Copyright © 2006 American Scientific Publishers Journal of

All rights reserved Nanoscience and Nanotechnology

Printed in the United States of America Vol. 6, 1–8, 2006

Pore Architecture of Diatom Frustules:

Potential Nanostructured Membranes for
Molecular and Particle Separations
Dusan Losic,1
2
∗ Gary Rosengarten,3 James G. Mitchell,2 Nicolas H. Voelcker1
1
School of Chemistry, Physics and Earth Sciences, Flinders University, Bedford Park, Adelaide 5001, SA, Australia
2
School of Biological Sciences, Flinders University, Bedford Park, Adelaide 5001, SA, Australia
3
School of Electrical and Computer Engineering, RMIT University, Melbourne 3000, Vic, Australia

Diatoms produce diverse three-dimensional regular silica structures with nanometer to micrometer
dimensions and hold considerable promise for biological and biomimetic fabrication of nanostruc-
tured materials and devices. In the present work, we describe the ultrastructural characterisation
of porous structures in diatom biosilica and discuss their potential as membrane filters for diffusion
based separations. The frustules of two centric diatom species Coscinodiscus sp. and Thalassiosira
eccentrica were investigated using scanning electron microscopy and atomic force microscopy. Their
morphological features including pore size, shape, porosity, and pore organization are described.
We observed that although pore organization in frustules of Thalassiosira eccentrica and Coscin-
odiscus sp. is in reverse order, a striking commonality is the size range of the smallest pores in
both species (around 40 nm). The consensus lower pore size suggests that frustule valves have a
common function at this size of excluding viruses or other deleterious particles and that the pore
size and organization is optimised for this purpose. We suggest and implement an experimental

RESEARCH ARTICLE
approach to study the potential of diatom frustules for diffusive separation of molecular or nanopar-
ticular components in microfluidic or lab-on-a-chip environments.
Keywords: Diatoms, Frustules, Porous Nanostructures, Membranes, Molecular Separation,
Particle Separation, Microfluidics.

1. INTRODUCTION amorphous silica.9 Each of the estimated 105 diatom spe-

The exploitation of biological processes, in particular
biomineralisation, can lead to interesting alternative routes
towards high tech materials.1–2 Since biological fabrication
operates on the nanoscale in a bottom-up fashion, nano-
structured materials with unusual properties (mechanical
or optical) are frequently found in nature.3–4 Nanocharac-
terisation tools are just starting to unveil such structural
elements that give rise to these properties of biological
materials.5 The obtained data sets enable researchers to
suggest models proposing so far unknown functions.6 In
some cases these functions have been confirmed in an
experiment with microfabricated analogues or products of
biomimetic synthesis.3 In virtually all aqueous environ-
ments an exceptional variety of silica-based structures are
generated by aquatic micro-algae known as diatoms.7–8
These diatoms are single-celled photosynthetic microor-
ganisms that possess exoskeletons (frustules) comprised of
∗
Author to whom correspondence should be addressed.
cies has a specific frustule shape decorated with a unique
pattern of nano-sized features (pores, channels, ridges,
spikes, spines etc.).7
A variety of potential applications for diatom frustules
including optics, photonics, catalysis, nanofabrication, bio-
sensing, drug delivery, filtration, bioencapsulations, and
immunoisolations have been proposed and explored.10–19
Most of these applications rely on using the unique and
ordered porous structures of frustule valves. Porous matter
is omnipresent in nature for functional reasons and the
material properties of pores provide a large scope for
exploitation. Micro- to macroporous membranes are fre-
quently used for separation processes, and there is
currently intensive research exploring several novel nano-
materials based on zeolites, metallo-organic composites,
inorganic oxides, and polymers.20–22
Considerable effort has gone into studying and creating
microstructures within microfluidic channels both for fil-
tering, trapping, and sorting nanoparticles.23–24 One partic-
ularly challenging problem is the incorporation of porous

J. Nanosci. Nanotechnol. 2006, Vol. 6, No. 4 1533-4880/2006/6/001/008 doi:10.1166/jnn.2006.174 1

 Pore Architecture of Diatom Frustules: Potential Nanostructured Membranes for Molecular and Particle Separations
membranes into microfluidic components or lab-on-a chip
systems.25–26 In contrast, the biological production of
amorphous silica proceeds under mild physiological con-
dition, and yet leads to a remarkable diversity of nano-
to microstructured frameworks.9 Diatoms can routinely be
grown to more than 106 cells/ml of culture therefore offer-
ing the possibility of cheaply producing nonporous silica.
In addition, because diatoms live in a complex environ-
ment with continuous flow where they are exposed to
living and nonliving Brownian particles, ranging from
nutrient molecules to deleterious bacteria and viruses, their
nanomorphologies are preadapted and evolved to passively
sort a myriad of particles. Diatom domination of marine
environments is testament to their successful and adap-
tive metabolism and their resistance to infection, which is
likely to be related to an optimised filtration and separa-
tion of particles. Hence, diatoms are promising candidates
for membrane applications, especially in combination with
microfluidic systems.27
In our previous work, we have shown that particle sorting
and separation is feasible on diatom frustules.27–29 It was
found that the microtopography of frustule surfaces con-
trols particle movement in very specific ways that appear
consistent with surface-induced drag. The surface struc-
ture alters the diffusion and advection of submicrometer
RESEARCH ARTICLE
Fig. 1. Acid cleaned frustule valves of two centric diatoms, (a) Coscin-
odiscus sp. and (b) T. eccentrica.
2 J. Nanosci. Nanotechnol. 6, 1–8, 2006
Losic et al.
particles, resulting in localized concentrations of particles
on ridged areas surrounding areolae. A second, prima
facie, approach consists of using frustule nanopores and
their membrane properties as the basis for molecular or
particle separation. Following this rationale, diatom frus-
tules or corresponding biomimetic structures could find use
as separation components in micro-fluidic and lab-on-a-
chip systems. To prove this concept, we firstly investigated
the detailed porous architecture of centric diatom frustules
over a range of scales from the whole valve (Figs. 1a–b),
to the nanometer size pores. The detailed description and
comparison of the frustules’ porous structures is essen-
tial for understanding the principles of separation through
the frustule membrane. Using these results, we identified
a geometrical model of frustule pores for the two diatom
species and we designed an experiment which could be
applicable for microfluidic systems. We present results for
molecular and particle diffusive transport through a single
frustule valve within a microfluidic environment.
2. EXPERIMENTAL DETAILS
Two diatom species Coscinodiscus sp. and T. eccentrica
were obtained from CSIRO, Marine Division (Hobart, Tas-
mania, Australia). The cultures were maintained at 20  C
using a 12 hour light/12 hour dark cycle. GSE medium was
used to culture Coscinodiscus sp. and Guillard’s medium
(f/2) was used for T. eccentrica.30 The live diatoms were
harvested after 2–3 weeks of culturing and cleaned using
a procedure described elsewhere.31 Organic material from
the frustule surface was removed using concentrated sul-
phuric acid or surfactant (2% SDS in 100 mM EDTA).32
These treatments cause the separation of the diatom frus-
tule into the 2 frustule valves and girdle band. Cleaned
frustule valves for both procedures were stored in 100%
ethanol. The structural characterisation of frustules was
performed by using scanning electron microscopy (SEM)
and atomic force microscopy (AFM). One drop of cleaned
frustule suspension was deposited onto freshly cleaved
mica or silicon wafers previously coated with polylysine
(from 0.01% aqueous solution of polylysine). Samples
were dried in a gentle stream of nitrogen for 1 minute.
Frustules settled on the substrate surface exposing their
inner surface (concave) or exterior surface (convex).
SEM images were acquired using a Philips XL 30
field-emission scanning electron microscope operated at
2–10 kV. Silicon wafers with diatom frustules were
mounted on microscopy stubs with carbon sticky tape and
coated with a thin platinum layer.
AFM imaging was performed using a Nanoscope IV
Multimode SPM (Veeco Corp., Santa Barbara, USA).
Images were acquired in air using contact and tapping
modes. Oxide-sharpened silicon nitride probes (NP-S,
Veeco Corp.) of spring constant k = 0
15 N/m were used
for contact mode experiments. Olympus silicon probes
Losic et al. Pore Architecture of Diatom Frustules: Potential Nanostructured Membranes for Molecular and Particle Separations
(TESP, Veeco Corp.) were used for tapping mode experi-
ments. These probes were cleaned in water plasma (6 mA
source, 0.05 Torr water) for 3 minutes and characterised
using a silicon calibrating grating (TGT 01, Silicon-MDT
Ltd., Russia).33 Different parts of the frustule from the cen-
tral to the peripheral areas were scanned on the inner and
outer frustule surface. A minimum of ten different frus-
tules of each species was imaged to account for minor
shape variations. All values of pore size are the average
of at least fifty individual measurements performed on
at least five different frustules ± the standard deviation.
Image processing was performed using Nanoscope off-
line software (Veeco Corp.) and SPIP (Image Metrology,
Denmark).
For the permeation experiment, cleaned diatom frustules
(Coscinodiscus sp.) were attached on the top of a micro-
capillary tube (single hole optical fibre with internal diam-
eter of 15 m, supplied by Optical Fibre Technology
Centre, Sydney, Australia) and glued with UV curable glue
(NOA-60, Norland, USA). Manipulation and attachment
of an individual frustule onto the end of a microcapil-
lary tube was initially performed using a Newport Auto
Aligner (USA). The attachment using a micromanipulator
(Micromanipulator Co, USA) and an inverted microscope
equipped with two cameras (bottom and side view) proved
to be sufficient to perform this procedure. For the mea-
surement of molecular transport of small molecules across
a microcapillary tube mounted frustule, a fluorescein
(Sigma-Aldrich, USA) solution (1 mM) was used. The per-
meation experiment toward particles was performed using
fluorescent nanoparticles (FluoSpheresTM ) with 20 nm and
100 nm diameters supplied from Molecular Probes, Invit-
rogen, USA. The permeation of molecules and parti-
cles across the frustule was monitored using a minia-
ture fibre optic fluorescence spectrophotometer USB 2000-
FLG (Ocean Optics, USA) and an LED light source with
emission maximum at 470 nm (Ocean Optics). The trans-
port data are processed as plots of moles of fluorescein
molecules in the permeate versus permeation time and the
permeation flux is calculated from the slope.
3. RESULTS AND DISCUSSION
3.1. Characterisation of the Coscinodiscus sp. Frustule
SEM and AFM images of the Coscinodiscus sp. frustule
from the external to the internal surface including their
profile structures are presented in Figure 2. Three porous
layers are identified in this frustule. The outer porous layer,
known as the cribellum is followed by a second porous
layer, called the cribrum which is connected to the internal
(third) porous silica plate containing the so called foramen
holes. Structural details of these layers and their porous
properties are given below.
Figures 2a–e show a series of SEM and AFM images
of the delicate cribellum layer, which is preserved upon
J. Nanosci. Nanotechnol. 6, 1–8, 2006
gently cleaning of frustules but removed by acid treatment
as described in the experimental part. Figure 2a shows a
whole frustule with intact valves, followed by two SEM
images (Figs. 2b–c) with details of the cribellum surface
and its pores. The cribellum surface consists of about
2 m diameter porous domes hexagonally packed across
the frustule. A more detailed AFM image of a dome is
presented in Figure 2d. The basic porous feature of this
layer is an array of 5–7 small pores, as shown in Figure 2e.
There are approximately 20 arrays per one dome with each
array surrounded by six other ones giving rise to a hexag-
onal pattern. The average pore diameter is determined as
45 ± 9 nm and a pore-to-pore distance of 68 ± 12 nm.
This corresponds to a porosity of only 7.5 ± 1.2%. In
Coscinodiscus sp., the cribellum represents the sieve plate
in contact with the external environment, hence the porous
features may have evolved for optimal nutrient uptake and
for defence against bacterial and viral attack. The observed
pore size suggests that this layer is a mechanical barrier or
filter that allows passage for particles smaller than 40 nm.
Particles below this size are unlikely to be of viral nature
and more likely to be salubrious.
A second porous layer, the cribrum, is exposed on
frustules prepared by acid cleaning which removed the
thin (<50 nm) and delicate cribellum layer only weakly
bonded to the underlying silica layer. A series of SEM and
AFM images of the cribrum are shown in Figures 2f–k.
The SEM image of a whole frustule valve displaying the
RESEARCH ARTICLE
cribrum surface is shown in Figure 2f, followed by detailed
images (Figs. 2g–i). In these images a honeycomb topog-
raphy of hexagonally packed holes is evident covered by a
dome-shaped porous silica membrane slab 1620 ± 130 nm
in diameter. The pores are shown with more detail in AFM
image (Fig. 2j) followed by an AFM image of a single pore
(Fig. 2k). About 20 cribrum pores with a pore diameter of
192 ± 35 nm decorated each dome and domes are self-
similar to the cribellum pore arrays. The porosity of the
cribrum layer (25.2 ± 2.5%) is significantly higher than
that of the cribellum. Cribrum pores and cribellum pore
arrays appear to be overlaid. The high-resolution AFM
image of the cribrum membrane (Fig. 2k) also shows that
the biosilica consists of fused nodules with average size of
74 ± 8 nm.34 These are presumably the result of colloidal
silica incorporated into the biomineralising frustule.8
SEM and AFM images of the third porous (internal)
layer are depicted in Figures 2l–r. The large-scale image
in Figure 2m shows the presence of the large pores (fora-
men), which are radially organised in lines from the cell
centre to the edge. The foramen holes are 1150 ± 130 nm
in diameter with an average spacing of 603 ± 41 nm.
This yields a porosity of 35 ± 3%. The porosity hence
increases from the outer to the inner membrane. At the
bottom of the foramen holes the cribrum layer is visible
in both SEM (Figs. 2n–o) and tapping mode AFM images
(Figs. 2p–r). Cross-sections imaged by AFM allowed us
3
 Pore Architecture of Diatom Frustules: Potential Nanostructured Membranes for Molecular and Particle Separations Losic et al.
RESEARCH ARTICLE

Fig. 2. Structural characterisation of frustule Coscinodiscus sp. SEM images of the outer layer (cribellum) of a whole frustule valve (bar scale 20 m)
in (a) and pore structures with more details in (b–c); AFM image (tapping mode in air) shows cribellum pores in (d) with details of an individual pore
array in (e); SEM images of the second layer (cribrum) of a whole frustule valve, (bar scale 20 m) in (f) and pore structures with more details in
(g–i); AFM images show cribrum pore structures in (j) with detailed structure of one pore in (k); SEM images of internal surface of frustule show a
whole frustule valve (bar scale 20 m) in (l) and images of foramen openings in (m–o); AFM image of pore structure (foramen) with corresponding
cross-section graphs is shown in (p–r); SEM cross-sections are shown in (s–u).

to visualise chambers (areolae) of 359 ± 62 nm depth
formed by the foramen holes and capped by the cribrum.
Interestingly, AFM images also show that the edge of the
foramen holes are elevated by 75 ± 14 nm above the silica
plane (Fig. 2p). These circular elevations enclose channels
radiating from the centre of the frustule to the periphery.
Whether these elevations fulfil a specific function or are
artefacts of the biomineralisation process is not clear at
this point.
Images of the internal porous structure taken from the
gently (surfactant) cleaned frustule are shown in Figure 2s
and those of acid cleaned frustule are shown in
Figures 2t–u. These images also confirm that all porous
layers (cribellum, arrow 1, cribrum, arrow 2, and foramen,
arrow 3) are integrated into a single multilayer membrane
of 1–1.2 m thickness.
3.2. Characterisation of the T. eccentrica Frustule
SEM and AFM images of the outer surface of the frus-
tule of T. eccentrica are presented in Figures 3a–e. The
first image (Fig. 3a) shows the whole frustule valve, fol-
lowed by higher resolution SEM images (Figs. 3b–d).
Here, foramen pores are organised in a hexagonal pat-
tern. Their diameter is determined from AFM images as
770 ± 38 nm, spaced 473 ± 55 nm apart (Fig. 3e). The
porosity of this layer is 37 ± 3%. This external frustule
structure is very similar to the internal frustule structure
of Coscinodiscus sp. in terms of the porosity and the pore

4 J. Nanosci. Nanotechnol. 6, 1–8, 2006

Losic et al. Pore Architecture of Diatom Frustules: Potential Nanostructured Membranes for Molecular and Particle Separations

RESEARCH ARTICLE
Fig. 3. Structural characterisation of frustule T. eccentrica. SEM images of the external layer show a whole frustule valve (bar scale 10 m) in (a)
and pore structures with more details in (b–d); AFM image (tapping mode in air) shows structure of an individual pore in (e); SEM images of internal
frustule surface show whole valve (bar scale 10 m) in (g) and closer view of pores structures in (h–i); AFM image of pore structure (foramen) with
a corresponding cross-section graph is shown in (j–k) and SEM cross-sections in (l–m).

patterns. However, T. eccentrica foramen pores are not
radially organised like the foramen holes on Coscinodiscus
sp. (Fig. 2m). Rather, they follow a concentric arrange-
ment (Fig. 3b). High-resolution AFM images of the silica
wall between two openings show granular nanostructured
topography with individual silica nodules of 40 ± 6 nm
diameter.34 We attempted to image the inner silica plate
at the bottom of the foramen openings by AFM, but this
could not be achieved due to the significant depth of the
foramen pore in T. eccentrica and the insufficient aspect
ratio of the AFM tips used.
SEM and AFM images of the inner surface of the frus-
tule of T. eccentrica are presented in Figures 3f–k. The first
image (Fig. 3f) shows a whole internal frustule surface,
followed by SEM images at higher resolution (Figs. 3g–i).
Characteristic arrays of pores with size 780 ± 90 nm diam-
eter, arranged in hexagons but with rather irregular bound-
aries are observed (Fig. 3h). The pattern is centripetal in
relation to the foramen openings seen on external part of
frustule (Figs. 3a–e). The irregularities of the array shape
(circular, ellipsoidal, quasi-rectangular) might be related
to suboptimal culturing condition. High-resolution SEM
and AFM images and cross-sectional graphs of the porous
part are shown in Figures 3i–k. The pores (43 ± 6 nm
diameter) are highly ordered into hexagonal packing as the
Fourier transform analysis from the AFM images shows.
Due to the large spacing between pore arrays, the poros-
ity of this layer is only 10 ± 2.5%, again similar to the
porosity of the cribellum layer of Coscinodiscus sp.
Cross-sections of T. eccentrica frustules are presented
in the SEM images in Figures 3l–m. The foramen walls
form chambers, which are capped on one side by the pore
arrays of the internal membrane. The internal shape of
pores is hexagonal with sharp edges, which is in stark con-
trast to the internal pore shape of Coscinodiscus sp. where
rounded shapes are dominant. The thickness of the frustule
structure ranged between 0.9–1 m, slightly thinner than
the Coscinodiscus sp. frustule.

J. Nanosci. Nanotechnol. 6, 1–8, 2006 5

 Pore Architecture of Diatom Frustules: Potential Nanostructured Membranes for Molecular and Particle Separations
At 30–40 m in diameter, T. eccentrica are approx-
imately half the size of Coscinodiscus sp. although the
thickness of the silica plate is similar (around 1 m). We
note that the foramen hole size in T. eccentrica is also
approximately 50% of the size of foramen holes in Coscin-
odiscus sp. However, the most prominent differences are
the presence of three porous layers in the Coscinodiscus
sp. frustule versus two porous layers in T. eccentrica and
the inversion of their structural layers as discussed earlier.
Whilst the Coscinodiscus sp. frustule pore size decreased
from the outside of the cell (cribellum, smallest pores) to
the inside (foramen, largest pore size), in T. eccentrica the
order is reversed. Figure 4 summarises these differences
showing a schematic model of a pore structure on the
surfactant cleaned and acid cleaned frustule of Coscin-
odiscus sp. (Figs. 4a–b) as well as on the T. eccentrica
frustule (Fig. 4c). Another major difference is the orga-
nization of the pores for these two species in the sieve
RESEARCH ARTICLE
Fig. 4. Schematic model of frustule pore architecture for the two inves-
tigated diatoms. (a) Gently cleaned Coscinodiscus sp.; (b) acid cleaned
Coscinodiscus sp. and (c) acid cleaned T. eccentrica.
6 J. Nanosci. Nanotechnol. 6, 1–8, 2006
Losic et al.
plate. Pores in T. eccentrica are grouped together in large
groups with hundreds of pores per array. In Coscinodiscus
sp. frustule pores are organised in groups of 5–7 pores per
array (cribellum layer), which are regularly dispersed. It
is intriguing to postulate that the different nanostructured
features on the diatom frustule and their organization has
been optimized by evolution for tasks such as uptake of
nutrition, sorting particles, defense, communication or a
combination of the above. But how can this be reconciled
with the fact that there are such significant differences in
the organization of the layers as the complete inversion
of the pore gradient? Possibly, the thick silica plate form-
ing the areolar chambers is responsible for the mechanical
stability of the frustule whilst the comparatively thin mem-
branes of cribrum and cribellum have developed for filtra-
tion and separation. Interestingly, for both diatom species
investigated here the size of the small pores (ca 40 nm)
is consistent. This consensus might imply that this pore
size is the species comprehensive size exclusion limit for
separation of salubrious particles and molecules (<40 nm)
and deleterious particles like virus and bacteria (>40 nm).
Therefore the location of the small pore layer (internal or
external) might not be critical to the sorting and separation
process.
3.3. Potential of Diatom Frustules for
Membrane Based Separations
The structural characterisation of the frustules for both
diatoms shows the potential as a blue-print for advanced
nanostructured membranes. The dimensions of the small-
est pores, their organization on the membrane and the
thickness of this membrane are arguably optimised for
sorting and separation. However, the shape of the areo-
lar chamber and/or the curvature of the domes might also
affect these processes.
Due to the tiny dimensions of the diatoms, and because
they live in very slow moving environments the Reynolds
number associated with flow around the frustules and
the particles and macromolecules that diatoms digest is
extremely low (<<1). Viscous drag is then the only sig-
nificant competing force with thermal (Brownian) motion.
This paves the way for diffusion based transport phenom-
ena for filtering that can be achieved using a tangential
flow geometry ensuring the correct flow velocities. There
is also a possibility of a Brownian ratchet-type separa-
tion mechanism occurring, with ocean pressure fluctua-
tions and diatom tumbling motion producing a periodic
driving potential, and the three-dimensional pore structure
with large diameter variations offering the asymmetry that
is required.35
Without doubt, the complex architecture of frustule
pores is quite distinctive from the simple architecture of
artificial membranes with long cylindrical or conical pores
and a moderate to high dispersity in size and density.36–37
Losic et al. Pore Architecture of Diatom Frustules: Potential Nanostructured Membranes for Molecular and Particle Separations

It is recognised that more complex shapes such as asym-

metrical pores can improve molecular separation efficiency
and selectivity.38 In the area of microfluidics, the large
surface area to volume ratios and need for small pore
sizes in separation layers creates large back pressures, fre-
quent clogging or very low flow rates. The structure of
the diatoms suggests how tangential flow systems could be
implemented in a microfluidic device where separation is
achieved through diffusion based transport phenomena.
The using of diatom frustules for selective membrane
separation has been suggested in the literature, but has not
yet been explored to our knowledge.10 13 17 The key to
exploit the diatom’s separation and transport properties is
the ability to manipulate individual diatom frustules and
integrate them in microfluidic channels. Our first attempt
to manipulate frustule in PDMS channels using a micro-
manipulator were not met with success due the high
fragility of the frustule structure. However, we have instead
achieved the attachment of an individual frustule to the
end of a microcapillary tube (Fig. 5a). A schematic of our
experimental set-up is shown in Figure 5b. We have subse-
quently been able to measure diffusion based transport of
molecules and nanoparticles through frustules. Figure 6a
shows the typical transports plot versus time of fluorescein
molecules permeating through the frustule membrane.
From the slope of these graphs, the flux of fluores-
cein molecules was calculated as 1.02 ± 0.20 nmoles
min−1 cm−2 (n = 3). We also have calculated theoreti-

RESEARCH ARTICLE
cal flux of fluorescein through diatom membrane using
modified Fick’s law and membrane model with a similar

Fig. 6. Permeation study of molecules and nanoparticles through

diatom frustule. (a) Time-lapse transport study of permeation of fluores-
cein; (b) Fluorescence emission spectra of 20 nm fluorescent particles
in feed solution and (c) in permeate solution after 8 hours of diffusion
through diatom membrane; (d) Fluorescence emission spectra of 100 nm
fluorescent particles in feed solution and (e) in permeate solution after
8 hours of diffusion.

size and area of pores to those of the diatoms.39 The result-

Fig. 5. (a) Diatom frustule (Coscinodiscus sp.) mounted on the top of
a microcapillary tube; (b) Experimental set-up for separation study using
diatom frustules in a microfluidic system.
ing flux of 5.5 nmoles min−1 cm−2 is in reasonable agree-
ment to the measured flux from the experiment, which
shows that the behaviour of the complex geometry of the
diatom frustule is consistent with diffusive theory.
To probe the size selectivity of diatom frustule toward
nanoparticles with different size, we used fluorescent nano-
particles with two sizes (20 nm and 100 nm) in the feed
solution. A single-component permeation experiment is
performed for 8 hours and fluorescence spectra in feed and
permeate solution are presented in Figures 6b–e. The fluo-
rescence signal of smaller particles (20 nm) were detected
in the permeate solution after 8 hours (Fig. 6c), but not

J. Nanosci. Nanotechnol. 6, 1–8, 2006 7

 Pore Architecture of Diatom Frustules: Potential Nanostructured Membranes for Molecular and Particle Separations
for larger particles (100 nm) where only background sig-
nal is observed (Fig. 6e). These results can be considered
as the first demonstration of size exclusion properties of
an diatom frustule. The details of these studies will be
presented in another publication.
4. CONCLUSIONS
The pore structures of the frustule valves of two
diatom species (Coscinodiscus sp. and T. eccentrica) were
resolved at both nano- and micro-level using a combina-
tion of SEM and AFM. Fundamental differences between
these two species were found in terms of the inversion of
sieve plate and the foramen layers. Other morphological
differences in frustules include the number of layers, the
size, shape, density, and geometric arrangement of pores.
However, both species show a consensus lower size limit
for pores in the frustule (about 40 nm), which might point
to a critical size exclusion limit that is beneficial to both
nutrient uptake and exclusion of bacteria and virus. The
complex architecture of frustule pores is quite distinctive
from the simple architecture of artificial membranes with
simple cylindrical pores currently used for nanofiltration.
Diffusive transport through diatom frustules would be
of great interest for microfluidic environments and lab-on-
a-chip systems. To that respect we have carried out pre-
liminary experiments showing that diatom frustules can be
micromanipulated onto microcapillaries and integrated into
RESEARCH ARTICLE
a microfluidic device. Initial results demonstrated success-
ful diffusion based transport of organic fluorophores and
fluorescent nanoparticles through a single diatom frustule
mounted on microcapillary tubes.
Acknowledgments: The authors acknowledge support
from Flinders University and the Australian Research
Council (ARC). We thank RMIT Micromaterial Technol-
ogy Centre (Melbourne) for the use of the AutoAligner and
Dean Pavlickovski (RMIT) for his help with the diatom
manipulations. We thank to Shane Huntington and Katya
Lytikainen for providing microcapillary tubes.
References and Notes
1. C. R. Lowe, Curr. Opin. Struct. Biol. 10, 428 (2000).
2. S. I. Stup and P. V. Braun, Science 277, 142 (1997).
3. M. Srinivasarao, Chem. Rev. 99, 1935 (1999).
4. M. Sarikaya, C. Tamerler, A. K.-Y. Jen, K. Scholten, and F. Baneyx,
Nature Mater. 3, 577 (2003).
Received: 10 November 2005. Revised/Accepted: 18 January 2006.
8 J. Nanosci. Nanotechnol. 6, 1–8, 2006
Losic et al.
5. G. S. McCarty and P. S. Weiss, Chem. Rev. 99, 1983 (1999).
6. A. R. Parker, R. C. McPhedran, D. R. McKenzie, N.-A. Nigorovici,
and L. C. Botten, Nature 409, 36 (2001).
7. F. E. Round, R. M. Craford, and D. G. Man, The Diatoms, Cam-
bridge University Press, Cambridge (1990).
8. R. Wetherbee, S. Craford, and P. Mulvaney, Biomineralisation:
From Biology to Biotechnology and Medical Application, edited by
E. Beurlein, Wiley-VCH, Weinheim (2000), p. 189.
9. R. Gordon and R. W. Drum, Int. Rev. Cytol. 150, 243 (1994).
10. J. Parkinson and R. Gordon, Trends Biotechnol. 17, 190 (1999).
11. D. E. Morse, Trends Biotechnol. 17, 230 (1999).
12. E. G. Vrieling, T. P. M. Beelen, R. A. van Santen, and W. W. C.
Gieskes, J. Biotechnol. 70, 39 (1999).
13. R. W. Drum and R. Gordon, Trends Biotechnol. 21, 325 (2003).
14. M. R. Weatherspoon, S. M. Allan, E. Hunt, Y. Cai, and K. H.
Sandhage, Chem. Commun. 651 (2005).
15. T. Fuhrmann, S. Landwehr, M. E. Rharbi-Kucki, and M. Sumper,
Appl. Phys. 78B, 257 (2004).
16. N. L. Rosi, C. S. Thaxton, and C. A. Mirkin, Angew. Chem. 116,
5616 (2004).
17. C. E. Hamm, J. Nanosci. Nanotechnol. 5, 108 (2005).
18. M. Hilderbrand, J. Nanosci. Nanotechnol. 5, 146 (2005).
19. D. Losic, J. G. Mitchell, and N. H. Voelcker, Chem. Commun. 4905
(2005).
20. M. E. Kosal, J.-H. Chou, S. R. Wilson, and K. S. Suslick, Nature
Mater. 1, 118 (2002).
21. H. Li, M. Eddaoudi, M. O’Keefe, and O. M. Yaghi, Nature 402,
276 (1999).
22. T. C. Merkel, B. D. Freemen, R. J. Spontak, Z. He, I. Pinnau,
P. Meakin, and A. J. Hill, Science 296, 519 (2002).
23. D. S. Paterson, Lab on a Chip 5, 132 (2005).
24. C. Chou, R. Austin, O. Bakajin, and J. Tegenfeldt, Electrophoresis
21, 81 (2000).
25. M. P. MacDonald, G. C. Spalding, and K. Dholakia, Nature 426,
421 (2003).
26. T. Sano, N. Iguchi, K. Ilda, T. Sakamoto, M. Baba, and H. Kawaura,
Appl. Phys. Lett. 83, 438 (2003).
27. M. S. Hale and J. G. Mitchell, Aquatic Microbial Ecology 24, 287
(2001).
28. M. S. Hale and J. G. Mitchell, Nano Lett. 1, 617 (2001).
29. M. S. Hale and J. G. Mitchell, Nano Lett. 2, 657 (2002).
30. R. R. L. Guillard and J. H. Ryther, Can. J. Microbiol. 8, 229 (1962).
31. R. Hasle and G. A. Fryxell, Trans. Amer. Micros. Soc. 89, 469
(1970).
32. F. Noll, M. Sumper, and N. Hampp, Nano Lett. 2, 91 (2002).
33. D. Losic, K. Short, J. J. Gooding, and J. G. Shapter, J. Serb. Chem.
Soc. 69, 93 (2004).
34. D. Losic, R. J. Pilar, T. Dilger, J. Mitchell, and N. H. Voelcker,
J. Porous Mater. (2006), in press.
35. S. Matthias and F. Müller, Nature 424, 53 (2003).
36. K. B. Jirage, J. C. Hulteen, and C. R. Martin, Science 278, 655
(1997).
37. N. Li, S. Yu, C. Harrell, and C. R. Martin, Anal. Chem. 76, 2025
(2004).
38. Z. Siwy, P. Apel, D. Dobrev, R. Neumann, R. Spohr, C. Trautmann,
and K. Woss, Nucl. Instr. Meth. B 208, 143 (2003).
39. H. C. Berg, Random Walks in Biology, Princeton University Press,
Princeton (1983), p. 142.