Devanathan and Stachurski (DS) proposed an electrochemically sensitive system with instantaneous recorder of hydrogen permeation rate in an electrolytic process through a metallic foil, sheet or membrane. The specimen of a metal membrane is clamped to two separate cells. Its structural model is shown in Fig. 1^[30].

Fig. 1

	Figure Option ViewDownloadNew Window
	Fig. 1 Schematic structural model of Devanathan-Stachurski double cell. C: cathodic side of working electrode; A: anodic side of working electrode; PA, PC: auxiliary electrode; r: reference electrode; W: working electrode; R: variable resistor.

One side of the membrane acts as the cathode and the other as the anode. In Fig. 1, the left cell (input side) contains a solution with high concentration of hydrogen ions like H₂SO₄ solution, while in the right side cell (exit side), it is necessary to have a basic solution (usually KOH) in order to avoid corrosion on the steel surface which affects hydrogen permeation detection. Previous results have indicated that it is necessary to cover the membrane with palladium or nickel to avoid corrosion and to provide catalysis for a complete oxidation of diffused hydrogen atoms^{[31], [32], [33] and [34].} The mechanism of I-sensor could be expressed as follows:

Anodic reaction: H + OH⁻ → H₂O + e⁻

Cathodic reaction: H₂O + e⁻ → OH⁻ + H

Cell reaction: H → 1/2H₂

Hydrogen atoms penetrate through cathodic Ni or Pd foil in sensor and are recombined to H₂ on the external surface of sensor and to escape to the atmosphere finally.

The distribution of hydrogen in steel profiles could be calculated by Fick's first diffusion law according to the permeation rate of hydrogen from the internal surface in the reactor steel wall to the nickel electroplated interface in the sensor.

In the steady state, the H concentration at inner wall of metal membrane, i.e, entrance side, c₀ (mol cm^{− 3}), could be calculated from the following equation:

equation(1)

where i_∞ (A cm^{− 2}) is the permeation current density at the steady state; L (cm) is the specimen thickness; F is the Faraday constant (96500 A s); D (cm² s^{− 1}) is the hydrogen diffusion coefficient in the metal specimen, calculation with 6.6 × 10^{− 5} cm² s^{− 1} for iron membrane at 298 K ^[35].

The E-sensor could be described in terms of the following galvanic cell:

(− )Ni(H)|0.2 mol dm^{− 3} KOH|NiO₂(s)(+)

Anode reaction: 2H + 2OH⁻ = 2H₂O + 2e⁻

Cathode reaction: NiO₂(s) + 2H₂O + 2e⁻ = Ni(OH)₂(s)+2OH⁻

The overall cell reaction may be represented as: NiO₂ (s) + 2H = Ni(OH)₂ (s).

The electromotive force generated when the cell anode is exposed to hydrogen atoms, may be represented by the Nernst's equation:

equation(2)

The concentration of hydrogen in the metal could represent the activity when the concentration of hydrogen is very low, then a_H = c_H (c_H is the concentration of hydrogen atoms at the interface of catalytic coating contacting with electrolyte). Because Ni(OH)₂ and NiO₂ are soaNi₂(OH)lid, so

and a₂NiO are both considered as 1. According to the Fick's second law and through Laplace transform, the concentration of hydrogen atoms in the inner wall of metal membrane (c₀, mol cm^{− 3}) could be expressed as:

equation(3)

where E_∞ and E^Ө is the measured steady potential and standard electromotive force, respectively (Checking the standard electrode potential table we can know that E^Ө is 0.49 V); F is the Faraday constant (96500 A s); R is the gas constant (8.314 J mol K^{− 1}); T is the experimental temperature (298 K).

The iron membranes through which hydrogen was diffused were cut from commercial α iron sheets of 0.40 mm in thickness. Because this work focused on investigating the performance of the amperometric and potentiometric sensors, so permeating hydrogen source through iron or steel specimen was not restricted. In order to shorten the experimental period, α iron was chosen as the permeation specimen. The chemical composition of the α iron was (wt%): 0.012 C, 0.60 Mn, 0.05 S, 0.05 P and balance Fe. The membrane specimens were 150 mm × 150 mm in size. In order to acquire highly catalytic activity of H oxidation, the specimen surface was pretreated and electroplated. Indeed, it is difficult to electroplate Pd coating as well in the field. In literature^[36], it has been reported that nickel coating could replace Pd coating as an alternative. Using Ni coating can save costs, and thus improve economic efficiency. Another aspect to be outlined is that 1-2 μ m Ni catalytic coating is much thinner that the specimen and the steady data were obtained through detection. So we did not consider the H diffusion time through Ni coating basically. So nickel plating was chosen as the catalytic coating in the sensor in our researches.

The following pretreatment of specimens was described: mechanically grounded with carbimet (SiC) grinding paper of 1200 grit → cleaned with alkaline solution (50 g dm^{− 3} NaOH + 10 g dm^{− 3}Na₃PO₄· 12H₂O) → rust was removed with HCl solution (90 g dm^{− 3}) → electroplated nickel (in the plating solution composed of 215 g dm^{− 3} NiSO₄· 7H₂O, 11 g dm^{− 3} NaCl, 33 g dm^{− 3} H₃BO₃, 25 g dm^{− 3}Na₂SO₄, 35 g dm^{− 3} MgSO₄, pH 5-5.5, Room temperature, with 15 mA cm^{− 2} cathodic current) → weathered. Washing with distilled water was performed between each step.

The charging and detection setup for hydrogen permeation are schematically shown in Fig. 1. Cathodic polarization of the entrance side was achieved in a galvanostatic way. The cathodic compartment was filled with 0.2 mol dm^{− 3} NaOH + 1% Na₂S solution. Na₂S (in fact as the form of H₂S) was a poison agent for H atoms to combine into H₂ on the cathode side and made more H atoms permeate into specimen. Although Fe reacted with S²⁻ to form FexSy passive film, we can remove it each experiment via polishing in order to gain a good reproduction of charging hydrogen. The testing membrane was respectively charged with 0.2, 0.4, 0.6, 0.8, 1.0 mA cm^{− 2} cathodic current density. There was only 1 cm² nickel coating contacted with electrolyte of the sensor. All the experiments were carried out at 298 ± 1 K and the permeation curves were measured by a computer-controlled electrochemical working station (Interface 1000 Potentiostat/Galvanostat/ZRA, American Gamry instruments, Inc.).

Nickel was plated on iron sheet for 5 min at 298 K with the current density of 1 A dm^{− 2} at the working side of the sensor according to our previous research^[37]. 0.30 V_Ni (versus Ni electrode) was imposed as the oxidation potential of diffused hydrogen atoms. Hydrogen was charged with the cathodic current density of 0.2, 0.4, 0.6, 0.8, 1.0 mA cm^{− 2} at room temperature (298 K), respectively. Fig. 2 shows the dependence of the steady values (i_∞ and E_∞ ) measured with two types of sensors on charging current density, respectively. When the charging current density was 0.2 mA cm^{− 2}, the response of steady current density only reached 17.3 μ A cm^{− 2} and the E-sensor reached a steady potential of − 170 mV. When hydrogen atoms were continuously accumulated on the cathode chamber, H permeation rate through the Fe membrane increased. It demonstrated that the measured steady current gradually increased with increasing the charging current density. The potential measured by the E-sensor was reduced because of the permeated hydrogen in the iron membrane^{[38] and [39]}. With the rise of charging current density, permeated hydrogen atoms in the iron also increased and the steady potentials became more negative. Nevertheless, the potential decrements turned small. This change of signal could reflect the adsorption state of the hydrogen atoms on the entrance side in the charging chamber. The excessive hydrogen atoms were not adsorbed on the charging side of iron membrane, and a part of hydrogen atoms would combine into hydrogen gas, then escape from the cathodic solution. Hydrogen gas did not permeate into the iron membrane. Under the charging current density of 0.2 mA cm^{− 2}, i_∞ and E_∞ were low. Both i_∞ andE_∞ reflected the state of hydrogen atoms adsorbed on the charging side of iron membrane in the cathodic room.

Fig. 2

	Figure Option ViewDownloadNew Window
	Fig. 2 Steady signal values from I-sensor and E-sensor at various charging currents.

The H concentration at inner wall of metal membrane could be calculated through Eq. (1) and Eq. (3). The H concentration and measurement deviation by two sensors are shown in Table 1.

Table 1

Table 1

Table 1 H concentration on entrance side of Fe membrane and measurement deviation by two sensors Charging current density (mA cm− 2) i∞ (μ A cm− 2) c0, aa (mol cm− 3) E∞ (mV) c0, pa (mol cm− 3) σ Rb 0.2 17.3 1.1 × 10− 7 − 170 9.5 × 10− 8 13.6% 0.4 43.6 2.7 × 10− 7 − 194 2.8 × 10− 7 3.7% 0.6 49.4 3.1 × 10− 7 − 197 3.0 × 10− 7 3.2% 0.8 52.6 3.3 × 10− 7 − 200 3.2 × 10− 7 3.0% 1.0 64.6 4.1 × 10− 7 − 205 4.3 × 10− 7 4.9%

Table 1 H concentration on entrance side of Fe membrane and measurement deviation by two sensors

A c_{0, a} and c_{0, p} are the H concentrations at inner wall of metal substrate which are detected by I-sensor and E-sensor, respectively. i∞ is the permeation current density at the steady state, E_∞ is the measured steady potential.B Relative deviation σ R=|c0, a− c0, p|/c0, a× 100%.

The H concentration was raised with increasing the charging current density. The H concentration c₀calculated by two equations lay in the same order of magnitude. The relative deviation ranged from 3.0% to 13.6%. The H concentrations calculated with the response signals of two types of sensor were slightly different.

4.2.1. Response time of two types of sensors

The response time is defined as the corresponding time when the response signal reaches 90% of the steady-state value^[40]. Fig. 3 shows the curves of response time via two types of sensor under different charging conditions. As the charging current density was raised, the response time of the both sensors turned shorter. It slowly decreased when the charging current density was raised from 0.6 to 1.0 mA cm^{− 2} because cathodic chamber could provide more hydrogen atoms to penetrate the sample membrane. I-sensor exhibited a faster response rate than E-sensor as shown in Fig. 3. The difference value (D-value) of response time of the two types of sensor also decreased with the charging current density. Response time of the E-sensor is at least 1 h longer than that of the I-sensor under the same hydrogen permeation condition because the equilibrium established by E-sensor needed a longer time. For the I-sensor, hydrogen atoms could be quickly and completely ionized when they diffused to the active coatings, so the response rate of the I-sensor would be faster.

Fig. 3

	Figure Option ViewDownloadNew Window
	Fig. 3 Contrast curve of the response time under different charging currents.

4.2.2. Reproducibility of two types of sensors

Under the same charging condition, the stability and reproducibility of the two type sensors was examined. After reaching a steady current or potential, charging was stopped in order for the sensor to recover to the original baseline. Then it was charged again and the hydrogen permeation curve was measured. After five cycles of charging on and off, the experimental results are shown in Fig. 4. It demonstrated a good stability and reproducibility of the signal. Fig. 4 also shows the long-term performance of the I-sensor and E-sensor under the same charging condition. After five cycles of charging on and off, it was seen that both of the two type sensors have good reproducibility and long-term stability.

Fig. 4

	Figure Option ViewDownloadNew Window
	Fig. 4 Hydrogen permeation curve after five cycles of charging on-off experiment at 298 K with I-sensor and E-sensor, respectively (charged with 0.8 mA cm− 2).

An understanding and comparison in the performance of the two types of sensors would be important in guiding the selection and development of the most appropriate technology for a specific application of hydrogen sensor, in particular for a future hydrogen sensor used at high temperature. A summary of the general performance specifications which include cost, installation considerations, and operational requirements for each type of sensor was compared in the following parts. The current signals from I-sensor are of a high stability, high precision, so it could work in a long-term. However, the application of a potentiostat instrument makes the testing equipment or the design of the sensor more expensive. Inside the I-sensor, the open vent for discharging hydrogen gas produced from the cathode is easy to make the liquid solution in the sensor evaporate and shorten the running time of I-sensor. The application of I-sensors is also limited in some working environments without on-site larger power or a permit of electrical wires for the potentiostat instrument. E-sensors differ from I-sensors in that E-sensors can ideally run without external power supply, which only measure the potential difference or electromotive force between the sensing electrode and the reference electrode. Unlike I-sensors, the potential signal from an E-sensor is nearly independent of sensor size and geometry, which is an advantage from a view point of sensor miniaturization. However, potential signal measured by E-sensors is a logarithmic dependence of hydrogen concentration, which results in a lower accuracy at high concentrations as compared with that of the I-sensors for linear response of current. The response of the I-sensor is faster than that of the E-sensor. Moreover, the I-sensor has been successfully applied in engineering field at room temperature.

I-sensors used for field application are only composed of an anodic compartment in the DS cell. A thin layer of nickel (or palladium) was plated on the surface of the external wall of the equipment as the catalytic coating to oxidize hydrogen. Because the sensor is fixed on the equipment walls, the leakage of the liquid electrolyte in the sensor is a potential engineering problem. The accuracy of hydrogen measurement not only depends on the design of the sensors, but also on the reliable installation technique of the sensors. It is a best option to use high conductive and non-flowing electrolyte inside the sensor from a point of leakage for liquid electrolyte. Colloidal electrolyte could solve the leakage problem when temperature of the equipment is lower than 323 K. The hydrogen permeation curves with the colloidal electrolyte and the liquid KOH electrolyte in the sensor developed in our group^[37], are shown inFig. 5. It exhibited the same steady permeation current with two types of electrolyte in the sensor under the charging current density of 0.4 mA cm^{− 2}. The colloidal electrolyte sensor could effectively prevent liquid electrolyte from the leakage in the gap between the wall surface of the equipment and sensor.

Fig. 5

	Figure Option ViewDownloadNew Window
	Fig. 5 Permeation curves in different electrolytes.

As compared with liquid electrolytes, solid polymer electrolytes have the advantages against the leakage, corrosion and volatilization. Nafion (Dupont, USA), as one of the best proton conductors, has been widely used in fuel cells^{[41], [42] and [43]} and electrochemical sensors which responses hydrogen gas at 323-473 K^[44]. The electrochemical hydrogen sensors composed of ceramic solid electrolytes (5 mol% Yb₂0₃-SrCeO₃) could measure H permeation at elevated temperatures (around 723 K) ^[45]. It was actually a hydrogen gas sensor but not a really hydrogen permeation sensor because there is a space between the sensor and the surface of equipment wall. So the sensor was not conformed to the detection principle of the DS cell. It also had the problem of hydrogen gas leakage if sealing was poor.

The joint between the sensor and surface of the equipment wall not only needs a high electrical conductivity of solid electrolyte, but also requires a strong adhesion between them. If a kind of proton conductor adhesives which contains two functions described above is developed, it could effectively solve the installation problems of I-sensor and improve the accuracy of measurements. At normal temperature, colloidal electrolyte could be employed to solve the leakage problem of the sensor. Nevertheless, the proton conductor adhesives used at elevated temperatures could meet the above requirements, so it is a promising solid electrolyte used for the hydrogen sensor at high temperature.