- Email: [email protected]
A solid electrolyte sensor for trace gas analysis
G Model SNB-14744; No. of Pages 6
ARTICLE IN PRESS Sensors and Actuators B xxx (2012) xxx–xxx
Contents lists available at SciVerse ScienceDirect
Sensors and Actuators B: Chemical journal homepage: www.elsevier.com/locate/snb
A solid electrolyte sensor for trace gas analysis Matthias Schelter a,∗ , Jens Zosel a , Wolfram Oelßner a , Ulrich Guth a,b , Michael Mertig a,b a b
Kurt-Schwabe-Institut für Mess und Sensortechnik Meinsberg, Kurt-Schwabe-Strasse 4, 04720 Ziegra-Knobelsdorf, Germany Technische Universität Dresden, Department of Physical Chemistry, 01062 Dresden, Germany
a r t i c l e
i n f o
Article history: Available online xxx Keywords: Solid electrolyte sensor Yttria-stabilized zirconia Trace gas analysis Coulometry Gas chromatography
a b s t r a c t Coulometric solid electrolyte sensors based on yttria-stabilized zirconia were characterized with respect to the electronic conductivity of the solid electrolyte and to their electrochemical activity of oxygen, hydrogen and hydrocarbon conversion. Sensor parameters like the temperature dependency of the electronic conductivity as well as the elevated noise of the coulometric cell at the working temperature determine the limit of detection of coulometric titration. Here we describe investigations on these parameters and approaches to shift the lower detection limit into the ppb-range. Furthermore, it is shown that a coulometric solid electrolyte sensor, positioned behind a gas chromatographic separation unit, allows simultaneous, calibration-free and long-term stable detection of different oxidizable components and oxygen. © 2012 Elsevier B.V. All rights reserved.
1. Introduction Normally, the measurement of a broad range of very small amounts of volatile substances in gas mixtures with concentrations below 1 vol.-ppm requires the use of expensive, space-occupying and immobile mass spectrometers [1,2]. Although this well accepted and commercially available technology provides very low detection limits, low-cost systems could reach much broader application, for example in clinical routine analysis of human breath gas. Coulometric solid electrolyte cells are well known devices for measuring oxygen [3], hydrogen [4], hydrocarbons [5] and other oxidizable or reducible components of gaseous mixtures. They are successfully used for a variety of applications [6–8]. In most cases, the key component of such devices is an oxygen-ion-conducting material like yttria-stabilized zirconia (YSZ) with an ionic conductivity many orders of magnitude higher than its electronic conductivity. One noticeable advantage of the coulometric sensor principle consists in its strict operation due to Faraday’s law, providing high long-term stability without baseline drifting and thus a calibration-free operation. Up to now, the lower limit of detection of coulometric sensors for the relevant gases amounts to 10 vol.-ppm [9,10]. This limit is determined by • the electronic conductivity of the electrolyte material [11], • the noise of the current through the electrochemical cell, and
∗ Corresponding author. Tel.: +49 34327 608 120; fax: +49 34327 608 131. E-mail addresses: [email protected] (M. Schelter), [email protected] (J. Zosel), [email protected] (W. Oelßner), [email protected] (U. Guth), [email protected] (M. Mertig).
• the design of the circuits providing the constant potential and measuring the cell current.
In this work investigations on parameters influencing the lower limit of detection of coulometrically operated solid electrolyte cells are presented. Possibilities to shift this limit into the ppb-range are shown. Furthermore, a measurement system containing such an optimized coulometrically operated solid electrolyte cell is presented acting as detector for a gas chromatographic separation unit. It is demonstrated that even the very low concentration of atmospheric H2 can be determined with this kind of coulometric detector. The principle of the coulometrically operated solid electrolyte sensor, shown in Fig. 1, is based on the complete oxidation or reduction of the gas component to be measured at the working electrode of a solid electrolyte cell resulting in an electronic current. Detectable components are gases existing as ions in the solid electrolyte or are gases which could react with these ions. In case of YSZ as electrolyte material, the reactions of oxygen and hydrogen at the working electrode can be written in the Kröger–Vink notation according to Eq. (1) and (2). O2 (g) + 2V··O (YSZ) + 4e− (Pt) 2OxO (YSZ)
(1)
H2 (g) + OxO (YSZ) H2 O(g) + V··O (YSZ) + 2e− (Pt)
(2)
The amount of analyte molecules which is converted at the working electrode results with Faraday’s law from the electrolysis current I and the residual current IR caused by electronic conductivity. Considering also the flow of the measuring gas dV/dt, the
0925-4005/$ – see front matter © 2012 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.snb.2012.10.111
Please cite this article in press as: M. Schelter, et al., A solid electrolyte sensor for trace gas analysis, Sens. Actuators B: Chem. (2012), http://dx.doi.org/10.1016/j.snb.2012.10.111
ARTICLE IN PRESS
G Model SNB-14744; No. of Pages 6
M. Schelter et al. / Sensors and Actuators B xxx (2012) xxx–xxx
2
electrolysis cell
heater tube made of 8 mole-% YSZ measuring gas working electrode counter/reference electrode reference gas (air)
60
b
a
d
O2
UP
5.7
I c
Fig. 1. Schematic of a coulometric solid electrolyte sensor.
concentration x of the gas component to be measured is determined according to Eq. (3). x=
I − IR VM · zF dV/dt
(3)
F is Faraday’s constant, z is the number of electrons transferred per molecule and VM is the molar volume. Therefore, the lower detection limit of such a coulometric detector depends on the absolute value of IR and its noise, both resulting from the solid electrolyte material properties and the sensor temperature. Due to the fact that YSZ is already widely used as base material for sensors, fuel cells and oxygen generators, the properties of this material are well known [11,12]. The ionic and electronic conductivities of this material in the temperature range between 800 ◦ C and 1200 ◦ C are depicted in Fig. 2 showing the contributions of the electronic conductivity of holes (h) and of electrons (e) [11]. The electronic current IR caused by the hole conductivity flowing through a disc of YSZ with an area A and a thickness d is given by Eq. (4).
⎡
IR =
ART Fd
⎣h (pIO ) 2
⎧ ⎨ pII 1/4 O2
⎩ pIO2
R is the universal gas constant,
⎫⎤ ⎬ −1 ⎦ ⎭ pIO 2
(4)
is the oxygen partial pres-
sure in reference gas and pIIO is the oxygen partial pressure in 2 the measuring gas. Following this equation, a usual coulometric electrolysis cell (A = 1 cm2 , d = 1 mm, pIO = 20.6kPa and pIIO = 2
2
2.7 × 10−4 Pa) operated at 800 ◦ C with thermal fluctuations of ±1 K shows residual current deviations of ±37 nA. Those deviations cause an error of ±26 vol.-ppb in hydrogen detection at the gas flow dV/dt = 10 ml/min. Furthermore, the noise of the electrolysis current is influenced by turbulences of the measuring gas resulting in fluctuations in substance turnover at the working electrode. To get a constant reaction rate, this turbulence has to be avoided or minimized so that the frequencies of the generated noise spectrum are high enough to be 0
log(
S·cm -1)
-2
ion
1.63 10 2 exp ( 0.79eV / kT )
Fig. 3. Schematic drawing of the used coulometric sensor (measures in mm): (a) solid electrolyte tube made of 8 mol-% YSZ, (b) platinum working electrode, (c) platinum reference electrode, (d) measuring gas flow.
filtered out by low-pass filter. This parameter is difficult to predict but can be estimated by utilizing complex simulations taking the electrode morphology into consideration. Due to the above-mentioned advantages of coulometric sensors, they have been used as detectors in gas chromatographic systems already [13,14], where the gained sensitivities resulted in lower detection limits of >10 vol.-ppm. A chromatogram showing the sensor current I as a function of time t, allows to resolve peaks of different gas components. The peak area equals to the amount of charge Q transferred during the oxidation or reduction process of the analyte molecules. The amount of substance n of the gas component can be determined with Faraday’s law according to Eq. (5), where IB is the baseline current. 1 Q n= = zF zF
(I − IB ) dt
(5)
t1
2. Experimental notes The coulometric sensor used in this work is shown in Fig. 3. It consists of a tube made of 8 mol-% YSZ with the outer diameter 5.7 mm and the wall thickness 1 mm. The tubular platinum electrodes on the inner and the outer surface of this tube are 60 mm in length. Between the working (inner) and the reference (outer) electrode a constant potential of −400 mV is applied by a self-made potentiostat. As shown in Fig. 4, two different designs of electronic circuits of the self-made potentiostat were examined when combining the sensor with a gas chromatograph MG1 (SRI Instruments Europe GmbH, Bad Honnef, Germany). Fig. 5 shows a scheme of the measuring system, consisting of a sample injection port, one or more gas chromatographic columns and an YSZ cell acting as detector. A gaseous sample, containing volatile trace components, is injected into a carrier gas flow of the gas chromatograph either by using a sampling loop or by injecting with a syringe through a septum. A column packed with molecular sieve provides the separation of hydrogen, methane and other volatile components of the sample gas mixture within 10 min at 50–200 ◦ C column temperature in Ar carrier-gas flow of 20 ml/min.
1200 °C
-4
+
1000 °C 800 °C
-6
cS
-8
e
1.31 10 7 exp ( 3.88eV / kT )pO1/ 4
2.35 10 exp ( 1.67eV / kT )p 2
h
-3
-6
1/ 4 O
-9
-12
-15
-18
-21
-24
-27
-30
log(pO2 / atm) Fig. 2. Ionic and electronic conductivity of 8 mol-% YSZ depending on oxygen partial pressure at different temperatures according to [11].
UP
cS
RM
I
-10 0
t2
+
+
I
UA UA
UP
+ -
RM
A
B
Fig. 4. Different designs of electronic circuits for controlling the coulometric sensor according to Fig. 3: (A) current–voltage-converter (I = UA /RM ), (B) variable current measuring resistor at the working electrode (I = UA /RM ); coulometric sensor (cS), applied polarization voltage (UP ).
Please cite this article in press as: M. Schelter, et al., A solid electrolyte sensor for trace gas analysis, Sens. Actuators B: Chem. (2012), http://dx.doi.org/10.1016/j.snb.2012.10.111
ARTICLE IN PRESS
G Model SNB-14744; No. of Pages 6
M. Schelter et al. / Sensors and Actuators B xxx (2012) xxx–xxx
log(electrolysis current / µA)
1.1
Fig. 5. Schematic viewgraph of the measuring system consisting of a sample injection system, one or more gas chromatographic columns and a coulometric detector.
For investigations of the electronic conductivity and the noise behaviour of the cell, an electrochemical measuring system REF 600 (Gamry Instruments, Warminster, USA) was used. As measuring gas N2 or laboratory air as well as mixtures of N2 + H2 and N2 + O2 + H2 + CH4 mixed with mass flow controllers (Brooks Instrument Company, Hatfield, USA) were used. Laboratory air served as reference gas. 3. Results and discussion 3.1. Electronic conductivity of the solid electrolyte material The electronic conductivity of a planar disc of 8 mol-% YSZ has been described in literature for temperatures between 800 ◦ C and 1050 ◦ C [11]. To obtain the conductivities for tubular cell geometry of this material at 650 ◦ C and 750 ◦ C, in this work the sensor current was measured at different temperatures at a constant flow of N2 and a constant cell potential of −400 mV. Before these measurements were performed, the whole apparatus was proofed on leak-tightness by a helium leakage test indicating leakage rates lower than 10−10 mbar l/s. Due to the fact that a small amount of residual oxygen in nitrogen leads to a sensor current caused by ionic conductivity, this oxygen concentration was determined to correct the sensor current. Nitrogen was piped through the cell operated at 750 ◦ C at different flow rates. As shown in Fig. 6, the linear increase of the cell current proves that the complete amount of oxygen traces originates from the measuring gas. From the slope of the
electrolysis current / µA
17
15
13
11
I[µA] = 0.14·
dV [ml·min -1] + 8.9 dt
9 0
10
20
30
measuring gas flow /
40
50
ml·min -1
Fig. 6. Electrolysis current of the coulometric sensor at different flow rates of the measuring gas (N2 5.0), sensor temperature: 750 ◦ C.
3
750
measured 1 corrected 2 calculated 3
0.9 0.7 0.5
650 1
0.3
2
0.1 temperature / °C -0.1 0.96 0.98 1.00 1.02
3
1.04
1.06
1.08
1.10
1000·T -1 / K-1 Fig. 7. Arrhenius plots of the electrolysis current, the measured and corrected data at different temperatures compared to the calculated data from [11], measuring gas: 5 ml/min N2 5.0.
linear regression the oxygen content 0.5 vol.-ppm was calculated according to Faraday’s law. The Arrhenius plot of the measured and the corrected sensor currents as well as the data for hole conductivity derived from [11] and calculated with Eq. (4) is shown in Fig. 7. The diagram confirms the expected behaviour that under the chosen experimental conditions the residual current mostly originates from hole conductivity. Furthermore, the electronic conductivity of the used electrolyte material is consistent with the information for YSZ materials given in literature so far. Small differences between the measuring results and the calculated data from literature may come from not quantifiable deviations of the cell geometry as well as systematic errors in temperature measurement. 3.2. Noise behaviour of the coulometric sensor Fig. 8A shows the noise behaviour of the coulometric sensor operated at temperatures between 650 ◦ C and 750 ◦ C. The measurements were taken with the sample rate of 104 s−1 while all electronic filters of the REF 600 measuring system were disabled. These measurements proved that the maximum noise amplitudes depend on the sensor temperature and not on the flow rate of the measuring gas. The noise amplitudes measured are some orders of magnitude larger than the calculated amplitudes caused by thermal noise at the respective working temperature. As depicted in Fig. 8B, there are different maxima in the noise spectrum, for example at 0.4 kHz and at 2.6 kHz. The frequencies of these maxima do not depend on the flow rate which gives a hint that the noise comes due to the connection with the REF 600 and the electrical shield of the cell rather than disturbances in the gas flow. Most applications of coulometric cells require response times of t90 < 200 ms ideally, so that a lowpass filter can be used to eliminate the higher noise frequencies completely. Fig. 9 shows the fluctuations of the electrolysis current due to the control of the sensor heating unit. As it can be seen, the temperature control has in the low frequency range a significant influence on the sensor signal. This noise results completely from electronic conductivity of the solid electrolyte material given in Eq. (4). The fluctuations of the sensor current cannot be eliminated by a lowpass filter. A useful strategy to get rid of this noise influence is to install a heating unit operated at a constant current which fixes the cell temperature to a bandwidth of ±5 K. Afterwards the current signal can be corrected by the value of the real cell temperature.
Please cite this article in press as: M. Schelter, et al., A solid electrolyte sensor for trace gas analysis, Sens. Actuators B: Chem. (2012), http://dx.doi.org/10.1016/j.snb.2012.10.111
ARTICLE IN PRESS
G Model SNB-14744; No. of Pages 6
M. Schelter et al. / Sensors and Actuators B xxx (2012) xxx–xxx
4
0.5
16
signal deviation / %
electrolysis current / µA
20
12 8 4 0 0.00
750 650
temperature / °C
A 0.04
0.08
0.12
0.16
norm. amplitude
0.2 0.1
0.20
2
-0.1 0
15
30
45
60
measuring gas flow rate / ml·min -1
0.6
B
Fig. 10. Deviation of the electrolysis current from the complete hydrogen conversion at the working electrode of a coulometric sensor according to Fig. 3 depending on the measuring gas flow rate and the sensor temperature; hydrogen flow: 5.96 × 10−9 mol/s, electrolysis current: 1.15 mA.
0.4 0.2
3.3. Electrochemical activity of coulometric titration
0.0 0
1
2
frequency / kHz
3
4
Fig. 8. (A) Noise amplitude of the electrolysis current at two sensor temperatures (measuring gas: 5 ml/min N2 5.0). For a better overview, the sensor signal at 650 ◦ C is depicted in foreground from t = 0–0.7 s and the signal at 750 ◦ C from t = 0.7–2 s. (B) Frequency/power density spectrum for the signals from (A).
Depicted by curve 2 in Fig. 9, this method eliminates the noise influence caused by the temperature control. Furthermore, the signal noise is heavily influenced by the design of the electronic control unit of the electrolysis cell. Especially the part of the current measuring unit has huge effect on the noise amplitude. The current measuring circuit shown in Fig. 4A leads to clearly larger noise amplitudes than the circuit shown in Fig. 4B, where a resistor is connected in series to the coulometric sensor. The reason is the feedback of the operation amplifier output to its input resulting in a phase shift and at the worst, causing an oscillation of the potential output at specific frequencies. An example for the effect of the design of the electronic control unit is enumerated in Section 3.4.
15
electrolysis current / µA
1
0.3
0.0
time / s
14 13 1
12
2
11
1 unregulated 2 regulated 10 0
temperature / °C 650 (curve 1) 0.4 750 (curve 2)
250
500
750
1000
1250
time / s Fig. 9. Fluctuations of the electrolysis current depending on the control of the sensor heating unit. Curve 1: unregulated heating with a constant heating current resulting in a temperature bandwidth of 750 ± 5 K. Curve 2: thermocouple controlled heating with fluctuations of about ±1 K.
For a coulometric detector it is necessary to have 100% conversion of analyte molecules at the electrolysis electrode so that Faraday’s law is applicable. Due to the constant potential of the presented electrode, a complete conversion of analyte molecules is given only up to a specific limiting flow rate of measuring gas depending on the diffusion coefficient and the reaction velocity. Only in this gas flow range the coulometric sensor can be used without calibration because the relevant parameters do not influence the sensor signal. As shown in Fig. 10, the results for titration of hydrogen prove that a sensor with the given geometric parameters operated at a temperature of 650 ◦ C shows a decrease in conversion of 0.2% at flow rates greater than 35 ml/min. The reaction rate at 750 ◦ C is high enough to achieve complete conversion of hydrogen in the determined flow range. 3.4. Use of the coulometric sensor as detector for gas chromatography The results of the investigations presented in Sections 3.1–3.3 were used to determine optimal operation parameters for a lownoise measurement of trace components in gaseous mixtures mentioned in Section 2. For the use of coulometric sensors as detectors for gas chromatographic measuring systems, the parameters response time, sensitivity, detection limits and long-term stability are of main interest. Known from investigations on broadband lambda sensors and breath gas sensors, the response time of YSZ solidelectrolyte sensors is short enough for the use in chromatography (t90 < 100 ms) [15,16]. The maximum sensitivity coulometric sensors can reach is given by Faraday’s law, whereas the lower limit of detection is affected by the boundary condition presented in Sections 3.1 and 3.2. Regarding to the long-term stability, coulometric sensors take advantage of being free of calibration due to the fact that with Faraday’s law the amount of substance is determined directly from the amount of charge resulting from the peak area given in Eq. (5). This coherence is valid until the electrodes get inactivated by specific gas components like sulphur or lead. Two chromatograms of gases containing the same components but with different concentrations are compared in Fig. 11. The concentrations calculated from the peak areas by Eq. (5) are given in the diagram. The residual current described in Section 3.1 does not agree to the equation, because it is already included in the baseline
Please cite this article in press as: M. Schelter, et al., A solid electrolyte sensor for trace gas analysis, Sens. Actuators B: Chem. (2012), http://dx.doi.org/10.1016/j.snb.2012.10.111
ARTICLE IN PRESS
G Model SNB-14744; No. of Pages 6
M. Schelter et al. / Sensors and Actuators B xxx (2012) xxx–xxx
5
sensor current / µA
25
2 1
0
CH4
H2 O2
-25
gas concentrations / vol.-ppm curve 1 curve 2 1 10 H2 23 70 O2 10 30 CH4
-50 -75
-100
0
100
200
300
400
500
600
time / s
electrolysis current / µA
50
680 +/- 140 vol.-ppb H2 400 +/- 50 vol.-ppb H2
1 µA 1 2 Control circuit of the coulometric 4A (curve 1) detector according to figure: 4B (curve 2) 60
amount of charge / mAs
4.5 3.6
methane
100
120
time / s Fig. 13. Comparison of hydrogen peaks of the coulometric titration of 1 ml laboratory air measured with a solid electrolyte sensor equipped with two different control circuits according to Fig. 4A and B.
Fig. 11. Chromatogram of two different gas mixtures of hydrogen, oxygen, methane and nitrogen. The concentrations calculated from the measured peaks are given.
10
mA
0
electrolysis current / µA
current IB . The values agree well with the adjusted concentrations of the components within the given accuracy, indicating complete conversion at the working electrode. The behaviour of the gas chromatographic system at different concentrations up to 280 vol.-ppm is depicted in Fig. 12. Minor deviations of the measured oxygen and methane concentrations from the adjusted concentrations are caused by the used mass flow controllers. The resolution is solely limited by the high-frequent noise described in Section 3.2. Therefore, a signal filtering with an optimized limiting frequency is crucial for the optimal lowering of the limit of detection. The hydrogen peaks of the coulometric titration of 1 ml laboratory air, shown in Fig. 13, were derived with the chromatographic measuring system described in Section 2 equipped with a silica gel column and a molecular sieve column. The peak derived with the electronic circuit according to Fig. 4A shows a significant higher noise than the one taken with the circuit shown in Fig. 4B, while the precision increases by the factor 3. The higher absolute concentrations of hydrogen, depicted by curve 1, result from other experiments with hydrogen taken near by the chromatographic measuring system. The results show that this detection principle can be used for the determination of analyte concentrations smaller than 500 vol.-ppb. Corresponding to Eq. (5), the detection limit decreases linearly with increasing number of electrons z exchanged during electrode reaction. This is valid, e.g.,
80
2.328 0.004 2.328+/-+/0.004
-10
vol.-% O2 -20 -30 0 0
100
200
300
400
change of measuring range
330 +/- 70 vol.-ppb CH4
-10
-20
3.5 +/- 0.2 vol.-ppm H2 -30 0
100
200
300
400
time / s Fig. 14. A chromatogram of a gas mixture of hydrogen, oxygen, methane and nitrogen depicted in two different resolutions showing the ability of resolving very small amounts of hydrogen and methane with comparably much larger amounts of oxygen.
for gas components like hydrocarbons, which give more electrons than hydrogen. One more remarkable advantage of the current measuring unit is the implemented range selection over six decades. As shown in Fig. 14, very small amounts of hydrogen and methane can be resolved reliably besides much larger amounts of oxygen. 4. Conclusions
oxygen
2.7 1.8
hydrogen
0.9
dots = measured values lines = calculated 0.0 0
40
80
120
160
200
240
280
concentration / vol.-ppm Fig. 12. Peak areas and amounts of charge derived from chromatograms of different concentrated gas mixtures in comparison to the values calculated by means of Faraday’s law.
In this work, investigations on properties of coulometric solid electrolyte sensors for continuous gas flows are presented. According to Faraday’s law the signal of those sensors depends only on concentration of the component to be determined and flow rate, allowing calibration free operation. The parameters influencing the lower detection limit of those coulometric sensors were determined. One of these parameters is the residual current caused by electronic conductivity of the solid electrolyte material, which depends on temperature. It was shown that the temperature dependence of the electronic conductivity calculated from the measured residual current of the presented sensor is consistent with data presented in literature. Consequently, the sensor current was real-time corrected successfully
Please cite this article in press as: M. Schelter, et al., A solid electrolyte sensor for trace gas analysis, Sens. Actuators B: Chem. (2012), http://dx.doi.org/10.1016/j.snb.2012.10.111
G Model SNB-14744; No. of Pages 6
ARTICLE IN PRESS M. Schelter et al. / Sensors and Actuators B xxx (2012) xxx–xxx
6
by subtracting the residual current calculated from the measured temperature value. Significantly higher effort was made for decreasing the noise of the cell current, which covers the frequency band from a few mHz to some kHz. Low-frequency fluctuations originate from temperature oscillations caused by controlling the heater with a thermocouple-assisted controller. This low-frequency noise was eliminated by uncontrolled heater operation coupled with the above mentioned signal correction based on temperature measurement. High-frequency noise caused by ambient influences like electrical fields was eliminated by the application of low-pass filters. The limiting frequency of the filter was adjusted according to the requirements of optimum chromatographic peak resolution. Additionally, the noise amplitude of the sensor current is significantly influenced by the design of the electronic circuitry for the sensor control. It was found that particularly the high-frequency noise is amplified by a current follower originally used in the working electrode circuit. That noise amplification was successfully eliminated by replacing the current follower with a resistor for current measurement in the working electrode circuit. By this means the detection limit of coulometric flow-through sensors as detectors for gas chromatography in trace gas analysis could be decreased from some 10 vol.-ppm to the ppb-range. As an example, the hydrogen concentration in laboratory air of about 400 vol.-ppb was measured with an accuracy of ±50 vol.-ppb. Acknowledgements The presented work is based on projects supported by the German Federal Ministry for the Environment, Nature Conservation and Nuclear Safety (funding reference number 03KB045) and the German Federal Ministry for Food, Agriculture and Consumer Protection (funding reference number 22011110). The authors are responsible for the content of the paper and thank gratefully for financial support. Furthermore, the authors would like to thank SRI Instruments Europe GmbH, Bad Honnef, Germany, and Zirox – Sensoren & Elektronik GmbH, Greifswald, Germany, for professional und device-related support. References [1] A. Amann, G. Poupart, S. Telser, M. Ledochowski, A. Schmid, S. Mechtcheriakov, Applications of breath gas analysis in medicine, International Journal of Mass Spectrometry 239 (2004) 227–233, http://dx.doi.org/10.1016/j.ijms. 2004.08.010. [2] W. Miekisch, J.K. Schubert, From highly sophisticated analytical techniques to life-saving diagnostics: technical developments in breath analysis, Trends in Analytical Chemistry 25 (2006) 665–673, http://dx.doi.org/10.1016/j.trac. 2006.05.006. [3] J. Fouletier, M. Bonnat, J. Le Bot, S. Adamowicz, Calibration of a highly sensitive oxygen analyzer for biological applications using an oxygen pump, Sensors and Actuators B 45 (1997) 155–160, http://dx.doi.org/10.1016/S09254005(97)00289-X. [4] K. Katahira, H. Matsumoto, H. Iwahara, A solid electrolyte hydrogen sensor with an electrochemically-supplied hydrogen standard, Sensors and Actuators B 73 (2001) 130–134, http://dx.doi.org/10.1016/S0925-4005(00)00672-9. [5] V.S. Sevast’yanov, E.M. Galimov, N.E. Babulevich, A.A. Arzhannikov, Zirconium dioxide-based solid electrolyte as a means of oxidation of organic compounds during isotopic assay of carbon, Russian Journal of Electrochemistry 43 (2007) 448–453, http://dx.doi.org/10.1134/S102319350704012X.
[6] L. Heyne, Solid electrolyte cells, in: H. Degn, et al. (Eds.), Measurement of Oxygen, Elsevier, Amsterdam, 1976, pp. 65–69. [7] C. Franx, A dynamic oxygen sensor with zero temperature coefficient, Sensors and Actuators 7 (1985) 263–270, http://dx.doi.org/10.1016/02506874(85)80006-8. [8] P. Butschbach, F. Hammer, H. Kohler, A. Potreck, T. Trautmann, Extensive reduction of toxic gas emissions of firewood-fueled low power fireplaces by improved in situ gas sensorics and catalytic treatment of exhaust gas, Sensors and Actuators B 137 (2009) 32–41, http://dx.doi.org/10.1016/j.snb. 2008.12.007. [9] K. Teske, H. Ullmann, N. Trofimenko, Thermal analysis of transition metal and rare earth oxide system–gas interactions by a solid electrolyte-based coulometric technique, Journal of Thermal Analysis and Calorimetry 49 (1997) 1211–1220, http://dx.doi.org/10.1007/BF01983677. [10] V. Vashook, L. Vasylechko, J. Zosel, W. Gruner, H. Ullmann, U. Guth, Crystal structure and electrical conductivity of lanthanum–calcium chromites–titanates La1–x Cax Cr1–y Tiy O3 (x = 0–1, y = 0–1), Journal of Solid State Chemistry 177 (2004) 3784–3794, http://dx.doi.org/10.1016/j.jssc.2004.07.007. [11] J.-H. Park, N. Blumenthal, Electronic transport in 8 mole percent Y2 O3 –ZrO2 , Journal of the Electrochemical Society 136 (1989) 2867–2876, http://dx.doi.org/10.1149/1.2096302. [12] H. Ullmann, Keramische Gassensoren, in: Grundlagen-Aufbau-Anwendung, Akademie-Verlag, Berlin, 1993. [13] K. Teske, P. Popp, J. Baumbach, Solid-state coulometric cell as detector for gas chromatography, Journal of Chromatography 360 (1986) 417–420, http://dx.doi.org/10.1016/S0021-9673(00)91690-X. [14] B.K. Zuev, A.Y. Olenin, Solid electrolyte sensor as a detector for gas chromatographic determination of combustible contaminants in air, Journal of Analytical Chemistry 61 (2006) 147–152, http://dx.doi.org/10.1134/ S1061934806020109. [15] E.M. Logothetis, Automotive Oxygen Sensors, in: N. Yamazoe (Ed.), Chemical Sensor Technology, 3, Elsevier, Amsterdam, 1991, p. 96. [16] J. Zosel, D. Tuchtenhagen, K. Ahlborn, U. Guth, Mixed potential gas sensor with short response time, Sensors and Actuators B 130 (2008) 326–329, http://dx.doi.org/10.1016/j.snb.2007.08.008.
Biographies Matthias Schelter received his Master of Science in chemistry from the Dresden University of Technology in 2011. Since 2011 he has been working at the Meinsberg Kurt-Schwabe Research Institute where he is doing currently his PhD thesis on the development of solid electrolyte sensors for dissolved gases in complex media. Jens Zosel received his diploma in physics from the University of Greifswald in 1990 and his PhD from the University of Freiberg in 1997. Since 1992 he has been working at the Meinsberg Kurt-Schwabe Research Institute. His basic research interests are directed towards the behaviour of electrochemical sensors in liquid and gaseous flows and the development of solid electrolyte sensors for different applications. Wolfram Oelßner received his diploma in nuclear technology from the Dresden University of Technology in 1962 and his PhD in electrochemical measuring technique from the same University in 1968. Since 1963 he has been working at the Meinsberg Kurt-Schwabe Research Institute, where he now acts as a scientific consultant. In 1989 he was appointed to a lecturer for physical–chemical measuring technique as well as for environmental measuring technique and corrosion research at the Dresden University of Technology. His research activities are directed towards the development of electrochemical sensors and measuring instrumentation. Ulrich Guth received his PhD from the University of Greifswald in 1975. In 1993 he became a professor for solid state chemistry at the University of Greifswald. From 1999 to 2010 he had been working as the director of the Meinsberg KurtSchwabe Research Institute and as a professor for physical chemistry especially sensor and measuring technology at the Dresden University of Technology. His principal research interests are directed towards solid electrolyte sensors and new materials for these high temperature sensors. Michael Mertig received his diploma and PhD in physics from the Dresden University of Technology. His basic research interests are biomimetic materials synthesis as well as electrochemical and biological sensors. Since 2010 he is the director of the Kurt-Schwabe-Institute in Meinsberg.
Please cite this article in press as: M. Schelter, et al., A solid electrolyte sensor for trace gas analysis, Sens. Actuators B: Chem. (2012), http://dx.doi.org/10.1016/j.snb.2012.10.111
ARTICLE IN PRESS Sensors and Actuators B xxx (2012) xxx–xxx
Contents lists available at SciVerse ScienceDirect
Sensors and Actuators B: Chemical journal homepage: www.elsevier.com/locate/snb
A solid electrolyte sensor for trace gas analysis Matthias Schelter a,∗ , Jens Zosel a , Wolfram Oelßner a , Ulrich Guth a,b , Michael Mertig a,b a b
Kurt-Schwabe-Institut für Mess und Sensortechnik Meinsberg, Kurt-Schwabe-Strasse 4, 04720 Ziegra-Knobelsdorf, Germany Technische Universität Dresden, Department of Physical Chemistry, 01062 Dresden, Germany
a r t i c l e
i n f o
Article history: Available online xxx Keywords: Solid electrolyte sensor Yttria-stabilized zirconia Trace gas analysis Coulometry Gas chromatography
a b s t r a c t Coulometric solid electrolyte sensors based on yttria-stabilized zirconia were characterized with respect to the electronic conductivity of the solid electrolyte and to their electrochemical activity of oxygen, hydrogen and hydrocarbon conversion. Sensor parameters like the temperature dependency of the electronic conductivity as well as the elevated noise of the coulometric cell at the working temperature determine the limit of detection of coulometric titration. Here we describe investigations on these parameters and approaches to shift the lower detection limit into the ppb-range. Furthermore, it is shown that a coulometric solid electrolyte sensor, positioned behind a gas chromatographic separation unit, allows simultaneous, calibration-free and long-term stable detection of different oxidizable components and oxygen. © 2012 Elsevier B.V. All rights reserved.
1. Introduction Normally, the measurement of a broad range of very small amounts of volatile substances in gas mixtures with concentrations below 1 vol.-ppm requires the use of expensive, space-occupying and immobile mass spectrometers [1,2]. Although this well accepted and commercially available technology provides very low detection limits, low-cost systems could reach much broader application, for example in clinical routine analysis of human breath gas. Coulometric solid electrolyte cells are well known devices for measuring oxygen [3], hydrogen [4], hydrocarbons [5] and other oxidizable or reducible components of gaseous mixtures. They are successfully used for a variety of applications [6–8]. In most cases, the key component of such devices is an oxygen-ion-conducting material like yttria-stabilized zirconia (YSZ) with an ionic conductivity many orders of magnitude higher than its electronic conductivity. One noticeable advantage of the coulometric sensor principle consists in its strict operation due to Faraday’s law, providing high long-term stability without baseline drifting and thus a calibration-free operation. Up to now, the lower limit of detection of coulometric sensors for the relevant gases amounts to 10 vol.-ppm [9,10]. This limit is determined by • the electronic conductivity of the electrolyte material [11], • the noise of the current through the electrochemical cell, and
∗ Corresponding author. Tel.: +49 34327 608 120; fax: +49 34327 608 131. E-mail addresses: [email protected] (M. Schelter), [email protected] (J. Zosel), [email protected] (W. Oelßner), [email protected] (U. Guth), [email protected] (M. Mertig).
• the design of the circuits providing the constant potential and measuring the cell current.
In this work investigations on parameters influencing the lower limit of detection of coulometrically operated solid electrolyte cells are presented. Possibilities to shift this limit into the ppb-range are shown. Furthermore, a measurement system containing such an optimized coulometrically operated solid electrolyte cell is presented acting as detector for a gas chromatographic separation unit. It is demonstrated that even the very low concentration of atmospheric H2 can be determined with this kind of coulometric detector. The principle of the coulometrically operated solid electrolyte sensor, shown in Fig. 1, is based on the complete oxidation or reduction of the gas component to be measured at the working electrode of a solid electrolyte cell resulting in an electronic current. Detectable components are gases existing as ions in the solid electrolyte or are gases which could react with these ions. In case of YSZ as electrolyte material, the reactions of oxygen and hydrogen at the working electrode can be written in the Kröger–Vink notation according to Eq. (1) and (2). O2 (g) + 2V··O (YSZ) + 4e− (Pt) 2OxO (YSZ)
(1)
H2 (g) + OxO (YSZ) H2 O(g) + V··O (YSZ) + 2e− (Pt)
(2)
The amount of analyte molecules which is converted at the working electrode results with Faraday’s law from the electrolysis current I and the residual current IR caused by electronic conductivity. Considering also the flow of the measuring gas dV/dt, the
0925-4005/$ – see front matter © 2012 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.snb.2012.10.111
Please cite this article in press as: M. Schelter, et al., A solid electrolyte sensor for trace gas analysis, Sens. Actuators B: Chem. (2012), http://dx.doi.org/10.1016/j.snb.2012.10.111
ARTICLE IN PRESS
G Model SNB-14744; No. of Pages 6
M. Schelter et al. / Sensors and Actuators B xxx (2012) xxx–xxx
2
electrolysis cell
heater tube made of 8 mole-% YSZ measuring gas working electrode counter/reference electrode reference gas (air)
60
b
a
d
O2
UP
5.7
I c
Fig. 1. Schematic of a coulometric solid electrolyte sensor.
concentration x of the gas component to be measured is determined according to Eq. (3). x=
I − IR VM · zF dV/dt
(3)
F is Faraday’s constant, z is the number of electrons transferred per molecule and VM is the molar volume. Therefore, the lower detection limit of such a coulometric detector depends on the absolute value of IR and its noise, both resulting from the solid electrolyte material properties and the sensor temperature. Due to the fact that YSZ is already widely used as base material for sensors, fuel cells and oxygen generators, the properties of this material are well known [11,12]. The ionic and electronic conductivities of this material in the temperature range between 800 ◦ C and 1200 ◦ C are depicted in Fig. 2 showing the contributions of the electronic conductivity of holes (h) and of electrons (e) [11]. The electronic current IR caused by the hole conductivity flowing through a disc of YSZ with an area A and a thickness d is given by Eq. (4).
⎡
IR =
ART Fd
⎣h (pIO ) 2
⎧ ⎨ pII 1/4 O2
⎩ pIO2
R is the universal gas constant,
⎫⎤ ⎬ −1 ⎦ ⎭ pIO 2
(4)
is the oxygen partial pres-
sure in reference gas and pIIO is the oxygen partial pressure in 2 the measuring gas. Following this equation, a usual coulometric electrolysis cell (A = 1 cm2 , d = 1 mm, pIO = 20.6kPa and pIIO = 2
2
2.7 × 10−4 Pa) operated at 800 ◦ C with thermal fluctuations of ±1 K shows residual current deviations of ±37 nA. Those deviations cause an error of ±26 vol.-ppb in hydrogen detection at the gas flow dV/dt = 10 ml/min. Furthermore, the noise of the electrolysis current is influenced by turbulences of the measuring gas resulting in fluctuations in substance turnover at the working electrode. To get a constant reaction rate, this turbulence has to be avoided or minimized so that the frequencies of the generated noise spectrum are high enough to be 0
log(
S·cm -1)
-2
ion
1.63 10 2 exp ( 0.79eV / kT )
Fig. 3. Schematic drawing of the used coulometric sensor (measures in mm): (a) solid electrolyte tube made of 8 mol-% YSZ, (b) platinum working electrode, (c) platinum reference electrode, (d) measuring gas flow.
filtered out by low-pass filter. This parameter is difficult to predict but can be estimated by utilizing complex simulations taking the electrode morphology into consideration. Due to the above-mentioned advantages of coulometric sensors, they have been used as detectors in gas chromatographic systems already [13,14], where the gained sensitivities resulted in lower detection limits of >10 vol.-ppm. A chromatogram showing the sensor current I as a function of time t, allows to resolve peaks of different gas components. The peak area equals to the amount of charge Q transferred during the oxidation or reduction process of the analyte molecules. The amount of substance n of the gas component can be determined with Faraday’s law according to Eq. (5), where IB is the baseline current. 1 Q n= = zF zF
(I − IB ) dt
(5)
t1
2. Experimental notes The coulometric sensor used in this work is shown in Fig. 3. It consists of a tube made of 8 mol-% YSZ with the outer diameter 5.7 mm and the wall thickness 1 mm. The tubular platinum electrodes on the inner and the outer surface of this tube are 60 mm in length. Between the working (inner) and the reference (outer) electrode a constant potential of −400 mV is applied by a self-made potentiostat. As shown in Fig. 4, two different designs of electronic circuits of the self-made potentiostat were examined when combining the sensor with a gas chromatograph MG1 (SRI Instruments Europe GmbH, Bad Honnef, Germany). Fig. 5 shows a scheme of the measuring system, consisting of a sample injection port, one or more gas chromatographic columns and an YSZ cell acting as detector. A gaseous sample, containing volatile trace components, is injected into a carrier gas flow of the gas chromatograph either by using a sampling loop or by injecting with a syringe through a septum. A column packed with molecular sieve provides the separation of hydrogen, methane and other volatile components of the sample gas mixture within 10 min at 50–200 ◦ C column temperature in Ar carrier-gas flow of 20 ml/min.
1200 °C
-4
+
1000 °C 800 °C
-6
cS
-8
e
1.31 10 7 exp ( 3.88eV / kT )pO1/ 4
2.35 10 exp ( 1.67eV / kT )p 2
h
-3
-6
1/ 4 O
-9
-12
-15
-18
-21
-24
-27
-30
log(pO2 / atm) Fig. 2. Ionic and electronic conductivity of 8 mol-% YSZ depending on oxygen partial pressure at different temperatures according to [11].
UP
cS
RM
I
-10 0
t2
+
+
I
UA UA
UP
+ -
RM
A
B
Fig. 4. Different designs of electronic circuits for controlling the coulometric sensor according to Fig. 3: (A) current–voltage-converter (I = UA /RM ), (B) variable current measuring resistor at the working electrode (I = UA /RM ); coulometric sensor (cS), applied polarization voltage (UP ).
Please cite this article in press as: M. Schelter, et al., A solid electrolyte sensor for trace gas analysis, Sens. Actuators B: Chem. (2012), http://dx.doi.org/10.1016/j.snb.2012.10.111
ARTICLE IN PRESS
G Model SNB-14744; No. of Pages 6
M. Schelter et al. / Sensors and Actuators B xxx (2012) xxx–xxx
log(electrolysis current / µA)
1.1
Fig. 5. Schematic viewgraph of the measuring system consisting of a sample injection system, one or more gas chromatographic columns and a coulometric detector.
For investigations of the electronic conductivity and the noise behaviour of the cell, an electrochemical measuring system REF 600 (Gamry Instruments, Warminster, USA) was used. As measuring gas N2 or laboratory air as well as mixtures of N2 + H2 and N2 + O2 + H2 + CH4 mixed with mass flow controllers (Brooks Instrument Company, Hatfield, USA) were used. Laboratory air served as reference gas. 3. Results and discussion 3.1. Electronic conductivity of the solid electrolyte material The electronic conductivity of a planar disc of 8 mol-% YSZ has been described in literature for temperatures between 800 ◦ C and 1050 ◦ C [11]. To obtain the conductivities for tubular cell geometry of this material at 650 ◦ C and 750 ◦ C, in this work the sensor current was measured at different temperatures at a constant flow of N2 and a constant cell potential of −400 mV. Before these measurements were performed, the whole apparatus was proofed on leak-tightness by a helium leakage test indicating leakage rates lower than 10−10 mbar l/s. Due to the fact that a small amount of residual oxygen in nitrogen leads to a sensor current caused by ionic conductivity, this oxygen concentration was determined to correct the sensor current. Nitrogen was piped through the cell operated at 750 ◦ C at different flow rates. As shown in Fig. 6, the linear increase of the cell current proves that the complete amount of oxygen traces originates from the measuring gas. From the slope of the
electrolysis current / µA
17
15
13
11
I[µA] = 0.14·
dV [ml·min -1] + 8.9 dt
9 0
10
20
30
measuring gas flow /
40
50
ml·min -1
Fig. 6. Electrolysis current of the coulometric sensor at different flow rates of the measuring gas (N2 5.0), sensor temperature: 750 ◦ C.
3
750
measured 1 corrected 2 calculated 3
0.9 0.7 0.5
650 1
0.3
2
0.1 temperature / °C -0.1 0.96 0.98 1.00 1.02
3
1.04
1.06
1.08
1.10
1000·T -1 / K-1 Fig. 7. Arrhenius plots of the electrolysis current, the measured and corrected data at different temperatures compared to the calculated data from [11], measuring gas: 5 ml/min N2 5.0.
linear regression the oxygen content 0.5 vol.-ppm was calculated according to Faraday’s law. The Arrhenius plot of the measured and the corrected sensor currents as well as the data for hole conductivity derived from [11] and calculated with Eq. (4) is shown in Fig. 7. The diagram confirms the expected behaviour that under the chosen experimental conditions the residual current mostly originates from hole conductivity. Furthermore, the electronic conductivity of the used electrolyte material is consistent with the information for YSZ materials given in literature so far. Small differences between the measuring results and the calculated data from literature may come from not quantifiable deviations of the cell geometry as well as systematic errors in temperature measurement. 3.2. Noise behaviour of the coulometric sensor Fig. 8A shows the noise behaviour of the coulometric sensor operated at temperatures between 650 ◦ C and 750 ◦ C. The measurements were taken with the sample rate of 104 s−1 while all electronic filters of the REF 600 measuring system were disabled. These measurements proved that the maximum noise amplitudes depend on the sensor temperature and not on the flow rate of the measuring gas. The noise amplitudes measured are some orders of magnitude larger than the calculated amplitudes caused by thermal noise at the respective working temperature. As depicted in Fig. 8B, there are different maxima in the noise spectrum, for example at 0.4 kHz and at 2.6 kHz. The frequencies of these maxima do not depend on the flow rate which gives a hint that the noise comes due to the connection with the REF 600 and the electrical shield of the cell rather than disturbances in the gas flow. Most applications of coulometric cells require response times of t90 < 200 ms ideally, so that a lowpass filter can be used to eliminate the higher noise frequencies completely. Fig. 9 shows the fluctuations of the electrolysis current due to the control of the sensor heating unit. As it can be seen, the temperature control has in the low frequency range a significant influence on the sensor signal. This noise results completely from electronic conductivity of the solid electrolyte material given in Eq. (4). The fluctuations of the sensor current cannot be eliminated by a lowpass filter. A useful strategy to get rid of this noise influence is to install a heating unit operated at a constant current which fixes the cell temperature to a bandwidth of ±5 K. Afterwards the current signal can be corrected by the value of the real cell temperature.
Please cite this article in press as: M. Schelter, et al., A solid electrolyte sensor for trace gas analysis, Sens. Actuators B: Chem. (2012), http://dx.doi.org/10.1016/j.snb.2012.10.111
ARTICLE IN PRESS
G Model SNB-14744; No. of Pages 6
M. Schelter et al. / Sensors and Actuators B xxx (2012) xxx–xxx
4
0.5
16
signal deviation / %
electrolysis current / µA
20
12 8 4 0 0.00
750 650
temperature / °C
A 0.04
0.08
0.12
0.16
norm. amplitude
0.2 0.1
0.20
2
-0.1 0
15
30
45
60
measuring gas flow rate / ml·min -1
0.6
B
Fig. 10. Deviation of the electrolysis current from the complete hydrogen conversion at the working electrode of a coulometric sensor according to Fig. 3 depending on the measuring gas flow rate and the sensor temperature; hydrogen flow: 5.96 × 10−9 mol/s, electrolysis current: 1.15 mA.
0.4 0.2
3.3. Electrochemical activity of coulometric titration
0.0 0
1
2
frequency / kHz
3
4
Fig. 8. (A) Noise amplitude of the electrolysis current at two sensor temperatures (measuring gas: 5 ml/min N2 5.0). For a better overview, the sensor signal at 650 ◦ C is depicted in foreground from t = 0–0.7 s and the signal at 750 ◦ C from t = 0.7–2 s. (B) Frequency/power density spectrum for the signals from (A).
Depicted by curve 2 in Fig. 9, this method eliminates the noise influence caused by the temperature control. Furthermore, the signal noise is heavily influenced by the design of the electronic control unit of the electrolysis cell. Especially the part of the current measuring unit has huge effect on the noise amplitude. The current measuring circuit shown in Fig. 4A leads to clearly larger noise amplitudes than the circuit shown in Fig. 4B, where a resistor is connected in series to the coulometric sensor. The reason is the feedback of the operation amplifier output to its input resulting in a phase shift and at the worst, causing an oscillation of the potential output at specific frequencies. An example for the effect of the design of the electronic control unit is enumerated in Section 3.4.
15
electrolysis current / µA
1
0.3
0.0
time / s
14 13 1
12
2
11
1 unregulated 2 regulated 10 0
temperature / °C 650 (curve 1) 0.4 750 (curve 2)
250
500
750
1000
1250
time / s Fig. 9. Fluctuations of the electrolysis current depending on the control of the sensor heating unit. Curve 1: unregulated heating with a constant heating current resulting in a temperature bandwidth of 750 ± 5 K. Curve 2: thermocouple controlled heating with fluctuations of about ±1 K.
For a coulometric detector it is necessary to have 100% conversion of analyte molecules at the electrolysis electrode so that Faraday’s law is applicable. Due to the constant potential of the presented electrode, a complete conversion of analyte molecules is given only up to a specific limiting flow rate of measuring gas depending on the diffusion coefficient and the reaction velocity. Only in this gas flow range the coulometric sensor can be used without calibration because the relevant parameters do not influence the sensor signal. As shown in Fig. 10, the results for titration of hydrogen prove that a sensor with the given geometric parameters operated at a temperature of 650 ◦ C shows a decrease in conversion of 0.2% at flow rates greater than 35 ml/min. The reaction rate at 750 ◦ C is high enough to achieve complete conversion of hydrogen in the determined flow range. 3.4. Use of the coulometric sensor as detector for gas chromatography The results of the investigations presented in Sections 3.1–3.3 were used to determine optimal operation parameters for a lownoise measurement of trace components in gaseous mixtures mentioned in Section 2. For the use of coulometric sensors as detectors for gas chromatographic measuring systems, the parameters response time, sensitivity, detection limits and long-term stability are of main interest. Known from investigations on broadband lambda sensors and breath gas sensors, the response time of YSZ solidelectrolyte sensors is short enough for the use in chromatography (t90 < 100 ms) [15,16]. The maximum sensitivity coulometric sensors can reach is given by Faraday’s law, whereas the lower limit of detection is affected by the boundary condition presented in Sections 3.1 and 3.2. Regarding to the long-term stability, coulometric sensors take advantage of being free of calibration due to the fact that with Faraday’s law the amount of substance is determined directly from the amount of charge resulting from the peak area given in Eq. (5). This coherence is valid until the electrodes get inactivated by specific gas components like sulphur or lead. Two chromatograms of gases containing the same components but with different concentrations are compared in Fig. 11. The concentrations calculated from the peak areas by Eq. (5) are given in the diagram. The residual current described in Section 3.1 does not agree to the equation, because it is already included in the baseline
Please cite this article in press as: M. Schelter, et al., A solid electrolyte sensor for trace gas analysis, Sens. Actuators B: Chem. (2012), http://dx.doi.org/10.1016/j.snb.2012.10.111
ARTICLE IN PRESS
G Model SNB-14744; No. of Pages 6
M. Schelter et al. / Sensors and Actuators B xxx (2012) xxx–xxx
5
sensor current / µA
25
2 1
0
CH4
H2 O2
-25
gas concentrations / vol.-ppm curve 1 curve 2 1 10 H2 23 70 O2 10 30 CH4
-50 -75
-100
0
100
200
300
400
500
600
time / s
electrolysis current / µA
50
680 +/- 140 vol.-ppb H2 400 +/- 50 vol.-ppb H2
1 µA 1 2 Control circuit of the coulometric 4A (curve 1) detector according to figure: 4B (curve 2) 60
amount of charge / mAs
4.5 3.6
methane
100
120
time / s Fig. 13. Comparison of hydrogen peaks of the coulometric titration of 1 ml laboratory air measured with a solid electrolyte sensor equipped with two different control circuits according to Fig. 4A and B.
Fig. 11. Chromatogram of two different gas mixtures of hydrogen, oxygen, methane and nitrogen. The concentrations calculated from the measured peaks are given.
10
mA
0
electrolysis current / µA
current IB . The values agree well with the adjusted concentrations of the components within the given accuracy, indicating complete conversion at the working electrode. The behaviour of the gas chromatographic system at different concentrations up to 280 vol.-ppm is depicted in Fig. 12. Minor deviations of the measured oxygen and methane concentrations from the adjusted concentrations are caused by the used mass flow controllers. The resolution is solely limited by the high-frequent noise described in Section 3.2. Therefore, a signal filtering with an optimized limiting frequency is crucial for the optimal lowering of the limit of detection. The hydrogen peaks of the coulometric titration of 1 ml laboratory air, shown in Fig. 13, were derived with the chromatographic measuring system described in Section 2 equipped with a silica gel column and a molecular sieve column. The peak derived with the electronic circuit according to Fig. 4A shows a significant higher noise than the one taken with the circuit shown in Fig. 4B, while the precision increases by the factor 3. The higher absolute concentrations of hydrogen, depicted by curve 1, result from other experiments with hydrogen taken near by the chromatographic measuring system. The results show that this detection principle can be used for the determination of analyte concentrations smaller than 500 vol.-ppb. Corresponding to Eq. (5), the detection limit decreases linearly with increasing number of electrons z exchanged during electrode reaction. This is valid, e.g.,
80
2.328 0.004 2.328+/-+/0.004
-10
vol.-% O2 -20 -30 0 0
100
200
300
400
change of measuring range
330 +/- 70 vol.-ppb CH4
-10
-20
3.5 +/- 0.2 vol.-ppm H2 -30 0
100
200
300
400
time / s Fig. 14. A chromatogram of a gas mixture of hydrogen, oxygen, methane and nitrogen depicted in two different resolutions showing the ability of resolving very small amounts of hydrogen and methane with comparably much larger amounts of oxygen.
for gas components like hydrocarbons, which give more electrons than hydrogen. One more remarkable advantage of the current measuring unit is the implemented range selection over six decades. As shown in Fig. 14, very small amounts of hydrogen and methane can be resolved reliably besides much larger amounts of oxygen. 4. Conclusions
oxygen
2.7 1.8
hydrogen
0.9
dots = measured values lines = calculated 0.0 0
40
80
120
160
200
240
280
concentration / vol.-ppm Fig. 12. Peak areas and amounts of charge derived from chromatograms of different concentrated gas mixtures in comparison to the values calculated by means of Faraday’s law.
In this work, investigations on properties of coulometric solid electrolyte sensors for continuous gas flows are presented. According to Faraday’s law the signal of those sensors depends only on concentration of the component to be determined and flow rate, allowing calibration free operation. The parameters influencing the lower detection limit of those coulometric sensors were determined. One of these parameters is the residual current caused by electronic conductivity of the solid electrolyte material, which depends on temperature. It was shown that the temperature dependence of the electronic conductivity calculated from the measured residual current of the presented sensor is consistent with data presented in literature. Consequently, the sensor current was real-time corrected successfully
Please cite this article in press as: M. Schelter, et al., A solid electrolyte sensor for trace gas analysis, Sens. Actuators B: Chem. (2012), http://dx.doi.org/10.1016/j.snb.2012.10.111
G Model SNB-14744; No. of Pages 6
ARTICLE IN PRESS M. Schelter et al. / Sensors and Actuators B xxx (2012) xxx–xxx
6
by subtracting the residual current calculated from the measured temperature value. Significantly higher effort was made for decreasing the noise of the cell current, which covers the frequency band from a few mHz to some kHz. Low-frequency fluctuations originate from temperature oscillations caused by controlling the heater with a thermocouple-assisted controller. This low-frequency noise was eliminated by uncontrolled heater operation coupled with the above mentioned signal correction based on temperature measurement. High-frequency noise caused by ambient influences like electrical fields was eliminated by the application of low-pass filters. The limiting frequency of the filter was adjusted according to the requirements of optimum chromatographic peak resolution. Additionally, the noise amplitude of the sensor current is significantly influenced by the design of the electronic circuitry for the sensor control. It was found that particularly the high-frequency noise is amplified by a current follower originally used in the working electrode circuit. That noise amplification was successfully eliminated by replacing the current follower with a resistor for current measurement in the working electrode circuit. By this means the detection limit of coulometric flow-through sensors as detectors for gas chromatography in trace gas analysis could be decreased from some 10 vol.-ppm to the ppb-range. As an example, the hydrogen concentration in laboratory air of about 400 vol.-ppb was measured with an accuracy of ±50 vol.-ppb. Acknowledgements The presented work is based on projects supported by the German Federal Ministry for the Environment, Nature Conservation and Nuclear Safety (funding reference number 03KB045) and the German Federal Ministry for Food, Agriculture and Consumer Protection (funding reference number 22011110). The authors are responsible for the content of the paper and thank gratefully for financial support. Furthermore, the authors would like to thank SRI Instruments Europe GmbH, Bad Honnef, Germany, and Zirox – Sensoren & Elektronik GmbH, Greifswald, Germany, for professional und device-related support. References [1] A. Amann, G. Poupart, S. Telser, M. Ledochowski, A. Schmid, S. Mechtcheriakov, Applications of breath gas analysis in medicine, International Journal of Mass Spectrometry 239 (2004) 227–233, http://dx.doi.org/10.1016/j.ijms. 2004.08.010. [2] W. Miekisch, J.K. Schubert, From highly sophisticated analytical techniques to life-saving diagnostics: technical developments in breath analysis, Trends in Analytical Chemistry 25 (2006) 665–673, http://dx.doi.org/10.1016/j.trac. 2006.05.006. [3] J. Fouletier, M. Bonnat, J. Le Bot, S. Adamowicz, Calibration of a highly sensitive oxygen analyzer for biological applications using an oxygen pump, Sensors and Actuators B 45 (1997) 155–160, http://dx.doi.org/10.1016/S09254005(97)00289-X. [4] K. Katahira, H. Matsumoto, H. Iwahara, A solid electrolyte hydrogen sensor with an electrochemically-supplied hydrogen standard, Sensors and Actuators B 73 (2001) 130–134, http://dx.doi.org/10.1016/S0925-4005(00)00672-9. [5] V.S. Sevast’yanov, E.M. Galimov, N.E. Babulevich, A.A. Arzhannikov, Zirconium dioxide-based solid electrolyte as a means of oxidation of organic compounds during isotopic assay of carbon, Russian Journal of Electrochemistry 43 (2007) 448–453, http://dx.doi.org/10.1134/S102319350704012X.
[6] L. Heyne, Solid electrolyte cells, in: H. Degn, et al. (Eds.), Measurement of Oxygen, Elsevier, Amsterdam, 1976, pp. 65–69. [7] C. Franx, A dynamic oxygen sensor with zero temperature coefficient, Sensors and Actuators 7 (1985) 263–270, http://dx.doi.org/10.1016/02506874(85)80006-8. [8] P. Butschbach, F. Hammer, H. Kohler, A. Potreck, T. Trautmann, Extensive reduction of toxic gas emissions of firewood-fueled low power fireplaces by improved in situ gas sensorics and catalytic treatment of exhaust gas, Sensors and Actuators B 137 (2009) 32–41, http://dx.doi.org/10.1016/j.snb. 2008.12.007. [9] K. Teske, H. Ullmann, N. Trofimenko, Thermal analysis of transition metal and rare earth oxide system–gas interactions by a solid electrolyte-based coulometric technique, Journal of Thermal Analysis and Calorimetry 49 (1997) 1211–1220, http://dx.doi.org/10.1007/BF01983677. [10] V. Vashook, L. Vasylechko, J. Zosel, W. Gruner, H. Ullmann, U. Guth, Crystal structure and electrical conductivity of lanthanum–calcium chromites–titanates La1–x Cax Cr1–y Tiy O3 (x = 0–1, y = 0–1), Journal of Solid State Chemistry 177 (2004) 3784–3794, http://dx.doi.org/10.1016/j.jssc.2004.07.007. [11] J.-H. Park, N. Blumenthal, Electronic transport in 8 mole percent Y2 O3 –ZrO2 , Journal of the Electrochemical Society 136 (1989) 2867–2876, http://dx.doi.org/10.1149/1.2096302. [12] H. Ullmann, Keramische Gassensoren, in: Grundlagen-Aufbau-Anwendung, Akademie-Verlag, Berlin, 1993. [13] K. Teske, P. Popp, J. Baumbach, Solid-state coulometric cell as detector for gas chromatography, Journal of Chromatography 360 (1986) 417–420, http://dx.doi.org/10.1016/S0021-9673(00)91690-X. [14] B.K. Zuev, A.Y. Olenin, Solid electrolyte sensor as a detector for gas chromatographic determination of combustible contaminants in air, Journal of Analytical Chemistry 61 (2006) 147–152, http://dx.doi.org/10.1134/ S1061934806020109. [15] E.M. Logothetis, Automotive Oxygen Sensors, in: N. Yamazoe (Ed.), Chemical Sensor Technology, 3, Elsevier, Amsterdam, 1991, p. 96. [16] J. Zosel, D. Tuchtenhagen, K. Ahlborn, U. Guth, Mixed potential gas sensor with short response time, Sensors and Actuators B 130 (2008) 326–329, http://dx.doi.org/10.1016/j.snb.2007.08.008.
Biographies Matthias Schelter received his Master of Science in chemistry from the Dresden University of Technology in 2011. Since 2011 he has been working at the Meinsberg Kurt-Schwabe Research Institute where he is doing currently his PhD thesis on the development of solid electrolyte sensors for dissolved gases in complex media. Jens Zosel received his diploma in physics from the University of Greifswald in 1990 and his PhD from the University of Freiberg in 1997. Since 1992 he has been working at the Meinsberg Kurt-Schwabe Research Institute. His basic research interests are directed towards the behaviour of electrochemical sensors in liquid and gaseous flows and the development of solid electrolyte sensors for different applications. Wolfram Oelßner received his diploma in nuclear technology from the Dresden University of Technology in 1962 and his PhD in electrochemical measuring technique from the same University in 1968. Since 1963 he has been working at the Meinsberg Kurt-Schwabe Research Institute, where he now acts as a scientific consultant. In 1989 he was appointed to a lecturer for physical–chemical measuring technique as well as for environmental measuring technique and corrosion research at the Dresden University of Technology. His research activities are directed towards the development of electrochemical sensors and measuring instrumentation. Ulrich Guth received his PhD from the University of Greifswald in 1975. In 1993 he became a professor for solid state chemistry at the University of Greifswald. From 1999 to 2010 he had been working as the director of the Meinsberg KurtSchwabe Research Institute and as a professor for physical chemistry especially sensor and measuring technology at the Dresden University of Technology. His principal research interests are directed towards solid electrolyte sensors and new materials for these high temperature sensors. Michael Mertig received his diploma and PhD in physics from the Dresden University of Technology. His basic research interests are biomimetic materials synthesis as well as electrochemical and biological sensors. Since 2010 he is the director of the Kurt-Schwabe-Institute in Meinsberg.
Please cite this article in press as: M. Schelter, et al., A solid electrolyte sensor for trace gas analysis, Sens. Actuators B: Chem. (2012), http://dx.doi.org/10.1016/j.snb.2012.10.111