Quantitative analysis of the sensitivity of

porous silicon optical biosensors

Cite as: Appl. Phys. Lett. 88, 163108 (2006); https://doi.org/10.1063/1.2196069
Submitted: 04 November 2005 . Accepted: 07 March 2006 . Published Online: 19 April 2006

Huimin Ouyang, Christopher C. Striemer, and Philippe M. Fauchet

ARTICLES YOU MAY BE INTERESTED IN

The structural and luminescence properties of porous silicon

Journal of Applied Physics 82, 909 (1997); https://doi.org/10.1063/1.366536

Porous silicon microcavities as optical chemical sensors

Applied Physics Letters 76, 2523 (2000); https://doi.org/10.1063/1.126396

Vapor sensing using the optical properties of porous silicon Bragg mirrors
Journal of Applied Physics 86, 1781 (1999); https://doi.org/10.1063/1.370968

Appl. Phys. Lett. 88, 163108 (2006); https://doi.org/10.1063/1.2196069 88, 163108

© 2006 American Institute of Physics.

 APPLIED PHYSICS LETTERS 88, 163108 共2006兲

Quantitative analysis of the sensitivity of porous silicon optical biosensors

Huimin Ouyang,a兲 Christopher C. Striemer, and Philippe M. Fauchet
Department of Electrical and Computer Engineering and the Center for Future Health,
University of Rochester, Rochester, New York 14620
共Received 4 November 2005; accepted 7 March 2006; published online 19 April 2006兲
The optical response of porous silicon affinity biosensors depends strongly on their
nanomorphology because the sensing species does not completely fill the pores but is instead
attached to the pore walls. The performance and sensitivity of porous silicon microcavity biosensors
are calculated as a function of pore size using a simplified effective medium approximation.
Excellent agreement is obtained between the model predictions and the experimental results after
binding of aminopropyltriethoxysilane and glutaraldehyde in mesoporous and macroporous silicon
microcavities. Detection of layers thinner than 0.01 nm should be possible. © 2006 American
Institute of Physics. 关DOI: 10.1063/1.2196069兴

The large internal surface area 共⬃100 m2 / cm3兲 of po- dex of PSi and causes a redshift of the resonance dip
rous silicon 共PSi兲 makes it a good host material for the sen- 关Fig. 1共a兲兴. To quantitatively analyze ⌬npore 共the increase of
sitive detection of gases,1 liquids,2,3 and biological refractive index of the pores due to the binding of a thin
molecules.4,5 PSi-based label-free affinity optical biosensors layer of molecules on the pore wall兲 a simplified effective
with a variety of configurations such as single layer Fabry- medium approximation based on volume ratios is used:18
Pérot cavities,6 Bragg mirrors,7 rugate filters,8 and
microcavities4,9,10 have been experimentally demonstrated
for the detection of toxins,11 DNA,5 and proteins.6,12 The size
⌬npore = npore
after before
− npore =4 冉 t
D D
t2
冊
− 2 共nlayer − npore
before
兲

of molecules captured inside the pores can be controlled by tⰆD

tuning the PSi nanostructures 共porosity and pore size兲, which
can be achieved by varying the etching parameters.13 How-
ever, a quantitative understanding of how the performance
and sensitivity of PSi optical biosensors depend on the nano-
structures 共porosity and pore size兲 is still lacking.
The figure of merit describing the sensitivity of affinity
sensors is ⌬␭ / ⌬n, where ⌬␭ is the wavelength shift resulting
from the capture of biological or chemical molecules and ⌬n
is the change of the ambient or pore refractive index. Our
simulations show that for PSi affinity sensors with an aver-
age porosity of 75%, ⌬␭ / ⌬n ⬃ 550 nm at a detection wave-
length of 800 nm.14 When the refractive index inside the
pores is increased homogenously, ⌬␭ / ⌬n is the same for
sensors with different pore sizes but the same porosity. How-
ever, in biosensing applications, the sensing species are at-
tached only to the internal surfaces 共pore walls兲 instead of
completely filling the pores. In this case, the pore size be-
comes an important parameter, because for a PSi layer of a
given porosity, the internal surface area 共or the total number
of available binding sites兲 decreases as the pore size in-
creases. In this letter, we quantitatively characterize the per-
formance of PSi microcavity biosensors by modeling the
wavelength shift resulting from binding events in sensors
with different pore sizes. Experiments were performed to
demonstrate the influence of the pore size on the optical re-
sponse and to verify the theoretical predictions.
The reflectance spectrum of PSi microcavity, which is
characterized by a sharp resonance dip,15 was simulated by
the transfer matrix method.16 The effective refractive index
of a PSi layer is defined by its porosity and the refractive
indices of silicon and the pores.17 Binding of biological mol-
ecules on the pore wall increases the effective refractive in-
a兲
Electronic mail: ouyang@ece.rochester.edu
t
⬇ 4 共nlayer − npore
before
兲, 共1兲
D
where D is the diameter of the pores, t is the thickness of the
layer binding on the pore wall, nlayer is the refractive index of
before after
the binding layer, and npore and npore are the refractive indi-
ces of the pore before and after binding 关see Fig. 1共b兲 inset兴.
In our case, the pores are filled with air before binding, thus
before
npore = nair = 1. Using this equation we simulated the
redshift of the spectra for microcavities with fixed porosities
共75% for the high porosity layer and 65% for the low
porosity layer兲 but different pore diameters ranging from
20 to 180 nm. The original resonance wavelength of the mi-
crocavity is 800 nm. The redshift of the resonance wave-
length ⌬␭ due to binding on the pore walls is plotted in Fig.
1共b兲 as a function of the optical thickness of the coating
L = tnlayer and the pore diameter D.
For a 30 Å thick binding layer, a PSi microcavity with
20 nm pores produces a redshift of 78 nm, while a microcav-
ity with 180 nm pores gives a redshift of only 10 nm. Thus,
to optimize the sensitivity, the pore size should be as small as
possible while still allowing for easy infiltration of the bio-
logical material. If the minimum detectable wavelength shift
and minimum acceptable pore diameter are known, one can
calculate the minimum detectable coating thickness, i.e., the
sensitivity of the sensor. As shown in Fig. 1共c兲, for a detec-
tion system capable of resolving a wavelength shift of
0.1 nm, the minimum detectable coating optical thickness is
approximately 0.03 Å for a microcavity with 20 nm pores,
whereas for a microcavity with 80 nm pores, it is approxi-
mately 0.13 Å.
To experimentally verify the simulation, we used two
different types of PSi microcavities: mesoporous silicon mi-
crocavities with an average pore diameter of ⬃20 nm and
macroporous silicon microcavities with an average pore di-
ameter of ⬃120 nm.12 Figure 2 shows the morphology of the

0003-6951/2006/88共16兲/163108/3/$23.00 88, 163108-1 © 2006 American Institute of Physics

163108-2 Ouyang, Striemer, and Fauchet
FIG. 1. 共Color online兲 共a兲 Illustration of the microcavity reflectance spec-
trum redshift upon binding. 共b兲 Simulation of microcavity spectrum redshift
as a function of coating layer optical thickness 共L = tnlayer兲 for different pore
sizes. The inset illustrates the coating of a thin layer on the pore wall.
共c兲 Redshift of the spectra with subangstrom coating thickness.
mesoporous and macroporous silicon microcavities. Both
microcavities have layers with porosities of approximately
65% and 75%, corresponding to refractive indices of 1.60
and 1.33 in the infrared calculated by the Bruggeman effec-
tive medium approximation.
FIG. 2. 关共a兲 and 共b兲兴 Top view and cross sectional scanning electron micros-
copy 共SEM兲 images of a mesoporous silicon microcavity with 20 nm pore
size etched in 共100兲 p-type 共0.01 ⍀ cm兲 silicon wafers using a solution of
15% hydroflouric acid 共HF兲 in ethanol; 关共c兲 and 共d兲兴 Top view and cross
sectional SEM images of a macroporous silicon microcavity with 120 nm
pore size etched in 共100兲 n-type 共0.01 ⍀ cm兲 silicon wafers using a solution
containing 5.5% HF in water. These PSi microcavities consist of two Bragg
mirrors 共a periodic stack of layers with two different porosities and quarter-
wavelength optical thickness兲 and a defect layer 共half-wavelength optical
thickness兲.
Appl. Phys. Lett. 88, 163108 共2006兲
FIG. 3. 共a兲 Redshift increase as the concentration of APTES increases for
both mesoporous and macroporous silicon microcavities. The redshift satu-
rates when one monolayer of APTES is formed inside the pores. 共b兲 Redshift
increase due to the binding of a two-layer coating made of glutaraldehyde
and APTES. A fixed amount of glutaraldehyde was applied to sensors that
were treated with different APTES concentrations. The redshift saturates
when two layers of molecules completely coat the pores.
Aminopropyltriethoxysilane 共APTES兲 and glutaralde-
hyde are common coupling agents that promote adhesion
between oxidized silicon surfaces and organic molecules.19
Furthermore, APTES and glutaraldehyde are small molecules
that can form uniform thin layers on the internal surface of
PSi, which makes them good model systems to quantitatively
characterize the optical response of PSi biosensors. The mi-
crocavities were first thermally oxidized at 900 ° C to create
an oxide layer on the PSi internal surface to promote cova-
lent attachment of APTES. After oxidation, the resonant
wavelength of the microcavity is ⬃800 nm. Then the micro-
cavities were exposed to APTES solutions with different
concentrations 关APTES was diluted in a H2O:methanol 共1:1兲
mixture兴. Each sensor was exposed to the solutions for
20 min then rinsed with de-ianized 共DI兲 water and methanol
and dried with nitrogen. The silanized sensors were then
baked in an oven at 100 ° C for 10 min before the optical
measurement.
The reflectance spectra were taken using an Ocean Op-
tics HR 2000 spectrometer with a R200-7 reflection probe
共six illumination fibers around one read fiber兲. As shown in
Fig. 3共a兲, the redshift of the resonance wavelength increases
as the concentration of APTES increases for both mesopo-
rous and macroporous silicon microcavities. The curves satu-
rate when all the available binding sites are occupied, i.e.,
when a monolayer of APTES coats the pore walls.
After the silanization, the microcavities were exposed to
a 2.5% glutaraldehyde solution in 20 mM HEPES buffer
共pH = 7.4兲 for 30 min, then rinsed with DI water and dried
with nitrogen. Glutaraldehyde reacts with the amino groups
on the silanized surface and coats the internal surface of the
pores with another thin layer of molecules. It can be seen
from Fig. 3共b兲 that after the binding of glutaraldehyde to the
silanized surface, the maximum total redshift is twice that
163108-3 Ouyang, Striemer, and Fauchet
FIG. 4. Redshift of microcavities due to the presence of a thin coating layer
as a function of pore size and layer thickness. The solid curves are the
calculated redshifts as a function of the pore size using t = 8 and 15 Å thick
coating layers with nlayer = 1.46. The data points are the measured redshift for
mesoporous 共20 nm兲 and macroporous 共120 nm兲 microcavities after coating
of APTES and APTES+ glutaraldehyde. The error bars are from multiple
trials.
due to the binding of APTES alone shown in Fig. 3共a兲. This
confirms the specific binding between APTES and glutaral-
dehyde, and, in the saturation region, the formation of one
uniform APTES monolayer inside the sensors. Because ⌬␭ is
linearly related to t as shown in Fig. 1, it also suggests that
the thickness ratio of the APTES and glutaraldehyde layers is
⬃1 : 1. For the same coating thickness inside the pores, the
redshift is approximately six times larger with 20 nm meso-
pores than with 120 nm macropores, which is in a good
agreement with Eq. 共1兲.
To compare the experimental results and the simulation,
the optical thickness of APTES and glutaraldehyde was de-
termined by variable angle spectroscopic ellipsometry on
planar surfaces. Crystalline silicon wafers with a thin oxide
layer were first treated with 5% APTES in H2O and metha-
nol 共1:1兲 followed by 2.5% glutaraldehyde using the same
procedure as described above. We measured the refractive
index of APTES and glutaraldehyde to be 1.46± 0.06. The
physical thickness of one APTES monolayer was determined
to be 8 ± 1 Å, and the total thickness of the APTES
+ glutaraldehyde layer to be 15± 1 Å. These results confirm
the thickness ratio of APTES to glutaraldehyde to be ⬃1 : 1.
Using the refractive index and physical thickness of APTES
and glutaraldehyde measured by ellipsometry, the redshift of
the spectra was simulated using the effective media approxi-
mation described above. The simulated redshift as a function
of the pore size is plotted as the solid curves in Fig. 4. The
top curve corresponds to a coating layer with t = 15 Å, and
the bottom curve corresponds to a coating layer with
t = 8 Å. As shown in Fig. 4, the experimental results obtained
Appl. Phys. Lett. 88, 163108 共2006兲
using mesoporous and macroporous microcavities are in ex-
cellent agreement with the simulations.
In conclusion, PSi optical biosensors with different pore
sizes have been quantitatively characterized using a simpli-
fied effective medium approximation. The wavelength shifts
measured with microcavities after the binding of thin layers
of molecules with different thicknesses were in excellent
agreement with simulations. Simulations show that binding a
very small amount of biological targets, equivalent to a con-
tinuous layer less than 0.1 Å thick, can be detected with a
high Q 共⬎2000兲 microcavity and a measurement system ca-
pable of resolving a wavelength shift of 0.1 nm.
The authors would like to thank Dr. Benjamin L. Miller
and Dr. Lisa A. DeLouise of the University of Rochester for
helpful discussions, and acknowledge the financial support
from the National Science Foundation and the Infotonics
Center of Excellence. The authors thank the Laboratory for
Laser Energetics at the University of Rochester for providing
access to the spectroscopic ellipsometer.
1
A. Motohashi, M. Kawakami, H. Aoyagi, A. Kinoshita, and A. Satou,
J. Appl. Phys. 91, 2519 共2002兲.
2
S. L. Letant and M. J. Sailor, Adv. Mater. 共Weinheim, Ger.兲 13, 335
共2001兲.
3
Z. Gaburro, N. Daldosso, L. Pavesi, G. Faglia, C. Baratto, and G.
Sberveglieri, Appl. Phys. Lett. 78, 3744 共2001兲.
4
S. Chan, S. R. Horner, P. M. Fauchet, and B. L. Miller, J. Am. Chem. Soc.
123, 11797 共2001兲.
5
S. Chan, Y. Li, L. J. Rothberg, B. L. Miller, and P. M. Fauchet, Mater. Sci.
Eng., C 15, 277 共2001兲.
6
K.-P. S. Dancil, D. P. Greiner, and M. J. Sailor, J. Am. Chem. Soc. 121,
7925 共1999兲.
7
P. A. Snow, E. K. Squire, P. S. J. Russell, and L. T. Canham, J. Appl. Phys.
86, 1781 共1999兲.
8
F. Cunin, T. A. Schmedake, J. R. Link, Y. Y. Li, J. Koh, S. N. Bhatia, and
M. J. Sailor, Nat. Mater. 1, 39 共2002兲.
9
L. Pavesi, Riv. Nuovo Cimento 20, 1 共1997兲.
10
H. Ouyang, M. Christophersen, and P. M. Fauchet, Phys. Status Solidi A
202, 1396 共2005兲.
11
H. Sohn, S. Letant, M. J. Sailor, and W. C. Trogler, J. Am. Chem. Soc.
122, 5399 共2000兲.
12
H. Ouyang, M. Christophersen, R. Viard, B. L. Miller, and P. M. Fauchet,
Adv. Funct. Mater. 15, 1851 共2005兲.
13
L. Canham, Properties of Porous Silicon 共INSPEC, London, UK, 1997兲.
14
H. Ouyang and P. M. Fauchet, Proc. SPIE 6005, 600508-1 共2005兲.
15
S. Chan and P. M. Fauchet, Appl. Phys. Lett. 75, 274 共1999兲.
16
E. Hecht and A. Zajac, Optics 共Addison-Wesley, Menlo Park, CA, 1979兲.
17
W. Theiss, Surf. Sci. Rep. 29, 91 共1997兲.
18
The change in refractive index of the cylindrical pore after infiltration
of a uniform layer coating is estimated by ⌬npore = 关Vlayer / Vporenlayer + 共1
− Vlayer / Vpore兲npore
before
兴 − npore
before
and Vlayer / Vpore = ␲共D / 2兲2 − ␲共D / 2
− t兲2 / ␲共D / 2兲2 = 4共t / D − t2 / D2兲, where Vlayer is the volume of the coating
layer and Vpore is the volume of the pore.
19
G. Hermanson, Biocojugate Techniques 共Academic, New York, 1996兲.