This work discusses the potential of combining non-thermal plasmas and conducting membranes for in situ resource utilization (ISRU) on Mars. By converting different molecules directly from the Martian atmosphere, plasmas can create the necessary feed-stock and base chemicals for processing fuels, breathing oxygen, building materials, and fertilizers. Different plasma sources operate according to different principles and are associated with distinct dominant physicochemical mechanisms. This diversity allows exploring different energy transfer pathways leading to CO dissociation, including direct electron-impact processes, plasma chemistry mediated by vibrationally and electronically excited states, and thermally driven dissociation. The coupling of plasmas with membranes is still a technology under development, but a synergistic effect between plasma decomposition and oxygen permeation across conducting membranes is anticipated. The emerging technology is versatile, scalable, and has the potential to deliver high rates of production of molecules per kilogram of instrumentation sent to space. Therefore, it will likely play a very relevant role in future ISRU strategies.
I. INTRODUCTION
Spaceflight programs are ever expanding, and prospects of humans colonizing space to become a multi-planetary species are on the horizon.1–4 To open the solar system for human exploration and colonization, several ongoing and planned space missions target the Moon and Mars.5–15 Their goals vary from pure scientific research to the development of the technologies needed to live and work on another world. A landmark is NASA’s Artemis III mission,15 which foresees landing humans on the surface of the Moon in 2024, the first astronauts to walk on the Moon in over 50 years.15 The long-term goals of the Artemis program are to establish the first sustained presence on the Moon and to use the knowledge acquired on and around the Moon to take the next giant leap: sending the first astronauts to Mars.
In situ resource utilization (ISRU) is crucial in this endeavor. Harnessing of resources in the exploration site instead of bringing them from Earth builds toward the self-sufficiency of space-bases and missions and reduces the logistics, expenses, and risks to the crew. The highest-impact ISRU products that can be used early in human operations are mission consumables, including propellants, fuel cell reactants, and life support commodities.16 This is also the case for Mars exploration, where, at least on a first phase, ISRU will most likely be limited to Mars atmosphere resources for mission-critical applications.16
The main component of the Martian atmosphere is carbon dioxide (95.9%), with smaller percentages of Ar (1.9%), N (1.9%), and other gases. The abundant CO can be converted directly from the atmosphere into oxygen (O ) and carbon monoxide (CO). O can then be collected and made available for breathing, feeding indoor environments. Portable breathing devices that directly convert CO from the atmosphere into O , allowing astronauts to wander outside without the need for O storage, can also be dreamed of. In addition, both CO and O can be used in a propellant mixture in rocket vehicles.17 Decomposition can be further pursued to arrive at carbon (C), of use for in situ manufacturing of carbon structures and for the synthesis of different organic molecules. Besides, this route doubles the oxygen production, albeit at an extra energy cost. Carbon is also a fertilizer,18 and a carbon feed-stock is required for future Martian agriculture. Nitrogen (N ) is another constituent of the atmosphere of Mars of interest for ISRU. It is essential for life support as a breathing gas and, once oxygen has been made available, can be used for the local production of NO for fertilizers.19,20
The only concrete proposal for oxygen production on Mars to date is the MOXIE experiment,21–25 under the framework of the Mars 2020 (Perseverance) mission.26 It is based on solid oxide electrolysis cells (SOECs), in which, under electric bias, electrode materials offer catalytically active sites for CO reduction to CO and oxygen evolution reactions, respectively, while the electrolyte provides the required pathway for transporting oxide ions from the cathode to the anode. MOXIE’s SOEC operates at temperature above 1000 K and pressures in the range of 260–760 Torr.25 Despite being a readily available technology, SOEC has several technical downsides regarding CO conversion. The difficulty to break the carbon–oxygen double bond poses challenging requirements on electrode materials. The most commonly used electrode material is based on Ni cermet that is susceptible to degradation under reduction oxidation cycles and carbon deposition, requiring protective CO atmosphere and operating conditions to maintain stable operation.27–29 Particularly, under the Martian scenario, the need for high temperatures27 calls for a careful design of the thermal insulation system; the process of collecting and compressing the thin atmosphere of Mars (about 160 times thinner than Earth’s atmosphere) prior to the decomposition process imposes a technological limit that calls for additional pumping equipment to compress the gas inside the reactor, increasing the mass and power requirements of the system.25
A new and complementary approach is proposed here, linking two emerging technologies: non-thermal plasmas and conducting membranes. Non-thermal plasmas are highly reactive gas media sustained by electrical discharges, which allow the coexistence of energetic electrons ( eV) with relatively cold gas molecules. Under these conditions, far from thermodynamic equilibrium, the discharge power can be selectively channeled, offering unique ways to break the strong bond by taking advantage of the energy stored in the internal degrees of freedom. The combination of plasma with oxygen separation membrane-based systems may rely on either pure oxygen ion conducting30,31 or mixed oxygen ion electron conducting32–34 membranes. Either way, the system allows a compact design and direct separation of the conversion products.
Other strong points in favor of plasma technologies are that they are compact, scalable, reliable, versatile, do not require the use of expensive materials, operate on (renewable) electricity and can be powered by solar panels and batteries, can instantaneously start and stop operation (being thus perfectly adapted to a power supply from intermittent renewable energy sources), and can operate directly under Martian conditions without the need for compression (as the Martian pressure is ideal for plasma ignition35) or external heating on the O production step.
Non-thermal plasmas have gained much attention in the last decade in the context of climate change and CO valorization on Earth due to their potential to activate CO at reduced energy cost.36–38 Dissociation of CO is one of the critical steps on the way to produce green synthetic fuels and platform molecules for the chemical industry and has been the focus of many studies, exploring electron impact,39,40 vibrationally driven,41,42 and thermal dissociation routes.43–45 As an extension of the vast investigation addressing the production of solar fuels on Earth (see Pietanza et al.46 and George et al.38 and references therein), it was recently noted that the knowledge acquired on Earth can be transferred to some extent to ISRU on Mars.47 The very strong case supporting plasma-based production of oxygen on Mars presented by Guerra et al.47 predicted theoretically that the Martian atmospheric conditions of pressure, temperature, and gas composition are very favorable to ignite a plasma system and to achieve efficient CO decomposition under a non-thermal regime. It was further noted that the presence of Ar and N from the Martian atmosphere contributes to further enhancing plasma dissociation and that the required power for discharge operation is typically W and can be as low as W at gas flows in the range 2–10 sccm,48,49 perfectly attainable on Mars.
The feasibility of oxygen production on Mars using plasma technologies was subsequently corroborated in two experimental campaigns, carried out at Laboratoire de Physique des Plasmas, Ecole Polytechnique, France, and the Dutch Institute For Fundamental Energy Research (DIFFER). In the first campaign, a DC glow discharge was cooled down to Martian temperatures by inserting a plasma reactor inside a cold bath of dry ice and ethanol.49 The cold Martian temperatures did increase the CO vibrational degree on non-equilibrium, a beneficial effect of Ar and N was observed, and the reactor provided dissociation fractions in the range of 10%–30%. These results are very encouraging considering that the plasma setup used was designed for fundamental research and is far from suited to the development of a prototype. The second experimental campaign showed that microwave (MW) discharges, operating under Martian composition and pressure, can provide CO conversions of about 35% for a power around 300 W and gas flow of 50 sccm,50 bringing the plasma results to a very competitive standard.
The other piece required to produce mission consumables is the separation of the products produced in the plasma. For pure oxygen ion-conducting membranes, the oxygen separation step may leverage on SOEC, using the same principle as in MOXIE. The plasma may be generated on top of the solid oxide membrane, to assist CO conversion, and thus relax electrode requirements. This configuration is expected to increase the flow of oxygen through the membrane since the production of O atoms in a plasma is much larger than on the surface of a conventional cathode. Furthermore, it has enhanced durability and brings cost reduction. Yttria (or Scandia) Stabilized Zirconia YSZ (ScSZ) seems to be good candidates to be used in an oxygen ion-conducting membrane. A heat integration is also possible in order to use the heat losses of plasma to heat the SOEC. Preliminary results, obtained in a DC hollow cathode discharge used to study O separation in a plasma + YSZ reactor, show evidence of increased durability of the SOEC process when used with a plasma enhanced gas stream, observed already with a dissociated flow of CO /O /CO.30 Furthermore, a synergistic effect due to the presence of electrical charge was also evinced by plasma exsolution of nanoparticles,31 which brings additional electro-catalytical pathways to improve the process.
A separation stage based on mixed ionic-electronic conduction (MIEC) membranes is also to consider. MIEC membranes are less complex than SOEC but require additional pumping, as the oxygen permeation is driven by its partial pressure gradient across the membrane.32–34 These membranes are also good candidates to benefit from plasma-assistance since the plasma can enhance the surface exchange kinetics owing to the presence of free electrons and oxygen radicals, which lead to increased oxygen permeation fluxes.33 Challenges associated with MIEC are their stability in CO and the possibility of back reactions during operation (that lead to a reduction of the total permeation flux).
The preliminary results already obtained for plasma CO dissociation in DC and MW discharges support the estimation that an optimized system may produce oxygen at a rate of 14 g/h using a 6 kg plasma reactor, amounting to 2.3 g of oxygen produced per hour per kg of equipment sent to Mars.49,50 These values are about six times larger than obtained by the current operation of MOXIE and are discussed in Sec. V.
The structure of this paper is as follows. Section II describes the types of plasma reactors more commonly used in CO conversion studies and their characteristics. Section III reviews the available solutions for O separation and anticipates the strategy foreseen for a coupled plasma-membrane system. Section IV addresses the current status of plasma chemistry in CO plasmas and brings new results and insight into the operation of low-temperature plasmas in a synthetic Martian atmosphere. Section V gives an overview of the figures of merit for the perspective of using plasmas for in situ resource utilization on Mars and how they compare with MOXIE, establishing the high potential of the emerging technology. Finally, Sec. VI summarizes the main results, gives directions for future research and concludes the paper.
II. PLASMA REACTORS
Plasma reactors use an external electric field to generate and accelerate electrons, which then share their energy with the rest of the gas through electron-impact excitation, dissociation, and ionization collisions, creating a very reactive environment. This reactivity is of particular interest for CO conversion, as CO is a highly stable molecule, which typically requires substantial energy input and active catalysts for chemical reactions to take place.51 From a practical perspective, important metrics of performance for plasma reactors are energy efficiency and conversion rate. From a fundamental perspective, plasma reactors can be classified in terms of their reduced electric field, , where is the electric field, is the gas density, and electron density, .37 Here, we part from the fundamental parameters and connect them to the observed or expected performance of different plasma reactors for CO conversion.
The reduced electric field determines the average electron energy or electron “temperature,” which in turn dictates which electron-impact reactions are favored and how the energy is shared with the gas, as it is shown, e.g., in Fig. 1 of Grofulović et al.52 and in Fig. 11 of Ogloblina et al.53 The value of in the discharge is closely related with the energy efficiency of the process. In gaseous CO , nonequilibrium conditions arise for Td (1 Td is equivalent to 250 V applied across a 1 cm gap at standard pressure and temperature): beyond this value, electron temperatures greater than the temperature of the gas can be accessed and efficient energy pathways of CO dissociation are favored.54,55 Between –100 Td, CO dissociation mainly follows from electron-impact excitation of low vibrational levels, (0.1–1 eV), which interact among themselves through vibrational–vibrational (V–V) energy exchange processes, climbing the vibrational ladder, and reaching energy levels that can result in dissociative quenching.51 Gliding arcs (GA), microwave (MW) discharges, and radio-frequency (RF) discharges are different types of plasmas that can access this range of reduced electric fields, which explains why the literature on CO conversion, seeking high energy efficiency, has greatly focused on these plasma sources.56 Figure 1 summarizes the typical parameter space of operation of various types of discharges.57–61
Overview of different plasma sources and typical values of electron temperature, reduced electric field and electron density reached in their operation: CO DC glow at 1 mbar (photo credit: O. Guaitella),57 2% CO in argon RF plasma at 1 bar (photo credit: T. Gans, O. Guaitella), CO MW plasma at 200 mbar—bottom image (photo credit: F. Peeters), CO MW plasma at 7 mbar—top image (photo credit: T. Silva),59 CO surface DBD at 5 mbar (photo credit: O. Guaitella), CO /N DBD plasma at 200 mbar (photo credit: O. Guaitella), postcombustion products NPD DBD plasma at 1 bar (photo credit: C. Pavan, C. Guerra-Garcia),62 CO fast ionization wave (FIW) inside a capillary tube at 10 mbar (photo credit: G. Pokrovskyi).60,63 All photos included with permission from the authors.
Overview of different plasma sources and typical values of electron temperature, reduced electric field and electron density reached in their operation: CO DC glow at 1 mbar (photo credit: O. Guaitella),57 2% CO in argon RF plasma at 1 bar (photo credit: T. Gans, O. Guaitella), CO MW plasma at 200 mbar—bottom image (photo credit: F. Peeters), CO MW plasma at 7 mbar—top image (photo credit: T. Silva),59 CO surface DBD at 5 mbar (photo credit: O. Guaitella), CO /N DBD plasma at 200 mbar (photo credit: O. Guaitella), postcombustion products NPD DBD plasma at 1 bar (photo credit: C. Pavan, C. Guerra-Garcia),62 CO fast ionization wave (FIW) inside a capillary tube at 10 mbar (photo credit: G. Pokrovskyi).60,63 All photos included with permission from the authors.
Early experiments with MW reactors report energy efficiencies of (80%–90%;36 yet, these typically require both low pressure (100 mbar) and low-temperature conditions so that vibrational–translational (V–T) relaxational processes and reverse oxidation of CO into carbon dioxide play a minor role.46 In practice, these high efficiencies have not been reproduced in the past decade. It is now recognized that for GA, MW, and RF plasmas, jointly referred to as warm plasmas, the gas temperature typically exceeds a few thousand kelvin and vibrational–translational (V–T) relaxation processes, as well as backward reactions, cannot be neglected,64 lowering the energy efficiency to the 40%–60% range.51 In these regimes, thermal dissociation in the discharge core has a dominant role, and optimization of the postdischarge environment, rather than the plasma environment, offers most opportunities for improvement of the plasma-chemical conversion process due to the importance of kinetics and turbulent transport in the cooling trajectory.44,45
Below Td, the energy channeled to ionization in CO plasmas is negligible compared with the energy spent in excitation, with the electron energy being essentially consumed in the excitation of the vibration levels.52,53 At Td, electron temperatures exceed a few eV and electronic excitation starts to dominate over the vibration modes,53,65 leading to higher-energy consumption channels of CO dissociation. However, the higher degree of nonequilibrium, lower gas temperature, and higher ionization degree all favor higher conversion rates. So far, the most popular plasma operating in the high regime has been the dielectric barrier discharge (DBD). DBDs are driven by AC voltage, and at least one of the electrodes is covered by a dielectric to limit the current flow in every half cycle of operation. Despite its modest energy efficiencies, typically below 15%,51 the popularity of the DBD is justified by its operation at atmospheric pressure and room temperature, scalability, ease of operation, and reasonable dissociation yields of up to 40%.51,66
Although the results with MW, RF, GA, and DBD plasmas are already very promising, we believe that improvements are possible in terms of energy efficiency and/or conversion through further engineering of the plasma sources. For instance, DBD reactors, which operate at low temperature, offer the possibility to incorporate catalyst materials in the plasma to control the selective production of value-added compounds when considering co-reactants.38 Finally, the question arises if an increased yield justifies a penalty on energy efficiency for reasonable power consumption. In particular, it should be kept in mind that an increased dissociation facilitates the process of separation of the products through the membrane (see Sec. III) and that an energy balance for the overall process should be done and not only for the plasma decomposition.
From this Perspective, of particular interest is the excitation of the gas using ultra-short high-voltage pulses so that there is not enough time to significantly change the electron concentration.54 Under such fast raising voltage pulses, the gap can endure an over-voltage that is higher than the static breakdown condition,68,69 and high , in the range of hundreds of Td, can be sustained. These energy pathways have proven to be extremely attractive in the related fields of plasma-assisted combustion and plasma-assisted ignition (PAC/PAI) by providing efficient energy deposition into electronic excitation.54,62,70 Experiments in air and nitrogen using nanosecond pulsed discharges (NPDs) have shown that electronic excitation by electron impact results in oxygen dissociation via collisions with electronically excited atoms and molecules67 and in fast gas heating, which is heat release by collisional relaxation of electronically excited species at rates faster than V–T relaxation, at rates around 50 K/ns at atmospheric pressure.67,71
Recent experiments using fast ionization waves (FIWs) in capillary tubes triggered by NPD operating at 10–20 mbar have confirmed that these mechanisms are also present in CO , in reactions of quenching of CO( ), O( ), and O( ) metastable states at sub-microsecond timescales.60,63 FIW are an interesting regime of pulsed breakdown, reached for very high over-voltages.69,72 In the single pulse regime, energy efficiencies and dissociation fractions of around 20% were demonstrated at reduced fields of 200–400 Td.60,63 More interestingly, the conversion fraction was increased to over 90% in a repetitively pulsed regime, (100 Hz–1 kHz).63 Although the results are promising, the relative role of gas heating and dissociation following electronic excitation is not clear at present, as well as how these results will translate to NPD regimes at lower over-voltages, which appear more readily than FIW when using nanosecond repetitively pulsed (NRP) discharges at kHz frequencies using pin-to-pin electrode configurations at atmospheric pressures and above.72
Recent work at Centrale Supelec73 and Università di Trento61 have addressed experiments of NRP in atmospheric pressure CO using pin-to-pin electrode configurations. The experiments by Maillard et al. used a very small electrode gap of 3 mm to generate a fully ionized filament, despite the ns-pulsed voltage, with electron density in the order of cm and measured temperatures of 30 000 K.73 This regime is very interesting as it yields reasonable electron temperatures and high electron number densities. The experiments by Ceppelli et al. also observed the fully ionized regime using a similar 5 mm gap and a pin-to-pin electrode setup.74 They characterized the discharge in terms of the NRP voltage waveform, which typically displays multiple reflections due to mismatch of the impedance between the power supply and the variable-impedance plasma load, identified as two discharge types.61 The first one, on the first region, corresponds to breakdown, has relatively low charge density, high energy electron, requires high voltage for ignition, and dissipates most of the discharge energy. The second type, characteristic of the consecutive temporal regions of the pulse, has a peak electron density of cm , a low electron energy, and requires a much lower voltage to ignite, which corresponds to a spark and offers favorable conditions for vibrationally enhanced dissociation.61 Additional studies reveal a significant dependence of both conversion fraction and energy efficiency with the inter-pulse time. A burst mode (short inter-pulse times) was found to perform better than the continuous mode, for the same total energy, with maximum conversion (downstream) around 20% and maximum energy efficiency close to 60%.75 Notice that the results on conversion exhibit a peak of dissociation around 70%,74 which decays to the mentioned 20% during the afterglow. This suggests that an adequate quenching of the dissociation products in the afterglow may largely increase dissociation and energy efficiency, in line with the results in MW discharges.44,45
Ultimately, more sophisticated electrical schemes can be devised by combining independent voltage waveforms to circumvent the otherwise limited degree of control over and species generation. NRP/RF hybrid plasmas in molecular gases, namely, nitrogen and its mixtures with H , CO, and CO , have recently demonstrated that selective electronic excitation of N and ionization during the NRP waveform, followed by vibrational excitation of the ground electronic state molecules during the RF waveform, is indeed possible.76 Considering the temporal waveforms of RF and NRP sources, this scheme can be sustained with a single pair of electrodes so that the complexity of the reactor is not increased. The approach can also be extended to any reacting molecular gas mixture to generate electronically excited molecules as well as atomic species (by the pulsed discharge) and vibrationally excited molecules (by the sub-breakdown RF discharge).76 Other hybrid strategies have also been proposed, including sub-breakdown RF fields + laser-induced plasmas to increase the plasma volume compared to the laser-only strategy;77 multi-frequency RF voltage waveforms to control the electron energy distribution function and, therefore, the plasma-produced chemical activity;78 and DC + NRP plasmas to separately control the vibrational (low ) and electronic excitation (high ) in the discharge, respectively.79–81 These results open the possibility of somewhat decoupling conversion rate and energy efficiency and constitute a new and open field of research, which still requires further investigation. Eventually, hybrid plasmas might enable the optimization of the strategy to either maximize conversion rate, efficiency, or some combined figure of merit in response to the mission’s needs.
III. PRODUCT SEPARATION
Aside from sophisticated plasma excitation schemes to improve gas-phase conversion and efficiency, another path toward enhanced ISRU is the combination of plasma with (solid) membrane gas-separation methods. For the case of plasma operated in pure CO , the products are always a mixture of CO , CO, and O , necessitating a means of separating these products for further utilization. While conversion and separation can be detached from one another in time and space, synergistic effects can be expected to occur if the two are operated in tandem. Similar to introducing a catalytic surface into a DBD plasma to control product yields,82,83 an oxygen-permeable membrane bounding the plasma will naturally shift the equilibrium of the CO dissociation reaction toward increased levels of CO and O.84
Three types of oxygen-permeable membranes can be distinguished:
Non-electrochemical membranes: materials that do not rely on charged species to achieve transport of oxygen but are nonetheless selectively permeable to oxygen species. An example is silver, which has a high diffusion coefficient for atomic O, especially along grain boundaries.85
Mixed ionic-electronic conductors (MIECs): materials that show electrical conductivity for both electrons and oxygen ions. Depending on the material and the operating conditions (e.g., temperature and oxygen partial pressure), electrical conductivity between the two conducting species may vary significantly, but generally, the conductivity of electrons is higher in perovskites. A broad variety of (ceramic) MIEC materials exist, showing oxygen ion conductivities in the range of 1–1000 S/m at K.32 Notice that conductivity can affect the surface reactions if any of the steps involve charge transfer.
Pure oxygen ion conductors: materials that show electrical conductivity dominantly via oxygen ions moving between vacancies in the material. To be classified as a purely ionic conductor, the ionic conductivity must be 2 orders of magnitude larger than the electronic conductivity .86 Due to substantial energy barriers for diffusion of atoms through the lattice, significant electrical conductivity requires high operating temperatures. Yttria or scandia-stabilized zirconia (YSZ, ScSZ) are O conductors at 0.1–1 S/m at temperatures of K.86
Bulk-to-surface mass transfer of gaseous oxygen (feed side to the membrane surface).
Dissociation (surface exchange). The oxygen molecule is adsorbed on the (i) SOEC electrode or (ii) the MIEC membrane surface and dissociates catalytically first to oxygen atoms and then in oxygen ions (O ). On the feed side, this can be expressed, using the Kröger–Vink notation, as 1/2O + 2e + V O , where V refers to oxygen vacancies in the membrane of the SOEC or MIEC and O to oxygen ions (O ) occupying the oxygen lattice.
Ionic transport (bulk diffusion). The oxygen ions diffuse through the membrane lattice (mainly oxygen vacancies, but also other defects) under (i) an electric bias for the case of SOEC and (ii) a pressure gradient between the feed and permeate side for the case of MIEC. To maintain electrical neutrality, electrons are transported at the same time in the opposite direction via (i) the external SOEC electrical circuit and through the electrode and (ii) the MIEC membrane.
Association (surface exchange). The oxygen ions recombine to form oxygen molecules and desorb from the surface of the membrane. The reaction involved in this step is the opposite of step 2.
Surface-to-bulk mass transfer. Adsorbed oxygen on the SOEC electrode or the MIEC membrane surface is desorbed to gaseous oxygen in the permeate side.
While Eqs. (3), (5), and (6) are all simplified cases excluding surface reactions, they do convey the general transport behavior: in all cases, thinner membranes and increased diffusivity can produce higher fluxes. The latter can be achieved by higher operating temperatures since can typically be described by an Arrhenius-type behavior with an exponential factor , with being an activation energy. Furthermore, both non-electrochemical and mixed conductors depend on a species gradient as the driving force, while purely ionic conductors depend on an externally applied voltage .86
A. Plasma with non-electrochemical membranes
The best example of a non-electrochemical membrane, designed with Mars ISRU in mind, is represented by the work of Outlaw and co-workers.85,92–95 Identifying silver as a membrane with selectivity toward oxygen species, the thermally activated nature of the diffusion coefficient of O atoms in Ag was mapped in the temperature range of 673–1073 K,85 followed by combination of these membranes with both O and CO glow discharges. For a 320 m Ag membrane at 923 K, the O flux was improved by a factor of 6.5 when comparing an inert O gas and a 17.5 W O plasma on one side of the membrane.92 This improvement is attributed to the rate-limiting step of O needing to stick to Ag and subsequently dissociating into O atoms prior to transport, being alleviated by the O-atom flux from the plasma.
In a follow-up study by Wu et al. using 350 m Ag0.05Zr,93 the upstream O-atom concentration in the membrane, or in Eq. (3), was determined with and without an O glow discharge; see Fig. 3. These data demonstrate that higher throughput can be achieved at lower temperature with the aid of plasma through an increase in the species gradient. For K, however, the (sub-)surface O-atom concentration (as well as the flux ) declines to the value without plasma. This is attributed to O atoms rapidly recombining to O and desorbing back into the gas phase at these higher temperatures. Nevertheless, the experiment demonstrates that plasma allows for reduced operating temperatures without sacrificing performance (i.e., the same O flux could be achieved at 570 K compared with 715 K without plasma).93 With the same membrane, a similar performance was found for CO plasma and O plasmas.94
Sub-surface O-atom concentration in a Ag membrane, determined from measured O fluxes. Data taken from Fig. 8 in Wu et al.93
Sub-surface O-atom concentration in a Ag membrane, determined from measured O fluxes. Data taken from Fig. 8 in Wu et al.93
As Eq. (2) suggests, a clear route toward improved oxygen species transport is using thinner membranes. This was investigated by Premathilake et al., where the previous 350 m Ag membranes were reduced in thickness to 25 m.95 This 14-fold decrease in thickness was shown to increase O flux from 1×1014 molecules/s/cm to above 1×1015 molecules/s/cm on exposure to a CO glow discharge at the same operating temperature.94 While the conditions of the plasmas used in both studies are different in terms of applied power and electrode configuration, most of this improvement can be comfortably assigned to the decrease in membrane thickness.
B. Plasma with mixed conductors
A hollow fiber mixed conductor (perovskite-type, La Ca Co Fe O or LCCF: a dominantly electronic conductor) was placed in the effluent of a CO microwave plasma torch by Chen et al.33 The interior of the fiber was continuously flushed by pure Ar, ensuring the highest possible O partial pressure gradient over the 0.2 mm fiber wall. At K, oxygen permeation rates were 1.8×1018 molecules/s/cm in the plasma environment vs 5.7×1017 molecules/s/cm in an inert gas containing 6.5 O . Chen et al. also demonstrate that, with a wall thickness of 0.2 mm and without plasma, they operate in a regime where the surface reactions are rate-limiting; i.e., the concentration of oxygen ions achieved in the sub-surface in Eq. (6) reaches a plateau and is not proportional to the partial pressure of O outside the surface. The improved oxygen permeation rates in the plasma environment can, therefore, be attributed to O radicals from the plasma being more readily incorporated into the membrane material than molecular O , thus increasing in Eq. (6) compared to the case without plasma.
A mixed conductor (La Sr Co Fe O , or LSCF, another dominantly electronic conductor) of 1 mm thickness was used as an end cap on a cylindrical DBD by Zheng et al.96 Operating the DBD at 15 W in air, at K, improved the oxygen permeation rate by a factor of nearly 30, to 2.8 × 1016 molecules/s/cm . At K, however, the oxygen permeation rate was only slightly increased by plasma, to a maximum of 1.2 × 1017 molecules/s/cm . Using data obtained at multiple temperatures, the apparent activation energy for oxygen permeation was shown to decrease from 1.42 to 0.45 eV/molecule. Since bulk diffusivities are unlikely to be affected by plasma species, this improvement can also be attributed to the easier incorporation of oxygen species from the plasma, such as O , in the membrane.
C. Plasma with oxygen ionic conductor systems
An even more intriguing possibility for merging plasma with electrolysis is to place both devices into a single series circuit and powering both using the same voltage source. In this scheme, one of the electrodes can be moved away from the electrolyzer membrane, creating a gas gap in which a plasma can be ignited. This configuration was investigated by Steinmüller et al. using pure O DC discharges.97 Plasma was successfully ignited in both a Pt YSZ plasma Pt and Pt YSZ plasma YSZ Pt configuration and compared against a Pt plasma Pt baseline. At K and an applied voltage of 1.2 kV, respective currents of 4.7, 6.0, and 5.7 mA were observed. With membrane(s) and plasma forming a series circuit, these currents were carried by O moving through the YSZ, which clearly does not form a severe limiting factor in sustaining the plasma. Furthermore, only a minor degradation of the 1 mm thick YSZ membranes was evident at temperatures above 570 K, for both polycrystalline and (100) single crystal YSZ, after 20 h of operation. Unfortunately, in the experiments of Steinmüller et al., the single-sided Pt YSZ plasma Pt configuration was tested only with the YSZ on the anode side of the plasma, meaning that O was supplied to the plasma volume and not extracted. However, since their highest currents are achieved in the experiment with YSZ on both sides of the plasma, the extraction of (charged) oxygen species from the cathode side of the plasma is likely the least rate-limiting.
Similar to reaction 9, fluxes of negative ions O and O , as well as electronically excited species O( D), O( S), O (a), and O (b), could undergo reaction to O on the plasma/YSZ interface at potentially higher rates than reaction 8 due to the energy already imparted to these species by the plasma. If reactions like 9 are indeed effective substitutes for 8, as the experiments of Steinmüller et al. suggest, in parallel with similar observations for non-electrochemical and mixed conductor membranes combined with plasma,33,92,93 two advantages of a hybrid plasma/SOEC approach emerge:
the electrolyzer no longer requires a potentially costly, and prone to degrading, catalytic electrode on its anode side (plasma cathode side) and
with the plasma supplying energetic species to the YSZ surface, the membrane could become effective at lower operating temperatures.
The advantage of ionic conductors compared with non-electrochemical or mixed conductor membranes is the free parameter [see Eq. (5) compared with Eqs. (3) and (6)]: the O flux can be controlled via the applied voltage and does not directly depend on the O concentrations on either side of the membrane. Essentially, an ionic conductor membrane like YSZ can act as a pump for selective transport of O . For Mars ISRU applications, this could be a distinct advantage since it alleviates the need for separate hardware to minimize the concentration of O on the downstream side of the membrane, as is the case for non-electrochemical or mixed conductor membranes. However, this gain comes with an added complexity of the design.
Table I summarizes the experimental values of O fluxes achieved with different types of membranes. It is worth noting that all these membranes rely on high temperatures to operate. Heating can potentially be supplied by the plasma or alleviated by the plasma effects, which is one of the interesting aspects of merging plasma and membrane technologies.
O2 fluxes (molecules/s/cm2) achieved with various membrane types, with and without plasma. MOXIE is included for comparison, although ScSZ membrane thickness is proprietary knowledge and remains unpublished. The − and + signs identify the anode and cathode, respectively, while ScSZ stands for “scandia-stabilized zirconia.” The results marked with * were obtained in an air mixture.
Type . | Membrane . | d (μm) . | T (K) . | ϕno plasma* . | . | . | p (mbar) . | Reference . |
---|---|---|---|---|---|---|---|---|
Non-electrochemical | Ag | 320 | 1173 | 4.40 × 1013 | 2.83 × 1014 | 1.74 × 1014 | 0.7 | 85 |
Ag0.05Zr | 320 | 723 | 3.40 × 1012 | 2.10 × 1013 | … | 4.0 | 93 | |
Ag | 350 | 723 | 1.90 × 1012 | 0.60 × 1014 | 1.0 × 1014 | 6.7 | 94 | |
Ag | 25 | 723 | … | … | 1.61 × 1015 | 6.7 | 95 | |
Mixed conductors | LCCF | 200 | 1173 | 5.74 × 1017 | … | 1.77 × 1018 | 1000 | 33 |
LCSF | 1000 | 873 | 9.9 × 1014 | 2.8 × 1016* | … | 1000 | 96 | |
1173 | 9.9 × 1016 | 1.2 × 1017* | … | 1000 | 96 | |||
Ionic conductors | YSZ (−) | 1000 | 820 | … | … | 1.27 × 1015 | 0.5 | 97 |
YSZ (+ and −) | 1000 | 820 | … | … | 1.33 × 1015 | 0.5 | ||
MOXIE | ScSZ | ? | 1100 | 2.38 × 1017 | … | … | 1000 | 22 |
Type . | Membrane . | d (μm) . | T (K) . | ϕno plasma* . | . | . | p (mbar) . | Reference . |
---|---|---|---|---|---|---|---|---|
Non-electrochemical | Ag | 320 | 1173 | 4.40 × 1013 | 2.83 × 1014 | 1.74 × 1014 | 0.7 | 85 |
Ag0.05Zr | 320 | 723 | 3.40 × 1012 | 2.10 × 1013 | … | 4.0 | 93 | |
Ag | 350 | 723 | 1.90 × 1012 | 0.60 × 1014 | 1.0 × 1014 | 6.7 | 94 | |
Ag | 25 | 723 | … | … | 1.61 × 1015 | 6.7 | 95 | |
Mixed conductors | LCCF | 200 | 1173 | 5.74 × 1017 | … | 1.77 × 1018 | 1000 | 33 |
LCSF | 1000 | 873 | 9.9 × 1014 | 2.8 × 1016* | … | 1000 | 96 | |
1173 | 9.9 × 1016 | 1.2 × 1017* | … | 1000 | 96 | |||
Ionic conductors | YSZ (−) | 1000 | 820 | … | … | 1.27 × 1015 | 0.5 | 97 |
YSZ (+ and −) | 1000 | 820 | … | … | 1.33 × 1015 | 0.5 | ||
MOXIE | ScSZ | ? | 1100 | 2.38 × 1017 | … | … | 1000 | 22 |
IV. PLASMA CHEMISTRY
An accurate description of plasma discharges sustained under a Martian environment calls for the detailed understanding of different kinetics (vibrational, chemical, electronic, etc.) involving the species that compose the atmosphere of Mars, including CO , N , and Ar, along with the decomposition products created in the reactor. The development and experimental validation of self-consistent models constitute a powerful tool to obtain insight into the coupled kinetics in these plasmas and to access quantities that are difficult to measure experimentally. Subsections IV A and IV B review the current status of plasma chemistry models focusing CO decomposition and NO production, respectively. Recent developments and new results are presented as well.
A. CO2 conversion
In the past few years, several works have been conducted addressing the kinetics of CO and its mixtures with N and/or Ar as a result of the increasing interest on Mars entry problems.99–102 This is a critical research topic for the understanding of heat generation on the surface of spacecrafts and gas dynamic parameters of space vehicles during atmospheric entry. CO plasma discharges produced in the laboratory and sustained under pressures close to the Martian atmosphere, as well as close to Earth’s atmospheric pressure, have also been modeled in recent works in the framework of investigations dedicated to solar fuel production and reforming of CO . In particular, the potential of plasmas to induce ladder-climbing mechanisms and molecular dissociation36 has motivated the investigation of the impact of vibrational excitation on CO dissociation. Kozák and Bogaerts41,103 have contributed to advance the topic through the development of a zero-dimensional kinetic model of CO splitting with 25 vibrational levels up to the dissociation limit of the molecule, while considering state-specific vibrational–translational (V-T) and vibrational–vibrational (V-V) relaxation reactions and the effect of vibrational excitation on chemical reactions. Additional contributions to this model were made by Pietanza and co-workers, paying particular attention to the self-consistent coupling between the electron energy distribution function (EEDF) and the vibrational distribution function (VDF).46,104–106
In a complementary line of research, joint computational simulations and plasma diagnostics have contributed to a better understanding of the fundamental properties of CO -containing discharges. A combination of modeling tools with experimental campaigns undertaken in DC glow discharges was used to study the relaxation of CO vibrationally excited levels,107 understand the transfer of electron energy toward CO vibrations,52 analyze the influence of N on the CO dissociation yield and vibrational distribution functions,108 validate the electron-impact dissociation cross sections of CO ,39 investigate the gas heating in the afterglow of pulsed CO discharges,109 elucidate the role of CO(a) electronically excited metastable states in CO dissociation and CO recombination,58,110 and define a reaction mechanism for the formation of CO dissociation products in vibrationally cold plasmas.48 Detailed information about these works can be found in Pietanza et al.,37 where a synthesis of the current state of knowledge on the physical chemistry of CO plasmas is provided. It is still worth noting that a macroscopic approximate approach for evaluation of the discharge parameters and conversion fraction was presented very recently by Naidis and Babaeva.40,111
A modeling study of discharges produced in the laboratory exclusively dedicated to the Martian environment and ISRU applications was presented recently (see Sec. I).47 The role of CO vibrations is revealed by demonstrating that the cold temperatures of Mars can preserve the asymmetric vibrations of CO , which accumulate energy for the decomposition of CO and subsequent oxygen production. An accompanying experimental campaign characterized DC glow discharges in which the gas was cooled down to Martian temperatures by inserting a plasma reactor inside a cold bath of dry ice and ethanol, both in pure CO and in a synthetic Martian atmosphere (in terms of composition, i.e., including the minor components N and Ar).49 Despite the limitations of DC glow discharges in terms of energy efficiency for CO conversion, these plasmas constitute ideal systems for fundamental studies given their simple geometry and the homogeneity of its positive column. These allow the study of CO plasmas using volume average 0D self-consistent kinetic models accounting in detail for the very complex plasma chemistry, as the one developed by Ogloblina et al.49 It describes the kinetics of electrons and heavy species in DC glow discharges sustained under Martian environment in terms of pressure and temperature, relying on the LisbOn KInetics (LoKI) simulation tool112 to solve the homogeneous two-term electron Boltzmann equation and the system of zero-dimensional (0D, volume averaged) rate balance equations for the most relevant charged and neutral species of a pure CO plasma. The simulation results were in very good agreement with the experimental data and showed that the pressure and temperature conditions of Mars can enhance the degree of vibrational non-equilibrium. Furthermore, the experiments have shown that the Martian atmospheric composition (the minor components N and Ar) has a positive effect on CO decomposition.
Here, we improve the model of Ogloblina et al.49 by including the (i) vibrational exchanges between N and CO,113,114 (ii) quenching of CO and CO vibrations by oxygen atoms,115,116 (iv) V–V and V–T exchanges involving CO molecules,117 (v) vibrational exchanges between CO and CO ,99 and (vi) several electron impact, vibrational energy exchanges, and reactions involving N 118 and Ar.119
Figure 4 shows the simulated EEDF for a DC glow discharge operated under a continuous regime at a pressure Torr and a discharge current mA, both for pure CO and 96%CO /2%Ar/2%N (synthetic Martian mixture). In both cases, we consider a gas temperature of 550 K based on the experimental values.49 The EEDFs reveal an enhancement of the tail of the EEDF for the Martian mixture in the region of energies above 5 eV (as it was also observed before49), related to the differences in the cross sections of the different gases considered in the model and the associated changes in the self-consistent sustaining reduced electric field (that increases from 59 Td in pure CO to 67 Td in the Martian CO /Ar/N mixture). As the threshold for electron-impact dissociation is above 6 eV,120 the effect justifies the observed enhanced dissociation in a Martian atmosphere in comparison with a pure CO discharge.
Electron energy distribution functions calculated for DC glow discharges at Torr and mA for pure CO and for a 96%CO /2%Ar/2%N mixture. The simulations assume a gas temperature of 550 K.49
Electron energy distribution functions calculated for DC glow discharges at Torr and mA for pure CO and for a 96%CO /2%Ar/2%N mixture. The simulations assume a gas temperature of 550 K.49
Vibrational distribution functions of N , CO, an asymmetric stretch mode of CO , and a bending mode of CO , calculated for a DC glow discharge operating at Torr and mA for a 96%CO /2%Ar/2%N gas mixture. The simulations assume a gas temperature of 550 K.49
Vibrational distribution functions of N , CO, an asymmetric stretch mode of CO , and a bending mode of CO , calculated for a DC glow discharge operating at Torr and mA for a 96%CO /2%Ar/2%N gas mixture. The simulations assume a gas temperature of 550 K.49
Measured and calculated values associated with vibrational temperatures and electron density (Ne) in a DC glow discharge operating at p = 5 Torr and I = 50 mA for a 96%CO2/2%Ar/2%N2 gas mixture. The simulations assume a gas temperature of 550 K.49
Parameter . | Experiment . | Model . |
---|---|---|
T12 (K) | 600 | 612 |
T3 (K) | 800 | 956 |
T (K) | 800 | 1063 |
T (K) | … | 1031 |
Ne (m−3) | … | 9.8 × 1015 |
Parameter . | Experiment . | Model . |
---|---|---|
T12 (K) | 600 | 612 |
T3 (K) | 800 | 956 |
T (K) | 800 | 1063 |
T (K) | … | 1031 |
Ne (m−3) | … | 9.8 × 1015 |
For completeness, the calculated concentrations of the dominant neutral species formed in the plasma are given in Table III. Along with the main CO decomposition products (CO and oxygen species), we observe the creation of other interesting products in the context of ISRU on Mars, namely, NO and NO , which are further discussed in Subsection IV B.
Calculated concentrations for the main species in the plasma in a DC glow discharge operating at p = 5 Torr and I = 50 mA for a 96%CO2/2%Ar/2%N2 gas mixture. The simulations assume a gas temperature of 550 K.49
Species . | Concentration (m−3) . |
---|---|
CO2 (X) | 4.88 × 1022 |
N2 (X) | 1.45 × 1021 |
Ar (1S0) | 1.47 × 1021 |
Ar (4s) | 3.29 × 1011 |
CO (X) | 2.16 × 1022 |
CO (a) | 2.98 × 1016 |
O2 (X) | 6.73 × 1021 |
O2 (a) | 4.37 × 1020 |
O2 (b) | 7.69 × 1017 |
O (3P) | 7.29 × 1021 |
O (1D) | 2.63 × 1016 |
O3 (X) | 8.62 × 1016 |
NO (X) | 3.13 × 1019 |
NO2 (X) | 8.09 × 1014 |
Species . | Concentration (m−3) . |
---|---|
CO2 (X) | 4.88 × 1022 |
N2 (X) | 1.45 × 1021 |
Ar (1S0) | 1.47 × 1021 |
Ar (4s) | 3.29 × 1011 |
CO (X) | 2.16 × 1022 |
CO (a) | 2.98 × 1016 |
O2 (X) | 6.73 × 1021 |
O2 (a) | 4.37 × 1020 |
O2 (b) | 7.69 × 1017 |
O (3P) | 7.29 × 1021 |
O (1D) | 2.63 × 1016 |
O3 (X) | 8.62 × 1016 |
NO (X) | 3.13 × 1019 |
NO2 (X) | 8.09 × 1014 |
It is worth highlighting that, despite their relatively small concentration, electronically excited states play a relevant role in the overall kinetics,48,49,60 as exemplified above for the case of CO(a). Together with vibrationally excited states, they can be important energy carriers and, depending on the operating conditions and discharge type, may participate in dissociation, recombination, and gas heating.
The DC discharges studied here are very useful for validation of kinetic schemes and gaining insight into the elementary processes taking place in the plasma. Future studies will need to consider other plasma sources and verify the applicability of these models to the higher regimes and fast-pulsed discharges. A thorough analysis and further validation of the results presented here will be addressed in a future publication.
B. NO production
One advantage of using plasma sources for in situ resource utilization is their versatility to synthesize different compounds. Indeed, by using different gases to produce various molecules of interest, the same power sources and principles outlined for CO decomposition can be adapted for other gas-conversion applications. For instance, exposing the transported O to activated N leads to the formation of NO , essential for the synthesis of fertilizers and nitrogen fixation. These processes are being pursued for agriculture and the food industry on Earth,122–124 and an adaptation using N from the Martian atmosphere (and O extracted from CO ) can be envisioned on Mars.
On Earth, nitrogen fertilizers are very often synthesized from ammonia (NH ) and aim to provide plants with the ammonium ion (NH ). If a sufficient source of hydrogen is available on Mars, ammonia can be synthesized with non-thermal plasmas. Gas-phase plasma chemistry induced by electrons is of great importance,122–124 but plasma-surface interactions should also be accounted for. For example, it was shown that in RF plasmas of N /H at a few mbar, NH is produced mostly on surfaces in direct contact with the plasma.125–127 However, many crops incorporate nitrate (NO anion) even faster than NH .128 Producing nitrate on Mars first requires the production of NO from atmospheric nitrogen and oxygen produced from CO dissociation and is the focus of this subsection.
On Earth, plasmas ignited in air are known to produce significant quantities of NO and NO . The optimization of NO production efficiency by plasma has been widely studied at atmospheric pressure. Energy efficiencies of the order of 5 MJ/mol of NO have been obtained with spark, gliding arc, or microwave plasma sources.129,130 These plasmas have a large energy deposition and induce high gas temperatures at atmospheric pressure. Nevertheless, NO production is also achievable at low pressure (albeit with lower energy efficiencies, e.g., 0.8 MJ/mol at 60 Torr in a MW discharge131) by taking advantage of the non-equilibrium character of nonthermal plasmas.
The kinetics of air plasmas has been studied for a long time for various applications, such as ozone production,132,133 indoor air pollution control,134–136 surface sterilization,137,138 or atmospheric re-entry shields.139,140 The mechanisms of NO formation in cold plasmas at reduced pressure are, therefore, relatively well known and have been described in various modeling works.141–144 A recent review article compiles the latest results on the description of the plasma kinetics of N /O , including the formation of NO .118
Reactions (16)–(19) concern creation/loss processes in the gas phase, but it is evident that surfaces play an important role in the kinetics of NO in plasmas at a few mbar. As a matter of fact, at these low pressures, the recombination of O and N atoms is mainly controlled by recombination at the walls,145 and these atomic species are directly involved in reactions (16)–(19). The probability of recombination of the atoms (noted or , respectively, for O and N) depends on the material, its roughness, the surface temperature, and whether or not the surface is directly exposed to the plasma, among other things. Typical values of and can vary from 10 to 10 .151
NO and NO can also be formed on surfaces.142,152–154 The strong interdependence between surface and gas-phase kinetics on the proportion of NO produced in the gas phase has already been investigated by modeling.142,143 However, the high variability of values and the lack of experimental data on the surface reactivity for different materials still demand for significant research efforts in order to allow accurate predictions of the influence of surfaces on NO production. Nevertheless, the possibility of forming NO and of converting NO to NO from oxygen atoms adsorbed on a surface even as inert as glass (SiO ) was evidenced experimentally with DC glow discharges, microwave post-discharges, or downstream arc discharges.155–157 The oxidation of NO into NO with adsorbed O atoms, following the reaction (19) with the wall playing the role of the third body, could even be used as a test case to demonstrate the existence of a continuous distribution of adsorption energies of O atoms on Pyrex by comparing modeling and experimental results.154 Therefore, a suitably engineered surface placed either in direct contact with the plasma or in a post-discharge, for instance, from a microwave source, could improve the NO production efficiency. In turn, an increase in NO/NO production efficiencies due to the presence of a MoO catalyst downstream of a microwave discharge was observed in early works on the subject,131 suggesting plasma-catalysis as another route to explore.
The production of NO on surfaces under direct plasma exposure can also benefit from the coupling with an ion-conducting membrane, as discussed for CO conversion in Sec. III. Indeed, a pure N plasma can be generated on one side of the membrane, while an oxygen flow resulting from the conversion of CO on the other side can diffuse through the membrane. The atomic oxygen adsorbed on the surface can then react with excited nitrogen species to form NO directly at the surface. This concept was already demonstrated with a YSZ membrane, but the oxygen was produced by water electrolysis instead of CO dissociation.158 The production of NO obtained was more than 3 orders of magnitude higher than the equilibrium concentration of NO for the same gas mixture and temperature conditions.158 It is not clear whether the NO formation at the membrane surface was due to N atoms or to vibrationally excited N molecules, but in any case, NO was only produced at the surface when the N plasma was ignited. In any case, the plasma-membrane approach can have several advantages: (i) by igniting the plasma in pure N , the vibrational excitation of the molecules is maximized by limiting the quenching of the vibrational states by oxygen species and (ii) the oxygen required for NO formation can be optimally tuned by playing on the voltage and temperature controlling the oxygen flow through the membrane in order not to waste oxygen previously produced by CO dissociation.
V. FIGURES OF MERIT
This section carries out an estimation of the performance attainable by a plasma-YSZ reactor for O production from the Martian atmosphere and establishes its high potential from the comparison with MOXIE. Clearly, a proof-of-principle reactor should meet the performance targets defined by MOXIE. In particular, it should
operate with a power supply of 300 W or lower;
operate directly in a synthetic Martian atmospheric gas mixture (96%CO /2%N /2%Ar);
have reduced dimensions, below cm ;
have reduced weight, below 15 kg; and
produce and separate O at a rate above 5.5 g/h.
As discussed in Sec. II, the energy efficiency for plasma conversion of CO varies depending on the plasma source and attained CO conversion, with the highest energy efficiencies observed at low CO conversion and vice versa. Typical values for MW and RF discharges, corresponding to a trade-off between energy efficiency and conversion, are around 45% energy efficiency at 30% CO dissociation.37,38,45 Preliminary results in MW discharges operating at a power of 300 W in the synthetic Martian atmosphere mixture attained 35% CO dissociation,50 compatible with the literature in pure CO . Herein, we focus our analysis assuming a (conservative) O production of 14 g/h.49
Regarding separation (see Sec. III), considering an YSZ ion conductivity of S/m, YSZ thickness mm, a cylindrical configuration with radius cm, and a voltage of 2 V ( W), the tube length required to reach the target 14 g/O /h would be cm. Assuming that all the O/O created from the plasma is extracted through such a tubular membrane, the energy efficiency required to produce the necessary O-atom concentrations with a plasma source operating at 200 W (to be added with the power used at the YSZ cell for a total of 300 W), is around 20%–25%, well within reach of current plasma technologies.37,38 How effective can a plasma-membrane system extract the oxygen is an open and critical research question.
The numbers above lead to a specific energy of kJ/g of produced O (or, equivalently, kJ/m , at Torr and K). This is the first figure of merit to be compared with kJ/g from the current production in MOXIE (or with the results of any other technologies having the same objective).
The target of 6 kg for the reactor is estimated from an optimized reactor, adding up to kg (see the last paragraph of Ogloblina et al.49 for a detailed analysis of this value), estimated for a solid-state MW reactor. A safety factor of 2 is considered to correct for the need of unforeseen elements or additional weight imposed by the optimized design configuration.
This analysis and the preliminary results already obtained allow one to estimate that oxygen can be produced at the rate of 14 g/h using a 6 kg reactor with dimensions cm (including main components and shielding),49,50 amounting to 2.33 g of oxygen produced per hour per kg of equipment sent to Mars. These values are the additional figures of merit to be compared with MOXIE’s 5.5 g/h at 300 W for a 15 kg reactor of cm , corresponding to 0.37 g of oxygen produced per hour per kg. Accordingly, a plasma-based technology can potentially outperform the current state-of-the-art technology defined by MOXIE.
It is still worth noting that the reliability of the plasma sources that can be used in a prototype has been proved in industry already for decades by their use in microelectronics159 and Hall-effect thrusters for space propulsion.160,161 These sources have a long lifetime, essentially limited by the lifetime of the power supply itself, estimated to be larger than 100 000 h. For MW and certain RF configurations,58 there is no contact between electrodes and the plasma, which prevents erosion and increases durability. Finally, the electromagnetic noise emitted by the sources can impose some constraints to the design of the system, but, if required, the system can be shielded with a simple and light grid to act as Faraday cage.
Additional hardware requirements will be needed for a flight qualification plan for a full-scale mission to Mars. Their realistic assessment can only be done after the demonstration of the integration of the technologies at higher TRL levels. In any case, ground testing should address durability, heat integration for smooth start and stop of the process, and submission of the prototype to relevant mechanical (compression, vibration, shock), thermal and radiation environments to survive launching, space travel, landing, and onsite operation.25 Another aspect to consider is the dust presence in the Martian atmosphere and how to filter and clean the system. A full-scale system should be able to produce 100 times more O than anticipated by MOXIE.22,25 This estimation is compatible with the requirement for a production rate of about 50 kg/day 2 kg/h of a propellant manufacturing system for a Mars ascent vehicle (MAV).162 Scalability studies should analyze oxygen production capacity, mass and volume, power requirements, and scaling of specific components. By the very nature of plasma technologies, scalability is not expected to become a major point of concern.
VI. CONCLUSIONS AND OUTLOOK
In situ resource utilization (ISRU) is the use of natural resources from the Moon, Mars, and other bodies for use in situ or elsewhere in the solar system. The most useful mission consumable products from ISRU are propellants, fuel cell reactants, and life support commodities (such as water, oxygen, and buffer gases). Making oxygen alone can provide significant mission savings, as it can be used for these purposes and also for the development of fertilizers for agriculture.16
This paper opens the perspective of combining non-thermal plasmas with conducting membranes to produce oxygen by decomposing CO directly from the Martian atmosphere and separate the products of CO dissociation, respectively. The current state of the art regarding O production on Mars is defined by MOXIE,21–25 which is based on solid electrolysis cells (SOEC), where electricity is provided between two electrodes to electrochemically convert CO into oxygen at temperatures above 1000 K and pressures of about 1 bar. MOXIE is a thrilling project and already operates on Mars. Nonetheless, the need to compress and heat the system increases the weight and power consumption of the system, and the erosion of the electrodes is a point of concern.
The approach suggested here brings a solution for ISRU on Mars that can be seen as an upgraded system when compared with MOXIE. Plasmas can enable activation of CO and thus enhance the performance of the SOEC electrode material toward an efficient oxygen separation and can be readily ignited at Martian ambient pressure.47,49 The combination of plasma with oxygen separation membrane-based systems holds the promise of a high performance, both in terms of the required specific energy (energy spent per kg of produced O ) and of the quantity of oxygen molecules produced per kg of instrumentation placed on Mars. It is very versatile due to its adaptability to work with different feed-stock and tunable reactivity depending on the desired outcome. As examples, if H is carried from Earth or produced from electrolysis on Mars, the same reactor used to produce oxygen and CO can be operated in a CO –H mixture to produce methane, which can be used directly as a fuel, in a N –O mixture to produce NO , or in a N –H mixture to produce NH ,163 the latter two of direct use for Martian agriculture. Additionally, a plasma system can fully decompose CO up to carbon, of interest for the manufacture of carbon structures.
Different plasma sources promote CO dissociation according to alternative concepts. For instance, in radio-frequency (RF) discharges, the CO vibrational kinetics is expected to have a relevant contribution, as the typical mean electron energy ( eV) favors vibrational excitation; in nanosecond repetitively pulsed (NRP) discharges, the high electric field increases the number of enough-energetic electrons able to directly dissociate CO , offering an alternative route to dissociation that does not rely on vibrational excitation; microwave (MW) discharges operate in yet another regime, where thermal dissociation plays a significant role.
The coupling of plasma with conducting membranes for product separation is an active field of research. Mixed electron–ion conducting (MIEC) membranes32–34,64 and ion-conducting membranes30,31,158 to be integrated in a plasma reactor are currently under development. The usage of plasma with a conducting membrane may not require the expensive, complex, and sensitive cathode materials needed to split oxygen molecules in a conventional solid oxide electrolyzer cell (SOEC). It should enhance the oxygen permeation through the membrane for two reasons: the prior dissociation of CO in the plasma and the heat produced by the plasma. On the one hand, plasma dissociation will significantly increase the density of O atoms on the surface of the membrane; in principle, a 100% dissociation of CO into CO and O is possible, provided a sufficiently high residence time is used.84 On the other hand, the ionic conductivity becomes significant above a certain temperature (about 600 K for YSZ); in the plasma-membrane combined setup, the membrane can be heated by the plasma itself without the need for additional pre-heating equipment, with another gain in the mass and power requirements of the system. Additionally, the presence of free electrons from the plasma may increase the surface kinetics and contribute as well to an increase of the permeation fluxes.33
A compromise must be found between the efficiency of dissociation in the plasma volume and the oxygen flow through the membrane: on the one hand, a long residence time of the plasma in the reactor benefits the separation process; on the other hand, the presence of O hinders further plasma dissociation. Furthermore, the temperature of the membrane can, in principle, be regulated through a combination of power and flow in the plasma, where an optimum working point that provides both a sufficient CO dissociation and the desired translation gas temperature for heating should be accessible. A simple system to impose a small gas flow in a reactor operating on Mars may then have to be included to control the gas residence time. Nonetheless, to know how effectively can a plasma-membrane system extract the oxygen is an open and critical research question.
Kinetic schemes describing the plasma chemistry in CO , N , Ar, and their mixtures are to a large extent already available. Validation in a Martian environment is still scarce,49 but those first results and the ones presented here are very encouraging. The simulations reveal the important role of vibrational excitation and of electronically excited states both in shaping the electron energy distribution function and in the very mechanisms of CO dissociation and CO recombination.37 Validated reaction schemes for N –O plasmas, of interest for nitrogen fixation and the production of fertilizers, are also available in the literature.118,124 Surface kinetics studies are less developed than their gas volume counterparts and should become an important subject of research in the next few years.37 These studies should improve the description of the elementary steps involved in surface adsorption, diffusion, and recombination under plasma exposure, as well as of the bulk diffusion through the membrane. Overall, the current degree of sophistication and the predictive power of plasma models hints at their importance in the design and optimization of future prototypes.
The successful association of plasmas with ion-conducing membranes has not been achieved to date, whereas coupling plasmas with MIEC are at their very first steps. Notwithstanding, the emerging plasma-membrane technology and the solutions outlined here are already founding a new active and probably long-lasting field of activity, associated with the development of a novel and highly versatile electrochemical conversion technology, which can be applied in different environments on Earth and to several ISRU applications.
ACKNOWLEDGMENTS
This work was partially supported by the Portuguese FCT-Fundação para a Ciência e a Tecnologia under Project Nos. UIDB/50010/2020, UIDP/50010/2020, MIT-EXPL/ACC/0031/2021 (CREATOR), and EXPL/FIS-PLA/0076/2021 (ROADMARS); the Portuguese Foundation for International Cooperation in Science, Technology and Higher Education through the MIT-Portugal Program (project IMPACT); and the European Space Agency (ESA) under Project No. I-2021-03399 (PERFORMER).
We are grateful to Benjamin Martell and Ahmed Ghoniem for their insightful comments and fruitful discussions.
AUTHOR DECLARATIONS
Conflict of Interest
The authors have no conflicts to disclose.
Author Contributions
V. Guerra: Conceptualization (lead); Methodology (lead); Project administration (lead); Writing – original draft (equal); Writing – review and editing (equal). T. Silva: Software (lead); Writing – original draft (equal). N. Pinhão: Writing – review and editing (equal). O. Guaitella: Writing – original draft (equal). C. Guerra-Garcia: Writing – original draft (equal); Writing – review and editing (equal). F. J. J. Peeters: Writing – original draft (equal); Writing – review and editing (lead). M. N. Tsampas: Writing – review and editing (lead). M. C. M. van de Sanden: Writing – review and editing (lead).
DATA AVAILABILITY
The data that support the findings of this study are available from the corresponding author upon reasonable request.