCP closepacked system. However, previous work by Kramer and co-workers shows that the
spherical BCP close-packed symmetry varies as a function of film thickness7, whereas
our structures are observed over a range of film thicknesses and also in short-anneal
studies (with minimal opportunities for long-range order).
A baseline technology for nanoporous media is phase inversion; a dynamic phaseseparation process that is difficult to control at the nanoscale8. The removal of labile
components from self-assembled materials offers an alternative control in porosity9, 10,
11. However, existing methods using self-assembling materials have their drawbacks;
for example, bicontinuous gyroid nanostructures require precise synthesis, and
cylindrical pores require alignment to be effective12, 13, 14, 15. Recent research has
highlighted the use of solvent swelling (for example, by organic or super critical fluids)
of a polymeric minority phase to mechanically deform its confining matrix, leading to
bicontinuous or close-cell porous materials16, 17. One distinction of the present
research is accessing much higher osmotic stresses because of the solvation of the
smaller, degraded polymeric components.
In an example of poration technology to develop low- k dielectrics, thermally degraded
minority components diffuse through the matrix, leaving behind closed or open foam
networks18. Such networks can collapse if the matrix mobility is affected by the labile
component. In our work, the solvent must meet the three criteria of being a selective
solvent to the degraded material, a non-solvent for the matrix, and yet able to permeate
through the nanoscale barriers between sphere layers. Acetic acid is found to work for
the PS- b-PMMA system, and its requirement as a co-solvent with ethanol to develop
the PS- b-PB systems suggests a strong coupling of acetic acid to the two styrenic BCPs
explored here.
There are several factors simultaneously at work during the COS process. The column
and pore multilayers result from plastic extrusion of the matrix upwards due to biaxial
compression and rupture of the dilating spheres19. The linear stages of expansion seen
in Fig. 2c suggest case II solvent ingression through a mechanically heterogeneous
material20. The matrix can also elastically return to its original dimension (on osmotic
stress and solute release) or resist deformation entirely. The observed optimal ultraviolet
dosage expresses the degree to which the solvent-mediated, ultraviolet-crosslinked
matrix is capable of plastically yielding under the generated osmotic stresses. Below
4
this optimal condition, a low degree of PMMA degradation generates insufficient
osmotic stress, and beyond this optimal condition, the crosslinked matrix is more
resistant to yield.
The polymer multilayer presents a unit cell made of two layers of different porosity and,
thus, it can behave as a one-dimensional photonic crystal in the ultraviolet range. This
was confirmed in the reflectance spectra of multilayer films of Fig. 3a (see
Supplementary Methods). A wide, intense primary maximum, characteristic of a
photonic crystal, is centred at λc≈350 nm. Secondary lobes, on both sides of the peak,
originate from the overall film thickness. The reflectance spectrum of the non-COStreated film shows only the low-intensity maxima typical of a homogeneous layer. From
the fitting of the reflectance spectra, the unit cell thickness is estimated to be ~120 nm,
with refractive indices nc=1.23 and np=1.52 in the column and pore layers, respectively.
This significant dielectric contrast for a purely polymeric photonic multilayer21, 22
gives rise to large peak intensities (I≈80%) and gap to midgap spectral ratios
(Δλ/λc≈25%) even for a small number of multilayer unit cells.
COS-modified BCP (82.5 kDa) coatings were tested as selective layers supported on
commercial ultrafiltration membranes (CM) using two common dyes (see Methods and
Supplementary Methods). The rejection (R) of malachite green oxalate (MGO,
Mw=927 Da) was 2,000 times purer from a COS-modified membrane (R=99.98% at
2.7 bar) than from the unmodified membrane (R=60% at 2.7 bar) (Fig. 3b,c). This
massive enhancement only required a doubling of the filtrate collection time and, in
fact, the pure water fluxes through these membranes are comparable to other BCP-based
filters (Fig. 3c)23, 24, 25. The surface pore size could be reduced by using a smaller
BCP (38 kDa) in the COS-modified coating. This was verified by observing that the
filtrates of a much smaller dye, methyl orange (MO, Mw=327 Da), were 250 times
purer (from 95% to 99.98% at 2.7 bar) using the smaller BCP film (Fig. 3b and
Supplementary Fig. S8). The diameters of the rejected dyes are ~1–2 nm; these
dimensions must relate to the surface pore size of the films as the internal film porosity
is much larger. As a result the COS-modified coatings enhance the capability of the
supporting membrane by three orders of magnitude without appreciably disrupting its
overall permeability. The rejected solutes are an order of magnitude smaller than those
reported for BCP materials23, 24, 25.
The perforated multilayer scaffolds can be transformed into metal oxide replicas using
sol–gel chemistry26 (see Methods). SEM cross-sections (Fig. 3d) show the calcined
sol–gel composites retaining the original scaffold architecture in an interconnected
lamellar sheet (20 nm thick) stacks of anatase titania (as determined by X-ray
diffraction); there is a 40% thickness reduction on calcination. Hybrid polymermetaloxide light emitting (HyLED) devices were prepared to demonstrate that the
optically transparent oxide multilayers could function as electrodes. The performance of
these electrodes has not been optimized for these devices. However, it is enough to
show, as we do here, that the electrodes contain significant pathways for electron
transport. These devices produced good, unoptimized, luminance efficiencies
5
(~6 cd A−1) and low turn-on voltages (2 V; see Supplementary Fig. S9), comparable to
contemporary reports27.
We propose collective osmotic shock as a powerful structure-formation method that
coordinates a series of stress-induced ruptures in organized nanomaterials to produce
periodic nanoporous architectures. Although requiring an appropriate combination of
materials and solvent, its significance relates to situations where sizeable osmotic forces
can be generated from small molecule solutes periodically trapped within a matrix.
Therefore, self-assembly is not a requisite because ordered heterostructures are possible
by alternative physical routes28. A simple extension to the observed systems may be
possible from ultraviolet treatment of similarly layered spherical aggregates of core–
shell colloids29 or, alternatively, following solvent development of photoresist material
exposed by multi-beam interference techniques to generate cubic structures30.
Methods
All BCPs used in the study were purchased from Polymer Source (Dorval, Canada) and
had polydispersities of <1.15. Five, close-packed-sphere-forming, polystyrene-blockpolymethyl methacrylate [PS- b-PMMA] BCPs with volume fractions of PMMA,
φPMMA, between 0.13 and 0.17 and total molecular weights, Mw, of 38–146 kDa were
used in our studies. A spherical composition of polystyrene-block-polybutadiene (PS- bPB; Mw=70 kDa, φPB~0.13) and a cylinder-forming PS- b-PMMA BCP (Mw 70 kDa,
φPMMA~0.29) were also studied. Representative BCP structural confirmation is given
in Supplementary Fig. S1. BCP films, spin cast from toluene solutions onto silicon
substrates, were annealed at 230 °C in a N2 atmosphere for 2 h to allow microstructure
formation. Films were exposed to a 254 nm ultraviolet light source (UVP LLC,
Cambridge, UK) in air, before immersion in glacial acetic acid for 15 min.
The air-dried films were observed by SEM (Hitachi S-5500) and their thicknesses were
measured with a profilometer (Dek-Tak, Veeco Instruments). The SEM cross-sections
were prepared by fracturing films on single crystal silicon substrates using a diamond
knife cut against the crystal plane of the substrates. We also exposed the film’s internals
by pressing Scotch tape firmly to the film surface (in a manner similar to the generation
of graphene sheets31) and repeatedly removing it to pull off the upper film surfaces.
The film reflectances for Fig. 2c were determined from 10 μm spots on the films and
collected every second during dialysis using a spectrometer (Ocean Optics QE6500)
coupled to an optical microscope (Olympus BX-51). Thin films of untreated and COStreated films were analysed for total reflectance (Fig. 3a) using an ultraviolet–visible
scanning spectrophotometer (Shimadzu UV-2101PC) attached to an integrating sphere.
The fitting of the reflectance spectra (dashed lines in Fig. 3a) was performed using a
code based on the transfer matrix method32.
Transmission small angle X-ray scattering (SAXS) and GiSAXS experiments were
performed on beamline 7.3.3 at the Advanced Light Source (ALS), Lawrence Berkeley
National Laboratory, with an X-ray energy of 10 keV (λ=1.24). Simulations of the data
6
are based on distorted-wave Born approximation methods discussed, for example, by
Stein and colleagues7.
A commercial ultrafiltration membrane (CM) sheet (Duramem 900, GMBH Evonik)
was used as the principal support material. This is made of a crosslinked polyimide
coating on a macroporous non-woven polyester cloth substrate (~250 μm). An 82.5 kDa
PS- b-PMMA BCP thin film (300 nm) was spin-coated onto a glass substrate and then
floated off (via water) onto 22 mm and 47 mm diameter CM circles. These BCP-coated
filters were dried, annealed at 180 °C for 3 h, and subsequently treated with an optimal
dose of ultraviolet light and acetic acid to produce the COS-modified filter membranes
(COS-CM). As a control, the uncoated commercial filter was also processed in the same
way. The filters were placed in a dead-end filtration cell and the de-ionized water and
filtrate flux through the membrane was measured at a series of pressures.
Titianium sol–gel precursors were made by mixing 1 M TiCl4 and 25 M absolute
ethanol at 4 °C, following a published recipe33. The precursor solution was added
dropwise to the COS-modified polymer scaffold (from 82.5 kDa PS- b-PMMA) and the
excess spin cast off. The sample was aged in air at room temperature for 24 h and then
sintered in a preheated oven (550 °C) for 60 min.
HyLED devices were prepared following a previously described method27. Briefly,
compact TiO2 (c-TiO2) layers (~50 nm) were fabricated on ITO substrates. 300 nm
thick mesoporous titiania (m-TiO2) was generated, by the COS method, on the c-TiO2
substrates and a layer of Cs2CO3 was spin-coated on this assembly, followed by
200 nm thick films of F8BT, poly[(9,9-dioctylfluoren-2,7-diyl)-co-(1,4-benzo-{2,1′-3}thiadiazole)] ( Mw=147 kDa). Samples were annealed at 155 °C under N2 to improve
the morphology of the F8BT. Finally, all these samples were coated, by thermal
evaporation, with MoO3 (10 nm) and Au (50 nm). Current density (Keithley 2400
source measurement unit) and brightness (Keithley 2000 multimeter) versus applied
voltage characteristics for the LEDs were measured in air using a calibrated reference Si
photodetector.
7
References
1. Gordeev, Y. N. & Entov, V. M. Osmotic mechanism of fracture growth in
saturated porous media. Fluid Dyn. 36, 964–970 (2001).
2. Olbrich, K., Rawicz, W., Needham, D. & Evans, E. Water permeability and
mechanical strength of polyunsaturated lipid bilayers. Biophys. J. 79, 321–327
(2000).
3. Sargent, J. P. & Ashbee, K. H. G. The contribution of osmosis to dimensional
instability of adhesive joints. J. Adhes. 17, 83–93 (1984).
4. Fozza, A. C., Klemberg-Sapieha, J. E. & Wertheimer, M. R. Vacuum ultraviolet
irradiation of polymers. Plasmas Polym. 4, 183–206 (1999).
5. Kramer, E. J., Yokoyama, H. & Mates, T. E. Structure of asymmetric diblock
copolymers in thin films. Macromolecules 33, 1888–1898 (2000).
6. Menelle, A., Russell, T. P., Anastasiadis, S. H., Satija, S. K. & Majkrzak, C. F.
Ordering of thin diblock copolymer films. Phys. Rev. Lett. 68, 67–70 (1992).
7. Stein, G. E., Kramer, E. J., Li, X. F. & Wang, J. Layering transitions in thin
films of spherical-domain block copolymers. Macromolecules 40, 2453–2460
(2007).
8. Smolders, C. A., Reuvers, A. J., Boom, R. M. & Wienk, I. M. Microstructures in
phase-inversion membranes. Part 1. Formation of macrovoids. J. Membr. Sci.
73, 259–275 (1992).
9. Hillmyer, M. A. Nanoporous materials from block copolymer precursors. Adv.
Polym. Sci. 190, 137–181 (2005).
10. Lu, Y. F. et al. Aerosol-assisted self-assembly of mesostructured spherical
nanoparticles. Nature 398, 223–226 (1999).
11. Melde, B. J., Holland, B. T., Blanford, C. F. & Stein, A. Mesoporous sieves with
unified hybrid inorganic/organic frameworks. Chem. Mater. 11, 3302–3308
(1999).
12. Hashimoto, T., Tsutsumi, K. & Funaki, Y. Nanoprocessing based on
bicontinuous microdomains of block copolymers: Nanochannels coated with
metals. Langmuir 13, 6869–6872 (1997).
13. Zalusky, A. S., Olayo-Valles, R., Wolf, J. H. & Hillmyer, M. A. Ordered
nanoporous polymers from polystyrene-polylactide block copolymers. J. Am.
Chem. Soc. 124, 12761–12773 (2002).
o
14. Chan, V. Z. H. et al. Ordered bicontinuous nanoporous and nanorelief ceramic
films from self assembling polymer precursors. Science 286, 1716–1719 (1999).
15. Crossland, E. J. W. et al. A bicontinuous double gyroid hybrid solar cell. Nano
Lett. 9, 2807–2812 (2009).
16. Wang, Y. & Li, F. B. An emerging pore-making strategy: Confined swellinginduced pore generation in block copolymer materials. Adv. Mater. 23, 2134–
2148 (2011).
17. Yokoyama, H. & Zhang, R. Fabrication of nanoporous structures in block
copolymer using selective solvent assisted with compressed carbon dioxide.
Macromolecules 42, 3559–3564 (2009).
18. Hedrick, J. L. et al. Polyimide nanofoams from aliphatic polyester-based
copolymers. Chem. Mater. 10, 39–49 (1998).
19. Eshelby, J. D. The force on an elastic singularity. Phil. Trans. R. Soc. A 244,
87–112 (1951).
8
20. Thomas, N. L. & Windle, A. H. A theory of case II diffusion. Polymer 23, 529–
542 (1982).
21. Sharp, J. S. & Bailey, J. Infrared dielectric mirrors based on thin film
multilayers of polystyrene and polyvinylpyrrolidone. J. Polym. Sci. Polym.
Phys. 49, 732–739 (2011).
22. Thomas, E. L. et al. Polymer-based photonic crystals. Adv. Mater. 13, 421–425
(2001).
23. Peinemann, K. V., Abetz, V. & Simon, P. F. W. Asymmetric superstructure
formed in a block copolymer via phase separation. Nature Mater. 6, 992–996
(2007).
24. Yang, S. Y. et al. Nanoporous membranes with ultrahigh selectivity and flux for
the filtration of viruses. Adv. Mater. 18, 709–712 (2006).
25. Phillip, W. A., O’Neill, B., Rodwogin, M., Hillmyer, M. A. & Cussler, E. L.
Self-assembled block copolymer thin films as water filtration membranes. ACS
Appl. Mater. Interfaces 2, 847–853 (2010).
26. Alberius, P. C. A. et al. General predictive syntheses of cubic, hexagonal, and
lamellar silica and titania mesostructured thin films. Chem. Mater. 14, 3284–
3294 (2002).
27. Kabra, D., Song, M. H., Wenger, B., Friend, R. H. & Snaith, H. J. High
efficiency composite metal oxide-polymer electroluminescent devices: A
morphological and material based investigation. Adv. Mater. 20, 3447–3452
(2008).
28. Rungsawang, R. et al. Magnetically induced pattern formation in phase
separating polymer–solvent–nanoparticle mixtures. Phys. Rev. Lett. 104,
255703 (2010).
29. Zentel, R. & Fleischhaker, F. Photonic crystals from core–shell colloids with
incorporated highly fluorescent quantum dots. Chem. Mater. 17, 1346–1351
(2005).
30. Turberfield, A. J., Campbell, M., Sharp, D. N., Harrison, M. T. & Denning, R.
G. Fabrication of photonic crystals for the visible spectrum by holographic
lithography. Nature 404, 53–56 (2000).
31. Novoselov, K. S. et al. Electric field effect in atomically thin carbon films.
Science 306, 666–669 (2004).
32. Lozano, G., Colodrero, S., Caulier, O., Calvo, M. E. & Miguez, H. Theoretical
analysis of the performance of one-dimensional photonic crystal-based dyesensitized solar cells. J. Phys. Chem. C 114, 3681–3687 (2010).
33. Perez, M. D. et al. Growth of gold nanoparticle arrays in TiO2 mesoporous
matrixes. Langmuir 20, 6879–6886 (2004).
Acknowledgements
We would like to acknowledge S. Vignolini, C. Wang (beamline 11.0.1.2, LBNL), S.
Alvarez, C. Lopez and Q. Song, and the insightful comments of U. Steiner and E. J.
Kramer. This work was funded by the Qatar Foundation (QNRF), the Engineering and
Physical Sciences Research Council (EPSRC), the Consejo Nacional de Ciencia y
Tecnología (CONACyT), the Spanish Ministerio de Ciencia e Innovación (MICINN,
Consolider HOPE) and the Government of Andalucía.
9
Figure captions
Figure 1. The microstructure of collective osmotic shock generated perforated
multilayers.
Figure 2. The role of ultraviolet dosage and acetic acid on collective osmotic shock.
Figure 3. The application and inorganic transformation of collective osmotic shock
materials.
10
Figure 1
a,b, Microstructure of PS- b-PMMA (82.5 kDa, φPMMA~0.14) film; fracture crosssection (a) and surface (b). c,d, Microstructure of a PS- b-PMMA (146 kDa,
φPMMA~0.14) film after peeling with tape; side view (c) and surface of its
intermediate layers (d). Both these films have received the optimal ultraviolet dosage. e,
Idealized schematic of the self-supported perforated multilayer. All scale bars are
200 nm.
11
Figure 2
a, Thickness variations versus ultraviolet dosage in air for spherical (filled circle
50 kDa, φPMMA~0.16), (filled square 82.5 kDa, φPMMA~0.14), (filled triangle
146 kDa, φPMMA~0.14) and cylindrical (open circle 70 kDa, φPMMA~0.29) PS- bPMMA. × denotes the 82.5 kDa BCP exposed to ultraviolet in nitrogen. The inset shows
the expansion of 82.5 kDa BCP films that received optimal dosage. b, SEM images of a
300 nm BCP film (82.5 kDa) that received minimal (left) or optimal (right) dosage. The
scale bar is 200 nm. Under minimal dosage, there is no expansion and the film thickness
corresponds to ~8 layers of PMMA spheres. Under optimal dosage, the film expands
and forms four perforated layers separated by four column layers. c, The expansion of
the film under optimal dosage is shown in the dynamic reflectance intensity colour map,
where five characteristic spectral bands are observed from the initial state (i) when the
film is first immersed in acetic acid via three intermediate states (1, 2, 3) to the final
state (f) with four perforated layers. d, Schematic of the osmotic shock process acting on
layers of spheres leading to the perforated multilayers.
12
Figure 3
a, Reflectance spectra of 82.5 kDa BCP films developed in acetic acid after being
treated with an optimal ultraviolet dosage (solid blue line) and no ultraviolet treatment
(solid orange line). Optimal transfer matrix data fits (see Supplementary Information)
are also plotted (dashed lines, same colour code). b, Photograph showing the high-purity
filtrate from a 500 ppm aqueous solution of MGO after passage through a COSmodified commercial membrane (COS-CM using 82.5 kDa PS- b-PMMA) using a
dead-end filtration system. Vials of collected MGO filtrates from the CM and COS-CM
at increasing pressures (left to right) are shown in the inset. c, Filtrate flux values; pure
water flux across a CM (+) and COS-CM (×), and MGO flux through the CM (open
circle) and COS-CM (open triangle). The slope of the dashed linear fits yield pressurenormalized fluxes of 37.0, 20.7, 4.5 and 2.0 L m−2 h−1 bar−1. Rejection values for
MGO through the CM (filled circle) and COS-CM (filled triangle), and MO through
COS-CM (filled square) are also shown. d, Cross-section of titania substrates. The white
scale bar is 200 nm.
13