Content uploaded by Peter Petrik
Author content
All content in this area was uploaded by Peter Petrik on Jan 05, 2015
Content may be subject to copyright.
RESEARCH ARTICLE
Copyright © 2010 American Scientific Publishers
All rights reserved
Printed in the United States of America
SENSOR LETTERS
Vol. 8, 1–6, 2010
In Situ Spectroscopic Ellipsometry Study of
Protein Immobilization on Different Substrates
Using Liquid Cells
Andrea Németh1∗, Péter Kozma12, Tímea Hülber1, Sándor Kurunczi1, Róbert Horváth1,
Péter Petrik1, Adél Muskotál2, Ferenc Vonderviszt12, Csaba H˝
os3, Miklós Fried1,
József Gyulai1, and István Bársony1
1Hungarian Academy of Sciences, Research Institute for Technical Physics and Materials Science,
P.O. Box 49, Budapest H-1525, Hungary
2Department of Nanotechnology, Research Institute of Chemical and Process Engineering, Faculty of
Information Technology, University of Pannonia, Veszprém, Hungary
3Department of Hydrodinamic Systems, Budapest University of Technology and Economics, Budapest 1111, Hungary
(Received: xx Xxxx xxxx. Accepted: xx Xxxx xxxx)
The influence of substrate materials on protein adsorption was studied by spectroscopic ellipsom-
etry (SE) and atomic force microscopy. For model proteins fibrinogen and flagellar filaments were
chosen and their kinetics of adsorption, surface coverage and adsorbed amount on virgin and chem-
ically activated SiO2and Ta2O5thin films were investigated. In case of flagellar filaments the SE
data were analyzed with an effective medium model that accounted for the vertical density distri-
bution of the adsorbed protein layer. Adsorption was measured in situ using flow cells with various
fluid volume. Compared to commercially available cells, a flow cell with significantly smaller volume
was constructed for cost-effective measurements. The development of the flow cell was supported
by finite element fluid dynamics calculations.
Keywords:
1. INTRODUCTION
Protein adsorption at liquid/solid interfaces is funda-
mental for several diverse areas of biotechnology, e.g.,
biosensing.1–4 In situ measurement of adsorption and
immobilization of proteins in liquid cells is a complex task
influenced by numerous parameters.5–8 Most of these fac-
tors (like temperature of the cell, purity of water, rate of
injection) are known to have little effect on the repeatabil-
ity of the experiment.5On the other hand, there are various
possible causes of uncertainty that need to be investigated
and as far as it is possible the experimental system needs
to be insulated from these undesired influences. Our pri-
mary aim is to increase our modeling and characterization
knowledge (thickness, surface coverage and kinetics) of
the adsorbing protein films. By monitoring in situ pro-
tein deposition we have more insight into the mecha-
nisms of protein layer formation. The adsorbed proteins
∗Corresponding author; E-mail: nemetha@mfa.kfki.hu
can be applied for instance as specific receptors for ligand
binding.34The first protein applied in our experiments
was fibrinogen (Fgn), a model protein that is compact but
large enough to perform optical measurements using SE.
Fgn was selected as one of the plasma proteins, since these
(HSA, Fgn, IgG, etc.) are well-characterized model and
test proteins.910 The second protein we studied was the
flagellar filament (FF). The bacterial flagella are tail-like
structures responsible for locomotion. Their extracellular
parts are the FFs. Every FF is composed of thousands of
protein subunits, called flagellin, by polymerization con-
stituting 5–20 m long helical filaments, the diameter of
which is ∼23 nm. By genetic engineering or directed evo-
lution, specific binding sites can be created in the central
parts of flagellin subunits which form the outer surface
of the filament. Flagellar filaments polymerized in vitro
from these receptor subunits are stable structures with very
high binding site density on their surface.11–13 Flagellin
(molecular weight 51.5 kDa) was gained from Salmonella
typhimurium wild type strain SJW1103 and purified as
Sensor Lett. 2010, Vol. 8, No. 5 1546-198X/2010/8/001/006 doi:10.1166/sl.2010.1338 1
RESEARCH ARTICLE
In Situ Spectroscopic Ellipsometry Study of Protein Immobilization on Different Substrates Using Liquid Cells Németh et al.
previously reported.14 Due to the procedure of producing
FFs the long filaments break up to shorter parts with a
length distribution of 300–1500 nm.
In this study, spectroscopic ellipsometry (SE) was
applied15–19 to follow the adsorption of proteins in situ in
liquid cells with various fluid capacities on activated and
non-activated tantalum pentoxide (Ta2O5and SiO2sub-
strates. Final surface mass densities and kinetics of deposi-
tion were determined and compared for the different cases.
After SE measurements the samples were dried and ana-
lyzed using atomic force microscopy (AFM).
2. EXPERIMENTAL DETAILS
2.1. Spectroscopic Ellipsometry (SE)
The possibility of performing in situ and non-destructive
measurements is one of the major advantages of ellipsom-
etry in life sciences.161720 Ellipsometry detects changes
in polarization of an electromagnetic wave reflected from
the sample. A Woollam M2000DI multichannel rotating
compensator ellipsometer was used with different flow-
through cells for in situ investigations. The fixed angle of
light incidence was 75for both cells, a value close to the
Brewster’s angle of silicon.
Typical measured and fitted (CompleteEASE, Woollam
Co.) (, spectra are shown in Figure 1 as a func-
tion of wavelength (). In our optical models, additionally
to the refractive index Refs. [20, 21], effective medium
approximation (EMA) and Cauchy dispersion function
were used.22
2.2. Sample Preparation
Thermally oxidized silicon substrates were used, and
onto half of the sample pieces a ∼190 nm thick Ta2O5
Fig. 1. Typical measured and fitted ellipsometric spectra over the wave-
length range of 360–1000 nm in the case of silicon substrate with thermal
oxide and Ta2O5film. The spectra were recorded during in situ SE mea-
surement in the small flow cell during PBS flow.
film was evaporated. Before the measurements all wafers
were immersed for 10 minutes in “Piranha” solution (1:3
mixture of 30% hydrogen-peroxide and concentrated sul-
furic acid), rinsed with Milli-Q water and dried by nitro-
gen blow. In order to compare activated and non-activated
surfaces only half of the sample pieces with SiO2and
Ta2O5were activated with 3-aminopropyl-triethoxy-silane
(APTES) and glutaraldehyde crosslinker (GA).52324 Due
to the silanization, the surface changed from hydrophilic
to hydrophobic.25–27
2.3. Flow Cell Design with Ultra-Small Capacity
The capacity of the commercially available flow cell of
Woollam Inc. is 5 ml. In order to minimize the amount
of the solutions needed, we designed a smaller liquid cell
with the capacity of 0.2 ml. Reducing the sample volume
has two major advantages. First, the time duration of the
transient when switching from pure buffer to protein solu-
tion can be reduced, allowing a more detailed characteri-
zation of the initial phase of the layer build up. Secondly
the measurements are much more cost-effective (smaller
amount of proteins and chemicals are needed), which is
important for several expensive proteins and reagents. The
properties of both flow cells are compared in Table I.
During the design process several important aspects had
to be considered,17 like the constructible lowest capacity,
applicable flow rate, and flow cell’s UV-grade window
properties. The effect of flow rate on the interaction
between surface and macromolecules had to be analyzed
to determine whether the flow is laminar or turbulent
within the liquid cell. On the account of the complexity
of the flow cell geometry Computational Fluid Dynamics
(CFD) by the commercial software ANSYS CFX 11.0 was
applied. Numerical simulations were performed to study
water flow dynamics in the small liquid cell at 25 C. The
analysis was the following:
(1) computational model of the flow field was constructed,
(2) the flow field was divided into discrete cells,
(3) the boundary conditions were defined and the equa-
tions of flow dynamics were solved for the finite elements,
(4) the result of the evaluation was visualized by stream-
lines (Fig. 2).
Table I. Comparison of the large and small flow cells.
Attributes Large flow cell Small flow cell
Developer Woollam Home made
Capacity 5 ml 0.2 ml
Angle of incidence 7575
Delta offset (window effect) Negligible Small
Flow rate 2.5 ml/min Below 0.5 ml/min
Solution amount for 150 ml 30 ml
1 hour (at least)
Windows diameter 12.5 mm 4 mm
Sample size (O-ring) 50 ×13 mm 27 ×7mm
2Sensor Letters 8, 1–6, 2010
RESEARCH ARTICLE
Németh et al. In Situ Spectroscopic Ellipsometry Study of Protein Immobilization on Different Substrates Using Liquid Cells
Fig. 2. Visualization of the flow field within the small flow cell rep-
resented by streamlines. The picture shows the results of dynamic flow
simulations in case of 0.5 ml/min volume flow. The liquid flows from the
right-hand side to the left.
2.4. Experiments in Flow Cells
Before any kind of treatments, all of the substrates were
measured by SE in their original state in the wavelength
range over 190–1700 nm. It is important to note that every
time the sample changed, a new ellipsometric spectrum
was taken and this spectrum was analyzed in order to
follow the evolution of the surface and investigate our
procedure. Birefringence of the cell windows can be com-
pensated by fitting the delta offset measuring a reference
sample in the cell without liquid.
Protein deposition was followed in situ by SE. The con-
centration of the dissolved FF and Fgn was 100 g/ml
in 20 mM phosphate buffered saline (PBS: pH 7.4) solu-
tion. A measurement consists of two steps: (1) the cell
is washed through by PBS, and the PBS base-line is
recorded, (2) the protein solution is injected, the protein
deposition proceeded until saturation. The flow rate was
below 0.5 ml/min and 2.5 ml/min in the case of the small
and large cells, respectively.
In order to compare the results, the deposition of the
model protein (Fgn) was injected in both flow cells. Fgn
adsorption181925 was investigated at first within the small
and large flow cells to make sure that the extremely low
capacity of the small cell doesn’t affect the mechanism
of the adsorption. After reproducing and comparing the
results of Fgn adsorption in both flow cells, we began to
investigate the FF adsorption in the small cell.
In the test period of the small cell the effect of the
flow rate on Fgn adsorption onto silica wafers was also
investigated. An extremely low (0.13 ml/min) and high
(5 ml/min) flow rate was chosen for the analysis of the
adsorption.
The most relevant information of the deposition of pro-
tein layers is the surface mass density () as a function
of time. is a robust measure (from the thickness and
refractive index at the wavelength of 632 nm) using de
Feijter’s equation eliminating the uncertainties of the deter-
mination of both layer thickness and refractive index.28
The kinetics of protein deposition was quantitatively char-
acterized by fitting exponential functions to the adsorption
curve.29
3. DISCUSSION
3.1. Optical Model
The (, spectra were fitted in the wavelength ranges of
360–1000 nm and 250–1000 nm for depositions on Ta2O5
and SiO2, respectively, due to the absorption of light in
water and in the Ta2O5layer. To construct the most proper
optical model for the multilayer sample structure refer-
ence refractive indices were applied from the literature for
the substrate and for the thermal oxide layer. For fitting
the Ta2O5layer the Cauchy dispersion function was used.
The surface roughness of the deposited Ta2O5film and the
APTES, GA layers were fitted commonly with a subse-
quent Cauchy function. The protein layers were modeled
with EMA (single in case of Fgn and mulitlayers for FF)
combining the refractive index of protein (Cauchy) and
ambient (the refractive index of PBS has been determined
by a measurement on a reference wafer in PBS, taking
into account the small dispersion of PBS using a Cauchy
fit). The schematic graph of a multilayer structure with the
typical thickness values is shown in Figure 3.
To find a suitable optical model for FF is more compli-
cated, because it does not form a well defined and dense
layer due to the size distribution of the filaments and their
multiple-binding to the substrate. In contrast to FF, Fgn
adsorbs in a nearly monolayer formation, thus a simpler
Fig. 3. Schematic graph of the measured multi-layer structure (typical
values of thicknesses are given in parentheses).
Sensor Letters 8, 1–6, 2010 3
RESEARCH ARTICLE
In Situ Spectroscopic Ellipsometry Study of Protein Immobilization on Different Substrates Using Liquid Cells Németh et al.
optical model can be constructed. Therefore Fgn was used
as a model protein to test the flow cells.
3.2. Numerical Results and Test of the Small
Flow Cell
At seven different values of volume flow, the flow field was
determined by the finite element calculations. The results
of these calculations indicated that the Reynolds-number
is below 1 (near the surface, below 1 m) in every investi-
gated case and the flow field is no longer symmetric above
the volume flow of 0.5 ml/min (mass flow: 8.3 mg/s). Near
the surface, along the adsorbed FFs the flow is laminar.
offset caused by birefringence of the UV-grade win-
dows of the Woollam flow cell was negligible. In case of
the smaller cell, offset was calibrated by a measurement
on a reference sample without liquid.
To compare the cells, Fgn adsorption onto thermally
oxidized silica substrates was investigated within the
smaller and the larger flow cells, simultaneously. The sur-
face mass density of the adsorbed protein layer was found
to be the same for both cases (0.27 g/cm2. Calculating
the surface mass density as a function of time the curves
were fitted by the following exponential function: =−A·
exp−t/ +0, where Ais the amplitude, tis the time,
and is the time constant. The value was also found to
be the same for both flow cells (1.5 min).
The effect of flow rate on protein adsorption was also
investigated in the small flow cell using silicon substrates
and Fgn. Comparing the values we have quantitative
information about the relation between the flow rate and
the kinetics: the higher the flow rate the faster the kinet-
ics (is smaller) and the lower the fit quality. The results
are represented numerically in Table II.). Note that fitting
the Fgn adsorption curves by double exponential func-
tion resulted in higher fit quality than the fitting of the
single exponential suggesting that probably two processes
(transport limited and adsorption limited) occur during
deposition.3031 In case of FF adsorption fitting by one
exponential function revealed good agreement showing
that the adsorption limitation is the dominant process. The
authors intend to perform more detailed investigations in
Table II. Comparison of the surface mass densities (, the exponential
time constants ( and the correlation coefficients for a fit using one
exponential function (R2, denoting the fit quality) for the flow rates. The
exponential functions were fitted in the same range of time. The values
were determined at the end of the fitted curves, and the saturation was
probably reached later.
[g/cm2][min] R2(fit quality for 1 exp.)
Flow rate Fgn FF Fgn FF Fgn FF
0.1 ml/min 0.33 — 1.83 — 0.984 —
0.5 ml/min 0.26 — 1.50 — 0.983
5 ml/min 0.30 — 0.74 — 0.924 —
Fig. 4. Surface mass densities as a function of time for two protein
adsorptions, FF and Fgn. The circles and triangles mark the deposition
of FF and Fgn, respectively. The lines denote the results of the fitted
exponential functions. In case of Fgn double exponential and for FF one
exponential fit was needed for a good agreement.
the future to reveal the theoretical background of the Fgn
and FF adsorption kinetics.
3.3. Adsorption onto Activated and Non-Activated
Surfaces
The testing procedure was followed by Fgn and FF adsorp-
tion experiments on activated and non-activated SiO2and
Ta2O5substrates. The adsorbed protein surface mass den-
sity was found to be consistently higher and the satura-
tion is reached later in the case of FF (around 1 g/cm2,
more than one hour) than in the case of Fgn (around
0.35 g/cm2, 20 minutes). A typical measurement can be
seen in Figure 4. The activation has propulsive effect on
Fig. 5. Surface mass densities during FF depositions onto activated and
non-activated silicon substrates, with thermal oxide film. The circles and
the triangles mark the increase of surface mass density as a function of
time and the lines denote the fitted exponential functions.
4Sensor Letters 8, 1–6, 2010
RESEARCH ARTICLE
Németh et al. In Situ Spectroscopic Ellipsometry Study of Protein Immobilization on Different Substrates Using Liquid Cells
Table III. Comparison of the surface mass densities (, the exponen-
tial time constants ( and the correlation coefficients for a fit using one
exponential function (R2, denoting the fit quality) for the materials. The
exponential functions were fitted in the same range of time. The values
were determined at the end of the fitted curves, and the saturation was
probably reached later.
[g/cm2][min] R2(fit quality for 1 exp.)
Material Fgn FF Fgn FF Fgn FF
Activated
SiO20.34 0.39 0.83 32 0.902 0.998
Ta2O50.27 0.58 1.62 101 0.940 0.996
Not activated
SiO20.29 0.36a1.58 128 0.991 0.999
Ta2O50.45 0.26a4.18 145 0.972 0.998
aThe optical signal (also the adsorbed amount) increased constantly.
the FF adsorption kinetics. By Fitting the exponential func-
tion mentioned above to (t, the value of Aobtained is
more than two times higher and the value obtained is
more than four times lower in the case of activation. In
other words a faster FF deposition and higher adsorbed
amount was observed (Fig. 5). In case of Fgn this rela-
tion is not so clear. The results for different materials are
compared in Table III.
The increase of phase shift ( as a function of time
during protein adsorption from base-line to saturation was
around 7 degrees for SiO2. In contrast, the phase shift for
Ta2O5increased by 2 degrees from base-line to satura-
tion (at 600 nm). This means that better sensitivity can be
reached in case of SiO2(mostly because using this layer
structure the angle of incidence of 75used in the cell is
closer to the pseudo Brewster angle). The effect of activa-
tion on the adsorbed protein amount seems to be more sig-
nificant for Ta2O5than for SiO2, but further investigations
are needed for better understanding. Without activation the
Fig. 6. AFM image about a non-activated silicon substrate covered by
FF. The protein was adsorbed in the small flow cell. After SE mea-
surement the substrate was dried with nitrogen blow and the image was
recorded by AFM in air.
optical signal increased constantly even after 100 minutes,
while in case of activation a plateau has been reached after
80 minutes.
To reveal the morphology, the samples were measured
in air using AFM after adsorption (Fig. 6). The structure
of the FF layer under buffer is like grass, after drying with
nitrogen blow, the strains lie down. This is in agreement
with previous finding for shorter filaments.32
4. CONCLUSIONS
Well-controlled in situ measurements were performed in
a flow cell of extremely low capacity. The dynamic flow
simulations for the small flow cell revealed that the flow is
laminar for the applied flow rates and showed that above
0.5 ml/min the profile of the flow field is no longer sym-
metric. The increase of surface mass density of FF was
analyzed during SE measurements using multilayer effec-
tive medium models. Good repeatability of the adsorbed
amount of Fgn was found by subsequent measurements.
The effect of the surface activation for immobilization and
the substrate material was also analyzed, and a higher
adsorbed amount and a faster saturation were observed
in the case of activation. The adsorbed FF layer can be
divided into sublayers with a higher adsorbed amount near
to the surface (100 nm, around 0.5 g/cm2and another
thicker one with lower mass density (around 500 nm,
around 0.3 g/cm2.
Acknowledgments: Support from the Hungarian Sci-
entific Research Fund (OTKA Nos. K61725, K81842,
PD 73084, and T046238) is greatly acknowledged.
This work was also supported by the European
Commission–Research Infrastructure Action, under the
FP6-Program “Structuring the European Research Area,”
through the Integrated Infrastructure Initiative “Euro-
pean Integrated Activity of Excellence and Network-
ing for Nano and Micro-Electronics Analysis,” contract
no. 026134(RII3)ANNA and by the OPTIBIO 231055
FP7-Program.
References and Notes
1. M. A. Cooper, Nat. Rev. Drug Discov. 1, 515 (2002).
2. J. J. Ramsden, J. Mol. Recognit. 10, 109 (1997).
3. M. Malmsten, Protein Architecture: Interfacing Molecular Assem-
blies and Immobilization Biotechnology, edited by Y. Lvov and
H. Möhwald, Marcel Dekker, New York (2000), pp. 1–23.
4. G. J. Szöll˝
osi, I. Derényi, and J. Vörös, Physica A 343, 359 (2004).
5. P. Kozma, N. Nagy, S. Kurunczi, P. Petrik, A. Hámori, A. Muskotál,
F. Vonderviszt, M. Fried, and I. Bársony, Phys. Stat. Sol. 5, 1427
(2008).
6. L. G. Castro, D. W. Thompson, T. Tiwald, E. M. Berberov, and J. A.
Woollam, Surf. Sci. 601, 1795 (2007).
7. O. Joshi, H. J. Lee, J. McGuire, P. Finneran, and K. E. Bird, Colloids
and Surfaces B: Biointerfaces 50, 26 (2006).
8. O. Santos, T. Nylander, M. Paulsson, and C. Tragardh, J. Food Eng.
74, 468 (2006).
Sensor Letters 8, 1–6, 2010 5
RESEARCH ARTICLE
In Situ Spectroscopic Ellipsometry Study of Protein Immobilization on Different Substrates Using Liquid Cells Németh et al.
9. J. L. Orgtega-Vinuessa, P. Tengval, and I. Lundstrom, Thin Solid
Films 324, 257 (1998).
10. J. S. Kavanaugh, W. F. Moo-Penn, and A. Arnone, Biochemistry
32, 2509 (1993).
11. F. Vonderviszt, H. Uedaira, S. Kidokoro, and K. Namba, J. Mol.
Biol. 214, 97 (1990).
12. A. Sebestyén, B. Végh, A. Szekrényes, S. Kurunczi, and
F. Vonderviszt, Biokémia 30, 4 (2006).
13. K. Namba and F. Vonderviszt, Quart Rev. Biophys. 30, 1 (1997).
14. I. Yamashita, F. Vonderviszt, T. Noguchi, and K. J. Namba, Mol.
Biol. 217, 293 (1991).
15. M. Fried, T. Lohner, and P. Petrik, Handbook of Surfaces and Inter-
faces of Materials, Academic Press, San Diego (2001), Chap. 6,
Vol. 4, p. 335.
16. H. Arwin, Thin Solid Films 377–378, 48 (2000).
17. H. Arwin, Ellipsometry in life sciences, Handbook of Ellipsometry,
edited by H. G. Tompkins and E. A. Irene, William Andrew Publ.,
Norwich, NY (2005).
18. S. Lousinian and S. Logothetidis, Thin Solid Films 516, 8002 (2008).
19. S. Lousinian and S. Logothetidis, Proceedings of International Con-
ference on Nanomedicine, Porto Carras Grand Resort, Chalkidiki,
Greece, September (2007).
20. H. Arwin, Thin Solid Films 313–314, 764 (1998).
21. J. Voros, Biophys. J. 87 (2004).
22. H. Arwin, Appl. Spectrosc. 40, 313 (1986).
23. S.-O. Molin, H. Nygeren, and L. Dolonius, The Journal of Histo-
chemistry and Chytochemistry 26, 412 (1978).
24. E. T. Vandenberg, L. Bertilsson, B. Liedberg, K. Uvdal,
R. Erlandsson, H. Elwing, and I. Lundström, Colloid Interface Sci.
147, 103 (1991).
25. M. Malmsten, Colloids Surf., B: Biointerfaces 3, 297 (1995).
26. R. J. Marsh, R. A. L. Jones, and M. Sferrazza, Colloids Surf., B:
Biointerfaces 23, 31 (2002).
27. Z.-H. Wang and G. Jin, Journal of Immunological Methods 285, 237
(2004).
28. J. A. de Feijter, J. Benjamins, and F. A. Veer, Biopolymers 17, 1759
(1978).
29. A. K. Bajpai, J. Mater. Sci.: Mater. Med. 19, 343 (2008).
30. M. A. Brusatori, Protein adsorption kinetics under an applied electric
field: An optical waveguide lightmode spectroscopy study, Disserta-
tion, Graduate School of Wayne State University, Detroit, Michigan
(2001).
31. Y. Tie, C. Calonder, and P. R. Van Tassel, J. Colloid Interface Sci.
268, 1 (2003).
32. S. Kurunczi, R. Horvath, Y.-P. Yeh, A. Muskotál, A. Sebestyén,
F. Vonderviszt, J. J. Ramsden, J. Chem. Phys. 130, 011101 (2009).
6Sensor Letters 8, 1–6, 2010
