Ammonia (NH3) is the second-largest global chemical product, mainly used to produce agricultural fertilizer for sustaining nearly 40% of the world’s population [1]. Additionally, ammonia has been increasingly seeing as a viable carbon-free fuel for hard-to-decarbonize sectors, such as long-haul aviation and maritime transport. Currently, ammonia is mainly produced by the Haber-Bosch (H-B) process [2], which generally requires relatively high temperature (e.g., 400° to 500° C), high pressure (e.g., 150 to 350 atm), and rely on large-scale installations [3]. H-B operations are responsible for almost 2% of the global energy consumption and for approximately 1.6% of global carbon dioxide emissions [4]. After over 100 years of development since the invention of the H-B process, H-B operations have almost reached the theoretical maximum performance [5]. Alternatively to thermochemical approaches, such as the H-B process, NH3 can be synthesized by electrochemical [6], photochemical [7], and plasmachemical [5] methods.
Plasmachemical nitrogen fixation, which includes the synthesis of ammonia or other nitrogen-containing molecules, can be traced back to the Birkeland-Eyde (B-E) process [8], which predated the H-B process but depicted lower energy efficiency. The B-E uses an electric arc discharge to generate thermal plasma for the synthesis of nitric oxide (NO) as the nitrogen-carrier molecule. In thermal plasma, free electrons and heavy-species (i.e., molecules, atoms, ions) are in local thermal equilibrium at a relatively high temperature (typically above 1 eV ~ 11600 K). Therefore, thermal plasma processes generally operate at far higher temperatures than thermochemical ones. In contrast, in nonthermal plasma, the electrons are at a relatievly high temperature (e.g., ~ 1–10 eV), whereas the heavy-species are at a significantly lower temperature (e.g., from near room temperature to a few hundreds or thousands degrees Celsius). The use of nonthermal plasma for nitrogen fixation has been experiencing increased interest due to the potential for processes that depict high efficiency (due to limited gas heating) and are relatively compact and respond rapidly (thanks to being powered directly by electricity), making them particularly attractive for small-scale and distributed installations utilizing intermittent electricity from renewable energy sources. Moreover, plasma-based ammonia synthesis has a lower theoretical minimum energy cost, of approximately 0.2 MJ/mol [9], than the H-B process (~ 0.48 MJ/mol) [10, 11]. Examples of nonthermal plasma sources used for nitrogen fixation include glow discharge [12], radio-frequency discharge [13, 14], microwave discharge [15], dielectric-barrier discharge (DBD) [16], and plasma-liquid systems [17]. Among these technologies, DBD integrated with catalyst appears as one of the most promissing given its potential for increased efficiency and selectivity [18–26].
DBDs are configured by the incorporation of a dielectric medium within the discharge gap, usually covering at least one of the electrodes. The discharge gap is typically small, between 1 and 5 mm [27]. DBDs can be generated in relatively simple set-ups that are suitable for modular (linear) scalability and can operate at relatively low temperature (typically up to a few hundred degrees Celsius) even under atmospheric pressure conditions, making them particularly suitable for distributed and point-of-demand operations [28–30]. Moreover, DBDs are particularly suitable for the integration of a catalyst within the discharge to increase the performance (energy efficiency and/or production rate) of chemical conversion processes.
The integration of catalysts within DBDs for the synthesis of ammonia has been explored in both, packed-bed reactors [16, 23, 31–33] and membrane reactors [20, 34, 35]. In a packed bed reactor, the electrode gap is partially or fully filled with dielectric beads or pellets loaded with (generally metal) catalyst [36]. For example, Peng and coworkers [23] demonstrated that the use of ruthenium as catalyst with carbon nanotube support led to the highest ammonia production efficiency among a wide range of catalysts. Gershman and collaborators [32] used a DBD reactor packed with mesoporous silica catalyst, showing that the catalyst enhanced conversion, especially at lower plasma densities. Barboun et al. [33] investigated the use of Ru, Co, and Ni as metal catalysts loaded on γ-alumina in a DBD packed-bed reactor, showing that Co catalysts provided the highest yield.
In a membrane reactor, a membrane is incorporated to exploit its chemical and/or physical properties to enhance the performance of the chemical conversion process, e.g., by increasing reactivity or limiting the transport of certain species [37]. Importantly, the incorporation of a membrane can concurrently enhance both, chemical reactions and gas separation, leading to increased chemical conversion [38]. Examples of the use of membrane reactors for ammonia synthesis include the work by Mizushima and collaborators [20, 34], who used a porous alumina tube as membrane and catalyst support. The membrane was loaded with Ru, Fe, Ni, and Pt catalyst. Their result revealed that greater conversion is achieved by the incorporation of a metal catalyst (compared to just alumina), and that the Ru catalyst led to the highest ammonia yield. Carreon and co-investigators [35] investigated the use of a porous organic cage (CC3) on a porous alumina tube support as membrane. The CC3 membrane had the dual roles of gas adsorption and separation, avoiding in-situ decomposition of ammonia and leading to enhanced process performance. The investigators’ findings advanced the rational design of porous catalyst and membranes for plasma catalysis-driven ammonia synthesis.
The performance of a membrane reactor crucially depends on the properties of the membrane, such as pore size, structure, porosity, and permeability. Pore size refers to the empty space within a porous material, typically characterized by the equivalent diameter of a representative pore. Pores are often categorized into three types based on their size as micropores (less than 2 nm), mesopores (between 2 and 50 nm), and macropores (more than 50 nm) [39]. Porosity, calculated by dividing the pore volume by the total volume, represents the proportion of empty space within the solid material [40]. Permeability quantifies how easily fluid can flow through the material and depends on factors such as pore structure, connectivity, and distribution. Diverse geometric analogies can be drawn between the porous medium in a membrane reactor and that in a packed bed reactor (e.g., both depict certain pore size and porosity due to the pores within the material or the space between particles). Nevertheless, in practical terms, packed bed reactors often face challenges related to the packing of the catalyst uniformly and within the small discharge gap [44–46]. In contrast, membrane reactors can generally ensure consistent catalyst packing, while additionally provide separation of gas streams potentially increasing process yield [34].
Here, we present the design and evaluation of a membrane DBD (mDBD) reactor for ammonia synthesis. The reactor design rationale is schematically depicted in Fig. 1. Plasma is generated by alternating-current (AC) power between concentric cylindrical electrodes. The powered electrode is a porous metal tube that distributes hydrogen radially through a porous alumina membrane, with pore size dp and porosity f, surrounding it. The alumina membrane acts as dielectric-barrier and as distributor of hydrogen. We evaluated the mDBD reactor's operation with membranes with dp = 0.1, 1.0, and 2.0 µm (macropores), and also in conventional DBD mode using a non-porous (i.e., f ~ 0) dielectric. The radially-transported H2 interacts with the axially-transported N2 within the plasma region. This configuration leads to greater residence of N2 within the plasma (compared to pre-mixing the N2 and H2 streams), favoring its decomposition. Concurrently, it leads to the distributed availability of H2, which requires significantly less energy to decompose. The greater amount of atomic nitrogen together with the greater availability of atomic hydrogen (and their ions and electronically-excited counterparts) can lead to the increased formation of NH and NH2 radicals, and potentially to greater NH3 production. It is to be noted that no catalyst has been incorporated in the reactor in order to more clearly assess the role of the membrane. Nevertheless, this reactor configuration allows the subsequent integration of catalysts through different approaches, such as the loading of the membrane itself, or the packing of the discharge gap.
This paper is structured as follows: section 2 presents the design of the mDBD reactor; section 3 its evaluation by a computational thermal-fluid model; section 4 describes the experimental setup; section 5 addresses the characterization of the reactor operation; and section 6 presents results of its performance in the synthesis of ammonia; finally, conclusions are drawn in section 7.