Exploration of electrochemical hydrogen pumping with ultralow feed concentration down to 1 ppm

Abstract

Electrochemical H₂ pumping with ultralow H₂ concentrations (such as 1–10 ppm) was experimentally demonstrated for prominent current response (0.42–3.1 µA), under a low cell voltage at a high flow rate, supported by Multiphysics simulation. The findings in this work may extend the application of H₂ pumping to H₂ detection and other advanced electrochemical systems.

---

Electrochemical H₂ pumping is a process that works to move H₂ molecules from one electrode of the electrochemical cell to the other, through H₂ splitting (H₂ → 2H⁺ + 2e⁻, or H₂ oxidation) on the anode and recombination (2H⁺ + 2e⁻ → H₂, or H₂ evolution) on the cathode.^1–5 Since the first concept introduced by Maget in 1964, H₂ pumping has been developed for practical applications including purification/recovery,^6,7 compression,^8–10 and kinetics study.^11,12 Process modeling was also investigated for H₂ pumping systems.^13–15 The H₂ pumping with low concentration (≤1%) of H₂ warrants scientific interest and potential industrial implementation. 1% of H₂ in N₂ was evaluated for H₂ compression,¹⁶ while 1% of H₂ in N₂ or CO₂ was studied for H₂ recovery,¹⁷ both of which employed a Nafion-based H₂ pump at 60 °C. Very recently, 0.1% of H₂ in He was reported in a ceramic electrolyte-based H₂ pump at elevated temperatures (450–550 °C) for H₂ extraction.¹⁸ Herein, we present the realization of an H₂ pump, operating at room temperature, fed with an ultralow concentration of H₂ in air down to 1 ppm (Fig. S1 and S2), while providing prominent current response.

With delicate cell design and engineering, a prominent “current response (ΔI)” from a very-low-concentration H₂ stream in air can be obtained. As defined by the difference in the electric current from the electrochemical H₂ pump with and without H₂ presence (background current, or I₀), the current response was evaluated by means of alternating the low-concentration H₂ supply to the anode feed every 150 seconds. Specifically, a current response of 3.1 µA (or 62 nA cm⁻²) was observed from the H₂ pump fed with 10 ppm of H₂ in air at a (total) flow rate of 5000 sccm (to the anode), under the following conditions (0.070 V of applied cell voltage, 200 sccm of flow rate of pure air to the cathode, and 26 °C for the hydration tank connected to the cathode) (Fig. 1). Unsurprisingly, the current response was recurring and reproducible, as shown in the four continuous identical H₂-pumping cycles, with a fairly flat background current (2.5 µA). Under the same testing conditions, the current responses were 1.1 and 0.42 µA with 5 and 1 ppm of H₂ in air to the anode, respectively. It could be clearly observed that the current response proportionally followed the H₂ concentration, while having a similar background current (1.5 µA) (Fig. 1). Note that the generated H₂ gas on the cathode chamber was indeed detected quantitatively, driven precisely by the pump operation (both feed concentration and operation time).

Fig. 1 Four-cycle current response (ΔI) from the H₂ pump at three typical H₂ concentrations to its anode (a complete cycle consisting of two arms with 150 seconds each). Test conditions: 5000 and 200 sccm as the total flow rate for the anode feed and cathode feed, respectively, and 0.070 V of applied cell voltage from the anode to the cathode.

In fact, a linear relationship was true between the current response and H₂ concentration with an average regressed order of 0.915 (R² = 0.995) from the double-logarithmic plot across a wide range of H₂ concentrations (1–25 ppm) and flow rates (5000–20 000 sccm) (Fig. 2A). In contrast with H₂ concentration, the current response came after the flow rate in a fractional order (0.453 on average with R² = 0.979) across a wide range of flow rates (1000–20 000 sccm) with three typical H₂ concentrations (1, 5, and 10 ppm) (Fig. 2B). The observed fractional power is essentially governed by the complex flow conditions where the local H₂ concentration depends on flow rate, flow-channel geometry, and media properties, as discussed in the working principle in the SI and the background current interpretation (Fig. S3).

Fig. 2 Impact of H₂ concentration and flow rate on the current response (ΔI) of the H₂ pump. (A) H₂ concentration (C_anode) and (B) flow rate (Q_anode). Typical test conditions: 5000 and 200 sccm as the total flow rate for the anode feed and the cathode feed, respectively and 0.070 V of applied cell voltage from the anode to the cathode.

Note that the applied cell voltage (0.070 V) was carefully chosen for the high-performance pumping operation with ultralow H₂ concentrations. Unlike common H₂ pumps, here the applied cell voltage is critical to the cell operation for a low-concentration H₂ pump where air is present on both sides. Unexpectedly, the current response decreased with increasing cell voltage applied between the anode and the cathode: from 4.8 to 0.30 µA under −0.450 to 0.300 V, accordingly, or a 15-fold drop (Fig. 3A), whereas the background current responded as expected (from −480 to 39 µA, Fig. S4A). Applying more positive cell voltage (>0.300 V), the current response diminished, even though the background current continued to increase. In particular, the background current reached zero from negative to positive under a cell voltage around 0.025 V, when the cell function switched from “diffusion cell” (protons moving from the cathode to the anode) to “electrochemical pumping”. To maintain the functionality of electrochemical pumping as well as retain the current response, a very narrow range of cell voltages (0.050–0.100 V) serves as our choice here. Besides Pt (60% Pt/C) as the anode catalyst in the H₂ pump, similar trends were also observed on other anode catalysts, including Pd (60% Pd/C, Fig. S5) and PtRu alloy (60% PtRu/C, Fig. S6). Despite the declining trend with applied cell voltage, the operation of the H₂ pump under a constant voltage was very stable, as evidenced with a merely 5% performance drop (ΔI: from 4.4 to 4.2 µA) after a continuous 50-cycle test (Fig. S7). In addition, the H₂ pump also showed a strong tolerance against CH₄ interferant in the anode feed (25 ppm CH₄vs. 5 ppm H₂) (Fig. S8).

Fig. 3 Current response (ΔI) of the H₂ pump along with the cell voltage applied. (A) Air electrode (200 sccm) as the cathode (H₂ in air as the anode feed); (B) H₂ electrode (RHE, 50 sccm) as the cathode (either H₂-in-air or H₂-in-N₂ as the anode feed). Test conditions: 5 ppm of H₂ concentration at 5000 sccm of total flow rate for the anode feed for all cases.

To understand why the current response decreases when the applied cell voltage increases, a reversible hydrogen electrode (RHE) was established for the cathode of the H₂ pump by providing ultrapure H₂ gas with a sufficient flow rate (50 sccm). A similar declining trend of current response was shown with the RHE as the cathode (from 73 to 3.8 µA under 0.800 to 1.100 V vs. RHE, accordingly, Fig. 3B, and its background current, Fig. S4B), solidifying our finding with the air cathode. Note: the H₂ diffusion across the Nafion membrane could slightly increase the background current, but such H₂ crossover is insignificant here. Specifically, the H₂ concentration brought by the diffusion crossover across the membrane is equivalent to 1.6 ppm in the anode feed, based on the H₂ permeation of 1.4 × 10⁻¹⁰ mol m⁻¹ s⁻¹ bar⁻¹,¹⁹ which corresponds to an insubstantial increase of 0.35 µA in the background current.

With the RHE cathode, the reversible potential of the H₂-containing yet air-dominating anode of the H₂ pump was found to be 1.130 V vs. RHE, based on cyclic voltammetry (Fig. S9). Such a high electric potential is understandable, because of the air dominance (the standard reduction potential = 1.229 V vs. RHE for O₂ + 4H⁺ + 4e⁻ → 2H₂O). The declining trends observed from two different cathodes are consistent fundamentally with each other. Under the same anode conditions (i.e., 1.130 V vs. RHE and 0 V vs. air electrode), the current responses are reasonably close: 3.8 and 1.6 µA, respectively (while both background currents are near zero).

The exact surface condition of the Pt catalyst used in the anode electrode is of relevance here, and full-range cyclic voltammograms (CVs, from 0.050 to 1.300–1.600 V vs. RHE) were measured on the anode fed with ultrapure N₂ and the RHE as the cathode (Fig. S10). CV currents under 0.050–0.400 V vs. RHE are related to the underpotential deposition (UPD), where Pt–H is of the surface species. The flat CV currents under 0.400–0.600 V vs. RHE represent a “clean” Pt surface. Starting at 0.600 V vs. RHE, Pt–O is formed and densified, as evidenced by the stronger desorption CV currents of Pt–O around 0.700–0.800 V vs. RHE with scans towards more positive potential. Clearly, the working potential of the H₂ pump (0.070 V vs. air electrode, which is 1.200 V vs. RHE) leads to strong and dense Pt–O coverage. Meanwhile, the range of the working potential (0.800–1.100 V vs. RHE) for the H₂ pump with the RHE cathode falls entirely within the Pt–O formation (>0.600 V vs. RHE).

Here, we speculate that the intrinsic catalytic activity of H₂ oxidation on the Pt–O surface is gradually weakened with its continuous growth and densification, and this weakening catalytic activity well explains the observed declining trend of current response. The dramatic reduction (20–40 fold) in the catalytic activity of Pt–O under 1.100 V vs. RHE compared with that under 0.800 V vs. RHE, strongly suggests that the HOR activity on Pt–O is about 1.5 orders of magnitude lower than that on the pristine Pt surface. Instead of being an inert spectator species, oxygen in air also promotes the Pt–O formation, contributing to HOR activity suppression. Under the same applied anode potential (0.800–1.100 V vs. RHE), the observed ΔI with H₂-in-N₂ as the anode feed was 20–100 times, or 1.3–2.0 orders, as much as that with H₂-in-air as the anode feed, in the same H₂ concentration (5 ppm) and total flow rate (5000 sccm) (Fig. 3B with background currents shown in Fig. S4B). The current response of the H₂ pump with 10-ppm-H₂-in-N₂ as the anode feed under a low range of applied cell voltages (0.100–0.600 V vs. RHE, Fig. S11) revealed a well-expected growing profile that reached a “limiting current” of 3,532 µA (under 0.500 V vs. RHE) or 54% of maximum theoretical current response (6572 µA, by the mass balance of H₂). It should be noted that a high utilization (63.8%) of Pt surface was observed on the prepared anode by a similar CV method, derived from the UPD measurement (1.205C, 0.050–0.500 V vs. RHE, Fig. S12).

To better understand the H₂ pumping process, a Multiphysics simulation was conducted using the COMSOL platform under typical conditions: 5 ppm of H₂ in pure air was fed to the anode inlet at a 5000 sccm flow rate, and pure air containing a regulated humidity was introduced to the cathode at a 200 sccm flow rate, under a 0.050 V constant cell voltage. The reported HOR/HER exchange-current density (i₀) from in situ Nafion-based electrodes in our previous work was adopted here (1.7 × 10² mA cm⁻²),¹² and the detailed simulation methods are discussed in the SI with parameters and modules/interfaces shown in Tables S1 and S2. Fig. 4A shows, at steady state, the contour map of H₂ concentration inside the anode chamber consisting of a flow channel, an electrode substrate, and a catalyst layer. Evidently, contour H₂-concentration lines are developed, starting at the edge of the anode inlet and electrode and following the flow direction. A similar profile for such contour concentration-line was theorized by a fractional order (1/3) in 1969 albeit from a different research field (electrodialysis).²⁰ The contour map of H₂ concentration inside the cathode is presented in Fig. 4B, where the H₂, generated from the catalyst layer of the cathode, accumulates and builds up to form its concentration profiles along with the flow direction, reaching 26 ppm as the maximum concentration at the cathode outlet.

Fig. 4 Simulation finding from the H₂ pump under typical conditions. (A) H₂ concentration profile in the anode chamber (both flow channel and the anode); (B) H₂ concentration profile in the cathode chamber (both flow channel and the cathode); (C) I_x (average horizontal current density, equivalent to current response) vs. H₂ concentration in the anode; and (D) I_xvs. average velocity (v) of the anode feed (1 m s⁻¹ equivalent to 360 sccm here). Typical test conditions: 0.050 V of applied cell voltage and 200 sccm of flow rate to the cathode feed; 5 ppm as the H₂ concentration in the anode and 5000 sccm of flow rate for both (A) and (B).

The impact of (anode) H₂ concentration on the current response was also examined by simulation with three different flow rates (5000, 10 000, and 20 000 sccm), yielding a straightforward linear relationship (average regressed order of 0.963, R² = 0.999; note: the imperfect R² is due to the convergence error in simulation) (Fig. 4C). Interestingly, the specific relationship between the flow rate and the current response varies along with the flow rate itself: a descending order from 0.720 (R²: 0.992) at 1 m s⁻¹ (360 sccm) to 0.188 (R²: 0.996) at 100 m s⁻¹ (36 000 sccm) (Fig. 4D). Around the same flow rate range (1000–20 000 sccm or 2.78–55.55 m s⁻¹), the regression order from the simulation (0.454 on average, Fig. 4D) is highly consistent with the experimental observation (0.453 on average, Fig. 2B). The velocity contours of both the anode chamber and cathode chamber are depicted in Fig. S13.

Of particular interest, the average H₂ concentration inside the catalyst layer of the cathode was simulated against the applied cell voltage (Fig. S14A and B), which not only proves the desired functionality for the H₂ pump but also serves as the basis for determining the reversible cathode potential. To further investigate the air electrode, simulation was conducted on the H₂ pump with the RHE cathode (Fig. S14C). A remarkable consistency was found in the current response against anode overpotential of the H₂ pump between the air cathode and the RHE cathode (Fig. S14D), consolidating the preceding simulation results for the air electrode.

It should be noted that the simulated current response (14.15 mA) under 0.050 V vs. RHE (pristine Pt surface) is over four orders of magnitude that of the current response experimentally obtained (1.1 µA) under 0.070 V vs. air electrode (i.e., 1.200 V vs. RHE, or Pt–O surface), under the same test conditions (5 ppm of H₂ concentration in air on the anode at 5000 sccm of flow rate, and pure air on the cathode at 200 sccm of flow rate). This contrast is in accordance with the observation of the HOR activity suppression by Pt–O formation in the air electrode. To further examine the suppressed HOR activity, the i₀ on the Pt catalyst of our H₂ pump was evaluated here under dry anode conditions at room temperature, giving rise to 0.71 ± 0.01 mA cm⁻² with both electrodes fed with ultrapure H₂ (Fig. S15). Compared with the i₀ at 60 °C in full hydration (1.7 ± 0.2 × 10² mA cm⁻²),¹² the measurement in this study is a 2.4-order of magnitude decrease, in concert with the HOR activity observed in the H₂-in-N₂ study (Fig. 3B).

In summary, we presented the realization of a H₂ pump, operating at room temperature, fed with ultralow concentrations of H₂ in air down to 1 ppm, while providing prominent current response following the concentration: 0.42–3.1 µA with 1–10 ppm of H₂ accordingly. A strong relation between the hydrogen response and the flow rate was observed in accordance with a fractional order (n = 0.453), supported by COMSOL Multiphysics simulations. The applied cell voltage was found to be critical for the high response of the hydrogen pump fed with ppm-level H₂ in air, leading to a narrow optimum range of 0.050–0.100 V. The findings in this work may extend the application of H₂ pumping to H₂ detection and other electrochemical systems.

We are grateful for the financial support by the U.S. Department of Energy through the Office of Energy Efficiency and Renewable Energy (#DE-EE0010742). This work was also supported by the U.S. National Science Foundation through both the ECO-CBET program (#2219172) and the EPSCoR FEC program (#2316482).

Conflicts of interest

The authors have no conflict of interest to declare.

Data availability

All data presented in this study are available upon reasonable request to the corresponding author.

Supplementary information (SI) is available. See DOI: https://doi.org/10.1039/d5cc03890a.

Notes and references

1. F. Barbir and H. Görgün, J. Appl. Electrochem., 2007, 37, 359–365.
2. K. Fishel, G. Qian, G. Eisman and B. C. Benicewicz, High temperature polymer electrolyte membrane fuel cells, 2016, p. 527.
3. J. Zou, N. Han, J. Yan, Q. Feng, Y. Wang, Z. Zhao, J. Fan, L. Zeng, H. Li and H. Wang, Electrochem. Energy Rev., 2020, 3, 690–729.
4. G. N. B. Durmus, C. O. Colpan and Y. Devrim, J. Power Sources, 2021, 494, 229743.
5. D. Marciuš, A. Kovač and M. Firak, Int. J. Hydrogen Energy, 2022, 47, 24179–24193.
6. J. M. Sedlak, J. F. Austin and A. B. LaConti, Int. J. Hydrogen Energy, 1981, 6, 45–51.
7. M. Trégaro, M. Rhandi, F. Druart, J. Deseure and M. Chatenet, Chin. J. Catal., 2020, 41, 770–782.
8. B. Rohland, K. Eberle, R. Ströbel, J. Scholta and J. Garche, Electrochim. Acta, 1998, 43, 3841–3846.
9. P. Bouwman, Fuel Cells Bull., 2014, 2014, 12–16.
10. P. J. Bouwman, J. Konink, D. Semerel, L. Raymakers, M. Koeman, W. Kout, W. Dalhuijsen, E. Milacic and M. J. J. Mulder, ECS Trans., 2014, 64, 1009.
11. A. Huth, B. Schaar and T. Oekermann, Electrochim. Acta, 2009, 54, 2774–2780.
12. M. D. Woodroof, J. A. Wittkopf, S. Gu and Y. S. Yan, Electrochem. Commun., 2015, 61, 57–60.
13. S. Toghyani, E. Afshari and E. Baniasadi, J. Energy Storage, 2020, 30, 101469.
14. M. Aziz, U. R. Aryal and A. K. Prasad, Int. J. Hydrogen Energy, 2022, 47, 33195–33208.
15. Z. Yang, Q. Du, Z. Jia, C. Yang, J. Xuan and K. Jiao, Appl. Energy, 2019, 256, 113959.
16. K. Onda, K. Ichihara, M. Nagahama, Y. Minamoto and T. Araki, J. Power Sources, 2007, 164, 1–8.
17. K. Onda, T. Araki, K. Ichihara and M. Nagahama, J. Power Sources, 2009, 188, 1–7.
18. Y. Jiang, H. Yang, K. Qin and T. Luo, Ceram. Int., 2024, 50, 40602–40613.
19. M. Schalenbach, T. Hoefner, P. Paciok, M. Carmo, W. Lueke and D. Stolten, J. Phys. Chem. C, 2015, 119, 25145–25155.
20. A. Solan and Y. Winograd, Phys. Fluids, 1969, 12, 1372–1377.

---