Abiotic ammonium formation in the presence of Ni-Fe metals and alloys and its implications for the Hadean nitrogen cycle

Experiments with dinitrogen-, nitrite-, nitrate-containing solutions were conducted without headspace in Ti reactors (200°C), borosilicate septum bottles (70°C) and HDPE tubes (22°C) in the presence of Fe and Ni metal, awaruite (Ni80Fe20) and tetrataenite (Ni50Fe50). In general, metals used in this investigation were more reactive than alloys toward all investigated nitrogen species. Nitrite and nitrate were converted to ammonium more rapidly than dinitrogen, and the reduction process had a strong temperature dependence. We concluded from our experimental observations that Hadean submarine hydrothermal systems could have supplied significant quantities of ammonium for reactions that are generally associated with prebiotic synthesis, especially in localized environments. Several natural meteorites (octahedrites) were found to contain up to 22 ppm Ntot. While the oxidation state of N in the octahedrites was not determined, XPS analysis of metals and alloys used in the study shows that N is likely present as nitride (N3-). This observation may have implications toward the Hadean environment, since, terrestrial (e.g., oceanic) ammonium production may have been supplemented by reduced nitrogen delivered by metal-rich meteorites. This notion is based on the fact that nitrogen dissolves into metallic melts.


Introduction
Ammonia (NH 3 ) or ammonium (NH 4 + ), henceforth NH 3 /NH 4 + , are necessary precursors for reactions associated with prebiotic syntheses, such as the Strecker synthesis. It has been experimentally shown that NH 3 /NH 4 + environments are more efficient in organic synthesis than those dominated by dinitrogen (henceforth N 2 ) in both aqueous and gaseous environments [e.g., [1]] [2,3]. This notion is not unexpected, considering that, the strong triple bond (948 kJ.mol -1 ) of the N 2 would presumably result in large reaction activation barriers (i.e., low conver-sion rates), even if the overall reaction is thermodynamically favored.
Several possible pathways to abiotic NH 3 /NH 4 + on early the Earth have been proposed: reduction of NO 2 -/NO 3by Fe ++ /FeS in the ocean [e.g., [4]] [5,6]; atmospheric production from N 2 and HCN [e.g., [7]] [8]; release from rocks and minerals [e.g., [9]]; photoreduction on mineral surfaces [e.g., [10]] [11,12]; and hydrothermal aqueous reduction from N 2 in the presence of minerals under conditions typical of submarine hydrothermal systems [e.g., (page number not for citation purposes) [13]] [14,15]. Each of the mechanisms relies on a different set of assumptions and none of the proposed mechanisms has, in our opinion, gained universal acceptance in the scientific community as the predominant source of abiotic NH 3 /NH 4 + .
In this scientific contribution we focus on the catalytic properties of Ni and Fe metals and their alloys which can form in submarine hydrothermal systems (SHS), especially those driven by exothermic hydration reactions (e.g., serpentinization) in an off-axis tectonic setting. Upon dissolution of Ni-containing rock-forming minerals (e.g., olivine, pyroxene, amphibole), released Ni and Fe can react to form metals and alloys under extreme reducing conditions imposed on the system by the serpentinization processes [16][17][18]. The conditions are commonly reducing enough to stabilize Ni-Fe alloys (e.g., awaruite -Ni 3 Fe), metallic nickel (Ni 0 ) and even iron (Fe 0 ). These minerals occur regularly, albeit in small quantities in both ancient and modern serpentinites [19][20][21][22][23][24][25][26]. A compilation of representative chemical analyses of metals and alloys found in serpentinites is presented in Fig. 1. The observations from natural systems have been corroborated by laboratory experiments [27,28].
An active global tectonic cycle is not required for the formation and operation of serpentinization-driven SHS and hence we assume that these environments were commonly present on the Hadean Earth. Moreover, the lack of oxygen and the possible presence of significant amounts of hydrogen gas in the Hadean atmosphere (and consequently in the ocean water) may have further enhanced the stability of base metals and their alloys [29,30].
The most abundant reactant for abiotic NH 3 /NH 4 + formation in the Hadean was N 2 dissolved in the seawater from the N 2 -rich atmosphere. NO 2 -and NO 3 are also thought to have been available, although likely in low concentrations. These oxidized N species could have formed in high energy events such as lightning, corona discharge and/or impacts and subsequently rain out into the ocean [31][32][33][34].
In this contribution we report the results of an experimental study undertaken to evaluate the hypothesis that abiotic NH 4 + formation from dissolved N 2 , NO 2and NO 3 -in the presence of Ni 3 Fe, NiFe, Ni 0 and Fe 0 was an operative synthetic route at anaerobic conditions potentially present in the Hadean Ocean. Furthermore, we attempt to quantify global NH 4 + yields in the Hadean Ocean produced by investigated mechanisms.

Reduction experiments
Three sets of experiments were conducted at three different temperatures: 200°C (runs 1-36), 70°C (runs  and 22°C (runs . The choice of reactors was based on experimental temperature: 15 mL passivated HIP ® Titanium 64 tube reactors (200°C); 20 mL I-Chem ® borosilicate vials with PTFE/Si septum caps (70°C) and 15 mL BD HDPE Falcon ® tubes (22°C). Reactors were kept at constant temperature in a heated water bath (20, 70°C) or an Isotemp ® oven (200°C). All experiments lasted 24 hours and were conducted in the absence of headspace (e.g., no gas phase). No additional pressure other than that of expanding liquid was imposed on the vessels (~400 psi/ 27 bars with Ni 0 to ~800 psi/55 bars with Fe 0 at 200°C).
To ensure clean and fresh mineral surfaces (e.g., free of oxidation products and/or atmospheric sorbed gases), all metal/alloys were ultrasonically cleaned for 1 hour in 0.06 M HCl immediately preceding the experiments. Subsequently they were washed three times with the designated reacting solution and loaded into reactors in the form of slurry. This "wet loading" procedure eliminated sorption of gases from the atmosphere onto freshly cleaned metal surfaces which was especially important in blank experiments.
Background NH 4 + production (e.g., release from reactants, reaction vessels, etc.) was assessed in blank experiments with Ar-saturated solutions and no added N source (runs 1,9,17,25,37,43,49,54,60,66,72,78). NO solutions were prepared by dissolution of ACS reagent grade KNO 3 and KNO 2 , respectively. The pH was not buffered and was allowed to change as a result of solutionmetal/alloy interactions and was recorded before and after the experiment. After pH measurements, all samples were acidified with 0.2 M HCl to ensure the conversion of NH 3 to NH 4 + and to prevent the formation of Fe precipitates. The samples were stored at 1°C and analyzed within 24 -48 hours. The summary of all experimental conditions is presented in Tab. 1.

Analysis of solids, their surfaces, and reaction products
Metals and alloys representing an fcc solid solution of Ni in Fe were purchased from Alfa Aesar ® and Goodfellow. All starting and selected reacted solids were characterized by X-Ray Diffractometry, X-Ray Photoelectron Spectroscopy, Scanning Electron Microscopy, B.E.T. surface analysis, and Electron Microprobe. The results of metal/alloy characterization are summarized in Tab. 2 and Fig. 2, 3.
Scanning Electron Analysis (SEM) was performed on the LEO 1550 SFEG scanning electron microscope equipped with an EDAX energy dispersive X-ray spectrometer (EDS) using an accelerating voltage of 15 kV and a 30 μm aperture.
Oxidation state of Ni, Fe and the presence of N in alloys were determined by X-Ray Photoelectron Spectroscopy (XPS). The data were acquired with unmonochromatized Mg Kα and Al Kα radiation at 1253.6 eV and 1486.7 eV using a Physical Electronics source controller in a vacuum chamber with a base pressure of 1 × 10 -9 Torr. A VG Microtech hemispherical analyzer was used to obtain the energy distribution of the photoemitted electrons at pass energies of 50 and 75 eV. The binding energies were calibrated by fixing the Au 4p3/2 and 4f7/2 peaks (546.3 eV, 87.5 eV) from a gold standard, and the metallic Fe 2p3/2 and Ni 2p3/2 cores (707.0 eV, 852.3 eV). Selected particles were sputtered with Ar + accelerated to 2 kV to expose their inte-  Total nitrogen content of metals/alloys was analyzed by IMR Test Labs (Lansing, NY) by inert gas fusion [35]. During the analysis, N is released from the metal at 1900°C into the stream of He gas and analyzed in a thermal conductivity cell.
No other compounds were analyzed. It is expected, however, that other reaction products and/or intermediates (e.g., NO) may have formed during a complex sequence of electron transfer reactions.
Geochemical equilibrium modeling was performed with the Geochemist's Workbench 5 [36] software package with the thermo.com.V8.R6.full thermodynamic database complemented with data for Ni 3 Fe and NiFe [37]. "n.a." denotes "not analyzed, "-" not applicable. * denotes NH 4 + concentration formed in the experiment normalized to 1 m 2 of surface area of the metal/alloy. ** denote residual concentration after the experiment was completed.

Dinitrogen reduction
The results of NH 4 + formation from N 2 and the effect of added H 2 and KCl at 200°C (normalized to 1 m 2 surface area) are shown in Fig. 4 and summarized in Tab. 1 (runs 1-32). All results are compared with respect to blank experiments conducted with Ar and no N 2 added. The blank experiments thus represent the background NH 4 + production from the metal/alloy involved and reactor catalysis. Error bars were calculated by propagating errors from solution dilutions and B.E.T. and IC analyses. Due to relatively large error bars, only results differing from the blank (or each other) by more than the calculated error will be discussed.
It is important to note that at 200°C both NO 2and NO 3decomposed in the absence of metal/alloys as well (Tab. 1, runs 34, 34). 42% of the initial 497 μmol.kg -1 KNO 2 solution was converted into NO 3 -(~15%), NH 4 + (~9%) and other N compounds (~18%) that were not analyzed. Of the initial 444 μmol.kg -1 KNO 3 solution, 71% was converted into NO 2 -(~61%), NH 4 + (~1%) and about 9% corresponds to other unanalyzed N compounds. It is not clear if this is a result of thermally induced decomposition, catalysis or reaction by/with the titanium reaction vessel, or a combination of all; nevertheless, NH 4 + was not the dominant reaction product. At 70 and 22°C, both NO 2and NO 3solutions were found to be stable in experiments without metals or alloys during the 24-hour reaction period.

Metal/alloy alteration
Generally, the extent of alteration of Fe-containing metal/ alloy increased with temperature, as demonstrated by the presence of secondary minerals (Fig. 6). Magnetite (Fe 3 O 4 ) was the most abundant alteration product, predominantly forming euhedral to subhedral crystals up to several μm in size (Fig. 6a, b, d, e). Pseudomorphoses of magnetite after reacted spherical Fe 0 particles were common (Fig. 6d). The second most common alteration phase were Fe-(oxy)hydroxides (e.g., lepidocrocite) usually of amorphous appearance or forming needle-like (Fig. 6c) and platy crystals several tens of nm thin and several μm long. Both magnetite and Fe-(oxy)hydroxides commonly occur simultaneously in all reacted samples (SEM) ( 6a,b); however, only magnetite is identified by the XRD method (Fig. 7). This suggests that the Fe-(oxy)hydroxides either lack long range order (e.g., "X-ray amorphous") and/or their abundance is less than 5%, the approximate detection limit of XRD. In general, the lower the Fe content, the lower the extent of alteration (Fig. 6e). In contrast, reacted Ni 0 exhibited no microscale (SEM) evidence of reaction (Fig. 6f, compare with Fig. 3a), as corroborated by the XPS spectra documenting the presence of residual zero-valent Ni species on the surface (Fig. 8). As a result of solution interactions with metals/alloys, the resulting pH in most experiments was higher than the starting value (see discussion) (Tab. 1).

Nitrogen in alloys and metals
All the metals and alloys investigated in this work were found to contain N, which resulted in a background production of NH 4 + . The presence of atomic N was based on two lines of evidence: 1) The presence of the N1s peak in the XPS spectra of starting metals and alloys even after "cleaning" the surface by sputtering with Ar + ions (Fig. 9) and; 2) quantitative analysis of N tot content of the starting materials by inert gas fusion with a thermal conductivity detection (Tab. 3).
Commercially available Ni, Fe metals/alloys may contain N due to the manufacturing process which employs either inert (N 2 , Ar) or reducing (NH 3 , H 2 ) atmospheres to prevent oxidation [38,39] [Alfa Aesar, Goodfellowpers.comm]. For example, during the synthesis, N 2 chemisorbs on the surface of molten/hot metal and dissociates (Equation 1).
Ni 2p3/2 peaks showing the speciation of Ni on the surface Ni metal powder at various stages of the experiment In the subsequent step N enters the structure via diffusion or convection to form primarily monoatomic interstitial and to lesser extent substitutional solid solutions. This process is governed by the Sievert's law (Equation 2), which predicts that diatomic gases such as N 2 dissolve in metals (c N ) proportionally to the square root of the partial pressure (p N2 ) in the coexisting gas phase [40,41].
Assuming homolytic N 2 bond cleavage, each N atom would have three unpaired electrons available for bonding with the surrounding metal atoms. XPS spectra collected from unreacted metals/alloys in our experiments point to nitride (N 3-, 398.6 eV) as the likely N species (Fig.  9) [42]. Consequently, upon release into the solution, N 3is expected to react with available protons to form NH 3 / NH 4 + (Reaction 3, 4) and contribute to their high background productivity.
However, undissociated N 2 gas may get trapped in the molten metal (e.g., in inclusions) as well.

Dinitrogen reduction
Although batch-type experiments, such as the ones described above, provide little insight into the kinetics of a reaction and much less the reaction mechanism, the results do allow one to compare the amount of NH 3 /NH 4 + formed under different conditions. In addition the results can be placed in the context of previously published research related to the reduction of N-species to NH 3 / NH 4 + . Due to the immense importance of NH 3 /NH 4 + in industry and agriculture, several decades of research exist on its synthesis and production from N 2 gas [e.g., [43]]. The industrial Haber-Bosch process utilizes Fe 0 catalyst at high temperatures and pressures (~500°C, ~100 bars) to synthesize NH 3 from H 2 and N 2 gas (Reaction 5).
In brief, the reaction proceeds as follows: sorption of H 2 and N 2 gases on the surface is followed by the formation of atomic H ads and N ads (dissociative sorption). Fe 0 then facilitates electron transfer from H ads to N ads (e.g., N reduction), followed by the formation of NH 3 gas on the surface and subsequent desorption. The dissociative chemisorption of N 2 is generally taken to be the rate-limiting step [44]. For comparison, modern life overcomes the reaction's activation barrier using the enzyme nitrogenase composed of dinitrogenase (MoFe center) and dinitrogenase reductase (Fe center) proteins. In Reaction 6, Fd stands for ferredoxin, the electron-transfer protein [45,46]. N 2 (g) + 8Fd 0 red + 10H + → 2NH 4 + + 8Fd + ox + H 2 (g) Analogous overall reactions of abiotic N 2 reduction can be written for aqueous solutions (Reaction 7, 8), although it is important to note that since this reaction requires a transfer of multiple electrons, several reaction intermediates must be involved. Once in solution, NH 3 and NH 4 + exist in a pH dependent equilibrium (Equation 9; valid for 25°C). N 2 (aq) + 2H + + 3H 2 (aq) → 2NH 4 + N 2 (aq) + 3H 2 (aq) → 2NH 3 (aq) We hypothesize that reactions between H 2 , N 2 and the metal/alloy surface are taking place in our experiments;  0.0009 ± 0.0005 however, they are orders of magnitude slower than those described above in the gas phase. Undoubtedly, this is due to the fact that aqueous reactions occurring in our experiments are not only more complex but also take place at much lower T, P conditions and H 2 /N 2 concentrations than those typical for Haber-Bosch synthesis. Such kinetic constraints could explain low NH 4 + yields, even though the thermodynamic equilibrium models predict NH 4 + to be the dominant N species (Tab. 4).
We speculate that in the presence of Fe 0 most of the NH 4 + was rapidly formed in the first few hours of the experiment when unreacted surface was still available for reaction. In this scenario (e.g., run 2) most (and possibly all) H 2 is formed in situ as a result of interactions between the pristine Fe 0 surface and H 2 O (Reaction 10) (e.g., mH 2 = 0 at t 0 ); however, the simultaneous Fe oxidation passivates the surface, reduces the availability of suitable H 2 /N 2 sorption sites and the overall yield of the N 2 -reduction reaction.
This notion is corroborated by experiments carried out with conditions in which H 2 was present in the system from the start of the reaction (e.g., run 3), as a result of purging the solution with a H 2 /N 2 mixture prior to loading (e.g., mH 2 > 0 at t 0 ). Abundant H 2 in this run correlates with greater NH 4 + production as the overall N 2 conversion rate increases from 2.5 to 10% (Tab. 2). Assuming that in both cases Fe 0 surface passivates at the same rate, then the H 2 purged system produces more NH 4 + per unit of time because it does not depend on the Fe 0 alteration process (Reaction 10) to provide H 2 . This circumstance may be more typical of natural serpentinization-driven SHS where H 2 can be provided by a number of processes, especially by Fe 2+ oxidation during alteration of rock-forming silicates [17,47].
Similar assumptions can be made about Ni 50 Fe 50 and Ni 81 Fe 19 assuming that Fe atoms exposed on the surface played a role in the reduction process. Due to good corrosion resistance, Ni 0 reacted to a much lesser extent and consistent with this lower activity is our experimental observation that the Ni 0 surface was not significantly altered (e.g., by precipitation of neoformed phases) throughout the experiment. Different modes of metal/ alloy participation in studied reactions are discussed below.
The addition of KCl into the Fe 0 -H 2 O-N 2 system in our experiments resulted in higher NH 4 + yield (Fig. 4a). While it may be intriguing to draw parallels with the Haber-Bosh process, where K is added to improve sticking coefficients and to help stabilize sorbed species [44], the apparent promoting effect of KCl may be partially or entirely caused by the presence of chloride ion (Cl -) in the solution. Clcan react with dissolved iron in the solution (Reaction 11) and remove products from the Fe 0 dissolution reaction (Reaction 10).
Fe 2+ + 2Cl -→ FeCl 2 (11) Such a complexation reaction would result in an equilibrium shift towards the product side and further drive the dissolution process and release of structurally bound reduced N species into the solution (Reactions 3, 4). This is in agreement with the results of Reardon [48] who observed an increase in Fe 0 corrosion rates in low ionic strength (~0.02 m) anaerobic NaCl, NaHCO 3 and Na 2 SO 4 solutions compared to DI water. The NH 4 + content of the KCl reagent solution at concentrations used in our experiments was found to be below the detection limit of ion chromatography.

Nitrite and Nitrate Reduction
Due to their status as environmental contaminants, NO 2 -/NO 3reduction has been extensively studied, especially focused on the reduction of NO 3 - (13) Analogous reactions can be written for other metals/alloys as well as NO 2 -(Reaction 13), although they have been by comparison less studied. The absence of resonance structures in the NO 2molecule makes it easier to reduce than NO 3 - [58], which is reflected in reduction reaction rates. For example, NO 3 reduction to NH 4 + in the presence of Fe 2+ was found to be a factor of 8 slower than that of NO 2 - [5]. Because the reduction from NO 3to NH 4 + requires transfer of at least 8 electrons, several intermediates must be formed in the process. Moreover, the formation of any N-N bonds must be avoided because it is effectively inert under all but the highest temperatures investigated here.
It is widely recognized, however, that NO 2is a reaction intermediate in NO 3reduction [e.g., [51]] [57,59]. For example, Wärna et al [59] proposed a reaction sequence from NO 3and NO 2through nitric oxide (NO), imidogen (HN) and aminyl radical (H 2 N¨) to NH 3 /NH 4 + on the surface of Fe 0 . Several studies with NO 2 -/NO 3 as well as some organic compounds suggest that Fe 2+ , Fe 0 , and Fe 2+ sorbed on neoformed Fe minerals (e.g., magnetite) are likely electron donors for the reduction reactions [e.g., [60]] [61,62]. The intriguing consequence of such a reaction mechanism in natural systems is that precipitation of secondary (neoformed) Fe minerals further along the flow path followed by surface sorption of Fe 2+ would provide additional reaction sites for the reduction of NO 2 -/NO 3 -[e.g., [63]].

The role of alloys/metals
Based on the XPS (oxidation state of Ni) and SEM (abundance of Fe-and the absence of Ni alteration phases) results it is possible to construct an order of apparent stability of studied alloys and metals where Ni 0 is most-and Fe 0 is least stable under the studied experimental conditions. This enables us to generalize that the higher the Fe content, the higher the reactivity towards potential oxidizing agents (e.g., H 2 O, NO 2 -, NO 3 -) and thus the higher extent of alteration. Metals and alloys typically undergo reductive dissolution (e.g., Reaction 10); however alloys frequently dissolve the less noble metal preferentially, leaving the surface enriched in the more noble metal [e.g., [64]]. For example, the reductive dissolution of Ni 50 Fe 50 alloys is expected to result in preferential release of Fe and a concomitant increase in the Ni:Fe ratio of the residual alloy.
Our findings are in agreement with metallurgical studies in which it has been demonstrated that Ni 0 is more corrosion resistant than Fe 0 , a notion that serves as a basis for their frequent alloying [38,65]. Unlike Fe 0 , Ni 0 reacts to a lesser degree in aqueous environments (reaction produces H 2 and Ni 2+ ), especially under reducing conditions. The presence of an oxidizing agent is usually required for significant corrosion; however, a protective oxide film may develop and impede further reactions [38,66]. Ni 0 with a combination of catalytic properties and corrosion resistance (e.g., slow dissolution kinetics) may serve as a basis for a unique mechanism of N 2 reduction, where Ni acts both as a reactant and a catalyst. We hypothesize that Ni 0 reacts with H 2 O to produce H 2 , a portion of which may stay adsorbed on the surface in its atomic form (H ads ) ( Reaction 14). If N 2 is also dissociatively chemisorbed (Reaction 15), surface-mediated reduction reactions may proceed (Reaction 16).
Ni + 2H 2 O → 2H ads + Ni 2+ + 2OH - N 2 (aq) → 2N ads (15) 2N ads + 2H + + 6H ads → 2NH 4 + (16) This set of reactions may operate until all plausible sorption sites are exhausted and/or deactivated. By analogy, we argue that if the experimental conditions were approaching the stability field of Fe metal (e.g., at sufficiently high fH 2 ) it could behave in a similar manner.
There exist; however, "true" catalytic systems for NO 2 -/ NO 3reduction, such as bimetallic Cu-Pt and Cu-Pd, Ag-Pd, Ag-Pt, which couple a noble metal and an oxidizable promoter. The reactions take place on the surface of Cu 0 which acts as an electron donor for the reduction of N species and as an acceptor of electrons from dissociative sorption of H 2 on the surface of Pt [67,68]. Even though natural alloys of platinum group elements (Pt, Pd, Ir, Os, Rh, Ru) are scarce on modern Earth and are almost exclusively limited to magmatic segregation deposits, placers, and meteorites [e.g., [69]] [70,71], their significance for prebiotic synthesis should not be overlooked [72].
The predominantly alkaline pH in reacted samples (Tab. 2) is likely a result of several pH controlling reactions such as reductive dissolution of metals (e.g., Reaction 10), mineral formation (e.g., magnetite), and the buffering reactions involving charged species including, but not limited to NO 3 -, NO 2 -, NH 3 , or NH 4 + . The fate of Fe 2+ in the experiments reported here is difficult to constrain. Assuming completely anoxic conditions, temperatures below 85°C and a negligible pCO 2 in our experiments, Fe(OH) 2 (white rust) could precipitate (Reaction 17) and due to its instability serve as a precursor to other Fe oxides and hydroxides, most notably Fe 3 O 4 (Reaction 18).
At higher temperatures and/or in the presence of trace levels of O 2 or other oxidizing agents (e.g., NO 3 -), mineral intermediates such as green rust (mixed-valence hydroxide) may have been involved [e.g., [73]] [74,75].

Implications for the Hadean Earth
Different modes of metal/alloy participation have different implications for natural systems, especially in terms of the amount of metal/alloy required to achieve the same NH 3 /NH 4 + production. A catalyst remains stable during the reaction and therefore a small amount can, in theory, catalyze the formation of large amounts of NH 3 /NH 4 + . Conversely, a reactant would have to be present in sufficient amounts and/or would have to be continually formed in order to achieve comparable NH 3 /NH 4 + production. While both mechanisms are plausible on the Hadean Earth, it is hard to assess which of the two would be prevalent. Based of equilibrium geochemical modeling, Smirnov [28] concluded that at 200°C Ni metal is stable at fH 2 orders of magnitude lower than Fe metal and even Ni 50 Fe 50 (tetrataenite) and Ni 81 Fe 19 (awaruite). Combined with results acquired from this study, it would appear that Ni metal is the most suitable candidate for a sustained long-term NH 3 /NH 4 + formation. Moreover, if we consider that the Hadean atmosphere may have had up to 30% H 2 [29], the primordial ocean would contain significantly higher concentrations of dissolved H 2 than today. In SHS, H 2 from advected seawater combined with H 2 formed by serpentinization could create conditions sufficiently reducing for stabilization of Fe containing alloys (e.g., awaruite, tetrataenite) and possibly even Fe 0 . Furthermore, as shown by Schoonen et al [72], seawater trapped in closed SHS (i.e., not open to seawater circulation) evolves to become extremely reduced as the partial pressure of hydrogen builds up.
The possibility of hydrothermal reduction of N 2 to NH 4 + permits us to attempt to constrain the total NH 4 + flux from Hadean off-axis SHS. The following set of assumptions and variables were taken into account: 1) The total heat production of the Hadean Earth was several times higher than today [81,82]. Because the exact value is unknown, we calculated scenarios for 2-to 8times the present day heat flow (PDHF = 4.3 × 10 13 W) [83] (Fig. 10); however, only values between 4 and 8 times PDHF are reported.
2) Because it is unclear if a global tectonic cycle was operational during the Hadean, we are unable to comment on the dissipation of Earth's internal heat, especially on the percentage of heat released through SHS. Therefore, two endmember scenarios are considered: a) heat is dissipated predominantly via volcanism (possibly through several supervolcanoes) and only 5% is released through hydro-thermal activity; and b) 80% of heat is dissipated predominantly through hydrothermal activity (Fig. 10). For comparison, presently about 20% of PDHF is released through hydrothermal activity [83].
3) Due to the increased heat flow and possibly due to the blanketing effect of the atmosphere [e.g., [84]], we assume the mean ocean water temperature to be 70°C. Although modern serpentinization-driven SHS commonly vent fluids below 100°C [85][86][87], we assume that the higher overall heat flow in the Hadean would also increase the fluid temperature of hydrothermal vents [e.g., [88]]. The temperature of the discharging fluid is thus assumed to be 200°C for the purpose of this calculation. Although the temperature of ambient seawater and of discharging fluid directly influences the total hydrothermal fluid mass flux (equation 19), their variations (± 20°C) only produced small changes in the final NH 4 + fluxes (usually within the same order of magnitude; data not shown). The heat capacity (c p ) of hydrothermal seawater at 200°C and P 100-600 bar is 4.1 J.g -1 K -1 [89].

4)
Ocean water is assumed to be in equilibrium with 1 bar of N 2 , resulting in a dissolved N 2 (aq) concentration of 0.481 mmol.kg -1 at 70°C [90]. For simplicity, no other gases and/or aqueous ions were taken into consideration.

5)
Even though experimental results reported in this study suggest a conversion of N 2 -to-NH 4 + 0.2 to 2.5% we calculate a variety of scenarios ranging from 0.1% to 10%. Although the 10% conversion may appear overly optimistic, our experiments suggest that the presence of advected H 2 and or K + may significantly improve the NH 4 + production (Tab. 2; Fig. 4a). Metals/alloys may act as either cata-lysts or reactants, however, if metals/alloys do react, it must be assumed that the rate of their destruction (e.g., passivation, poisoning) is equal to their rate of formation (e.g., via serpentinization). It is important to point out that for simplicity, we do not distinguish between respective metals/alloys and we are only concerned with their capability to facilitate the conversion of N 2 to NH 4 + (in %).
The mass flux of seawater through hydrothermal systems (F) can be estimated from heat flux (H) in Watts, heat capacity of seawater (c p at 200°C) in J.g -1 K -1 and temperature anomaly ΔT in Kelvin [83]: Using hydrothermal heat fluxes from Fig. 10 we can calculate annual seawater mass fluxes from SHS. Subsequently, using various N 2 -to-NH 4 + conversion percentages (0.1 to 10%), annual NH 4 + production of Hadean SHS is calculated (Fig. 11). Assuming the most conservative scenario with 0.1% conversion of N 2 to NH 4 + , the annual NH 4 + production would be between 5.9 × 10 8 mol (4 × PDTH) NH 4 + formation from N 2 in Hadean hydrothermal systems Figure 11 NH 4 + formation from N 2 in Hadean hydrothermal systems. Fluxes are calculated as a function of N 2 conversion between 1 and 10%. NH 4 + formation from NO 2 -/NO 3is not included in these calculations.
Hadean hydrothermal flow as a function of total Earth's heat flow (expressed as multiplicities of present day heat flow -PDHF) Figure 10 Hadean hydrothermal flow as a function of total Earth's heat flow (expressed as multiplicities of present day heat flow -PDHF). Each data line thus represents percentage of the total heat flow released through hydrothermal systems at a given value of Hadean heat flow. The shaded area represents assumed realistic scenarios for the Hadean. and 1.2 × 10 9 mol (8 × PDTH) if 5% of Earth's heat is removed via SHS and between 9.4 × 10 9 mol (4 × PDTH) and 1.9 × 10 10 mol (8 × PDTH) if 80% of heat is removed via SHS. Conversely, with a 10% N 2 conversion efficiency, the annual NH 4 + production would be between 5.9 × 10 10 mol (4 × PDTH) and 1.2 × 10 11 mol (8 × PDTH) if 5% of heat is removed via SHS and between 9.4 × 10 11 mol (4 × PDTH) and 1.9 × 10 12 mol (8 × PDTH) if 80% of heat is removed by SHS (Tab. 5). To place these modeled fluxes in context we can compare their magnitude to those of other proposed NH 3 /NH 4 + formation mechanisms (Fig.  11). An annual NH 4 + flux at 1% conversion efficiency, for example, would be comparable to that based on a homogeneous reaction (Reaction 20) of Summers and Chang [4] or to the flux calculated by Brandes et al [13], which was based on NH 3 formation in the presence of various minerals between 300 and 800°C. 6Fe 2+ + 7H + + NO 2 -→ 6Fe(III) + 2H 2 O + NH 3 (20) Although we cannot comment on the total NH 4 + content of the Hadean Ocean, we can estimate the contribution of hydrothermal N 2 reduction per unit of time. For timescales longer than 1 year, the following equation may be used (21): NH 4 + production in mol.yr -1 can be taken from Fig. 11 or Tab. 6 (or supplied from reader's own sources), t denotes the time period in years and V ocean is the total volume of Hadean Ocean in liters. We have calculated a scenario for one million years using the present-day global ocean vol-   . These results (Fig. 12, Tab. 7) represent an upper contribution limit of this reaction, because no sinks (ion exchange, photooxidation, loss to the gas phase, formation of organic molecules, etc) were taken into account. In the absence of a comprehensive Hadean Nitrogen Cycle model, it is difficult to quantitatively assess the annual loss of NH 3 /NH 4 + from the ocean; how-ever, the numbers in Fig. 12, Tab. 7 can be simply amended by assumption of loss expressed in %.
It is imperative to note that due to a large number of unknown and/or poorly constrained variables, these calculations should only be regarded as a first order approximation. However, it is clear that N2 reduction, albeit very inefficient, could have been a significant source of NH4+, especially in localized environments.
Because NO 2 -/NO 3are inherently easier to reduce than N 2 , its presence in advected seawater could have significantly increased the annual NH 4 + hydrothermal flux. It is unclear; however, how much NO 2 -/NO 3would be advected into the crust, especially in the presence of such significant sinks as the reduction by Fe(II) [5]. While this process is a viable pathway to abiotic NH 3 /NH 4 + , its operation is dependent on atmospherically-driven processes of NO 2 -/NO 3formation as well as chemical composition of the Hadean Ocean, especially pH and mFe 2+ . A change in one of the parameters (e.g., a shift in oceanic pH) may have negatively affected or completely halted this pathway. We assume that serpentinization-driven SHS would have been less affected by changes in ocean water chemistry because their physico-chemical conditions (e.g., pH, fH 2 ) are determined by fluid-rock interactions (e.g., availability of fresh rock) and possibly magmatic input rather than ocean composition. Moreover, the high temperature and pressure conditions combined with accumulations of suitable minerals would make these environments well suited for a long term, sustained NH 3 /NH 4 + production on the Hadean Earth.
Besides facilitating the production of NH 3 /NH 4 + , metals and alloys in SHS may have been involved in other reac-Estimated increase in NH 4 + concentration of the Hadean Ocean (in μmol.L -1 ) from the hydrothermal N 2 reduction reaction per 1 Ma as a function of N 2 -to-NH 4 + conversion percentages, heat flow and percentage of heat released via hydrothermal systems (5, 20 and 80%) Figure 12 Estimated increase in NH 4 + concentration of the Hadean Ocean (in μmol.L -1 ) from the hydrothermal N 2 reduction reaction per 1 Ma as a function of N 2 -to-NH 4 + conversion percentages, heat flow and percentage of heat released via hydrothermal systems (5, 20 and 80%). tions potentially important for prebiotic synthesis. In the context of environmental science, for example, Fe 0 was found to reduce nitrobenzene [91] or to facilitate reductive dehalogenation of carbon tetrachloride and chloroform [92][93][94]; Fe 2+ sorbed on Fe(III) minerals decomposes nitrobenzene [60]. Our future research will also assess SHS as potential sinks of prebiotic molecules during the late Hadean/early Archaean.
The notion that N is commonly present in Fe 0 in its reduced form presents a possibility of meteoritic delivery of reduced N species to Earth, especially during the periods of heavy bombardment. Fe 0 and its alloys (e.g., tetrataenite, awaruite, kamacite) are among the dominant mineral phases in iron meteorites and to a lesser extent in stony-iron meteorites [e.g., [95]] [96][97][98][99].
To assess the possible importance of meteoric delivery of reduced N to Earth, we submitted four octahedrites for inert gas fusion analyses (IMR Test Labs, Lansing, NY). The meteorites -Bogou (IAB), Sikhote Alin (IIAB), Canyon Diablo (IAB), N'Goureyma (Ungrouped) [100][101][102][103] (Stony Brook University's meteorite collection) contained 22 to 32 ppm of N TOT (Tab. 5). Although these analyses provided no insight into the oxidation state or speciation of nitrogen in these meteorites, it is likely to be present predominantly in the form of nitride (N 3-) as is the case in similar commercially available metals, alloys and known meteorite minerals (e.g., roaldite, carlsbergite). Hence in the following calculation we assume that all meteoriteassociated nitrogen is present as nitride. Nitride would be readily released from meteorites after falling into the Hadean Ocean as a result of the rapid and complete dissolution due to inherent instability of Fe 0 in aqueous solutions (even in O 2 -free solutions). Aqueous nitride is expected to react quickly with protons to form NH 3 /NH 4 + . Similar scenario for meteoritic delivery of phosphorus has already been proposed by Pasek and collaborators [104,105].
To constrain the influx of meteoritic N to the Hadean Ocean we have adapted the meteorite flux values used by Pasek et al [104,105]: 2 × 10 5 kg.year -1 (current flux of iron meteorites to Earth; 50% of total meteoritic flux by weight) and meteoritic flux 10 5 -10 6 times the present-day value during the Late Heavy Bombardment Period. Using these values we have created models for varying total reduced N content of iron meteorites: 5, 10, 15, 20, and 30 ppm. For comparison, average and median values for N TOT from 91 published analyses [106,107] and four analyses acquired in this study were 20.1 and 12.3 ppm respectively.
The results presented in Fig. 13 show that during the Late Heavy Bombardment Period, iron meteorites could have delivered ~10 3 to 10 5 mol.yr -1 of N TOT to Earth which is approximately six to nine orders of magnitude less than our estimates for hydrothermal production (Fig. 11). Although the N influx was likely negligible on the global scale it is possible that crater lakes associated with iron meteorite impacts [108] may have contained significant concentrations of NH 3 /NH 4 + . Environments containing NH 3 /NH 4 + and organic phosphorous compounds (e.g., phosphonates, organophosphates) from corroding iron meteorites ((Fe,Ni 3 )P) [104,105] thus could have created very favorable, spatially-restricted conditions for prebiotic synthesis, perhaps unparalleled on the prebiotic Earth. Conclusion 1) N 2 reduction to NH 4 + was found to be limited (up to 2.5% at 200°C) compared to NO 2 -/NO 3 -(100% at 200°C) 2) Metals are more effective at reducing NO 2 -/NO 3than alloys; NH 4 + is the dominant reaction product.
3) The reduction process exhibits a strong temperature dependence. 4) Fe 0 and Ni 0 were found to be least-and most resistant to alteration, respectively. 5) Ni 0 , Fe 0 , Ni 50 Fe 50 , Ni 81 Fe 19 were found to contain up to 124 ppm of nitrogen in their structures, some of which is released upon dissolution and reacts to form NH 4 + .
Estimated annual flux of nitrogen from iron meteorites Figure 13 Estimated annual flux of nitrogen from iron meteorites. Different line styles correspond to different average N content of iron meteorites. Shaded areas represent iron meteorite flux during the period of Late Heavy Bombardment.