Stability and reproducibility of solid electrolyte amperometry sensors at the analysis of hydrogen in nitrogen-containing gas mixtures

This paper illustrates the results of long-term tests on the stability of the output signal of the solid electrolyte amperometry sensor when measuring the hydrogen concentration in the H 2 + N 2 gaseous mixture. The obtained experimental data verify the stability and reproducibility of the sensor output signal for hydrogen concentration measurements in the nitrogen-containing gaseous mixture during > 8000 h of operation. The output signal drift, i.e., the limiting current value, was insignificant, less than ± 5 %. The sensor operation was performed at 3 temperature shifts with different time intervals; these changes did not have any impact either on the sensor integrity or on its operation. The structure of the solid electrolyte sensor, intermediate solid electrolyte / electrode layer and electrodes did not undergo any significant changes during operation. The dynamic characteristics of the sensor, the response time in particular, remained stable during the operation.


Introduction
Solid electrolytes with oxygen-ionic conductivity are widely used in sensors to analyze the content of free oxygen in the gas media and metallic melts [1][2][3][4][5][6][7][8][9].The majority of published research on solid electrolyte sensors is devoted to the study of such sensors with the mixed potential based on oxygen-conducting solid electrolytes with single gas space [10][11][12][13][14][15].When detecting the gas, for instance hydrogen in various gas media, the occurring processes result in the deviation of the concentration dependence from the Nernst dependence.This effect may be caused either by the occurrence of several simultaneous potential determining processes or by the accumulation of the process products on the three-phase boundary or by slow adsorption on the three-phase boundary.
The dependence of the generated EMF on the concentration of the inflammable compound in the analyzed gas of such sensors is described by the following equation: where  is an empirically derived constant.The value of the appeared mixed potential is determined mainly by a number of factors including the concentration and composition of the analyzed gas media, the material of the measuring electrode and operation temperature.
Potentiometric sensors based on solid electrolytes of the YSZ composition are the most widely known.These sensors, based on the solid electrolytes with the unipolar oxygen conductivity, proved to be efficient in the heat engineering, metallurgy, chemistry and motor industry.They are characterized by ability to determine wide oxygen concentration range, rapid operation, simplicity of the construction and simple processing of the obtained electrical signal, i.e., EMF, to the hydrogen concentration.The serious drawback of potentiometric sensors is that an etalon gas with known and stable oxygen concentration is required for accurate sensor operation.This fact makes the application of the sensor more complicated, as in some cases a separate etalon gas supply is hindered [16][17][18][19][20][21][22].
Recently solid electrolyte amperometric sensors with diffusion barrier aimed at the analysis of oxygen, hydrogen and moisture concentrations in inert and oxidizing gas media have been actively analyzed in a published literature [23][24][25][26][27][28][29].These sensors are typically composed of YSZ solid electrolytes.The data on the most promising amperometric sensors design, composition of electrolytes, electrode materials, the most appropriate operating temperatures for the data analysis and etc. are accumulated in the obtained results of the published works [30][31][32][33][34][35].The major part of the research is devoted to the determination of the hydrogen content in gas mixtures.
Among the vast published literature data, we would like to elucidate the results on the evaluation and practical usage of amperometic sensors based on solid electrolytes with proton conductivity by Katahira et al. [36].The authors developed and tested complex sensors composed of two electrochemical cells: hydrogen pump and potentiometric sensor based on the SrCe0.95Yb0.05O3electrolyte.Development of amperometric sensors design and detailed analysis of their characteristics are illustrated in papers by Yajima et al. [37,38].
Due to the growing interest towards amperometric sensors with diffusion barriers, our research group evaluated the long-term operation stability of the output signal of the sensor based on the proton solid electrolyte of the CaZr0.9Sc0.1O3-δcomposition to determine the hydrogen concentration in its mixture with nitrogen.

Samples preparation
In the present paper an amperometric sensor with a diffusion barrier, i.e. an electrochemical cell made on the basis of the proton-conducting solid electrolyte of the CaZr0.9Sc0.1O3-δcomposition, was studied.The sensor schematic and appearance are presented in Figures 1  and 2. It is made of two solid electrolyte plates of 25 mm length, 12 mm width, and 1 mm thickness.One of the plates is concave.The depth of the concave is 0.5 mm and diameter is 6 mm.Porous platinum electrodes with a platinum wire current lead of 0.1 mm in diameter are applied to the opposite surfaces of the solid electrolyte concaved plate.A ceramic capillary of 257 µm inner diameter and 20 mm length was located between the  electrolyte plates.The plates were glued by a thermoresistant glass.To provide a constant voltage on the electrochemical cell we used a GPS-18500 source of a constant current; the value of current passed through the cell was measured by a GDM-8246 type multimeter.We used H2 (2 %) + nitrogen as a test gas mixture.
Constant current voltage was applied to the sensor electrodes with different polarity to provide pumping hydrogen from the sensor inner cavity to the analyzed gas flow.Via the sensor capillary the analyzed gas mixture input and the output nitrogen from the cavity exchanged.When a definite value of the input voltage was reached these processes became equilibrium.This is testified by the appearance of the limiting current.The value of the limiting current allows calculating the amount of hydrogen in the analyzed gas according to Equation (2).
where 2 is the amount of electricity necessary for the transfer of one mole of hydrogen;  is the total pressure of the analyzed gas;  is the diffusion coefficient of hydrogen in nitrogen;  and  denote the cross section area and length of the capillary diffusion channel;  is the gas constant;  is the absolute temperature;   2 is the hydrogen concentration (volume fractions).

Experiment procedures
The electrochemical sensor, illustrated in Figure 1, was placed into a tube furnace with a nichrome heater.The volume of the tube furnace was 0.65 l.The temperatures in the furnace were maintained with the accuracy of ± 3 °C using a thermoregulator TP 703 Varta (Laboratory of ceramics Ltd., Moscow, Russia).The analyzed gas mixture was blown through the inner furnace area; the gas consumption was 20 ml/min.The gas mixture of the set composition (H2 (2 %) + nitrogen) was additionally purified from moisture and admixtures (zeolite).Gas batching was performed using the F-201C-33-V type gas consumption regulator.The sensor operated in the pulse mode, i.e., during the whole experimental period there were four cooling-heating cycles to verify the sensor thermal resistance.

Results and discussion
When the sensor was heated to 500 °C, the constant current voltage was applied to the electrodes.As the applied voltage increased, the current passing through the sensor solid electrolyte plate increased.When a definite value of the applied voltage was reached, the current reached the desired value and stabilized; this current value was considered to be the limiting current.At each hydrogen concentration in the analyzed gas mixture and under otherwise equal conditions including diffusion barrier parameters, temperature, gaseous mixture component composition, and pressure, the limiting current was constant.Figure 3 illustrates voltammetry characteristics of the sensor during the determination of the hydrogen content at the temperatures of 500 and 550 °C.It is seen that the limiting current appears at the voltage of about 1 V.The measured limiting currents of the sensor, observed for gas mixtures containing from 0 to 4 % of hydrogen, allowed obtaining a linear dependence of the limiting current on hydrogen, presented in Figure 4.
Apart from the sensor characteristics illustrated in Figures 3 and 4, we recorded dynamic characteristics of the sensor before resource and stability tests (Figure 5).The time of the sensor reaction and sensor signal of 90 % of the nominal value were determined.The time of the sensor initial response varied from 3 to 5 seconds, and the time of the sensor signal output of the 90 % nominal value was from 70 to 100 seconds (these values were obtained using the transport delay).The dynamic characteristics of the sensor were evaluated after 4-thousand-hour operation and after 8 thousand hours (Figures 6 and 7). Figure 8 illustrates the dependence of the sensor limiting current changes on the hydrogen concentration in air during 8000 hours.During the tests, there were 3 heating-cooling cycles of the different time length.These thermal changes did not cause any destruction, decompression, or fault in the reproducibility of the sensor measurements.The sensor demonstrated a stable operation and well-reproducible results.The absolute hydrogen measurement error ranged from 0 to 4 % and was ± 0.3, which is satisfactory for hydrogen analyzers, based on different measuring modes.For instance, the hydrogen analyzer in gases HY-OPTIMA by the Artvik Company (Almaty, Kazakhstan) is based on the analysis of the regularities of hydrogen dissolution in metals and has an absolute measurement error of ± 3 mA.The amperometric hydrogen analyzer in the inert gas EHDV-G produced by FSUE 'NII NPO "Luch" (Podolsk, Russia) is based on liquid electrolyte and has a measurement error of ± 15 %.

Conclusions
The stability of the amperometric solid electrolyte CaZr0.9Sc0.1O3-sensor with the diffusion barrier has been studied for measuring the concentration of hydrogen in the hydrogen-nitrogen gas mixture during 8000 hours.The sensor demonstrated stable operation and good data reproducibility.The performed 3 heatingcooling cycles did not cause any faults in the sensor operation.The sensor characteristics were found to be promising for its practical implementation.

Supplementary materials
No supplementary materials are available.

Funding
The present research was performed within the budget plan of the Institute of High Temperature Electrochemistry.

Figure 2
Figure 2 General view of the sensor before the tests (the sensor appearance after 8000 h tests remained unchanged).

Figure 3
Figure 3 Dependence of the sensor current on the applied voltage.Hydrogen concentration in the gas mixture was 1.4 %.

Figure 4
Figure 4 Dependence of the sensor limiting current on the hydrogen concentration.

Figure 5
Figure 5 Dependence of the sensor dynamic characteristics on the hydrogen concentration change at the beginning of the tests.

Figures 6 ,
7 and 8 illustrate that the sensor dynamic characteristics remain almost stable during the whole test period.These figures verify good reproducibility of the sensor results.The deviations the limiting current values during the tests were ± 5 % of the average value.

Figure 6
Figure 6 Dependences of the sensor dynamic characteristics on the hydrogen concentrations in nitrogen after the 4000-hour operation.

Figure 7
Figure 7 Dependences of the sensor dynamic characteristics on the hydrogen concentrations in nitrogen after the 8000-hour operation.

Figure 8
Figure 8 Changes in the sensor limiting current during the tests.