Highly sensitive detection of processes occurring inside nanoporous anodic alumina templates: A waveguide optical study

Article abstract of DOI:10.1021/jp0498567

We have applied nanoporous anodic alumina films as planar optical waveguides and studied changes in the effective dielectric constants of these thin films due to various processes occurring in the pores. We demonstrate the potential of the porous anodic alumina waveguide for high sensitivity (bio-) chemical sensing with bovine serum albumin adsorption and desorption at various pH values, with subangstrom sensitivity in the effective thickness of protein adsorbed. We also monitored pore widening (alumina dissolution) with subangstrom sensitivity, which is conceptually the reverse of detecting conformal film deposition on pore surfaces. Furthermore, we monitored the exchange of pore-filling media between phosphate buffer solution and ethanol, which produces qualitatively the same response as complete pore filling with other materials by various deposition techniques. Thus porous anodic alumina films may be employed simultaneously as deposition templates and as highly sensitive detectors of processes within the pores.

Full text of DOI:10.1021/jp0498567

10812
J. Phys. Chem. B 2004, 108, 10812-10818
Highly Sensitive Detection of Processes Occurring Inside Nanoporous Anodic Alumina
Templates: A Waveguide Optical Study
King Hang Aaron Lau,† Le-Shon Tan,† Kaoru Tamada,‡ Melissa S. Sander,† and
Wolfgang Knoll*,§
Institute of Materials Research and Engineering, 3 Research Link, Singapore 117602, and Departments of
Chemistry and of Materials Science, National UniVersity of Singapore, 10 Science DriVe 4, Singapore 117543
ReceiVed: January 12, 2004; In Final Form: April 12, 2004
We have applied nanoporous anodic alumina films as planar optical waveguides and studied changes in the
effective dielectric constants of these thin films due to various processes occurring in the pores. We demonstrate
the potential of the porous anodic alumina waveguide for high sensitivity (bio-) chemical sensing with bovine
serum albumin adsorption and desorption at various pH values, with subangstrom sensitivity in the effective
thickness of protein adsorbed. We also monitored pore widening (alumina dissolution) with subangstrom
sensitivity, which is conceptually the reverse of detecting conformal film deposition on pore surfaces.
Furthermore, we monitored the exchange of pore-filling media between phosphate buffer solution and ethanol,
which produces qualitatively the same response as complete pore filling with other materials by various
deposition techniques. Thus porous anodic alumina films may be employed simultaneously as deposition
templates and as highly sensitive detectors of processes within the pores.
I. Introduction
the alumina surface for specific (bio-) molecular detection
through, for example, biotin and amine coupled silane mol-
ecules6,13 can confer tailored chemical specificity for sensing
surface reactions (molecule-specific detection will be detailed
in a later account.)
Porous anodic alumina films have attracted great attention
in the areas of nanoscience and engineering due to their well-
ordered, densely packed, nanoscale pore structure that naturally
forms when aluminum films are anodized in an acidic electro-
lyte. These pores run straight through the film thickness, are
parallel to each other, and are in the range of 101 to 102 nm in
diameter. Research has focused on using porous anodic alumina
(PAA) films as templates for fabricating nanoarrays of particles
and nanowires of multifarious materials.1-5 Interest has also
been expressed in their use as biomimetic scaffolds for the self-
assembly of lipid membranes,6,7 and they have been patterned
using photolithography methods for micro-electro-mechanical
systems (MEMS).8 Excellent hexagonal pore ordering is possible
through 2-step anodization9 or by imprinting the initial Al
surface with a mold.10 The voltage and the acid electrolyte used
in anodizing the aluminum determine the pore diameter and
spacing, while acid etching fine-tunes the pore diameter through
pore widening.11,12
In this report, we apply PAA films as optical waveguides
and demonstrate the monitoring of processes occurring inside
the pores. Thus PAA films may be employed simultaneously
as a deposition template or scaffold, as well as an in situ monitor
of those processes. Very high sensitivity in the PAA sensor is
possible due to the presence of a large internal surface sensitive
to external stimuli (more than an order of magnitude increase
in surface area compared with a flat surface for films considered
in this study) and to the precise conditions necessary for guiding
confined optical modes. In addition, the ability to functionalize
The optical behavior of the composite pore-alumina layer
can be described by effective dielectric constants (ꢀeff) when
the pore diameter is controlled to <1/10 of the wavelength of
the guided light.14 This is easily achieved with proper film
processing and with common laboratory lasers. ꢀeff values have
been modeled1,14,15 and measured for both air- and metal-filled
pores,1,16 and a polymer-filled PAA channel waveguide has also
been demonstrated.17 When a PAA film is placed on a substrate,
the pores are open to the external environment at the top and
ꢀeff can be changed at will by filling the pores with different
media, which is detected as a change in the wave-guiding
conditions and constitutes the principle of operation of the PAA
sensor. For experimental simplicity, prism coupled optical
waveguide spectroscopy18 was used to measure ꢀeff of the PAA
waveguide. However, it is emphasized that other setups such
as channel waveguides and grating or end-fire coupling schemes
can be employed to suit the experimental environment.
We present three examples to illustrate the operation and
sensitivity of the PAA sensor. First, we examine the complete
exchange of the pore material by immersing the PAA waveguide
in phosphate buffer solution (PBS) and then in ethanol. PBS
and ethanol were chosen because of their importance in
biorelated experiments. This experiment also illustrates how the
sensor distinguishes changes in film dielectric properties from
a change in the surrounding media. In addition, substituting one
liquid with another in the pores is analogous to complete pore
filling with other materials such as metal or semiconductor by
deposition into the template, in which case a different set of
material-specific dielectric constants are involved. In the second
experiment, we illustrate kinetic tracking of fabrication processes
by monitoring pore widening, which is a key step in controlling
the pore diameter. In terms of sensor response, the removal of
* To whom correspondence should be addressed: knoll@mpip-main-
z.mpg.de.
† Institute of Materials Research and Engineering.
‡ Present address: Bio-Photonics Group, Photonics Research Institute,
National Institute of Advanced Industrial Science and Technology (AIST),
Tsukuba, Ibaraki, 305-8565 Japan.
§ Present address: Max-Planck-Institut fu¨r Polymerforschung, Acker-
mannweg 10, 55128 Mainz, Germany.
10.1021/jp0498567 CCC: $27.50 © 2004 American Chemical Society
Published on Web 06/23/2004

In Situ Sensing with Nanoporous Anodic Alumina
J. Phys. Chem. B, Vol. 108, No. 30, 2004 10813
Figure 1. Typical scanning electron micrograph of a porous anodic
alumina film (top view).
Figure 2. Schematic of the optical waveguide spectroscopy setup and
of the idealized field distributions of several guided modes. Excitation
of the surface plasmon and waveguide modes are observed as very
sharp dips when reflectivity (R) is measured as a function of the angle
of incidence (θ).
a layer of material is simply the reverse of conformal deposition
processes that coat the film and pore surfaces and can be just
as sensitively monitored. Thus conformal deposition processes
such as atomic layer deposition19 and layer-by-layer growth20
with PAA templates may potentially be directly analyzed in
situ. Last, bovine serum albumin (BSA) adsorption and desorp-
tion at various pH on the porous alumina surface are demon-
strated. BSA physisorption does not exceed a monolayer and
its kinetics and equilibrium behavior have been well-character-
ized for alumina particle surfaces,21-24 as well as for a range
of other surfaces.21,25,26 These reports in the literature form a
basis for comparison and verification of the utility of the PAA
film as a biomolecular sensor.
mounted on the prism assembly, the dielectric medium/
environment surrounding the film naturally operated as the top
cladding. We also clamped a Teflon flow cell onto the exposed
film to inject and extract the desired dielectric medium/reagents.
To observe optical waveguide modes, reflectivity (R) was
measured as a function of angle of incidence, θ. Both TE and
TM modes (corresponding to s- and p-polarizations, respec-
tively) can be excited in the PAA waveguide. In Figure 2, θc is
the critical angle for total internal reflection and at angles above
θc, guided modes can be observed as very sharp minima,
indexed according to the number of nodes in their field
distributions (e.g. m ) 0, 1, etc.). The angles of these modes
are completely dictated by the dielectric functions (ꢀ) and
thicknesses (h) of the waveguide layers, i.e., the glass layer,
the coupling gap metal layer, the PAA film, and the top
cladding/dielectric medium. The ꢀ of the metal coupling layer
was determined separately from surface plasmon measurements
with the same setup18 before aluminum deposition. ꢀmedium was
independently determined by θc and ꢀglass has been tightly
controlled by the manufacturer. As for the PAA film, the pore
structure dictates that one of the principal axes of the dielectric
function is parallel to the pore direction (normal to the film
surface), and two are parallel to the film surface, indicated by
ꢀeff ) {ꢀx, ꢀy, ꢀz}. From Figure 1, we see that the pore
distribution in the plane of the film is fairly isotropic and long-
range order is weak, thus ꢀx ) ꢀy. Therefore, only three
parameters, ꢀx, ꢀz, and hPAA, need be determined from the p-pol
and s-pol mode spectra to characterize the sensor response.
Although a surface plasmon resonance (SPR) mode is also
expected under p-polarization (m ) 0), the grazing angles
necessary to excite this SPR in the present setup preclude its
use for characterization purposes. Therefore, we have chosen
hPAA ∼ 1 µm such that at least two waveguide modes are always
observed in R vs θ measurements of both polarizations, yielding
independent information on ꢀx, ꢀz, and hPAA with a degree of
redundancy.
II. Experimental Section
1. Materials and SEM Characterization. Ethanol (min
99%) was purchased from Merck KGaA, Darmstadt, Singapore.
Samples of 5 mM PBS of different pH values were prepared
by dissolving appropriate amounts of sodium hydroxide, diso-
dium hydrogen phosphate, sodium dihydrogen phosphate mono-
hydrate, and 85% orthophosphoric acid (98% pure, APS Ajax
Finechem) in deionized water (18.2 MΩ‚cm). All inorganic
crystals were purchased from Merck KGaA, Darmstadt, Sin-
gapore (min 99%). Bovine Serum Albumin (66 kDa, Fraction
V, min. 96%, pH 7) was purchased from Sigma-Aldrich Co.
and was dissolved in PBS as specified above at a concentration
of 50 µM. A digital pH meter (Fisher Scientific AR 50) with
uncertainty of (0.1 between calibrations was used for all
solution preparations.
The pore structure was visualized with field emission
scanning electron microscopy (JEOL JSM-6700F) at an ac-
celeration voltage of 5 kV (Figure 1.) Samples were carbon
coated (Emitech K950X) to minimize charging artifacts. Pore
diameter and spacing were determined by software-assisted
analysis (Scion Image, Scion Corporation; Adobe Photoshop.)
2. Optical Waveguide Spectroscopy (OWS). We performed
OWS based on the Kretschmann prism coupling technique18
(Figure 2). Briefly, a thin gold coupling gap layer and aluminum
were successively deposited on glass substrates index matched
to the coupling prism. The aluminum was then anodized to form
a PAA film (described below) before the substrate was mounted
to the prism base with index matching oil. The glass prism and
substrate thus form an optically contiguous medium. A λ )
632.8 nm laser was used and a polarizer selected for either p-
or s-polarization. With the coupling gap layer and PAA film
Experimentally, Fresnel calculations were compared with
measurements to obtain the best-fit dielectric constants and layer
thicknesses.18 Such calculations assume infinite layers with
perfect interfaces and generate impressive agreement with data,
especially with layered assemblies comprised of atomically or
molecularly smooth surfaces such as mica films27 or Langmuir-
Blodgett assembled layers.28 In the case of PAA films, the

10814 J. Phys. Chem. B, Vol. 108, No. 30, 2004
Lau et al.
distribution in pore diameter and surface roughness are expected
to break up spatial coherence in waveguide coupling conditions
over probed areas of ∼1 mm2 corresponding to the width of
the incident laser beam. These effects cause the coupling dips
to broaden and decrease in amplitude in the data compared with
calculated curves.
For following the kinetics of an experiment, the angle at
which a reflectivity minimum characterizes a waveguide mode
is tracked with in-house instrumentation. Angular resolution
better than 0.005° is obtained.
3. Fabrication of Alumina Films on LaSFN9 Substrates.
LaSFN9 glass substrates (Hellma Optik, GMbH) were cut into
25 mm × 20 mm pieces and ultrasonically cleaned in 2%
Hellmanex solution (Hellma Optik, GMbH) followed by a final
rinse in absolute ethanol. To ensure adhesion to the substrate,
first a ∼3 nm Cr or Ti film was deposited onto the substrate,
then ∼46 nm Au and ∼1 µm Al (99.9999% Alfa Aesar) were
deposited. Al was deposited by electron beam evaporation
(Edwards, Auto306). Ti/Au was also deposited by electron beam
evaporation while Cr/Au was deposited by thermal evaporation
(R-DEC Co., Ltd., Japan). To anodize the Al film, the edges of
the samples were first coated with nail polish to protect them
from the solution, and the sample was placed in a beaker of
0.3 M oxalic acid solution at 2-5 °C opposite a Pt mesh counter
electrode. The Al was anodized at a constant potential of 40 V.
When anodization was nearly complete (after 30 min), there
was a distinct color change in the sample and a change in the
current. To ensure that the film was completely anodized, the
anodization was allowed to proceed 90-120 s past the color
change. After complete anodization, the pores were widened
by immersing the samples in 5 wt % of H3PO4 for 30 min. The
PAA thus formed is stable in the ambient, and X-ray crystal-
lography and thermogravimetric studies have confirmed that the
anodization process completely oxidizes the original aluminum
into an amorphous form, while annealing above 830 °C induces
formation of crystalline phases.29
4. BSA Adsorption and Desorption. BSA adsorption
proceeded in the pH range of 3.8-7.4. Before each adsorption,
the flow cell was infused with PBS at the chosen pH and a R
vs θ measurement was taken. The buffer was then exchanged
with BSA solution and the PAA waveguide was left immersed
in the BSA solution for a specific amount of time while the
adsorption kinetics (coupling angle shift) were tracked. To stop
the adsorption process, pure buffer at the same pH was
exchanged into the flow cell. Another R vs θ measurement was
taken before desorption was carried out with PBS at pH 8.5 for
another specific period of time. Desorption was stopped by
exchanging the cell with PBS at the pH chosen for the next
adsorption-desorption cycle. Both alkali22,30 and acid elution31
are effective in desorbing BSA. However, significant pore
widening occurs below pH 3.8 and above pH 8.5. No pore
widening was observed for pH 5.0 to 7.4 and acceptably slow
etching at a rate of <0.01° of shift/h was observed for
intermediate pH, consistent with Mardilovich et al.32
Figure 3. Waveguide mode patterns of PAA film filled with pH 7.4
PBS (filled circles) and ethanol (open circles) under s- (a) and
p-polarization (b). Kinetics of the exchange is shown in the inset. Lines
fitted to data are the corresponding Fresnel calculations. Parallel
shifts in waveguide mode angles are highly characteristic of a change
in the effective dielectric constant of the PAA due to an exchange in
pore material. Changes in the external media alone produce angle shifts
that are larger for higher order modes, simulated in part c. Excitation
under s-polarization shows identical behavior. Dielectric constants
of the external media are measured independently with θc-PBS and
θc-ethanol (c).
approximately parallel shifts in the mode coupling angles to
higher values (Figure 3a,b; ∆θ ) 0.92°), which is highly
characteristic of a change in ꢀeff. In contrast, a change in the
dielectric medium outside the pores alone produces a different
pattern of shift, simulated in Figure 3c. Here, ꢀeff remained
constant while the dielectric constant outside the pores changed
from ꢀPBS to ꢀethnaol as if the pores were isolated from the
environment. The pattern of angle shifts in this case is
characterized by smaller increases in the waveguide coupling
angles with higher order modes. In fact, the shift for the m )
1 mode is relatively minute due to the change in the cladding
medium. There is higher coupling efficiency under p-polariza-
III. Results and Discussion
1. Detection of a Complete Exchange of Pore Material.
Since the pores are open to the external environment, when the
PAA film is immersed in a wetting liquid media, the pores will
also be infused with the liquid. We began this experiment by
filling the pores with 5 mM PBS (pH 7.4) and then exchanged
the PBS with ethanol. ꢀPBS ) 1.774 and ꢀethanol ) 1.845, thus
ꢀeff of the PAA film increased together with the increase in
dielectric constant inside the pores. This was detected as

In Situ Sensing with Nanoporous Anodic Alumina
J. Phys. Chem. B, Vol. 108, No. 30, 2004 10815
tion, therefore the m ) 1 mode (p-pol) was chosen for kinetic
tracking of changes in the dielectric constant of the porous
alumina layer. The exchange between PBS and ethanol in the
pores necessarily goes together with an identical change above
the film. This change in the surroundings is independently
detected as a change in the critical angle for total internal
reflection from θc-PBS to θc-ethanol (Figure 3b.) Note that the
difference between ꢀPBS and ꢀethanol is a relatively drastic one
and we shall demonstrate in the following sections the detection
of much smaller changes with the PAA sensor.
The corresponding kinetic measurement is shown in the inset
of Figure 3b. Repeated exchanges show identical coupling angle
changes and indicate that the PAA film is stable in pH 7.4 PBS
and ethanol. In addition to the parallel angle shifts, this
repeatability indicates further that the exchanges were complete.
(It is highly unlikely that the pores were randomly “half-filled”
by the same amount during successive exchanges.) The sloped
steps in the inset, however, are due to the finite response time
of the instrumentation designed for angular resolution rather
than for the relatively large and fast responses shown.
Referring to parts a and b of Figure 3, the fit between data
and Fresnel calculations is excellent. As discussed in the
Experimental Section, decreases in amplitude and broadening
around coupling angle dips compared with the ideal of the
simulation are expected, but calculations of the pertinent
coupling angles produce exact fits with data. The values of ꢀeff
thus obtained are ꢀx ) ꢀy ) 2.317 and ꢀz ) 2.360 in PBS, and
ꢀx ) ꢀy ) 2.347 and ꢀz ) 2.385 in ethanol. These values are
intermediate between those of pure alumina and the respective
solutions and it is noted that ꢀz > ꢀx, ꢀy. A simple conceptual
framework of the dielectric response at pore-alumina bound-
aries oriented parallel and perpendicular to the incident electric
field may suffice to explain this difference: ꢀeff is generally
different from a volume-averaged value because screening
charge, which depends on the shape and size of inclusions,
develops at inclusion-matrix boundaries and sets apart the local
and the macroscopic applied fields.14 The minimum and
maximum deviations of ꢀeff from the volume-averaged value
are specified by the Wiener bounds:14
Figure 4. Kinetics of pore widening by phosphoric acid etching. The
angle shift and the decrease in the film dielectric constant with time
are highly linear except for an initial transient. Dashed curves show
Fresnel calculations with dielectric values at equally spaced intervals.
The corresponding pore widening is from 41 to 58 nm. Filled circles,
measurement before pore widening; open circles, after pore widening.
shows both the kinetic and R vs θ measurements. As expected,
the coupling angles were shifted to lower values after pore
widening, corresponding to a decrease in the ꢀeff of the PAA
film because of a decrease in the volume fraction of alumina:
ꢀalumina > ꢀPBS. Although ꢀalumina is unknown for the present
anodization conditions, it is expected to range between 2.2 (PAA
prepared by Hornyak et al.1) and 3.1 (fully dense film33). Fresnel
fits give ꢀz-start ) 2.36 and ꢀz-final ) 2.17. From measurements
of SEM micrographs, the pore diameter increased from ∼41 to
58 nm, and the pore volume fraction from 13% to 26%. Fresnel
fits also gave a film thickness change from 1150 to 1130 nm,
which is comparable to the amount of pore widening. However,
the m ) 1 (p-pol) shift is much more sensitive to a change in
ꢀeff than to film thickness, and we do not employ the effects of
the film thickness change for sensing purposes.
ꢀmin ) faꢀa + (1 - fa) ꢀb
1/ꢀmax ) fa/ꢀa + (1 - fa)/ꢀb
(1)
(2)
>
The inset of Figure 4 also shows that ∆θm)1 is linear with
the duration of pore widening, except for an initial transient.
Moreover, calculations of θm)1 for equally spaced values of
ꢀeff and film thickness between ꢀz-start and ꢀz-final show that
∆θm)1 varies linearly with ∆ꢀz (R2 ) 0.9999 for a linear fit.)
where fa is the volume fraction of component a, and ꢀmin
ꢀmax for 0 < fa < 1. ꢀmin is equal to the volume averaged value
and is applicable when the local and macroscopic fields are
identical everywhere. Such is the case, for example, if all internal
boundaries in a laminar composite are parallel to the electric
field. In contrast, ꢀmax refers to the situation when the electric
field has to penetrate the laminae oriented normal to the field
direction. In the PAA waveguide, ꢀx and ꢀy describe the dielectric
responses when an electric field is incident with the plane of
the film and, essentially, the field penetrates normal to the
cylindrical pore surfaces column after column, similar to the
situation for ꢀmax. On the other hand, ꢀz can be compared to
ꢀmin because ꢀz describes the response to an electric field normal
to the film surface and incident with the straight pores. Thus
ꢀmin ∼ ꢀz > ꢀmax ∼ ꢀx, ꢀy.
With respect to the sensitivity of the PAA sensor, we observe
θstart - θend ) 6.29°, while the change in ꢀz is 0.19, which gives
a resolution of 0.03∆ꢀz/deg, or a smallest experimentally
detectable change of ∆ꢀz ) 2 × 10-4, given the 0.005°
resolution in the instrumentation. In terms of physical dimen-
sions, with a measured change of 17 nm in pore diameter, we
are able to observe changes in average pore diameter of 2.7
nm/deg, which indicates a subangstrom detection of ∼0.01 nm.
Angle tracking has a virtually unlimited dynamic range, as long
as the waveguide mode being tracked is excited. For increased
sensitivity but reduced dynamic range, the reflectivity at fixed
angles where the R vs θ scan has a high gradient (near the
waveguide coupling angles) can be tracked instead. Here the
slope in R vs θ is approximately linear and shifts in the coupling
angle produce parallel translations of the curve. The change in
reflectivity at this fixed angle thus displays a response propor-
tional to the coupling angle shift but multiplied by the slope
within the linear region. The practical limit to sensitivity here
is the inherent stability of the laser intensity, and the root-mean-
2. Monitoring Pore Widening. Following the previous
experiment that left the PAA waveguide immersed in PBS (pH
7.4), we introduced 0.1 M phosphoric acid into the flow cell to
initiate pore widening and observed the shift in the m ) 1 (p-
pol) coupling angle. After 25 min, we rinsed the flow cell with
PBS pH 7.4 and measured the coupling angles again. Figure 4

10816 J. Phys. Chem. B, Vol. 108, No. 30, 2004
Lau et al.
square value of the intensity amplitude is easily <0.1% in
reflectivity. Given the sharpness of the waveguide coupling dips,
this would give a sensitivity in dielectric constant ∆ꢀz ≈ 10-6
and a corresponding increase in the sensitivity in thickness
changes. In comparison, an integrated Mach Zehnder interfer-
ometer demonstrated detection of a bulk dielectric constant
change of ∆ꢀ ) 9 × 10-6 and a layer thickness change of 0.02
nm, which is already at least an order of magnitude better than
obtainable with surface plasmon spectroscopy.34 Thus the
present configuration is capable of sensitivities at least com-
parable with integrated optics (IO) devices and surpasses that
of surface plasmon spectroscopy (SPS). Naturally, the sensitivity
of an optical sensor depends on the optical contrast between
the analyte and the background material.
Figure 5. Kinetics of repeated BSA adsorption and desorption. pH
values shown refer to those used during adsorption. Note that maximum
adsorption is always at the IEP of BSA at pH 5.0. Complete desorption
of BSA from alumina is not possible and an apparent baseline for the
regenerated alumina surface is established after the first desorption.
3. BSA Adsorption, Desorption, and Readsorption. BSA
is a protein of approximately ellipsoidal dimensions 12 × 3 ×
3 nm3.24,35 It physisorbs onto various surfaces through a range
of mechanisms that include electrostatic interactions,25,35 hy-
drogen bonding,26,35 and multipoint interaction with hydrophobic
residues.35 The isoelectric point (IEP) of BSA is around pH 5,
at which adsorption reaches a maximum,21,24,25 regardless of
the charge state of the adsorption surface.21 (Alumina is an
amphoteric oxide and its surface charge is dependent on pH in
aqueous solution. IEP for PAA membranes has been measured
to be pH ∼836 and agrees well with pH ∼9 for R-alumina
particles.21) Thus the alumina surface for our adsorption
experiments from pH 3.8 to 7.4 is always positively charged.
Monolayer coverage calculated for BSA varies from 0.4 to 2.4
to 8.5 mg/m2, depending on whether there is complete spreading
of BSA, or whether its globular conformation is intact and its
long axis is perpendicular or parallel to the surface.24 Multilayer
adsorption has not been observed in the literature nor in our
preliminary experiments (data not shown). Experimentally, the
maximum amount of BSA adsorbed on alumina particle surfaces
at BSA’s IEP ranges from 1.6 mg/m2 for protein concentrations
>17 µM24 to 4.7 mg/m2 for protein concentrations >30 µM.21
Using a BSA concentration of 50 µM in the present experiment,
we expect the maximum amount adsorbed at IEP to be within
that range. Desorption is achieved when BSA covered surfaces
are exposed to pH conditions far away from the IEP. In terms
of kinetics, both adsorption and desorption follow two stages
that can be described by two first-order rate equations.21-23,30
The initial stages dominate for the first ∼10 min. The second
stage for adsorption is a steady state that continues until
equilibrium is reached, while that for desorption is very slow
and takes 20 h to reach equilibrium, with residue amounts of
BSA still remaining on the surface.
reflected as a shift of coupling angles to higher values. As
discussed in the last section, there is excellent empirical linearity
between ∆θm)1 and ∆ꢀeff. We have also shown that ꢀz can be
compared to ꢀmin in eq 1 and ∆θm)1 is therefore linearly related
to ꢀa ) ꢀpore. The more protein is adsorbed on the pore surface,
the higher the ꢀpore is, therefore ∆θm)1 corresponds to the amount
of BSA adsorbed. Thus Figure 5 shows that BSA adsorption
proceeds at a decreasing rate. Given that multilayer BSA adsorp-
tion is not expected, we also deduce that ∆θm)1 is proportional
to the volume-averaged thickness of BSA adsorbed. Similarly,
desorption is also shown to proceed at a decreasing rate. These
observations are in accordance with the reported adsorption and
desorption behavior of BSA interacting with surfaces without
confined dimensions and of varying composition.21,24-26
After the first adsorption at pH 5, most of this protein layer
was desorbed and subsequent readsorption at the same pH
shifted the coupling angle (amount of BSA adsorbed) back to
nearly the same level. Figure 6a shows the R vs θ scans obtained
on the blank film and after the first readsorption at pH 5. Using
eq 1, we estimated the amount of BSA adsorbed. In the film
system before adsorption, the dielectric constant in the pores,
ꢀpore ) ꢀPBS, and the pore volume fraction (fpore) were measured
from SEM micrographs. However, the dielectric constant for
alumina anodized with our conditions, ꢀalumina, is unknown. After
adsorption, the pores were coated with BSA and ꢀpore increased
to an unknown value. Therefore, we obtain ꢀz from R vs θ scans
before and after adsorption and set up two simultaneous
equations of eq 1 to solve for both ꢀalumina and ꢀpore. Once ꢀpore
after adsorption is solved for, using known values of ꢀPBS
)
1.774 and ꢀprotein ) 2.1, the volume fraction of protein in the
pores is estimated by applying eq 1 to the protein-PBS system
within the pores. The effective thickness of BSA deposited on
the pore surfaces (hBSA) is then obtained by using this volume
fraction and the pore diameter measured from SEM micrographs
for this sample (∼31 nm). A value of ꢀalumina ) 2.5 was obtained
from this analysis and is very reasonable for porous anodized
alumina.1,33 The calculated value of hBSA is shown as the right-
hand scale bar in Figure 5. The maximum amount of BSA
adsorbed is approximately 7 nm thick and corresponds to a layer
of BSA oriented in mixed flat and upright positions. Estimating
from the ellipsoidal dimensions of BSA and calculated full
monolayer coverage, the maximum hBSA observed is ∼4.9 mg/
m2, which is consistent with values reported in the literature
for the 50 µM BSA solution used.
To verify the behavior of the interaction of BSA with the
PAA surface, adsorption was carried out over a range of pH
values centered around BSA’s IEP. Due to the high sensitivity
of the sensor, we carried out a series of adsorption-desorption
cycles at different pH values with the same PAA film to directly
compare the adsorbed amount without any possible external
influences such as differences in pore sizes or PAA film
thickness between samples.
Figure 5 shows the kinetics of one such series of experiments
measured by angle tracking of ∆θm)1 (p-pol). Adsorption steps
were 90 min long while desorption steps lasted 60 min. The
pH values shown refer to the adsorption condition while all
desorption steps were carried out at pH 8.5. In the first cycle,
BSA adsorbed at pH 5 formed a layer of material with ꢀprotein
) 2.1 (refractive index ) 1.4537) on the pore and film surfaces.
This displaced the volume originally occupied by PBS at ꢀPBS
) 1.774 and ꢀeff of the PAA waveguide increased, which is
The second desorption shown in Figure 5 did not proceed
long enough to take the regenerated surface back to the baseline
level indicated by the first desorption. However, the third, fourth,

In Situ Sensing with Nanoporous Anodic Alumina
J. Phys. Chem. B, Vol. 108, No. 30, 2004 10817
convex surfaces, which suggests that (bio-) chemically important
experiments may be carried out with this PAA sensor.
Parallel shifts in coupling angles shown in Figure 6a indicate
that a change in the effective dielectric constant was caused by
BSA adsorptionsBSA was adsorbed into and throughout the
pores. However, could BSA deposition/agglomeration on top
of the PAA film account for the angle shifts? First of all, the
dielectric constants of both pure and BSA loaded buffers are
1.774 for the low concentrations considered here. Thus only
ꢀeff is changed if BSA adsorbed onto pore surfaces. As discussed
previously, a change in ꢀeff produces parallel shifts in coupling
angles for different waveguide modes. In contrast, if BSA
adsorbed or accumulated on top of the PAA film, the coupling
angles for higher order modes are shifted much more than the
lower order modes: Figure 6b shows Fresnel calculations for a
blank PAA film filled with PBS and the same film with a 50
nm protein layer added on top, and demonstrates that appreciable
shifts in θm)1 can only be effected by an improbably thick
protein layer (50 nm in Figure 6b, compared with ellipsoidal
dimensions of 12 × 3 × 3 nm3 for BSA, which does not form
multilayers). Thus the parallel mode angle shifts in Figure 6a
confirm BSA adsorption onto pore surfaces. Alternatively, could
the shifts be produced by concentrated BSA adsorption down
only part of the pores? Again, the best fit between the data and
calculations is given by assuming the most straightforward case
of adsorption through the length of the pores. In addition, to
explain the angle shifts by partial filling of the pores would
impose tight restrictions on the distribution of such partial
deposition along the length of the pores and is considered
unlikely. Thus we conclude that we have demonstrated BSA
adsorption on PAA pore surfaces and high sensitivity detection
of the protein at the monolayer and submonolayer level.
Figure 6. (a) Waveguide mode patterns of blank film (filled circles)
and film with BSA readsorbed at pH 5.0 (open circles). BSA adsorption
is characterized by parallel shifts in mode coupling angles. (b)
Comparison of Fresnel calculations for the same PAA film if a 50-nm
protein layer were deposited aboVe the film instead of onto the internal
pore surfaces. The larger shifts for higher order modes distinguish this
external deposition from BSA adsorption into the porous film shown
in part a.
IV. Conclusions
We have adapted porous anodic alumina (PAA) as a highly
sensitive detector of surface reactions and processes by applying
it as a planar optical waveguide. Through a simple pore
widening experiment, we illustrated how the new sensor can
simultaneously assist in nanoscale fabrication processes and
monitor the same processes with angstrom level sensitivity. This
is also the first time that pore widening has been tracked in
situ. In another experimentsBSA adsorption, desorption, and
readsorptionswe demonstrated kinetic tracking of (bio-) mo-
lecular processes, and confirmed that BSA adsorbed onto the
porous anodic alumina in a manner analogous to its physisorp-
tion on other free surfaces in terms of both kinetics and response
to pH environment. High sensitivity is realized by the very large
internal surface area of the nanoporous structure, which ampli-
fies the amount of material deposited on the film, and by the
minute changes needed to shift waveguide coupling conditions.
Although we used prism coupling for guiding light into the PAA
film, grating or end-fire coupling and channel waveguide
configurations can also be envisioned.
The sensitivity to thickness changes is on the subangstrom
level and corresponds to ∆ꢀ ) 10-4 in the effective dielectric
constant of the PAA film. A simple switch in the experimental
setup to observing angle shift through reflected intensity change
would increase the sensitivity to ∆ꢀ ) 10-6. Further increases
in sensitivity may be achieved by fine-tuning the PAA fabrica-
tion conditions to improve film uniformity and to increase the
pore density (leading to an increase in internal surface area for
reactions to take place).
Figure 7. ∆θm)1 and thickness of BSA adsorbed of a second sample
(pore diameter ) 28 nm) plotted as a function of pH. Maximum
adsorption at pH 5.0 (IEP of BSA) is characteristic of the protein’s
physisorption on a range of surfaces, including alumina.
and fifth desorptions all did. We also note that there is a drop
in ∆θ between the first and last readsorptions at pH 5.0.
Subtleties in the kinetics of BSA adsorption and desorption are
beyond the scope of the present study, and we take the difference
in minimum angles reached as a safe margin of uncertainty for
∆θ measured. Even when assuming such large uncertainties, it
is seen that maximum BSA adsorption always occurred around
the IEP and decreases with pH deviating from the IEP. To
confirm this result, a second set of experiments on a different
film (pore diameter ∼28 nm) was carried out with longer
desorption cycles such that the regenerated baseline established
after the first adsorption-desorption cycle was always reached.
Figure 7 summarizes these data and reproduces the pattern of
BSA adsorbed vs pH plots seen in the literature on flat and
The examples presented show that anodic alumina films can
be employed as a highly sensitive device to monitor processes

10818 J. Phys. Chem. B, Vol. 108, No. 30, 2004
Lau et al.
(12) Li, F.; Zhang, L.; Metzger, R. M. Chem. Mater. 1998, 10, 2410-
occurring within the alumina pores in situ. These processes may
be related to the use of anodic alumina in nanotechnology as a
template or scaffold, or they may be related to other surface
reactions and processes. Indeed the present experiments only
minimally illustrate the full potential of the PAA sensor. One
important omission is in functionalization of the alumina surface
through silane-coupled molecules for specific molecular interac-
tion and detection. Analogous to the physisorption of a BSA
monolayer, growth of a self-assembled monolayer (SAM) can
be monitored as an average layer thickness change. Reactions
that are specific toward alumina surfaces may also be of interest,
and target molecules tagged with fluorophores may be employed
concurrently with fluorescence microscopy for multitarget, high-
resolution detection.
2480.
(13) Baselt, D. R.; Lee, G. U.; Hansen, K. M.; Chrisey, L. A.; Colton,
R. J. Proc. Inst. Electr. Electron. Eng. 1997, 85, 672-680.
(14) Aspnes, D. E. Thin Solid Films 1982, 89, 249-262.
(15) Granqvist, C. G.; Hunderi, O. Phys. ReV. B 1978, 18, 2897-2905.
(16) Foss, C. A., Jr.; Hornyak, G. L.; Stocket, J. A.; Martin, C. R. J.
Phys. Chem. 1994, 98, 2963-2971.
(17) Saito, M.; Shibasaki, M.; Nakamura, S.; Miyagi, M. Opt. Lett. 1994,
19, 710-712.
(18) Knoll, W. Annu. ReV. Phys. Chem. 1998, 49, 569-638.
(19) Elam, J. W.; Routkevitch, D.; Mardilovich, P. P.; George, S. M.
Chem. Mater. 2003, 15, 3507-3517.
(20) Liang, Z. J.; Susha, A. S.; Yu, A. M.; Caruso, F. AdV. Mater. 2003,
15, 1849-1853.
(21) Fukuzaki, S.; Urano, H.; Nagata, K. J. Ferment. Bioeng. 1996, 81,
163-167.
(22) Urano, H.; Fukuzaki, S. J. Colloid Interface Sci. 2002, 252, 284-
Acknowledgment. We thank Bernhard Menges at the Max
Planck Institute for Polymer Research, Mainz, and Xiaodi Su
at the Institute of Materials Engineering and Research, Sin-
gapore, for their excellent input concerning Fresnel modeling
of anisotropic films and protein properties, respectively. Part
of this material is based upon work supported by the Science
and Engineering Research Council, Singapore, under grants
R-152-000-037-303 (A*Star) and R-152-000-037-112 (MOE).
289.
(23) Sarkar, D.; Chattoraj, D. K. Ind. J. Biochem. Biophys. 1992, 29,
135-142.
(24) Hajra, S.; Chattoraj, D. K. Ind. J. Biochem. Biophys. 1991, 28, 114-
123.
(25) Elgersma, A. V.; Zsom, R. L. J.; Lyklema, J.; Norde, W. Colloids
Surf. 1992, 65, 17-28.
(26) Silin, V.; Weetall, H.; Vanderah, D. J. J. Colloid Interface Sci.
1997, 185, 94-103.
(27) Lawall, R.; Knoll, W. J. Appl. Phys. 1994, 76, 5764-5768.
(28) Hickel, W.; Duda, G.; Jurich, M.; Kroehl, T.; Rochford, K.;
Stegeman, G. I.; Swalen, J. D.; Wegner, G.; Knoll, W. Langmuir 1990, 6,
1403-1407.
(29) Mardilovich, P. P.; Govyadinov, A. N.; Mukhurov, N. I.; Rzhevskii,
A. M.; Paterson, R. J. Membr. Sci. 1995, 98, 131-142.
(30) Sarkar, D.; Chattoraj, D. K. J. Colloid Interface Sci. 1996, 178,
606-613.
(31) Pouliot, M.; Pouliot, Y.; Britten, M.; Ross, N. J. Membr. Sci. 1994,
95, 125-134.
(32) Mardilovich, P. P.; Govyadinov, A. N.; Mazurenko, N. I.; Paterson,
R. J. Membr. Sci. 1995, 98, 143-155.
References and Notes
(1) Hornyak, G. L.; Patrissi, C. J.; Martin, C. R. J. Phys. Chem. B
1997, 101, 1548-1555.
(2) Hoyer, P. AdV. Mater. 1996, 8, 857-.
(3) Sander, M. S.; Tan, L. S. AdV. Funct. Mater. 2003, 13, 393-397.
(4) Martin, C. R. Chem. Mater. 1996, 8, 1739-1746.
(5) Black, M. R.; Padi, M.; Cronin, S. B.; Lin, Y. M.; Rabin, O.;
McClure, T.; Dresselhaus, G.; Hagelstein, P. L.; Dresselhaus, M. S. Appl.
Phys. Lett. 2000, 77, 4142-4144.
(6) Proux-Delrouyre, V.; Elie, C.; Laval, J.-M.; Moiroux, J.; Bourdillon,
C. Langmuir 2002, 18, 3263-3272.
(7) Hennesthal, C.; Steinem, C. J. Am. Chem. Soc. 2000, 122, 8085-
8086.
(8) Routkevitch, D.; Govyadinov, A. N.; Mardilovich, P. P. Micro-
Electro-Mech. Syst. (MEMS) 2000, 2, 39-44.
(9) Masuda, H.; Satoh, M. Jpn. J. Appl. Phys. 2 1996, 35, L126-L129.
(10) Masuda, H.; Yamada, H.; Satoh, M.; Asoh, H.; Nakao, M.;
Tamamura, T. Appl. Phys. Lett. 1997, 71, 2770-2772.
(11) O’Sullivan, J. P.; Wood, G. C. Proc. R. Soc. London Ser. A 1970,
317, 511-543.
(33) Handbook of Optical Constants of Solids; Palik, E. D., Ed.;
Academic Press: San Diego, CA, 1998; Vol. 3.
(34) Weisser, M.; Tovar, G.; Mittler-Neher, S.; Knoll, W.; Brosinger,
F.; Freimuth, H.; Lacher, M.; Ehrfeld, W. Biosens. Bioelectron. 1999, 14,
405-411.
(35) Urano, H.; Fukuzaki, S. J. Biosci. Bioeng. 2000, 90, 105-111.
(36) Winkler, B. H.; Baltus, R. E. J. Membr. Sci. 2003, 226, 75-84.
(37) Spinke, J.; Liley, M.; Schmitt, J.; Guder, H. J.; Angermaier, L.;
Knoll, W. J. Chem. Phys. 1993, 99, 7012-7019.