Professional Documents
Culture Documents
Early Fire Detection Are Hydrogen Sensors Able To Detect Pyrolysis of House Hold Materials
Early Fire Detection Are Hydrogen Sensors Able To Detect Pyrolysis of House Hold Materials
Early fire detection: Are hydrogen sensors able to detect pyrolysis of house MARK
hold materials?
⁎
Simone Krügera, , Marie-Claire Despinassea, Tina Raspea, Kai Nörthemannb, Werner Moritzb
a
Division 7.5 Technical properties of polymeric materials, Bundesanstalt für Materialforschung und –prüfung, Unter den Eichen 87, 12205 Berlin, Germany
b
Humboldt-Universität zu Berlin, Brook-Taylor-Str. 2, 12489 Berlin, Germany
A R T I C L E I N F O A BS T RAC T
Keywords: We analysed the hydrogen generation during the smouldering of polymeric materials, which are typically used
Fire chemistry in the household, in the Smoke Density Chamber coupled to a new developed hydrogen sensor to detect early
Detection stages of fires. The results of hydrogen generation were compared with the emission of carbon monoxide and
Compartment fires smoke during the fire scenarios. Additionally, the results were compared with parameters used in traditional
Smoke
commercial detection systems. In this scenario, the hydrogen sensor showed encouraging results for the
detection of fires in earlier phase compared to traditional detectors. Furthermore, we tested the new developed
hydrogen sensor in a real room with different fire scenarios. We have also investigated interferences, e.g. steam
and cigarette smoke. The hydrogen sensor could detect hydrogen generation in the earliest stage of fire, even
before CO and smoke were developed in detectable amounts. Therefore, the hydrogen sensor can be applied for
early fire detection in case of pyrolysis. The sensors are quite good for detecting pyrolysis gases. But when it
comes to a fast ignition other techniques are more suitable for it. The sensors are best for combination with
other techniques, such as smoke detectors.
⁎
Corresponding author.
http://dx.doi.org/10.1016/j.firesaf.2017.04.035
Received 18 February 2017; Received in revised form 12 April 2017; Accepted 18 April 2017
Available online 28 April 2017
0379-7112/ © 2017 Elsevier Ltd. All rights reserved.
S. Krüger et al. Fire Safety Journal 91 (2017) 1059–1067
amount of hydrogen produced in different fires has been given contrary ranges, and number of calibration points, the flow rates and ultimately
values in the literature, and no clear trend in different fire scenarios a mathematical calibration curve model. The calibration of the MEIS
like in the Test Fire scenarios of the EN 54: Fire detection and fire sensor in different conditions has been evaluated previously [7]. This
alarm systems have been emitted. calibration was carried out with various hydrogen reference gas
There are various fire phases according to ISO 19706. Any organic concentrations taking into account this specified parameters (40%
material, when irradiated to a sufficient heat flux, will decompose, relative humidity, concentration range 1–60 ppm (1; 2; 5; 10; 20; 40
emits gases, and give off smoke. Smouldering is a process, which is and 60 ppm)) with the relation Um=U0+log(k c), the hydrogen con-
slow, has low-temperature and is a flameless form of combustion, centration c is correlated to the measured voltage Um by the acalibra-
caused by heat when oxygen directly attacks the surface of a con- tion parameters U0 and k. More details are published earlier [7,20,21].
densed-phase fuel [18]. In contrast little or no oxidation is usually The limit of detection of the hydrogen sensor was determined to be
involved in this gasification process during pyrolysis during pyrolysis, 0.5 ppm. The response time and accuracy of the sensor is depending on
and thus is mainly an endothermic process. Hydrogen generation is gas concentration. At concentrations smaller than 10 ppm the response
most likely to occur in pyrolysis processes. For example, an overview of time t50 (time to 50% of signal change) is smaller than 1 min and the
a pyrolysis mechanism of guaiacol as lignin model compound can be accuracy about 5%. A temperature correction was realized to correct
found in [19]. the influence of the temperature on the hydrogen signal. Each sensor
In this study we investigate a new developed technology for the was calibrated before use. To remove any influence of the calibration
detection of hydrogen by capacitance sensors. Therefore, we simulated used an amount of hydrogen calculated, the measured signal has been
the pyrolysis to detect hydrogen generation in the earliest stage of fire, given for most of the test, and the approximated hydrogen value
and compared it with the emission of CO and smoke. The concentra- calculated for the discussion part.
tions of the combustion gases components were quantified by FTIR
(Fourier transformed infrared) spectroscopy, and compared with the
hydrogen sensor signal. Interferences of fire gases on the sensor were 2.2. Bench-scale tests: coupling of the hydrogen sensor with the
evaluated, and conclusions were drawn on the different fire detection Smoke Density Chamber
systems.
We simulated different pyrolysis scenarios in the Smoke Density
2. Materials and methods Chamber (Fire Testing Technology, East Grinstead, UK) with polymeric
materials, which are typical present in living rooms, e.g. polyethylene
We simulated different scenarios in the Smoke Density Chamber (PE), polyurethane (PUR) and wood. The experimental set-up is shown
(SDC similar to ISO 5659-2) with polymeric materials, which are in Fig. 1.
typical present in living rooms, e.g. polyethylene (PE), polyurethane The SDC was used for determination of the optical smoke density
(PUR) and wood. The hydrogen evolution was monitored during the Ds according to ISO 5659-2. Additional determinations of qualitative
test by the new developed hydrogen sensor, which was coupled to the and quantitative composition of smoke gases in order to identify
SDC. we further simulated the scenarios with the same materials and interferences were measured by directly connecting a FTIR spectro-
household items in a two room house. In this case the hydrogen meter (Nicolet type 380, Thermo Fisher Scientific GmbH) to the SDC.
detectors had been mounted on the ceiling of the room. The effectivity The optical density and the smoke gas composition were measured
of the hydrogen sensors has been compared with commercially simultaneously. The sampled gas was conveyed to the FTIR spectro-
available smoke and carbon monoxide detectors. Furthermore, we meter (0.5 l gas cell heated at 180 °C) through a 1-micron particulate
investigated interferences with other emitted gases, humidity and filter (all heated to 180 °C) via a 2 m long heated transfer line. The
smoke. FTIR spectra were measured at a gas temperature of 180 °C, with a
resolution of 0.5 cm−1(16 scans pro spectrum) with a DTGS detector, in
2.1. The hydrogen MEIS sensor the domain 900–4250 cm−1. The hydrogen evolution was monitored
during the test by the hydrogen sensor, which was coupled to the SDC.
The hydrogen sensors used in this study are based on a metal/solid When hydrogen was present, the interaction with the palladium gate
electrolyte/insulator/semiconductor (MEIS) structure [7]. On the changed the potential. The increase in hydrogen concentrations
silicon wafer isolators silicon oxide and silicon nitride are grown, correlated with an increase in the related voltage shift.
followed by the solid electrolyte lanthanum trifluoride. This ionic
conductor amplifies the signal of the MEIS sensor. As gate material a
palladium layer is deposited on the top. A capacitance element results
from this structure. A capacitance element results from this structure.
When hydrogen is present, the chemical potential of the palladium gate
is changed which results in a change the capacitance of the MEIS
structure. By a feedback mode the capacitance was held constant
during the measurement. So the applied voltage shift is the change of
the chemical potential. The sensors were heated once prior to the test
for reactivation. This heating was performed via a set up based on a
platinum thin film around the palladium gate. The increase in hydro-
gen concentrations is correlated with an increase in the related voltage
shift. A logarithmic dependency between these two parameters exists.
Details of the measurement principle and concentration to voltage
conversion are published elsewhere [7,20,21].
In order to assess the hydrogen concentration during the fire tests,
the determined hydrogen signals in mV must be converted in the
corresponding concentration in ppm. For this purpose, a suitable
calibration is essential. This calibration is very complex and following
parameters must be considered, for example, humidity, other gases like Fig. 1. Experimental set-up in the Smoke Density Chamber coupled with a FTIR
CO as well as the temperature during the calibration, the concentration spectrometer and a hydrogen sensor.
1060
S. Krüger et al. Fire Safety Journal 91 (2017) 1059–1067
Fig. 2. Schematic representation of the two rooms used for the tests (dimensions in meter). a) plan (view from top), b) front elevation (view from side).
2.3. Real-scale tests: coupling of the hydrogen sensor and other gases concentrations. Commercially available CO detectors (gas cell)
detection systems in a 2-room apartment were located on the ceiling. The CO concentration values measured in
FTIR were compared to the indication of CO detectors (alarm or not).
We performed different fire tests in a 2-room apartment under In addition, a paramagnetic oxygen analyzer was used to measure
smouldering and flaming conditions with the same virgin materials the oxygen availability, and in order to identify problems with cross-
which were investigated in the bench-scale test, a combination of sensitivities. The temperature was measured by thermocouples (TE) at
materials as a mixed crib and furnishings. The structure of the room various positions to identify the influence of temperature on the
and the location of measurement technology are similar to EN 54, but hydrogen sensors. Furthermore, different commercially available
with reduced dimensions. The specimens stood in the middle of the smoke detectors were located on the ceiling, so a direct comparison
room on a hot plate (pyrolysis) or on the ground in a fire pan (flaming). with the hydrogen sensor in terms of the limits of use and functionality
In the middle of the room was a heating plate, which proved in in these particular fire scenarios was possible.
preliminary tests an appropriate pyrolysis source. Different tempera- To avoid false alarms, it must be excluded that the sensors react to
tures between 350 and 600 °C could be set. A Bunsen burner was used noises, such as water vapor. Therefore 300 ml of water were evaporated
for fire scenarios with flaming conditions. in a pan during the test. We investigated the influence of cigarette
The measurement technology was positioned on a ring-on the smoke on the hydrogen signal. For this purpose, four cigarettes and two
ceiling of the room with a radius of 1.8 m from the room center. The cigarillos were smoked directly underneath the ceiling sensors (CO,
used experimental set-up is shown in Fig. 2. smoke, H2).
The new developed hydrogen sensor MEIS [7] was used to measure
the emitted hydrogen concentration at four different positions. The CO 3. Results and discussion
emission was determined in Room 1 using a mobile FTIR spectrometer
(Gasmet Dx4000N, Ansyco GmbH). The emitted gases were collected 3.1. Hydrogen generation during pyrolysis of different materials in
through a removable probe sensor corresponding to the H2 sensors. the Smoke density chamber (SDC)
The sampling gas was conveyed to the FTIR spectrometer through a 1-
micron particulate filter, directly behind the probe tube. The gases were The hydrogen sensor could detect hydrogen generation in the
continuously collected and analysed during the fire tests. The FTIR earliest stage of fire, so quantitatively measurable hydrogen concentra-
spectra were measured online by a MCT detector with a resolution of tions were detectable with a heating source set in the temperature
8 cm−1 and 16 scans pro spectrum, keeping a gas temperature of 50 °C range of 300–400 °C. These temperatures correspond approximately to
in the gas cell (0.5 l) of the spectrometer in the domain 900– an external heat flux of 4.6–9.3 kW/m2. In addition to the specific
4250 cm−1. The path length was 5 m, and the sampling flow 3 l/min. optical density, the concentration of smoke gas components was
The spectrometer was calibrated for the relevant chemical substances evaluated and compared with the hydrogen generation. A comparison
contained in smoke, allowing the quantitative evaluation of smoke of the resulting H2 detection signal, the mass loss, the CO concentra-
1061
S. Krüger et al. Fire Safety Journal 91 (2017) 1059–1067
for wood, PE, PUR. The transmission value is above 80%, which means
no alert by smoke detector. The CO emission is very low, which means
no alert by CO detector. Carbon monoxide is produced in detectable
concentrations (above 3 ppm) at later time (9 kW/m2) compared with
the hydrogen signal (already at 5 kW/m2). Conclusion of this bench-
scale fire tests is, that the hydrogen sensor is suitable to detect the
decomposition of the material in an early phase of fire, that means
smouldering conditions, before ignition of the materials due to the
early release of hydrogen by thermal decomposition. The measured H2
signal voltage rises corresponded to a hydrogen concentration value
higher than 40 ppm, reached at source temperature of 350 °C for wood,
400 °C for PUR and 500 °C for PE.
Table 1
Results from pyrolysis at increasing heat fluxes in the SDC, no pilot flame.
1062
S. Krüger et al. Fire Safety Journal 91 (2017) 1059–1067
Table 2
Emitted gases from pure materials during the fire tests in the 2 room apartment.
Material Conditions Gas detected Max. CO [ppm] CO2 release H2O release Other gases
PE Flaming 600 °C CO, CO2, H2O, ethane, ethylene 25 1190 0.15 Ethane (16 ppm)
Ethylene (4 ppm)
Pyrolysis 450 °C H2 O – 0 0.17 –
during the duration of the test, but the hydrogen detection technology reduce the hydrogen signal during the fire tests. The proposed thresh-
was appropriate in this case for early fire detection, with increasing old for the hydrogen sensor in the range of 20–25 mV is not reached. In
signal over time. this case both smoke detectors triggered an alarm, so that should be the
The concentration measured by FTIR at the roof was under 25 ppm preferred technique.
for most of the tests, over the whole test duration. The CO concentra-
tion reached the 25 ppm limit only for wood after 40 min of the test 3.2.2. Reproducibility of the H2 sensors
(Fig. 4). The 25 ppm limit is the limit for an alarm defined by the BS The H2 sensors showed good reproducibility results in similar
EN 54–26:2015 Fire detection and fire alarm systems for carbon pyrolysis conditions with the same materials (Fig. 5). In case of wood
monoxide detectors. That means that only in the case of wood, an and polyurethane (PUR), the signal response to the pyrolysis is
alarm by a CO detector would occur. In our case, the two CO detectors respectively, similar, with a strong slope between the start and 15 min,
mounted on the ceiling did not given any alarm during the test times. and then levelled off after 25 min, or 30 min, respectively. The plateau
Furthermore, the two H2 sensors showed reproducible results, even occurred at different values between the two tests of the same material,
though they were placed diametrically opposite of each other (Position and a difference in the amount of hydrogen produced is expected.
1 and 2).
Bench-scale and large scale tests of the pure materials, apart from
3.2.3. Pyrolysis of mixed cribs
PE, show similar good results in terms of hydrogen detection, also
According to the fire tests of the pure materials, a combination of
compared to CO and smoke detectors. If we compare all the measure-
different materials in form of a mixed crib was pyrolysed to validate
ment of the pure materials at low heating plate temperatures, including
hydrogen sensors. The selected materials reflected a typical average
PE, we see the limit of the H2 sensor. The detection of hydrogen is not
living room and the mixed crib contained 50% wood, 20% PE, 20%
always possible, as demonstrated by the example of PE. Emitted gases
PUR and 10% PVC.
Fig. 4. Results of the different technologies used in a 2 room apartment: The H2 sensors (H2 signal), CO concentration measured by FTIR (for the CO detectors) and smoke detectors
(0=no alarm, 1=one detector alarm, 2=two detectors alarm) during pyrolysis of wood, PUR foam, PVC and a mixed crib.
1063
S. Krüger et al. Fire Safety Journal 91 (2017) 1059–1067
the smoke alarm was triggered in all four cases. In case of the
smouldering arm chair, the second smoke detector between the trigger
an alarm 20 min after the first one, even if they were positioned
equidistant from the fire position. The reproducibility of the smoke
detectors is not satisfactory in this case, whereas both hydrogen
sensors showed reproducible results in this test.
1064
S. Krüger et al. Fire Safety Journal 91 (2017) 1059–1067
Fig. 6. Results of the different technologies used in a 2 room apartment: the H2 sensors (H2 signal), CO concentration measured by FTIR (for the CO detectors) and smoke detectors
(0=no alarm, 1=one detector alarm, 2=two detectors alarm) during pyrolysis of house hold items.
Only after 11 min, when the cigarette was held directly under the
smoke detector, the smoke detector gave an alarm. CO2 concentration
rose twice; CO could not be detected during this time. The CO detectors
did not emit any warning signals. The hydrogen signals increased
during smoking and reached values of 17–37 mV. The hydrogen
sensors are able to detect the smouldering of the cigarettes, with
signals above the proposed threshold at the end of the test. This
unwanted effect is a limitation of the sensors, and cross-sensibility
need to be improved.
Table 3
Emitted gases from the household items during the fire tests in the 2 room apartment.
Material Conditions Temp. Sample [°C] Gas detected Max. CO [ppm] CO2 release H2O release Other gases
Carpet Forced ignition n/A CO, CO2, H2O 6 959 0.17 Ethane (9 ppm)
Forced ignition n/A CO, CO2, H2O, 12 1199 0.14 –
1065
S. Krüger et al. Fire Safety Journal 91 (2017) 1059–1067
Fig. 8. Representation of the three fire detection technologies used (smoke, CO and the new developed H2 detectors) and the corresponding extremum obtained over the tests time.
armchair and paper roll). The two most effective technologies are But due to the limitations of the hydrogen sensors, especially due to
smoke and hydrogen sensors. The developed hydrogen sensors are the need of further developments, the hydrogen sensors are not
competing with the commercially available smoke detectors in most of recommended for every fire scenario. When fast developing fire
the case. In seven fire scenarios the hydrogen sensor triggered an alarm scenarios are possible, the most successful system would be a combi-
earlier than the smoke detector, and in three fire scenarios, they trigger nation of smoke detector and hydrogen sensor for the highest safety.
an alarm at the same time than the smoke detectors, considering a
threshold of 25 mV for an alarm of H2 detector. The hydrogen sensors 4. Conclusion
are recommended for pyrolysis detection in an early stage. Reasons for
the good results of the hydrogen sensor can be: (1) H2 is released The hydrogen sensor could detect hydrogen generation in the
earlier and at lower temperature during the pyrolysis process and (2) earliest stage of fire, even before CO and smoke were developed in
H2 can be easier and more accurate detect. The early detection of detectable amounts. Therefore, the hydrogen sensor is adapted for
hydrogen is due to the volatility of H2 and the high sensitivity of the early fire detection in case of pyrolysis.
sensor. The limit of quantification of the hydrogen sensor is 0.5 ppm. In bench-scale tests we could identify advantages of the sensor for
This concentration is in the same range of environmental amount of the early detection of smouldering fires when the external irradiance is
Hydrogen in the atmosphere [20]. Due to the logarithmic dependency lower as 15 kW/m2 and the source temperature under 500 °C. Under
between Hydrogen concentration and measured signal the sensor is these conditions we could find an evidence of low concentrations of
able to detect even small concentration changes. hydrogen, before smoke and/or CO detectors could alert. The investi-
Regarding the pyrolysis, we could show in our experiments, that H2 is gated hydrogen sensors have a significant advantage in comparison to
released earlier than CO. The detection is depending on the sensitivity of the CO detectors in this case.
the sensors. The limit of quantification of CO is 11 ppm (FTIR measure- We could determine a maximum hydrogen concentration of 3–
ments). In contrast the threshold of most commercial CO detectors is 40 ppm, in bench-scale tests and in a 2-room container apartment the
mainly around 20–30 ppm. We orientated our interpretation on this fire tests with household items typical for a habitation. The concentra-
threshold. Under these circumstances the hydrogen sensor showed the tions depend on the fire conditions and the materials. The hydrogen
better results in comparison with the FTIR and the CO detectors, in sensor is a possible alternative or supplement to traditional smoke and
particular regarding practical applications. CO detectors for improved safety.
A theoretical work in [19] showed that CO is released earlier than H2 In some cases of forced flaming conditions, the hydrogen sensors
in the case of the pyrolysis of cellulose, hemi-cellulose and lignin. The signals showed strong cross-sensitivities, and the signal did not reach a
release of H2 and CO is dependent on the composition and structure of reliable value. This is a limitation of the sensors. No systematic
the material. These natural components contain oxygen groups in their correlations could be made between the change of the hydrogen signal
chains, and the pyrolysis of the chains releases CO as decomposition by cross-sensitivity and the oxygen concentration, the gas concentra-
product. The release of CO as pyrolysis product is not possible in tion of the by-products (measured by means of FTIR spectroscopy) and
material that do not contain oxygen, without additional oxidation. We the presence or absence of a flame. For the end use applications of the
used in our study a new generation of hydrogen sensor that are more detectors, the main influencing factors and cross-sensitivities need to
sensible that the existing sensors used in this study, and a direct be further evaluated and meliorated. Moreover, a combination with
comparison is not possible. Further investigations that use CO detecting smoke detectors is advised for an alarm in almost all scenarios tested.
methods with higher sensitivity and lower limit of quantification The MEIS hydrogen sensors have the advantage, that their fabrica-
combined with our hydrogen sensors will allow a better evaluation of tion is based on the semiconductor technology. Low production costs
the time of release of H2 compared to CO. If settings are lower e.g. 2– can be achieved, which gives a basis for the competitive improvement
3 ppm for CO sensors, maybe CO sensor can give also an early alarm. of fire detectors. Moreover, a combination with smoke detectors still
1066
S. Krüger et al. Fire Safety Journal 91 (2017) 1059–1067
provides a positive economical aspect. The results presented in this [10] I. Lundstrom, S. Shivaraman, L. Stiblert, C. Svensson, Hydrogen in smoke detected
by the Pd-gate field-effect transistor, Rev. Sci. Instrum. 47 (1976) 738–740.
study give an outlook for the further development of the hydrogen [11] O. Linden, H. Hölemann, Testing Methods for Gas Sensor Based Fire Detectors,
sensing technology, for an improved fire safety. Proceedings AUBE, 2001.
[12] D. Kohl, J. Kelleter, H. Petig, Detection of fires by gas sensors, Sens. Update 9
(2001) 161–223.
Acknowledgements [13] A. Sawada, T. Higashino, T. Oyabu, Y. Takei, H. Nanto, K. Toko, Gas sensor
characteristics for smoldering fire caused by a cigarette smoke, Sens. Actuators B-
The authors thank the Federal Ministry for Economics and Chem. 130 (1) (2008) 88–93.
[14] F. Zhou, J. Li, Y. Liu, H. Shao, Rules of variation in hydrogen during reignition of
Technology, Germany, for its financial support (KF 2201068DF3). underground fire zones of spontaneous coal combustion mining, Sci. Technol. 20
The authors thank Vinicius Da Silva Duarte and Benjamin Klaffke for (2010) 0499–0503.
the technical support during the room fire tests. [15] L. Boon-Brett, J. Bousek, G. Black, P. Moretto, P. Castello, T. Hubert, U. Banach,
Identifying performance gaps in hydrogen safety sensor technology for automotive
and stationary applications, Int. J. Hydrog. Energy 35 (2010) 373–384.
References [16] L. Boon-Brett, J. Bousek, P. Moretto, Reliability of commercially available hydro-
gen sensors for detection of hydrogen at critical concentrations: Part II – selected
[1] S. Krüger, A. Berger, U. Krause, Chemical-analytical investigation of fire products sensor test results, Int. J. Hydrog. Energy 34 (2009) 562–571.
in intermediate storages of recycling materials, Fire Mater. 36 (3) (2012) 165–175. [17] T. Hubert, L. Boon-Brett, G. Black, U. Banach, Hydrogen sensors – a review, Sens.
[2] S. Krüger, Brandanalytische untersuchungen von polymerwerkstoffen-toxizität von Actuators B-Chem. 157 (2011) 329–352.
rauchgasen am beispiel von polyvinylchlorid (PVC), MP Mater. Test. 52 (3) (2010) [18] G. Rein, Smoldering Combustion. SFPE Handb. Fire Prot. Eng., New York, NY,
124–131 (ISSN 0025-5300). Springer New York, 2016, pp. 581–603. 〈http://dx.doi.org/10.1007/978-1-4939-
[3] S. Krüger, Investigation of smoke gases and temperatures during car fire – large- 2565-0_19〉.
scale and small-scale tests and numerical investigations, Fire Mater. 6 (40) (2016) [19] H. Yang, et al., Characteristics of hemicellulose, cellulose and lignin pyrolysis, Fuel
785–791. 86 (2007) 1781–1788. http://dx.doi.org/10.1016/j.fuel.2006.12.013.
[4] S. Krüger, Experimental studies on the effect of fire accelerants during living room [20] C.E. Yver, I.C. Pison, A. Fortems-Cheiney, M. Schmidt, F. Chevallier, M. Ramonet,
fires and detection of ignitable liquids in fire debris, Fire Mater. 7 (39) (2015) A. Jordan, O.A. Søvde, A. Engel, R.E. Fisher, D. Lowry, E.G. Nisbet, I. Levin,
636–646. S. Hammer, J. Necki, J. Bartyzel, S. Reimann, M.K. Vollmer, M. Steinbacher,
[5] M. Jackson, I. Robins, Gas sensing for fire detection: measurements of CO, CO2, T. Aalto, M. Maione, J. Arduini, S. O’Doherty, A. Grant, W.T. Sturges, G.L. Forster,
H2, O2, and smoke density in European standard fire tests, Fire Saf. J. 22 (2) (1994) C.R. Lunder, V. Privalov, N. Paramonova, A. Werner, P. Bousquet, A new
181–205. estimation of the recent tropospheric molecular hydrogen budget using atmo-
[6] T. Amamoto, K. Tanaka, K. Takahata, S. Matsuura, T. Seiyama, A fire detection spheric observations and variational inversion, Atmos. Chem. Phys. 11 (2011)
experiment in a wooden house by SnO2 semiconductor gas sensors, Sens. Actuators 3375–3392. http://dx.doi.org/10.5194/acp-11-3375-2011 (http://www.atmos-
B, 1, 1990, pp. 226–230. chem-phys.net/11/3375/2011/).
[7] W. Moritz, V. Fillipov, A. Vasiliev, G. Cherkashinin, J. Szeponik, A. Field, Effect [21] M. Lang, U. Banach, K. Nörthemann, A.-K. Gerlitzke, M. Milstrey, R. Kaufer,
based hydrogen sensor for low and high concentrations, ECS Trans. 3 (2006) M. Woratz, T. Hübert, W. Moritz, Long-term stability of a MEIS low energy
224–230. hydrogen sensor, Sens. Actuators B: Chem. 187 (2013) 395–400. http://
[8] V. Antonenko, A. Vasiliev, I. Olikhov, Early fire detection. semiconductor gas dx.doi.org/10.1016/j.snb.2012.12.081.
sensors, Electronica B 4 (2001) 48–52.
[9] K. Nörthemann, W. Moritz, S. Krüger, M.-C. Despinasse, Forest fire detection using
Hydrogen Sensors, AUBE '14 – 15th International Conference on Automatic Fire
Detection, ISBN 978-3-940402-02-8, 1, 2014, pp. 247–254.
1067