Development of an electrochemical balance to measure quantitatively hydrogen generation during electrochemical processes

A new method to measure the amount of hydrogen generated from the surface of an immersed electrode is presented in this work. The method consists of a mechanical balance with a horizontal arm attached to two hydrogen-collecting containers that are submerged in two independent electrochemical cells. One of the two cells (the test cell) contains the electrode of interest, generating an unknown amount of hydrogen, whereas the other cell (the measurement cell) contains an inert electrode that is used to evolve an amount of hydrogen equal to that generated in the test cell, such as the mechanical equilibrium between the sides of the balance is constantly maintained. Adequate electrical connections and circuitry ensures that, as hydrogen is evolved from the electrode of interest in the test cell, the displacement of the balance arms activates an electrical contact, which triggers the hydrogen evolution in the measuring cell. Once a sufficient amount of hydrogen is evolved from the measuring cell, the horizontal arm is displaced in the opposite direction and the electrical contact to the measuring cell is interrupted. The measurement of the current flowing through the measuring cell enables precise estimation of the amount of hydrogen generated in the test cell. © The Author(s) 2017. Published by ECS. This is an open access article distributed under the terms of the Creative Commons Attribution 4.0 License (CC BY, http://creativecommons.org/licenses/by/4.0/), which permits unrestricted reuse of the work in any medium, provided the original work is properly cited. [DOI: 10.1149/2.0441712jes] All rights reserved.

Proton reduction, resulting in generation of hydrogen gas, is observed in many applications involving electrochemical reactions and often during corrosion of light alloys. It can be a side reaction during metal electrodeposition, 1-3 a parasitic reaction for batteries [4][5][6] and sacrificial anodes, [7][8][9][10] or the cathodic reaction for corrosion of metals with low electrochemical potential, such as for example magnesium [11][12][13][14][15][16][17][18][19][20][21][22][23] or aluminum. [24][25][26][27][28][29][30][31][32][33][34] The accurate measurement of the amount of hydrogen gas generated during an electrochemical process is important both to aid fundamental understanding and to optimize technological processes. The present work has originated from the authors' activity focusing on the measurement of the hydrogen evolved from magnesium during corrosion in aqueous environments; however the method presented here could be applied to virtually any study where quantitative measurement of gas evolving from an immersed electrode is required.
The most simple and most widely applied method to measure the hydrogen generated from an immersed surface is the (volumetric) hydrogen collection method, widely used in corrosion studies on magnesium and magnesium alloys. [11][12][13][14][15][16][17][18][35][36][37] The volumetric method involves the use of a graduated cylinder that is closed at the top and open at the bottom, and initially filled with test electrolyte. The bottom of the electrolyte-filled graduated cylinder is immersed in a cell containing the same electrolyte, and a specimen evolving an unknown amount of hydrogen is then placed below the cylinder (sometimes with a funnel to aid complete bubbles collection). The hydrogen bubbles generated from the specimen float toward the top of the graduated cylinder, where they are collected, and the gas volume can be measured. The advantage of such method is the simplicity and the use of inexpensive materials, enabling many long-term tests to be run in parallel with minimal expense and setup time. On the other hand, the main disadvantage is the relatively low sensitivity and precision, which is the consequence of the fact that a relatively large amount of hydrogen (at least fractions of a milliliter) must be accumulated at the top of the cylinder, before a reliable estimation of the volume can be made. Reducing the diameter of the cylinder can, to some extent, increase the precision of the volumetric measurement, but as the diameter of the cylinder becomes small, the relative error, associated with the hydrogen bubbles sticking to the cylinder walls instead of immediately floating at the top, increases.
A far more advanced, faster and precise method was proposed by Lebouil et al., 11 who used a micro electrochemical flow cell arranged in such way that the hydrogen evolved from the electrode was immediately pumped, together with the flowing electrolyte, in a small capillary. Once inside the capillary, images of the small hydrogen bubbles could be automatically acquired and their volume could be estimated by automated image analysis, almost in realtime. This method is precise and fast, but the small scale of the electrochemical cell poses a serious limitation for macroscopic corrosion studies, and the required hardware is relatively complex and expensive.
Recently, a gravimetric method based on the measurement of the weight of a submerged hydrogen collecting container was introduced by Curioni et al., 20,22,23,38 and subsequently improved by Fajardo et al. 38 The method is based on the continuous measurement of the weight of a submerged hydrogen-collecting container. Specifically, a hydrogen evolving specimen is mounted rigidly below a hydrogen collecting container, open at the bottom and closed at the top, and initially filled by the test electrolyte. Both specimen and container are completely immersed in a test cell, and connected rigidly to a laboratory scale. When hydrogen evolves from the corroding surface, it is collected in the submerged container, producing a buoyancy force since the hydrogen gas replaces and displaces some of the test electrolyte within the collecting container. Such buoyancy force is directly proportional to the amount of hydrogen evolved and can be easily recorded by the laboratory scale. The method has very high sensitivity and resolution (less than a mA of hydrogen current can be easily measured by using a scale with 0.1 mg resolution) and it enables the investigation of the real-time the behavior of freely corroding or electrochemically polarized macroscopic electrodes. Further, as long as the specimen is rigidly connected to the submerged collecting cylinder, no error is introduced when bubbles stick either to the specimen or to the container wall, since the buoyancy force is transmitted to the scale and accounted for immediately. However, a possible disadvantage is that the laboratory scale is relatively expensive and bulky, and therefore the setup it is not practical for running in parallel a large number of long-term tests.
This work presents a new method that uses the same principle, i.e based on the buoyancy force produced by the evolved hydrogen on a submerged collecting container, but instead of using a laboratory digital scale, exploits the principle of the mechanical balance to estimate the amount of hydrogen evolved.

Operating Principle
The method presented here exploits the principle of the mechanical balance, schematically illustrated in Figure 1. A horizontal arm with a central support is rigidly connected by insulated metallic rods to two hydrogen collecting containers. The electrodes are rigidly connected to the collecting cylinders and electrically connected to the metal rods (in the interest of clarity, detailed electrical connections are discussed later). The two containers-electrodes assembly are immersed in two independent electrochemical cells, referred to hereon as the "test cell" (containing the electrode producing an unknown amount of hydrogen) and as the "measuring cell", containing an inert electrode used to evolve hydrogen (Figure 1a). Above the horizontal arm, in the side of the arm above the measuring cell, a metallic needle is placed, such as when (an unknown amount of) hydrogen evolves from the specimen in the test cell, the horizontal arm moves upwards in the measuring cell side (and downwards in the other), until it contacts a metallic needle (Figure 1b). When the arm contacts the needle, adequate circuitry induces hydrogen evolution on the electrode in the measuring cell. Once sufficient hydrogen is evolved in the measuring cell, the balance arm moves upwards in the measuring cell side and downwards in the test cell side, opening the electrical contact and interrupting hydrogen evolution in the measuring cell ( Figure 1c). The mechanical system is self-equilibrating, i.e. it maintains equal amounts of hydrogen in both cells. The simple measurement of the current flowing in the measuring cell, by acquiring the potential difference across the measuring resistor R m , enables estimation of the amount of hydrogen produced in the test cell.

Electrical Circuitry
From the electrical viewpoint, the hydrogen balance requires 3 key components: i) the base (schematically represented in Figure 2a), ii) the balance arm (schematically represented in Figures 2b, 2c) and iii) the supported contact pin (schematically represented in Figures  2c, 2d). The base (Fig. 2a) consists of two coplanar conductive flat surfaces that are electrically insulated from each other. An electrical cable is soldered to each of the two surfaces, such that the 2 electrical cables are attached to the base, but there is no electrical connection between the two cables. On the base, the arm is suspended trough two metallic needles (Figures 1a, 1b), such that each of the two needles is in electrical contact with one of the two conductive surfaces of the base, but not with the other. This provides electrical continuity between the cables connected to the base (labelled as 1B and 2B in Figure  2a) and the cables placed on the balance arm (labelled as 1A and 2A respectively in Figure 2a), without introducing significant friction or spring back forces that might affect the response of the balance. In the configuration used for this work, only cables 2B and 2A are used to connect a potentiostat to the electrode placed in the test cell, and cables 1B and 1A are not connected. More advanced electrical configurations require connection of cables 1B and 1A, but they are not discussed here. Overall, the electrical connections can be seen in Figures 2b, 2c. Specifically, the electrode in the test cell is electrically connected to cable 2B, through the support and electrical contact needles, while the electrode in the measurement cell is electrically connected to the contact needle (when the balance's arm move upwards) through the contact pad placed on the balance arm ( Figure 2c).
From the electrical viewpoint, the overall principle of operation is simple; when the balance arm move upwards in the test cell side, the contact pad touched the contact needle and this contact closes the circuit in the measuring cell side, developing hydrogen. The contact is maintained until sufficient hydrogen is developed in the measuring cell to balance that developed in the test cell. The overall test circuit used in this work is schematically presented in Figure 3. An Ivium Vertex potentiostat is used to apply a measured current to a titanium electrode attached to the balance arm in the test cell side. A 9 V battery, connected in series with a measuring resistor, provides the voltage to generate hydrogen on a titanium electrode connected to the balance's arm in the measurement cell side. The potential difference across the measuring resistor is recorded by using a NI-USB6009 analogue to digital converter, acquiring the voltage difference across the measuring resistor at a rate of 5 kHz. The high-frequency data are subsequently averaged to produce a time record at 1 Hz.

Mechanical Considerations
From the mechanical viewpoint, the behavior of the balance can be schematically represented as in Figure 4. Ignoring all the balanced components of the forces, when hydrogen is generated in the test cell, a buoyancy force F 1 is generated at a distance b from the support point. Such force generates a moment M 1 The rotation angle of the arm (α), and the displacement d, are a function of the applied moment M 1 , of the distance (h) between the center of gravity and the support point, and of the weight of the arm (F p ). With reference to Figure 4, if the arm is in equilibrium before the application of the force F 1 , the center of gravity is located in position 1 and aligned vertically with the support point. When F 1 is applied, the arm rotates counter-clockwise and the center of gravity is displaced from position 1 to position 2. The force F p associated with the weight of the displaced center of gravity, has now a component, F P ⊥ , perpendicular to the line passing through the center of gravity and the support point. F P ⊥ generates a moment M 2, opposing to the moment M 1 , that can be calculated as: Where At the equilibrium, the two moments are equal (M 1 = M 2 ) and, considering that the displacement d can be written as it is possible to write calculate the displacement d as a function of the buoyancy force F 1 , the length of the arm b, the weight of the arm F p and the vertical distance between the support point and the arm's center of gravity, h: For example, considering a 200 mm arm that weights 200 g (1.961 N), and a change in weight of 7.6 mg (74.6 × 10 −6 N), which is equivalent to the quantity of hydrogen produced in a minute by 1 mA of cathodic current in water, the displacement (d) equals 0.152 mm if the center of gravity of the arm is located 10 mm below the support point and 1.52 mm if the center of gravity is located 1 mm below the support point. If the center of gravity of the balance is above the support point, the considerations above are not applicable, since the balance would be unstable. In this case, any horizontal displacement of the center of gravity from the support point would create a momentum in the same direction of the applied force F 1 causing the arm to fall.
Based on the mechanical analysis, two key conclusions can be made; 1) the center of gravity of the balance must be below the support point at all times, such that the balance is stable and the arm has a horizontal equilibrium position and 2) that for a given force applied to one side, the displacement of the arm increases (and hence the sensitivity of the balance) when the distance between the center of gravity of the arm and the fulcrum decreases. It follows that to improve the sensitivity of the hydrogen balance it is essential to be able to adjust the position of the center of gravity with respect of the position of the support point.

Practical Implementation
The hydrogen balance used to obtain the results presented in this work consists of a glass support, the conductive base, the balance arm with the attached hydrogen collecting containers and electrodes, and a holder for the contact pin. The two key components of the balance that deserve detailed discussion are the balance arm and the contact pin. The balance arm is graphically illustrated in Figure 2. The material used for constructing the arm was 2 layer copper clad glass fiber reinforced polymer with total thickness of 1.6 mm and 200 mm long. This is a material that is commonly used for prototyping printed circuit board and it is readily available on the market. Mechanical connections between the various parts of the balance arm were made by soldering, using solder wire (Ag 4%, Sn 85.5%, Cu 0.5%.) The aluminum rods, connecting the arm to the hydrogen collecting cylinders were connected to the arm by epoxy resin, and electrical connection was made directly using insulated copper wires as needed. The aluminum rods were insulated from the test electrolyte by heat-shrinking plastic sleeves. The hydrogen collecting cylinders were made of glass, 2 cm diameter by 3 cm height. The electrodes were connected to the rod by using insulated electric wire, and the connections were insulated from the electrolyte by epoxy resin.
In order to increase the precision and sensitivity of the balance, the key element is the center of gravity adjustment system. This is realized by attaching with epoxy resin two 8 mm internal diameter nuts sustaining a bolt in the horizontal direction and two similar bolts sustaining a screw in the vertical direction, as schematically illustrated in Figure 2b. Moving the horizontal bolt enables the adjustment of the position of the center of gravity in the horizontal direction, while the vertical bolt enables the adjustment of the position of the center of gravity in the vertical direction. Before starting a measurement, the position of the center of gravity must be carefully adjusted by turning the horizontal and vertical bolt such that, when the arm is horizontal, the center of gravity of the arm is as close as possible, but below, the support point.
The contact for the closure of the circuit was made with a pin, held either with a rigid support or a flexible aluminum support. The flexible aluminum support was obtained by cutting from a 0.9 mm 99.95 wt% Al sheet an isosceles triangle of 2 cm base and 5 cm height, with the contact pin mounted on the tip, as schematically described in Figure  2d. In both cases, the pin was connected to the negative pole of the 9 V battery by a copper cable. In order to measure the current the positive pole of the battery was connected to the counter electrode, placed in the measuring cell, through a measuring resistance.
The test electrolyte in the two electrochemical cells was a 3.5 wt.%NaCl aqueous solution, which is commonly employed for corrosion tests. Therefore, in both cells the only available cathodic half-cell reactions are hydrogen evolution and oxygen reduction. Due to the high potential difference across the working and the counter electrodes, it is possible to assume that oxygen reduction occurred under mass transport control, with a limiting current density at room temperature of ∼10 μA cm −2 , 39 thus being almost negligible with respect to the current measured during the experiments. Similarly to oxygen, the effects associated with the solubility of hydrogen in the solution are expected to be equivalent on both sides, and therefore to cancel out. This is due to the fact that the volume of the solution in the test cell is closely similar to that in the measurement cell, and the amount of hydrogen evolved in both cells is virtually identical.
High purity titaniumfoils were used both as anode and cathode materials in the two cells. The area of the hydrogen evolving cathodes was ∼1 cm 2 , while the area of the anodes was ∼100 cm 2 .

Experimental Validation
In order to verify the accuracy and the limitations of the hydrogen balance, the balance arm was prepared as described in Figure 3, such that a known current was applied by the potentiostat to (and hence a known amount of hydrogen could be generated from) the electrode placed in the measuring cell. Preliminary testing indicated that a value of measuring resistance of 100 ohm provided the most accurate results, and hence only results obtained with this value of measuring resistance are presented hereon. Two types of current-time profiles were applied to the electrode in the measuring cell. The first current-time cycle  ( Figures 5, 6 red lines) involved: i) no current applied for the first 600 seconds, ii) 0.5 mA applied for 3600 s, iii) no current applied for 3600 s, iv) 1 mA for 3600 s, v) no current applied for 3600 s, vi) 2 mA for 3600 seconds and vii) no current applied for 3600 s. The second current-time cycle (Figure 7) was similar, i.e. it comprised periods of 3600 s when the current was applied, separated by periods of 3600 s when the current was not applied. For the second current-time cycle, the values of applied current were varied from 0.5 mA to 0.1 mA at 0.1 mA intervals. The purpose of the first current-time cycle was to evaluate the accuracy of the response for relatively large values of current and the time required to the balance to respond to a change in current, while the purpose of the second cycle is to identify the lower limit of current detection. Figure 5 presents the results obtained by using a contact pin mounted on a rigid support. In Figure 5, the current applied in the test cell by the potentiostat (red curve) is overlapped on the current measured from the measuring cell (black). If the balance behaved ideally, the two curves should overlap. However, it is evident that the current measured by the hydrogen balance is relatively noisy, and does not follow immediately the current applied by the potentiostat, but a transient of about 400 seconds is observed when the current is increased or decreased stepwise. However, smoothing of the current  curves reveals that the average values are closely similar, albeit with some noise. Scrutiny of the time evolution of the charge, obtained by integrating the current over time, reveals that at the end of the experiment an error of 7% of the total charge passed is observed. Such error could be considered not excessive for most corrosion measurements. Figure 6 presents the results obtained in a similar experiment, conducted under nominally identical conditions, except for the fact that the contacting pin was not mounted on a rigid support but on the triangular aluminum sheet, such as the contact pin could move slightly upward after touching the contact pad on the balance arm. It is evident that the noise levels were significantly reduced at high current, and virtually no current was detected by the balance when no current was applied by the potentiostat. For the higher values of current, the transient time was also reduced. Comparison of the values of charge (Figure 6b) indicates that the charge measured by the balance was closely similar to that measured by the potentiostat (approximately 2% error). Figure 7 presents the result obtained by applying the second current-time cycle, with the contact pin mounted on the aluminum support, a setup identical to that of the data presented in Figure 6. It is evident that for currents below 0.3 mA, the noise in the signal recorded from the balance becomes substantial and it is difficult to  clearly resolve the values of current. However, integration of the current measured by the balance shows that, overall, the values of charge measured from the balance were closely similar to those measured from the potentiostat, even for the lower values of current. As the applied current decreased, the transients in the curve measured by the balance become longer, and hence it was difficult to separate the steps.

Discussion
The operation of the hydrogen balance relies on the buoyancy force generated by the evolved hydrogen on the two collecting cylinders. For relatively long experimental times the balance is able to equilibrate and the current flowing across the measurement cell reflects accurately the amount of hydrogen evolved in the test cell, regardless of the fine details of the experimental setup. However, the transient behavior of the balance is affected by several practical details, such as the value of the measurement resistor, the mechanical behavior of contact pin, and the position of the center of gravity relative to the support point.
The value of the measurement resistor, for a fixed voltage supplied to the measurement cell and provided that the electrodes in the measurement cell are not excessively small (as in the experiments presented here), is the main factor that limits amount current flowing through the measurement cell during the time when the contact pin touches the contact pad on the arm. Thus, for a constant contact time between pin and pad, the current (and hence the charge) delivered to the measurement cell increases with decreasing values of resistance. Similarly, for a constant value of measuring resistance, the charge delivered to the measurement cell per time unit increases with increasing contact time between pin and pad. However, the same contact time can be achieved equivalently by a low number of relatively long contacts or by a high number of short contacts. Overall, for a fixed position of the center of gravity, the contact time depends on the imbalance between the hydrogen present in the measurement cell and the hydrogen present in the test cell; the more the hydrogen in the test cell exceeds that in the measurement cell, the longer the contact time per time unit. The position of the center of gravity with respect to the support point also contributes to the overall behavior, since, for a given imbalance in hydrogen amounts between measurement and test cell, the displacement of the arm increases for decreasing distance between center of gravity and support point.
Another important consideration is that, when the contact pad connected to the balance arm moves upwards touching the contact pin, it has a tendency to bounce back (downwards) as a result of the (relatively elastic) collision between pad and pin and of the inertial forces. In addition, during the contact time, hydrogen is evolved in the measurement cell, producing a buoyancy force that tends to displace the contact pad downwards. Thus, the two effects (bouncing of the pad on the pin and generation of hydrogen in the measurement cell) act in the same direction and the balance shows some degree of oscillation. Such oscillation is barely visible by the naked eye, but it can be clearly resolved with an optical microscope, as illustrated in Figures 8a-8d. From the figure, the relative position of pin and pad can be evaluated by looking at the reflected image of the pin on the pad. When the tip of the pin touches the tip of the reflected image, pin and pad are in contact, whereas when the tip of the pin is visually separated from the tip of the pin in the reflected image, there is no electrical contact between pin and pad. In Figure 8, the red lines are added as a guide to the eye to identify the tip of the pin and the tip of the pin in the reflected image.
As a result of the combination of all the effects described above, the way in which the contact pin reacts to the impact with the moving balance arm, has a significant impact on the quality of the measurement, because the current is delivered differently to the measurement cell. Figures 8e-8f clearly illustrates this issue. In Figure 8e, the potential across the measuring resistor (and hence the current flowing to the test cell) has been recorded at high frequency (5 kHz) approximately in the middle of the 2 mA step in Figures 5 and 6. Figure 8e displays the behavior recorded for a contact pin mounted on a rigid support, while Figure 8f shows the behavior for the pin mounted on the flexible aluminum support, schematically described in Figure 2d. It is evident that (for approximately the same value of current measured) the rigid support produces a lower number of longer contacts, while the flexible support produces a higher number of shorter contacts, since the arm oscillates less. Unreported microscope videos of the pin contacting the pad also show that the amplitude of the oscillation of the balance arm is significantly larger when the pin is mounted on the rigid support.
Based on the considerations above, it can be concluded that an ideal hydrogen balance should be perfectly damped, such as the contact pin only touches the contact pad on the arm for the time required to produce enough hydrogen in the measurement cell, and once the hydrogen produced is sufficient to balance that in the test cell, the contact between arm and pad is interrupted by an infinitesimally small displacement of the arm. In this case, since the displacement of the arm would be infinitesimally small, the vertical distance between center of gravity and support point would have no effect.
In practice, the balance behaves non-ideally, since the mechanical and electrical behaviors are inherently interlinked by complex interactions as described above. For example, sometimes the hydrogen in the test cell might slightly exceed the hydrogen on measurement cell but the duration of a single contact between pad and pin might not be sufficient to exactly re-balance the amount of hydrogen in both sides because the pin might bounce on the pad before sufficient current is passed. Similarly, when the amount of hydrogen is equal in the two sides, the arm's oscillation might produce unwanted contact between pin and pad, unbalancing the amount of hydrogen in the two cells. As a result, the current measured by the balance is relatively noisy and transients are long. Similarly, when the current in the test cell is stepwise reduced to zero, residual oscillations of the arm will induce occasional contact between pin and pad. Each contact will produce an excess of hydrogen in the measurement cell side, and some oscillation of the arm, which might induce another (unwanted) contact, with associated hydrogen excess. If this process proceeds for some time, eventually the excess hydrogen in the measurement cell will prevent any further contact between pin and pad, but some current has been already measured even when the electrical current applied to the test cell was zero.
Based on the discussion above, it is clear that, for long measurement times (minutes to hours) and relatively constant rate of hydrogen evolved in the test cell (as in many free corrosion-related scenarios), the charge measurements obtained from the hydrogen balance are very reliable, since sufficient time is available to achieve a dynamic equilibrium between the oscillations and the current delivered in the measurement cell. However, if the current varies substantially within a short period of times (seconds to minutes), the hydrogen balance is not suitable to perform reliable measurements. Overall, the hydrogen balance appears to have substantially higher time and current resolution compared to the traditional volumetric hydrogen collection method and comparable current resolution compared the gravimetric method using a commercial precision scale. However, the hydrogen balance behaves significantly worse in terms of time resolution compared to the gravimetric hydrogen collection method using a digital laboratory scale, since transients of the order of hundreds of seconds are observed before a steady reading is attained. The main advantage of the hydrogen balance is the relatively low cost of the equipment. Further, the principle of measuring a weight change by measuring the current associated to an electrochemical reaction, introduced here for the first time and exemplified by the hydrogen balance, might be useful in a variety of other applications where detecting the variation of weight of an immersed specimen is of interest.

Conclusions
In this work, a new method to measure hydrogen evolving from a submerged electrode was presented. The method consists of a mechanical balance with a horizontal arm attached to two hydrogen-collecting containers that are submerged in two independent electrochemical cells. When hydrogen is developed in the test cell, the balance arm is displaced, closing an electrical contact that triggers hydrogen development in the other cell. Measuring the current flowing in the measuring cell enables accurate estimation of the hydrogen evolved in the test cell. The hydrogen balance is significantly more accurate and provides accurate estimation of hydrogen evolved that the usual volumetric hydrogen collection method. However, a time delay of the order of tenths or hundreds of seconds (depending on the detailed configuration) is present before the current can be accurately measured. This prevents the use of the hydrogen balance for real-time coupled, relatively short, electrochemical and hydrogen measurement. On the other hand, the hydrogen balance is simpler and less expensive than the gravimetric methods based on the use of a precise digital laboratory scale and could be used for all those applications where several long term tests must run in parallel.