Catalysts for the direct electrolysis of seawater

Research output: Book/ReportPh.D. thesis

86 Downloads (Pure)


The deployment of renewable energy sources has accelerated in recent years, and consequently, one of their inherent challenges became increasingly relevant: The main sources of renewable energy, wind and solar, are intermittent in nature, and the production capacity therefore does not match the energy demand at all times. As a result, they are only base-load capable if the energy produced can be stored and reconverted again when needed. Apart from energy production, a complete decarbonization of society also implies that CO2 free alternatives for industrial processes such as steel and cement production must be found. Hydrogen as an energy carrier can serve these purposes, and a general understanding has developed that green hydrogen production by means of water electrolysis will be a key technology to reach the above mentioned goals.
Although alkaline water electrolysis has been commercialized since the early 20th century, efficiency is still a major shortcoming of this technology. The work presented herein is therefore concerned with the development of highly active, noble metal-free catalysts for alkaline electrolysis and, more generally, the efficient exploration of the associated chemical spaces using machine learning guided experimentation.
The work conducted in the context of this thesis can be divided in three parts. In a first approach, a co-precipitation method was used to screen various combinations of single, binary and ternary metal hydroxides derived from eight different metals (Fe, Cr, Al, Co, Ni, Mn, Zn, Cu) with regard to their oxygen evolution reaction (OER) activity. The compounds are deposited directly onto Ni foam substrates, mimicking technical electrodes. Electrochemical tests were conducted in conventional and saline KOH to assess the suitability for direct seawater electrolysis. Samples containing Fe generally outperformed Fe free compositions, and NiFeCr showed the lowest overpotential of all samples with 247 mA cm−2 in 1 M KOH. OER activity was found largely similar between the two electrolytes, although some compositions are identified with considerable differences in overpotential such as NiMn.
Subsequently, an autonomous experimentation system was designed based on the same synthesis method to optimize the OER overpotential in complex multi metal hydroxides. A combination of lab robotics, custom made instrumentation and machine learning guided optimization allows for extremely high experimental throughput and efficient material screening without human interaction. The system has the capability to independently test and optimize the material composition under practical conditions in a closed loop. To the best of the authors’ knowledge, it is the first system of this kind applied to alkaline OER catalysts. First results illustrate the excellent data quality and reproducibility, as well as the general feasibility of the optimization approach. The trends known from previous manual experiments were reproduced with the autonomous system and multi metal doping showed high sensitivity of the overpotential with respect to the composition.
Lastly, highly porous Ni hydrogen evolution reaction (HER) catalysts are prepared by machine learning guided electrodeposition. The synthesis parameters current density, temperature, deposition time and ligand concentration are varied to increase the electrochemically active surface area and consequently HER activity. The best samples achieve 10 mA cm−2 at a very low overpotential of -117 mV, rivalling the most active reported Ni electrodes to date. The use of common laboratory instruments in connection with the human-in-the-loop approach makes this highly efficient workflow accessible to many researchers.
Original languageEnglish
Place of PublicationKgs. Lyngby
PublisherTechnical University of Denmark
Number of pages192
Publication statusPublished - 2023


Dive into the research topics of 'Catalysts for the direct electrolysis of seawater'. Together they form a unique fingerprint.

Cite this