You are on page 1of 5

Optimized broadband wide-angle absorber structures

Juan J. Monzn,1 Teresa Yonte,1 Luis L. Snchez-Soto,1,* and ngel Felipe2


1

Departamento de ptica, Facultad de Fsica, Universidad Complutense, 28040 Madrid, Spain


2

Departamento de Estadstica e Investigacin Operativa I, Facultad de Matemticas,


Universidad Complutense, 28040 Madrid, Spain
*Corresponding author: lsanchez@fis.ucm.es
Received 22 July 2008; revised 2 October 2008; accepted 30 October 2008;
posted 31 October 2008 (Doc. ID 99212); published 21 November 2008

By introducing the notion of wavelength- and angle-averaged absorptance, we assess in a systematic way
the possibility of achieving wide-angle absorptance in a spectral range. We determine the optimum thicknesses for which this broadband wide-angle absorption occurs for a representative example of infrared
detector. 2008 Optical Society of America
OCIS codes:
350.2450, 040.3060, 230.4170.

1. Introduction

In thermal radiation detectors, the incident


radiation absorbed by a structure leads to a temperature change that is measured by means of a variety
of physical effects. Typical examples are the
temperature-dependent electric resistivity in bolometers, gas pressure in Golay cells, the Seebeck
effect in thermopiles, and the pyroelectric effect in
pyroelectric sensors [1]. There are many advantages
about thermal infrared (IR) sensors, such as no
need for cooling equipment, low cost, and easy
integration [2].
The absorber structure should have high efficiency
and low thermal mass. Several solutions have been
suggested to meet these requirements [3], although
the so-called quarter-wavelength structures are
largely employed nowadays [412]. Indeed, for
room-temperature operation they offer the best performance because of their sensitivity and frequency
response. A number of variants have been proposed
in the literature, each with its own advantages, but
most of them can be roughly schematized as in Fig. 1.
A metal film is placed at the top, then the transmitted radiation goes through a quarter-wave dielectric layer and after that, at the bottom, is reflected by
a metal film and destructively interferes, giving zero
0003-6935/08/346366-05$15.00/0
2008 Optical Society of America
6366

APPLIED OPTICS / Vol. 47, No. 34 / 1 December 2008

reflection and 100% absorption (at least, in the


ideal case).
To the best of our knowledge, all these structures
have been designed at normal incidence and any variation of the absorption with the angle of incidence is
safely ignored. This strongly contrasts with the flurry of papers dealing with the possibility of achieving
perfect reflection at any polarization, at any incidence angle, and over a range of wavelengths, which
has been termed omnidirectional reflection [13]. It is
clear that the analogous notion of omnidirectional
absorption is unattainable because the behavior at
grazing angles limits seriously the performance of
any absorptive device. However, we suggest relaxing
this requirement and looking for approximate
conditions, which may be called broadband wideangle absorption.
To attack this problem we first need to define a
magnitude that catches all the appropriate directional aspects. Going beyond our previous work
[14,15], we thus propose to average the absorptance
over a range of incidence angles and over the wavelengths in the spectral interval of interest. In
consequence, we content ourselves when this wavelength- and angle-averaged absorptance takes a
value close enough to unit. The results of this paper
confirm that, when the arrangement is properly
optimized, this magnitude can indeed attain values
as high as 0.96 for such a wide aperture as 60.

complex since the material may be absorbing), is [16]


1
Mj
Tj
Here the layer
coefficients are

Fig. 1. (Color online) Scheme of the thin-film IR thermal absorber


considered in the text. The materials and thicknesses of the layers
are indicated in Table 1, and the semi-infinite substrate is made of
silicon.

In a real detector a number of essential physical


parameters, such as heat capacity, the value of the
electrical conductivity of the film to be used as an
electrode, and lateral heat conductivity, should be
taken into account. However, here we ignore these
problems and focus only on the optical performance.
2. Integral Estimator of Averaged Absorptance

To be specific, we take as a guiding example the absorption filter presented in Ref. [12], which fits into
the scheme of Fig. 1. This device is quite flexible and
potentially compatible with pixel-by-pixel tuning of
current uncooled focal plane array. It is designed
to couple the incident light into its own top mirror,
which is achieved by making this mirror both lightly
reflecting and absorbing, while the bottom mirror is
highly reflecting with negligible absorption.
The IR absorbing material is deposited on the top
of the upper plate, which is made of germanium (Ge).
Chromium (Cr) was chosen as the absorber because
its optical properties for IR absorption. The bottom
mirror is a modified distributed Bragg reflector,
made of a two-period structure of zinc sulfide
(ZnS) and germanium. These materials combine
the advantages of desirable optical properties and
manufacturing simplicity. A layer of gold (Au) with
an adhesion layer of chromium is included to enhance the mirror reflectivity.
The top mirror is supported by rigid metallic
tethers and is electrostatically actuated by applying
a dc voltage that allows one to change the air gap of
the cavity. For additional technical details and
further explanations about the manufacturing of
the structure, the reader is referred to the original
proposal [12].
To characterize the optical response we employ the
transfer matrix, which for a stack of m plane parallel
layers deposited on a substrate can be computed as
(we assume that the ambient, labeled by 0, is air with
N 0 1)
M

m
Y

Mj Msub ;

j1

where the transfer matrix for the single layer j, of


thickness dj and refractive index N j (in general,

Rj


Rj
T 2j R2j :

1
Rj

reflection

r0j 1 expi2j 
;
1 r20j expi2j

and

2
transmission

1 r20j expij

Tj

1 r20j expi2j

3
In these equations, j 2=N j dj cos j is the layer
phase thickness and j is the angle of
refraction, which is determined by Snells law. We
have taken the incident radiation as purely monochromatic, with a wavelength in vacuo . In addition,
r0j are the Fresnel reflection coefficient for the interface 0j, which, as it is well known, depend on the
basic polarizations (p or s) and on the angle of incidence. For the substrate, which we take transparent
and semi-infinite (in fact, in our scheme is silicon),
we have
Msub

1
rsub

tsub


rsub
;
1

where rsub and tsub are now Fresnel coefficients for


the ambientsubstrate interface.
The transfer matrix (1) for the system is also of the
general form (2). Hence, once it is computed, one immediately gets the overall reflection and transmission coefficients as R M 21 =M 11 and T 1=M 11 ,
M ij being the matrix elements of M. In terms of them
we obtain the corresponding reflectance and transmittance as
R jRj2 ;

N sub cos sub 2


jTj ;
cos

being the angle of incidence from the ambient.


From here we get the overall absorptance as
A1RT:

All these magnitudes depend on the basic polarizations. To avoid separate discussions for each one of
them, henceforth we work with A Ap As =2.
As mentioned before, we average the absorptance
over the wavelengths in the interval and
over the incidence angles in the range
(for simplicity, throughout all this work we take
0):
1

A;


Ad d;

and take this as an appropriate figure of merit to


assess the broadband wide-angle performance of
the structure. Apart from the parameters and
1 December 2008 / Vol. 47, No. 34 / APPLIED OPTICS

6367

, it is a function of the layer thicknesses (for fixed



materials). Note also that the behavior of A;

differs drastically from the absorptance A.


As pointed out in Refs. [17,18], when the cavity and
the substrate are transparent, it is possible to ascertain the absorptance in each mirror (denoted as Atop
and Abottom , respectively) as the difference between
the total incident and emergent fluxes. The resulting
expressions are


cos
1 jRbottom j2
Atop 1 R
T;
N sub cos sub
jT bottom j2


cos
1 jRbottom j2
Abottom
T T;
N sub cos sub
jT bottom j2

where Rbottom and T bottom are the reflection and


transmission coefficients for the bottom mirror sandwiched between the cavity and the substrate and can
be computed multiplying the associated layer matrices as in Eq. (1). Note that, because the ambient
and the cavity are made of identical media, the angles of incidence and refraction in the cavity are
the same.
Obviously, adding Eqs. (8) we get in a natural way
A Atop Abottom ;

and the same holds for the corresponding wavelength- and angle-averaged magnitudes.
3. Results and Discussion

As the spectral range of interest we have considered


7 m and 10 m. The optical properties of
all the materials have been taken from Ref. [19]
and interpolated, when necessary, in spectral steps
of 0:1 m to perform the numerical integration over
. The choice is mainly motivated because above
10 m we have not managed to find reliable values
for gold. In this spectral range the refractive indices
of Ge, ZnS, and Si (substrate) are fairly constant,
with variations almost unnoticeable around the central values of nGe 4:006, nZnS 2:231, and
nSi 3:422. For Cr and Au the corresponding optical
constants are plotted in Fig. 2.
For each we have employed an efficient quasiNewton algorithm (implemented in the Numerical
Algorithms Group Ltd. (NAG) numerical library)
in order to find the thicknesses optimizing the target
(7). The resulting values are listed in Table 1, as well
as the corresponding averaged absorptances.
Most of the merit functions commonly used in
standard multilayer design are encompassed by
the general form [20]
Z
MF

WD2 d;

10

where D stands for the difference between the target and the present values of the quantity of interest
and W is some spectral weight factor that permits
6368

APPLIED OPTICS / Vol. 47, No. 34 / 1 December 2008

Fig. 2. (Color online) Optical constants n (in red/dark gray) and


(in yellow/light gray) for Au (circles) and Cr (triangles) in the
spectral range considered in this work.

the D term to receive different amounts of attention


in the optimization. By a quantity of interest we
mean any property of the system (such as the absorptance) that can be evaluated by matrix methods once
the construction parameters have been given.
Because of the presence of the term D2 , one should
expect that the results obtained from merit functions
like (10) and from our proposal (7) may deviate
significantly.
A quick look at Table 1 immediately reveals a number of interesting features. First of all, one can check
that, for the bottom mirror, beyond certain values of
the thicknesses, significant variations (10% and even
more) in the Cr glue layer and the Au reflection layer
do not influence the results. Therefore, we have
decided to fix these thicknesses to 0:03 m and
0:5 m, respectively, as in the original proposal.
The optimum Cr top-layer thickness turns out to
be fairly independent on the aperture angle .
On the contrary, the parameters of the Bragg mirror
strongly depend on : the double-period mirror,
which is optimum at normal incidence, gives results
worse than a single ZnS-Ge bilayer in most of the

Table 1. Optimum Thicknesses (m) and Associated Values of the


Averaged Absorptances for the System Sketched in Fig. 1 for
Different Values of the Aperture Angle

Material
Cr
Ge
Air
Ge
ZnS
Ge
ZnS
Au
Cr
 top
A
 bottom
A

A

30

45

60

75

85

0.0020
0.0691
1.8468
0.0000
0.1145
0.0000
0.0575
0.5000
0.0300
0.9864
0.0045
0.9909

0.0020
0.0807
1.5233
0.0000
0.2639
0.0831
0.0000
0.5000
0.0300
0.9765
0.0065
0.9830

0.0020
0.0942
1.1848
0.0000
0.3473
0.1207
0.0000
0.5000
0.0300
0.9540
0.0098
0.9638

0.0020
0.1071
0.9513
0.0000
0.3724
0.1417
0.0000
0.5000
0.0300
0.9101
0.0124
0.9225

0.0020
0.1132
0.8980
0.0000
0.3564
0.1528
0.0000
0.5000
0.0300
0.8570
0.0127
0.8697

cases, except at 30. At first sight, this behavior may


seem counterintuitive and can be assigned to the
highly nontrivial interference phenomena arising
in these transparent layers. In this respect, we em
phasize that the function A;
depends in a
highly nontrivial way on the aperture angle , so
it may come as no surprise that for different values
of we get different optimum thicknesses. How guarantee that they
ever, the final results for A
are truly optimum. Notice also that air gap of the
cavity cannot be considered now as quarter-wavelength, since we are working with a range of both
wavelengths and angles of incidence.
 decreases as increases, as it
The absorptance A
may be expected. This variation is small (especially
for moderate values of ), and even at such wide
apertures as 85 one finds remarkable high va Moreover, it is important to observe that
lues of A.
 top decreases with , while A
 bottom increases.
A
For wide apertures one cannot strictly assign the
whole absorption to the top mirror (although it
absorbs 98.5%, confirming the results of the initial
design in Ref. [12]).
To gain further insights into these behaviors, in
Fig. 3 we have plotted the absorptance A for the optimum thicknesses corresponding to 75 as a
function of the angle of incidence and the wavelength . At the bottom plane we have also included
the contours of the regions where this absorptance is
greater than 0.99 and 0.95. This optimized function
is quite flat and very close to unity, except for big incidence angles ( 60).
Finally, in Fig. 4 we have plotted five sections of
Fig. 3 for equidistant values of as a function of .
The absorptance at shorter wavelengths is worse
at low angles of incidence, while they become the better ones for high values of . The converse is true for
the longer wavelengths.

Fig. 3. (Color online) Absorptance A as a function of the wavelength (in the interval between 7 m and 10 m) and the angle
of incidence (between 0 and 75) for the optimum thicknesses in
Table 1, in the case of 75. At the bottom plane, we show
the contour plots corresponding to absorptances 0.99 and 0.95.

Fig. 4. (Color online) Plot of the absorptance A as a function of


the angle of incidence (between 0 and 75) for the same system
as in Fig. 3 and the five values of the wavelength indicated in
the inset.

To sum up in a few words, we have exploited the


notion of wavelength- and angle-averaged absorptance to explore in a systematic way the performance
of IR thermal absorbers. Needless to say, our
approach is general and can be applied to other
materials and other spectral regions.
We wish to thank one anonymous reviewer for his
helpful comments on the optimization process. This
work has been supported by the grant FIS200506714 of the Spanish Research Agency.
References
1. W. L. Wolfe and P. W. Kruse, Thermal detectors, in OSA
Handbook of Optics, 2nd ed. (McGraw-Hill, 1995), Vol. I,
Chap. 20.
2. D. Baselt, B. Fruhberger, E. Klaassen, S. Cemalovic,
C. L. Britton, S. V. Patel, T. E. Mlsna, D. McCorkle, and
B. Warmack, Design and performance of a microcantileverbased hydrogen sensor, Sens. Actuators B 88, 120131 (2003).
3. E. L. Dereniak and D. G. Crowe, Optical Radiation Detectors
(Wiley, 1984).
4. A. D. Parsons and D. J. Pedder, Thin-film infrared absorber
structures for advanced thermal detectors, J. Vac. Sci. Technol. A 6, 16861689 (1988).
5. A. Hadni and X. Gerbaux, Infrared and millimeter wave absorber structures for thermal detectors, Infrared Phys. 30,
465478 (1990).
6. S. Bauer, S. Bauer-Gogonea, and B. Ploss, The physics of
pyroelectric infrared devices, Appl. Phys. B 54, 544551
(1992).
7. M. C. Larson, B. Pezeshki, and J. S. Harris, Vertical coupledcavity microinterferometer on GaAs with deformablemembrane top mirror, IEEE Photonics Technol. Lett. 7,
382384 (1995).
8. S. R. Manalis, Two-dimensional micromechanical bimorph
arrays for detection of thermal radiation, Appl. Phys. Lett.
70, 33113313 (1997).
9. G. Wu, R. H. Datar, K. M. Hansen, T. Thundat, R. J. Cote, and
A. Majumda, Bioassay of prostate-specific antigen (PSA)
using microcantilevers, Nat. Biotechnol. 19, 856860 (2001).
10. J. Numus, H. Botmer, C. Kunzel, U. Vetter, A. Lambrecht, J.
Schumann, and F. Volklein, Thin film based thermoelectric
1 December 2008 / Vol. 47, No. 34 / APPLIED OPTICS

6369

11.

12.

13.

14.

energy conversion systems, Twenty First International


Conference on Thermoelectrics (IEEE, 2002), pp. 523527.
P. G. Datskos, N. V. Lavrik, and S. Rajic, Performance of uncooled microcantilever thermal detectors, Rev. Sci. Instrum.
75, 11341138 (2004).
Y. Wang, B. J. Potter, and J. J. Talghader, Coupled absorption
filters for thermal detectors, Opt. Lett. 31, 19451947
(2006).
A complete and updated list of publications can be found in J.
P. Dowling, Photonic and sonic bandgap bibliography, http://
baton.phys.lsu.edu/ jdowling/pbgbib.html.
A. G. Barriuso, J. J. Monzn, L. L. Snchez-Soto, and A. Felipe,
Comparing omnidirectional reflection from periodic and quasiperiodic photonic crystals, Opt. Express 13, 39133920
(2005).

6370

APPLIED OPTICS / Vol. 47, No. 34 / 1 December 2008

15. A. G. Barriuso, J. J. Monzn, L. L. Snchez-Soto, and A. Felipe,


Integral merit function for broadband omnidirectional
mirrors, Appl. Opt. 46, 29032906 (2007).
16. P. Yeh, Optical Waves in Layered Media (Wiley, 1988).
17. J. J. Monzn and L. L. Snchez-Soto, On the definition of
absorption for a FabryPerot interferometer, Pure Appl.
Opt. 1, 219222 (1992).
18. J. J. Monzn and L. L. Snchez-Soto, Optical performance of
absorber structures for thermal detectors, Appl. Opt. 33,
51375141 (1994).
19. Handbook of Optical Constants of Solids, E. Palik, ed.
(Academic, 1998).
20. J. A. Dobrowolski, F. C. Ho, A. Belkind, and V. A. Koss, Merit
functions for more effective thin film calculations, Appl. Opt.
28, 28242831 (1989).

You might also like