A bubbler controls the steam concentration in the gas by controlling the partial pressure of water in the bubbler container via temperature. The partial pressure of water vapor in the gas is calculated using the vapor pressure of water, which is in turn calculated from the temperature of the water (Buck, 1981). The bubblers operate on the assumption that the water and vapor in the bubbler are in thermodynamic equilibrium. If the flow rate of gas into the bubbler exceeds the threshold in which water can evaporate, this assumption will be invalidated. However, in equilibrium, the concentration of water vapor leaving the bubbler will be controlled by the temperature of the water. The Buck equation, shown in Equation 1, describes the vapor pressure of water P in kPa, as a function of temperature T in K with good accuracy in the temperature range of 50 °C–100 °C (Buck, 1981). P = 0.61121 ⁡ exp 18.678 − T 234.5 T 257.14 + T (1)

Using the vapor pressure of water as the partial pressure of water in the gas stream, the concentration of water in the gas stream can be calculated by dividing by the total pressure. Importantly, the atmospheric pressure at the approximate elevation of the laboratory must be used, as this can significantly change the temperature of the water required for a given steam concentration. For example, at the Idaho National Laboratory, elevated at 4,700 ft above sea level with a pressure of 0.84 atm (85.1 kPa) (International Organization for Standardization (ISO), 1975), the temperature required to achieve 50% humidity is 77.6 °C, while at sea level, this would only achieve a concentration of 42.5%. At sea level, a temperature of 81.6 °C is required to achieve a composition of 50% humidity. The effect of altitude on the required temperature is more significant at higher steam concentrations, as the vapor pressure curve is exponential with temperature. Note that the pressure reported by meteorologists is normalized to sea-level and is not the true absolute pressure at the given location. Additionally, the measurement of the water temperature is vitally important, as minor deviations in temperature, which are common with thermocouple measurements, can substantially impact the expected versus actual steam flow (see section 3.1).

The primary purpose of the bubbler is to transfer water vapor into the gas stream. To ensure the steam concentration reaches the target value calculated via thermodynamic equations, it is critical that the steam concentration in the bubbles reach equilibrium. The mass transport from the water to the gas in the bubbler is a function of the gas flow rate, the water temperature, the liquid height, the bubbler diameter, and the flow area, among other properties (Mahmood et al., 2024). Larger column heights and smaller bubbles increase the mass transfer between the two phases. The mass transport limitations for the bubbler design presented in this paper have been experimentally determined to be beyond the flow regimes of interest to the typical single cell. For larger bubblers needing higher gas flow rates, similar principles apply. The regime anticipated in the bubbler design is homogenous bubbly flow (Mahmood et al., 2024). Care must be taken to avoid slug flow regimes in the bubbler. General guidelines include a liquid height to flow diameter ratio of 2 or greater, and a superficial gas velocity less than 30 cm/s (Lau et al., 2010), however tuning of column properties can allow deviation from these general guidelines. A detailed discussion of column design is out of scope for this paper, and the reader is referred to the extensive literature available to aid in the fine tuning of column properties, if desired (Kulkarni and Joshi, 2005; Lau et al., 2010; Chang et al., 2018; Eder et al., 2022; Fang et al., 2022; Gong et al., 2022; Mahmood et al., 2024).

The use of water, electrical components, high power furnaces, heat sources/high temperatures, and flammable gas require careful consideration of the safety systems and possible hazards. Best practices should be followed. NFPA 2: Hydrogen Technologies Code (National Fire Protection Association, 2023) should be followed as applicable. For laboratory operations, this includes chapters 4, 5, 6, 7, 8, and 16. Additional code compliance, such as ANSI/CGA G-5 (Compressed Gas Association, 2024) or similar, may be relevant and required depending on location and jurisdiction.

Oxygen-ion-conducting SOECs (O-SOEC) require a humidified hydrogen stream. Figure 2a shows the sketch of the basic O-SOEC bubbler design. The dashed box represents the repeat unit for a multiple bubbler system. The refill, overfill drain, and condensate collection do not need repeated and can be utilized by multiple connected bubblers. Figure 2b demonstrates a complete two-bubbler system, while Figure 3 highlights critical aspects of specific components. The design utilizes a sanitary pipe spool as the main bubbler container. The flat surface of the cap allows for national pipe thread (NPT) connections to the container using PTFE tape. Caps can be drilled and tapped or purchased with NPT connections. Extra connections may be created and capped for future use but are not necessary. Connections into the bubbler are made through compression fittings. Bored-through compression fittings (see item 26 and 27 in the bill of materials) allow for temperature probes and gas lines to penetrate the cap into the container. Note that the welding of NPT or compression fittings often distorts them and can ruin their ability to seal. The gas supply into the bubbler is released near the bottom of the container through a sparger (porous metal gas muffler) to induce bubble formation. Valves control the flow of water into (during normal operation) and out of (during blowdown) the bubbler.

There are 2 heating zones for each stand, as shown in Figure 3b. The first zone is the bubbler container, kept at the desired dew point of the supply. The second zone is the gas line between the bubbler and the cell/furnace, kept above the dew point to prevent condensation. The heat is supplied to the bubbler through heat tape or band heaters. Temperature is measured via two probes in the water, with one for the PID controller and the other for the overtemperature protection. For the heated line, the temperature is measured via probes fixed to the tube under the heat tape. Again, two separate probes are used for control and overtemperature protection. Note that insulation is not shown in the figures but is required. The heated bubblers and humidity line must be insulated using appropriate insulation, such as ceramic fibers. It is important that the heat source and insulation of the heated line be as uniform as possible to ensure similar temperatures throughout. The insulation and/or heat tape may be stopped prior to the furnace as exhaust gases and proximity to the furnace will keep the gas stream above the dew point. This may be necessary to avoid overheating any metal to ceramic seals close to the furnace.

The design includes an automatic water refill system which maintains the water level of the bubbler during long duration tests. Gravity maintains an even level between the bubblers and the float-containing container (float container, shown in Figure 3a), given all containers are mounted at the same height, and backpressure is minimized. This ensures a slow refill of the bubblers, decreasing temperature variations and maintaining a constant water level throughout the test. Multiple bubbler assemblies may be connected to the same refill assembly. A solenoid connecting the water supply to the bubblers is controlled by a float in an extra unheated container. Power to the solenoid is wired in series with the float switch, thus fills when the float is low, and shuts off when the water raises the float. The water can be supplied from either a water tank or deionized (DI) water loop/system. The water tank is the simpler iteration but requires more operator intervention. For the water tank, it must be mounted high enough to provide supply pressure. A peristaltic pump can aid in refilling if the mounting height prevents easy refilling by hand. Common button cell water usage rates are on the order of mL per hour, allowing significant time between refilling with a 10 L tank. For a DI water loop/system, the pressurized DI water can directly feed into the system.

The use of a water reservoir or water supply system creates the possibility of overfilling the bubbler. Depending on the quantity of water stored/available to the system, overfilling could result in a hazardous situation. With small quantities, the hazard is mostly insignificant and limited to cell damage. The hazard increases as the quantity of the supply increases. Namely, the control system on the water feed line could fail, allowing the entire bubbler to fill with water. In this event, water will leak from the refill atmosphere vent and can also entrained in the humified hydrogen line, potentially cracking the cell and leaking significant quantities of steam and hydrogen into the furnace. To prevent this, the system should be designed so that water can be released, ensuring the level stays below a set maximum. This can be achieved with an overfill drain, shown in Figure 3d, which limits the maximum height the water can achieve at low flow rates. Additionally, the overfill drain should be vented to prevent siphoning.

The continuous supply of steam produced by the bubbler requires the removal of condensed water downstream of the cell. This is performed with a condensate removal system, depicted in the top left of Figure 2a. The condensate is directed into a container (or tube with a diameter greater than ¼” (6.35 mm)) where it collects. The top of the container is connected to the ventilation system or hydrogen exhaust point to provide a direct path for hydrogen to exhaust. The bottom of the container contains a water trap, similar to the overfill drain. Water collects in the low point, preventing any gas flow, and water flows towards the drain when the water level exceeds the height of the trap. As with the overfill drain, a vent in the high point of the trap should be included to prevent siphoning. If no drain is available below the condensate removal system, a commercial condensate pump can be used to pump the water towards a drain.

The float container must be open to the atmosphere to prevent a pressure imbalance between the bubbler and float containers. The opening must be vented to a safe location. In the event of a loss of supply water, the bubblers can empty completely, breaking the water seal between the H₂ containing bubblers and the float container. This leads to a loss of containment of H₂. Thus, it is critical that the float container and the overfill drain vent are vented to a safe location.

The significant difference between bubblers for O-SOECs and proton-conducting SOECs (p-SOEC) is the electrode in which the steam is supplied. In O-SOECs, the steam is supplied to the fuel electrode, while in p-SOECs, it is supplied to the air electrode. The carrier gas for the steam supply for p-SEOCs is air, thus bubblers utilized for p-SOECs can omit hydrogen safety protocols, which leads to decreased flammability considerations and costs.

The maximum flow rate through the bubbler is limited by the volume and depth of the container. Flowing rates higher than the bubbler’s capability will decrease the humidity concentration as the kinetic limitations of evaporation prevent equilibrium. As larger flow rates are needed, the size of the bubbler container can be increased. For flow rates on the order of 1 slpm, a 3” (7.6 cm) diameter and 6“ (15.2 cm) tall spool may be used (see section 4). Increasing the spool to a 6” (15.2 cm) diameter and 12” (30.5 cm) height increases the allowable flow rates beyond 10 slpm. Furthermore, since evaporation requires substantial energy (∼2.26 J/g at 100 °C for water (Torquato and Stell, 1982)), it is important to ensure the thermal system can supply adequate heat without causing hot spots. The heat duty of the bubbler can be easily calculated from basic thermodynamic equations to estimate the required heater power.

The accuracy of the temperature sensor is critical to achieving the correct steam concentration, as the water temperature is the independent variable affecting the vapor pressure. Thus, it is suggested to use resistance temperature detectors (RTD) for PID control, which have standard accuracies of ±0.15 °C, as opposed to Type K thermocouples, which have standard accuracies of ±2.2 °C (ASTM International, 2024). The steam concentration increases exponentially with the temperature. At high steam concentrations, the steam concentration increases ∼3.5% for 1 °C of water temperature increase. Thus, a Type K thermocouple set to 92 °C could produce steam concentrations ranging from 80.8% (at 89.8 °C) to 95.3% (at 94.2 °C). This uncertainty highlights the necessity for accurate temperature measurement. At lower temperatures, the effect is less significant, and Type K thermocouples can be used.

A quick test to validate the temperature measurement and atmospheric pressure is to boil the water in the bubbler while measuring the temperature. The temperature will plateau at the boiling point, as all heat input to the two-phase water/steam equilibrium will go into vaporization as opposed to temperature rise. The measured temperature should match the calculated value for the given elevation.

Proper PID tuning is also critical for stable steam supply. Improperly tuned PIDs can swing multiple degrees, inducing oscillations in the steam concentration, Nernst potential, and electrochemical performance data. Additionally, the heat duty can be significantly different between flow conditions, requiring higher power at higher flow rates, and possibly a separate PID tuning. It is desirable to record the duty cycle of the heaters via the stand data acquisition system to diagnose any heater fluctuation impacts on cell performance.

The generation of steam outside of the furnace requires proper heating of the humidified gas lines in order to prevent condensation. Any condensation can have a detrimental effect on cell performance and typically manifests as spikes in the open circuit voltage (OCV) or constant current/voltage performance. The condensing water will typically coalesce until a large enough droplet forms, which can move due to gravity or gas flow. The droplet can then flash to steam when it touches a hotter section of the line. The sudden creation of steam temporarily increases both the gas flow rate and the concentration of steam in the gas, which decreases the OCV as per the Nernst equation. When operated in potentiostatic mode performing electrolysis, this decrease in OCV causes an increase in the overpotential, which results in an increase in the magnitude of the current density, as shown in Figure 4a. The behavior shown in the Figure 4a is characteristic of steam flashing, with an instantaneous increase in current density (or decrease in OCV) followed by a decay back to baseline, typically occurring periodically. Note that steam flashing may also cause a pressure spike which can crack or dislodge a cell from the test stand.

The heat loss required to induce condensation from a flow of water vapor is extremely low. The heat loss can be approximated from Equation 2, which shows the heat loss P in watts as a function of the gas flow rate Q in sccm, starting gas temperature T_s, dew point T_d, volume fraction of the hydrogen and steam y _H and y _s respectively, and molar heat capacity of the hydrogen and steam C_p,H and C_p,s respectively. P = Q * 7.43 * 10 − 7 mol s   sccm T s − T d y H C p , H + y s C p , s (2)

For a 200 sccm flow of 50% steam balance hydrogen at 110 °C with an 82 °C dew point (assuming 1 atm pressure), only 136 mW of heat can be lost until condensation begins. This calculation highlights the need for guard heaters and insulation on the tubing, as the allowable heat loss is practically zero.

In order to prevent condensation, it is critical that the entire humidified line remains above the dew point of the gas. This includes sections downstream of the cell, thus, it is beneficial to heat the outlet of the fixture. The tubing can be wrapped densely with heat tape followed by a layer of insulation to ensure even heating. It is best to heat every section of the line evenly, as opposed to leaving sections unheated and relying on conduction from hotter sections. Relying on conduction will require a higher temperature setpoint which may exceed the temperature limit of the seals. Thus, the aim should be a low temperature setpoint with uniform heating. A setpoint of 30 °C above the dewpoint is suggested, however, there is no issue with going hotter, given the temperature limit of the seals is not exceeded. If the system requires a high setpoint to prevent condensation, it is indicative of poor heating uniformity. Heat conduction and radiation out of the furnace, as well as hot exhaust gases, can keep the immediate area heated, however, it should not be relied upon to maintain a heated line, as the temperature of metal gas lines quickly drop to room temperature outside of a heat source. Calculating the flow velocity of the reactants can provide some clarity residence time where gas may condense. Additionally, it is easier to ensure uniform heating on shorter lines. Thus, it is beneficial to position the bubbler directly below the cell fixture. The use of flex hoses, such as convoluted tubing, can provide flexibility which is beneficial for laboratory operations, but their use increases the risk of leaks, condensation, and decoupling of the temperature sensor and heat source.

In the exhaust line, condensation will occur as the temperature of the gas drops, thus, a condensate trap is necessary to collect liquid water. Additionally, the near-room-temperature gas will be saturated with water vapor, meaning any further decreases in temperature will induce condensation. The line between the cell fixture and the condensate trap must be sloped downwards, to allow liquid water to easily flow into the trap. It is beneficial for the exhaust line leaving the condensate trap to be larger than 1/4” (6.35 mm), which allows for water to flow upstream of the gas flow and back into the trap. A vertical section of 1/4″ exhaust line with condensate can impose significant backpressure on the cell, which in turn affects the stability of the OCV/electrochemical performance. The surface tension of water within a 1/4″ tube effectively plugs the line and prevents gas motion past the droplet. Thus, it is suggested to use 3/8” (9.5 mm) tubing or larger for lines in which water moving towards the drain and gas moving towards the exhaust are in opposite directions, i.e., on an exhaust line traveling up towards ventilation. For these same reasons, any level sight should use 3/8″ or larger tubing. If the two mass fluxes are in the same direction and the line can be horizontal or sloped down towards the condensate collector, quarter inch line is acceptable.

Additionally, since the bubbler headspaces are maintained at atmospheric pressure and the float container is vented to the atmosphere, any backpressure in the bubbler headspace can cause the flow of heated water out of the bubbler into the float container. This is indicated by an increase in temperature of the water line in between the bubbler and float container as well as the float container itself. In normal operation, uninsulated tubing leaving the heated bubbler achieves room temperature ∼2 cm away from the insulated portions. If heat is detected further from the bubbler, it is indicative of movement of hot water out of the bubbler. An oscillating water level on the float container’s sight is typically caused by condensation related backpressure. This occurs as the water blocks the exhaust flow, building back pressure and driving water from the bubbler to the float container. When the pressure reaches a certain point, exhaust gases push past the water plug and release the pressure, which moves water back into the bubbler.

If no active cooling of the exhaust stream is utilized (such as with the design presented in this paper) the exhaust gas stream will be saturated with water. Thus, small variations in room temperature can induce condensation further down the exhaust line. It is critical that the system design considers this effect and follows the best practices in this paper to prevent apparent instability in the steam supply. This includes proper exhaust line sizes and sloping lines to prevent water traps. When the bubbler is not performing optimally, the system should be closely inspected for potential areas of condensation, as this is the most common cause of poor bubbler performance.

The bubblers are filled using a gravity fill system, with the headspace of the float container and the bubbler containers near atmospheric pressure. Thus, the bubbler headspace must remain at atmospheric pressure, otherwise, an imbalance in water levels will arise. Backpressure can occur through a variety of means downstream of the cell, from condensation to check valves. A 1/3 psi (2.3 kPa) check valve in the exhaust line, which increases the pressure in the bubbler to ∼9 in (22.9 cm) of water relative to atmosphere, will displace water from the bubbler until there is a ∼9″ difference between the bubbler and the float container. This essentially displaces all the water from the bubbler, reducing its effectiveness. Thus, the use of any component in the downstream that imposes backpressure should be avoided with this design. If absolutely necessary, the system can be modified to operate slightly above atmospheric pressure. The headspace of the float container may be connected to the bubbler (as opposed to vented to atmosphere) to ensure they remain at equal pressure, ensuring the water levels are identical. This is only possible for a single bubbler-refill system. This modification exposes the float container to hydrogen, requiring the float switch to posses the proper electrical/hazardous classifications and it is recommended to remove or protect the polymer tubing sight. Additionally, the water reservoir/supply must be held at higher pressures to ensure water will flow into the system.

During the operation of a bubbler, steam is continuously removed, with non-volatile contaminants remaining in the bubbler (Schäfer et al., 2022). Additionally, fresh water is fed into the bubbler, bringing in more contaminants. This process can concentrate contaminants in the bubbler, potentially increasing the volatility. Thus, it is important to implement blowdown procedures, commonly used in industrial boilers. In practice, draining the water from the bubbler and refilling twice can help remove contaminants. It is recommended to do this between every test due to the simplicity of the process. In extremely long tests (>5 kh) it is desirable to perform blowdowns at set intervals to avoid accumulation of contaminants.