Design of a NDIR gas sensor with two non-symmetric Fabry-Perot
absorber-structures working as IR-emitter and IR-detector
Johann Mayrw?ger*a, Wolfgang Reichl b, Christian Krutzler c, Bernhard Jakoby a
a Institute for Microelectronics and Microsensors, Johannes Kepler University Linz,
Altenberger Str. 69, A-4040 Linz, Austria
b E+E Elektronik, Langwiesen 7, 4209 Engerwitzdorf, Austria
c Integrate
d Microsystems Austria (IMA), Viktor Kaplan Str. 2/1, A-2700 Wiener Neustadt, Austria
ABSTRACT
Every gas (e.g. CO2) absorbs IR-radiation at individual gas specific IR-wavelengths. Non-dispersive infrared (NDIR) gas
sensors exploit this property for gas monitoring. Such sensors are used in various applications, e.g. for control of air
quality in office buildings or cars. This is a big market for low cost sensors. A NDIR sensor consists basically of three
components: an IR-emitter, a chamber containing the sample gas, and an IR-detector with a filter for the observed
wavelength. Commercially available systems use broadband IR-emitters (e.g.: micro-lamps) in combination with
thermopile or pyroelectric detectors fabricated with a narrowband gas-specific IR-filter, e.g., an interference filter. We
devised a concept for a simple and cost-effective NDIR-gas sensor based on two non-symmetric Fabry-Perot absorber-
structures as IR-emitter and as IR-detector where no additional interference filter is needed. The presented sensor
combines thin layer technology with optical sensing techniques. The system was first analyzed using ray tracing models
based on a Monte Carlo method in order to model the response function of the system’s sample chamber. For our results,
the sample gas is CO2 where the major absorption is centered around 4.26μm.
Keywords: G as monitoring, infrared radiation, bolometer
1.INTRODUCTION
Many gas molecules absorb electromagnetic radiation in the infrared region. The specific absorption spectrum can be
used to identify defined substances like CO2. In Figure 1 the major absorption peaks for some gases in the wavelength
section between 1μm and 8μm are displayed. The basic functionality of a NDIR gas sensor is very simple [1]: a (usually
broadband) IR-source emits infrared radiation, the sample-gas in the sensor cell absorbs radiation at its characteristic
wavelengths, and the IR-detector at the end of the optical path measures the remaining intensity of the radiation. The
decrease in the wavelength dependent infrared radiation caused by gas absorption follows the Lambert-Beer law [1]:
l c
I???
I
e
=α
?
(1)
This equation provides the remaining intensity I (with initial intensity I0) after passing through some absorbing media
characterized by an absorption coefficient α and concentration c, where the total length of the propagation path through
the medium is l.
To establish NDIR sensors in the fields of low budget mass marked applications, their overall costs have to be reduced. It
is easy to see from (1) that the achievable measurement resolution is predetermined by the optical length and therefore
linked to the geometry of the optical path, so the degrees of freedom in modifying the geometry are limited.
An interesting option for lowering the overall costs is the use of cheaper infrared components. We proposed and
investigated a low-cost concept for a non-dispersive infrared (NDIR) gas sensor consisting of two non-symmetric Fabry-
Perot absorber-structures (FPAS) as IR-emitter and as IR-detector. For the functionality of such a detector structure,
three layers are essential: a top metal mirror, a dielectric layer and a metal mirror at the bottom (see Figure 2). In the thin
*johann.mayrwoeger@jku.at; phone +43 732 2468 6263; fax +43 732 2468 6252; www.ime.jku.at
Optical Sensing and Detection, edited by Francis Berghmans, Anna Grazia Mignani, Chris A. van Hoof,
Proc. of SPIE Vol. 7726, 77260J · ? 2010 SPIE · CCC code: 0277-786X/10/$18 · doi: 10.1117/12.853516
top metal mirror the actual IR-absorption takes place. It is known (e.g., from [2]), that for optimum absorption and emission, respectively, the sheet resistance of the top metal mirror has to match the vacuum-impedance:
Ω==3770
00εμZ (2) The thickness of the dielectric layer defines the selectivity for a specific wavelength (the thickness has to be equal to (2n +1) λd /4 where n corresponds to the order of the absorption peak and λd is the wavelength in the dielectric layer) [3]. The metal mirror at the bottom acts as reflector for infrared radiation. According to Kirchhoff's law of thermal radiation, the characteristics for the IR-emitter are the same as for the IR-absorber.
Figure 1: Major gas absorption peaks in the wavelength region between 1μm and 8μm.
Figure 2: Sketch of a non-symmetric Fabry-Perot structure (dimensions not to scale). The thickness of the dielectric
layer is determined by the infrared absorption wavelength λd (wavelength in the dielectric layer) required for the
intended gas detection, where n is an integer number which represents the order of the selected absorption peak
associated with the Fabry-Perot structure and α defines the angle to the normal incidence to the surface.
An elementary point of our NDIR sensor concept is that the dielectric layers of the two FPAS have different thicknesses (IR-emitter: 3 λd /4, IR-absorber: 5 λd /4), a sketch of the sensor concept is depicted in Figure 3. The absorption maximum of the entire system, which is supposed to be equal to the selected absorption peak of the analyzed gas (e.g. CO 2: vacuum wavelength 4.26μm), corresponds to the 3 λd /4 and 5 λd /4 resonance peaks of the emitter and absorber (detector), respectively (see also below). In that manner, the spurious side-lobes of emitter and detector characteristics are located at different wavelengths such that they are attenuated in the overall characteristics. For the latter also the distribution of the emitted wavelength according to the temperature-depending blackbody-radiation and the emissivity has to be considered. Using this setup, the cross-sensitivity to the absorption of other gases (mainly water vapor but also CO and N 2O, laughing gas) is minimized. We furthermore present theoretical models (incl. ray tracing simulations) for the entire system performance.
Figure 3: Sketch of the presented NDIR gas sensor concept with its basic building blocks: IR-source fabricated as
3 λd/
4 structure, optical path containing the absorbing gas confined by infrared reflective walls, and IR-detector
fabricated as 5 λd/4 structure.
https://www.doczj.com/doc/1f17842173.html,YER DESIGN OF THE INFRARED ACTIVE COMPONENTS
The chosen materials for the two different FPAS are in both cases gold for the bottom mirror (about 120nm thick), germanium for the dielectric layer (thickness about 800nm for the IR-source with a 3 λd/4 structure and about 1330nm for the IR-detector with a 5 λd/4 structure, respectively), and titanium for the top metal mirror (about 15nm thick). The motivations for choosing this materials and the calculation of there thicknesses are described in [4] in more detail.
2.1Design of IR-emitter and IR-absorber
Even though the principle assembling of our IR-emitter and IR-absorber are identical, in their individual designs one has to pay attention to their different purposes. The IR-emitter needs an additional heating layer below the bottom metal mirror to achieve the temperature necessary for emitting the required IR-wavelength (the discussion of the design of the heating layer is beyond the scope of this paper). To obtain a better measurement resolution, the IR-absorber is designed as a semiconductor bolometer. The reason for this approach is the high (negative) temperature coefficient of the semiconductor germanium. When absorbing IR-radiation, the whole absorber will be heated, which changes the electrical resistance of all device layers. Since the change in resistance of the dielectric layer is the most distinctive, the lateral direction of the bottom metal mirror is interrupted so that the measurement current is forced to flow through the dielectric germanium layer (sketches of the two different designs are depicted in Figure 4). A detailed description of
semiconductor bolometer can be found in [5].
Figure 4: Sketches of the different designs for IR-emitting and IR-absorbing structure based on the principle design of Figure 2. The upper figure sketches the current-heated IR-emitter with its additional heating layer. The lower
figure shows the IR-absorber, which is working as semiconductor bolometer.
3.SIMULATION OF THE SPECTRAL CHARACTERISTICS
3.1Simulating the emitting and the absorbing structure
The center absorption wavelength associated with a Fabry-Perot structure is influenced by the incidence angle of the radiation onto the device surface. Therefore, it is necessary to take the influence of this parameter into account. To simulate this behavior, we applied the enhanced matrix approach [6]. The obtained absorption (or emission) from this method are depicted in Figure 5 for the 3 λd/4 structure and in Figure 6 for the 5 λd/4 structure.
Figure 5: Simulated absorption (or emission, according to Kirchhoff's law) of a 3 λd/4 structure similar to Figure 2. The
absorption depends on the IR-wavelength and the incidence angle of the IR-radiation.
Figure 6: Simulated absorption (or emission, according to Kirchhoff's law) of a 5 λd/4 structure similar to Figure 2. The
absorption depends on the IR-wavelength and the incidence angle of the IR-radiation.
3.2Spectral characteristics of the total system
For calculating the resulting wavelength-depending spectral IR-absorption of the entire system, three parameters have to be considered. The first is the blackbody radiation at a defined temperature (we use 600K as emitting temperature) [7]. For the latter two parameters, the results of the previous section 3.1, i.e. the wavelength- and angle-dependent emission and absorption, are required. In combining these parameters we can find the resulting transfer function, which gives the total IR-absorption of the whole system. The starting parameters and their corresponding dependence on the wavelength are illustrated in Figure 7. In this picture the area below the curves of the resulting total IR-absorption is proportional to
the achieved detector signal in the case of a non absorbing media where for this simple illustration only normal incidence on the IR-detector has been assumed. When taking oblique incidence into account, transfer functions for each indivudial ray have to be established using the characteristics shown in Figure 5 and Figure 6. The total transfer function is then obtained by summing up the contributions of all rays (ray tracing method, see also below).
Figure 7: The calculated resulting wavelength-depending absorption characteristics for the combination a 3 λd/4 structure as IR-source and a 5 λd/4 structure as IR-absorber is given by the product of the blackbody radiation
times the emission characteristics associated with the 3 λd/4 structure times the absorption characteristics of the
5 λd/4 structure. It is important to note that in this simple approach, only normal incidence to the surfaces of the
structures is assumed. To also take into account the change in emission and absorption characteristics due to
oblique incidence, the ray tracing method can be employed.
In Figure 8 the total IR-absorption spectrum (obtained with the approximate method discussed above) is compared to the absorption spectra of some gases in the wavelength region between 2μm and 8μm. This picture illustrate the basic spectral sensitivity of the setup shows at a major peak around 4.26μm, corresponding to the most important absorption line of CO2. The picture also shows that the cross-sensitivity to the absorption of other gases mainly water vapor but also CO and N2O (laughing gas) are damped without using an additional filter. Note that for many applications, water vapor is the main interfering substance as it can be present in significant concentrations. The present design efficiently suppresses the major absorption areas associated with water vapor, i.e. the regions around 2.7μm and 6μm.
Figure 8: Comparison of the resulting IR-absorption characteristics (see also Figure 7) to the characteristic infrared
absorption spectra of some gases.
4.MODELING DES TOTAL SENSOR SYSTEM WITH RAY TRACING
To demonstrate the applicability of our sensor concept for CO2 measurements, a sample cell visualized in Figure 9 has been simulated. For those simulations we used a ray tracing models based on a Monte Carlo method. The method (we used the ray tracing code ZEMAX [8]) is illustrated in [9] in more detail.
Figure 9: 3D-model of the simulated sensor geometry with the IR-emitter on the left, the optical path containing the sample gas, and the IR-detector on the right hand side. In this case we use a simple circular pipe as optical path. Its
length is 55mm and its diameter is 10mm. The geometrical dimensions of the IR-emitter and the IR-detector are
4mm×0.5mm on a 0.3mm substrate. Several emitted rays are shown as an example.
As the outcome of our ray tracing simulation, we obtain the normalized CO2 response for the geometry given in Figure 10. This resulting response function shows a shallower characteristic than commercially available NDIR gas sensors but is suitable for low cost mass marked applications. Commercially available sensors are commonly equipped with broadband micro-lamps as IR-source in combination with thermopile or pyroelectric detectors, which are combined with a narrowband gas-specific IR-filter, as IR-detector.
Figure 10: Simulated CO2 response of a gas sensor system cell using the assembling sketched in Figure 9.
5.FABRICATED NON-SYMMETRIC FABRY-PEROT ABSORBER STRUCTURES FOR
EMITTING AND ABSORBING INFRARED RADIATION
Both structures, IR-emitter and IR-absorber, are designed according to the simple structure shown in Figure 2, however, they feature different thicknesses of the dielectric layer. A scanning-electron-microscope picture of such a structure is shown in Figure 11.
Figure 11: Profile of a non-symmetric Fabry-Perot absorber structure designed according to Figure 2.
To control the IR-emission, it is necessary to recognize the temperature of the emitting area. For this reason we designed our IR-emitter (and also our IR-detector) together with two temperature sensors on the same substrate (microscope pictures of the fabricated structures: Figure 12 and Figure 13, respectively).
Figure 12: Top view of a IR-source (microscope picture). On the top and on the bottom two temperature sensors are placed, the emitting area is between the two sensors. The heater is designed as a meander structure.
Figure 13: Top view of a IR-absorber (microscope picture). On the top and on the bottom two temperature sensors are placed, the absorbing area designed as a semiconductor bolometer is between the two sensors.
The characterization of these devices is not yet finished at this time but is underway.
6.CONCLUSION
We demonstrated a new sensor concept for low cost NDIR gas sensor system. Instead of using broadband IR-emitters (e.g.: micro-lamps) in combination with thermopile or pyroelectric detectors combined with narrowband gas-specific IR-filters, we employ two non-symmetric Fabry-Perot absorber-structures as IR-emitter and as IR-detector where no additional interference filter is needed. We apply ray tracing simulations of the entire sensor system to demonstrate its feasibility for the measurement of CO2 as target gas and illustrate the applicability for mass marked applications.
ACKNOWLEDGEMENT
This work is supported by the Federal Ministry of Economics and Labor, the Federal Government of Lower Austria (Industrial Center of Competence program K ind, project KOFIRS 03301), and by the Austrian COMET (Competence Centers for Excellent Technologies) Program.
REFERENCES
[1]Staab, J., [Industrielle Gasanalyse], Oldenbourg Industrieverlag, Munich & Vienna (1994).
[2]Bauer, S., Bauer-Gogonea, S. and Ploss, P., …The physics of pyroelectic infrared devices”, Appl. Phys. B65, pp. 544-
551 (1992)
[3]Kasberger, J., Rauter, P. and Jakoby, B., …Wavelength Selectivity of a Thermal IR–Absorber as Part of a Fully
Integrated IR–Absorption Sensor”, Proc. IEEE Sensors Lecce, pp.996-999 (2008)
[4]Mayrw?ger, J., Reichl, W., Hauer, P., Krutzler, C. and Jakoby, B., …CO2 monitoring using a simple Fabry-Perot-
based germanium bolometer”, Sensors and Actuators B: Chemical, article in press
[5]Liddiard, K. C., …Thin-film resistance bolometer IR detectors-II”, Infrared Phys. 26 (1), pp. 43-49 (1986)
[6]Gaylord, T. K., …Stable implementation of the rigorous coupled-wave analysis for surface-relief gratings: enhanced
transmittance matrix approach”, J. Opt. Soc. Am. A 12, pp. 1077-1086 (1995)
[7]Hecht, E., [Optics], 4. ed. Addison-Wesley, San Francisco, 2009
[8]https://www.doczj.com/doc/1f17842173.html,
[9]Mayrw?ger, J., Hauer, P., Reichl, W., Schw?diauer, R., Krutzler, C., and Jakoby, B., …Modeling of Infrared Gas
Sensors Using a Ray Tracing Approach”, IEEE Sensors Journal, accepted for publication