Introduction

Phosphorus (P) is a critical element in modern biochemical systems, where it serves to store metabolic energy as ATP, forms the backbone of genetic material such as RNA and DNA, and separates cells from the environment as phospholipids. The demonstration of a prebiotically plausible route to forming phosphorylated biomolecules, or ancestral versions of these molecules, has been confounded by the poor solubility and low reactivity of phosphate minerals under early earth conditions (e.g., Keefe and Miller 1995).

Of the major biogenic elements (e.g., CHONPS), phosphorus alone lacks a significant volatile phase. Ultimately, the phosphorus that formed the first phosphorylated biomolecules had to come from a mineral source. The variety of phosphate minerals likely present on the early earth was quite small relative to the large variety present under modern earth conditions (Hazen 2013). The most likely phosphate minerals present on the early earth would have included calcium phosphate minerals such as apatite and whitlockite, and the rare earth element phosphate mineral monazite, as well as a few even rarer phosphate minerals (Hazen 2013). The mineral struvite, MgNH4PO4×6H2O, has shown the greatest promise as a plausible, prebiotic phosphate mineral (Handschuh and Orgel 1973), though this mineral was not likely abundant on the early earth (Gull and Pasek 2013). The formation of organophosphates from these minerals is not spontaneous: the equilibrium constant K is less than 0.001 for the formation of phosphorylated sugars using orthophosphate; other organic compounds are even less favorable. The most plausible route to forming organophosphates from phosphate minerals typically invokes reaction of these minerals with hot, acidic water, then heating of dissolved phosphate to form polyphosphates (Yamagata et al. 1991). Polyphosphates, the simplest of which is diphosphate or pyrophosphate, could then phosphorylate organic compounds in higher yields (e.g., Pasek and Kee 2011).

An alternative to a phosphate mineral source on the early earth is a meteoritic phosphide source—the mineral schreibersite—as proposed by Pasek and Lauretta (2005) and Bryant and Kee (2006). Recent results have shown that schreibersite could have influenced early earth P chemistry (Pasek et al. 2013), and that schreibersite is capable of spontaneously phosphorylating organic compounds, with phosphorylation of glycerol demonstrated thus far. In this paper we review the process by which the mineral schreibersite reacts with water, provide new estimates on the flux of meteoritic phosphorus to the surface of the earth, place constraints on the ultimate fate of schreibersite as it reacts over geologic time, and provide a brief summary on the history of the investigation of schreibersite as a potential prebiotic P source.

Review of the Reactions of Schreibersite with Water

Schreibersite is a phosphide mineral common to many meteorites, but is found on the earth only in exceedingly rare environments (e.g., Essene and Fisher 1986). Its formula is (Fe,Ni)3P, and it forms a solid solution with the mineral nickelphosphide (Ni,Fe)3P. Schreibersite was first described by Berzelius (1832), and is one of oldest “meteorite-only” minerals characterized by early mineralogists (Pasek 2014). Alloys containing Fe-Ni are amongst the earliest minerals to have condensed from the solar nebula (Grossman 1972; Kelly and Larimer 1977) and both schreibersite and perryite [(Fe,Ni)~8(Si,P)~3] are important minor phases therein (Lehner et al. 2010). Indeed, the presence of nickeliferous (>5 % Ni) schreibersite can be a diagnostic feature of a meteorite. Schreibersite forms as a result of the siderophilic (metal-loving) nature of the element phosphorus at high temperatures under reducing conditions. Though generally recognized as a lithophile (rock-loving element), on a cosmic scale phosphorus will tend to bind with iron before it oxidizes and reacts with calcium to form phosphate minerals. During planetary differentiation, or the separation of a planet into a lighter silicate mantle and a denser metal core, phosphorus will preferentially enter the core. In fact, about 97 % of the earth’s phosphorus is within the core, based on chondritic and mantle abundances of P (McDonough and Sun 1995).

The oxidation state of phosphorus in schreibersite is not obvious to a traditional student of chemistry (Fig. 1). An inorganic chemist who deals with synthetic phosphides such as AlP or Ca3P2 typically assumes an oxidation state for P of −3 in these materials. This oxidation state is consistent with the primary product of reaction of synthetic phosphides with water: the gas phosphine (PH3). However, the oxidation state of phosphorus in schreibersite appears to be closer to −1, based on electron binding energy curves of the P and iron atoms in the metallic mineral (Fig. 2, see also Bryant et al. 2013).

Fig. 1
figure 1

Schreibersite corrosion schematic. Schreibersite, a picture of which is provided as the inset (with surrounding material kamacite, an iron-nickel mineral alloy) corrodes to phosphite radicals and iron oxides in water, releasing H2 gas. These phosphite radicals terminate to form a variety of P oxyanions. These P oxyanions may react with organics via phospho-aldol reactions to form organophosphonates or organophosphinates, or by condensation reactions to form organophosphates

Full size image
Fig. 2
figure 2

Oxidation state of iron in schreibersite as determined by XPS. An increase in binding energy corresponds to an increase in oxidation state a XPS analyses of both Fe2P & Fe3P surfaces showing core 2P3/2 electron binding energy region of iron. The oxidation state of iron in the iron phosphide correlates to close to zero to slightly oxidizing (Descostes et al. 2000). The P oxidation state is −1 based on its internal binding energy of 129.5 eV (Pelavin et al. 1970; Bryant et al. 2013). b Fe-2p3/2 electron XPS line scan analysis traversing matrix-schreibersite-matrix regions of a sectioned sample of Sikhote-Alin. The increase in binding energy of iron corresponds to Increasing P-composition and locates the inclusion region that was sampled for XPS analysis

Full size image

The oxidation of schreibersite is exergonic under most conditions, and certainly within the modern atmosphere. The phosphorus within schreibersite is an efficient scavenger of oxidizing agents, and reacts with air, water, CO2, and most other oxygen-bearing molecules. Characterizations of the surface of schreibersite demonstrate that the mineral is coated with phosphorus oxides at thicknesses of 80 nm or more (Pirim et al. 2014). This coating serves to protect the schreibersite from further oxidation in air, and in many meteorite samples schreibersite is one of the last minerals to accumulate rust when stored in air.

The interesting chemistry occurs when schreibersite is placed within water (Fig. 1). Schreibersite reacts with water to oxidize the phosphorus and the iron to form P oxyanions and ferrous iron. There is a stoichiometric release of H2 gas during this process (Pasek and Lauretta 2005), carrying away extra electrons during this net oxidation. Bryant and Kee (2006) also demonstrated that the oxygen reacting with the phosphorus came from the water: when schreibersite was reacted with water labeled with the 18O isotope, the isotopomer ratio indicated that three of the 18O- oxygen atoms attached to the P oxyanions could be traced to water, the remaining oxygen to adventitious presence of dioxygen, perhaps bound to the surface (e.g., Pirim et al. 2014).

The major products of schreibersite corrosion are phosphite (HPO3 2−), hypophosphate (P2O6 4−), phosphate (PO4 3−), and pyrophosphate (P2O7 4−). The oxidation state of P in these anions is thus +3, +4, and +5 respectively (Pasek and Lauretta 2005; Bryant and Kee 2006; Pasek et al. 2007). In addition to these primary ions, which are generated in almost every schreibersite corrosion experiment, the other anions that may be present (Pasek and Lauretta 2005; Bryant and Kee 2006; Pech et al. 2011) depending on reaction conditions include the dimer of phosphite pyrophosphite (H2P2O5 2−), the trimer of phosphate triphosphate (P3O10 4−), a very reduced form of P as hypophosphite (H2PO2 −) and the cyclic trimer of phosphate trimetaphosphate (P3O9 3−). Note that the production of pyrophosphite during schreibersite corrosion (e.g., Pasek and Lauretta 2005) has been difficult to replicate, despite its utility in prebiotic P chemistry (Kee et al. 2013), but it is the known to be the principal product from dehydration of aqueous phosphate solutions at pH’s below 5, when heated to dryness at temperatures above 60 °C.

Intriguingly the production of these oxyanions proceeds by release of radicals from the schreibersite, specifically PO3 2− radicals (Pasek et al. 2007). These phosphite radicals were detected by electron paramagnetic resonance spectroscopy (EPR), and recombine to give three products (Schäfer and Asmus 1980) (Schema 1):

Schema 1
scheme 1

The reaction of two phosphite radicals results in the recombination product, hypophosphate, or the disproportionation products, phosphite and phosphate

Full size image

PO3 − and PO3 3− both react with OH− and H+ respectively to give HPO4 2− and HPO3 2−. The yield of 18 % hypophosphate relative to 82 % phosphate and phosphite is consistent with P NMR speciation, though this was not the path considered by Pasek et al. (2007). A radical reaction is also implicated in the production of pyrophosphate, triphosphate and trimetaphosphate, which occurs as the reduced P compounds phosphite or hypophosphite are oxidized (Pasek et al. 2008) in a Fenton reactor, which consists of adding H2O2 to a solution of Fe2+ to generate OH and OOH radicals. Similar condensed P-oxyanions are produced when phosphate and hypophosphite solutions are exposed to microwave plasmas (Pasek et al. 2008).

The formation of organophosphate compounds using schreibersite has only been reported with low yield (e.g., 1 % by Pasek et al. 2007, and 4 % in Pasek et al. 2013). These reactions most likely occur by dehydration reactions on the surface where there is lower water activity, and yields are enhanced when an organic such as urea is added, perhaps following the urea chemistry of Orgel (Österberg et al. 1973). Yields of organophosphates can be increased significantly using a few additional steps (Gull et al., in prep.).

Although the formation of organophosphates has only recently been demonstrated using schreibersite, the more general formation of organophosphorus compounds using schreibersite is more reliable and has occurred since the first experiments of Pasek and Lauretta (2005) and Bryant and Kee (2006). In these reactions, phosphite or hypophosphite reacts with simple aldehydes to generate phospho-aldol addition products (Fig. 3).

Fig. 3
figure 3

31P NMR spectrum resulting from reaction of acetaldehyde and Fe3P in water. The top of the spectrum has been truncated due to the height of the phosphite (+3.5 ppm) peak Decoupled spectrum is in black, coupled is green. The main organic product of this reaction is the phospho-aldol reaction product between H3CCHO and H2PO2 −, at about +30 ppm. The yield of this compound was 12 % based on 31P NMR integration, though quantitative NMR was not performed with this sample. The other major peak is hypophosphite (+7 ppm)

Full size image

However, this can occur even if no organic has been added to the solution (see Fig. 2 of Pasek et al. 2007). These organophosphorus compounds must have formed by introducing carbon from one of four sources. 1) The solutions of water are contaminated or have been incompletely cleaned. This is unlikely as experiments are washed in acid, and formation of organophosphorus compounds occurs even in new glassware, and does not occur even when other organics are left over. 2) The carbon comes from trace carbon dissolved in the synthetic Fe3P powder. Pirim et al. (2014) demonstrated carbon contaminated the surface of both natural and artificial schreibersite samples, though the origin of this carbon is unknown. 3) The carbon comes from the plastic coating of the stir bar. Since the Fe3P is an abrasive powder, the stir bar could be abraded itself and might be the ultimate source of carbon for this reaction. However, since the stir bars are coated with PTFE (Teflon), no aldehydes should be present. 4) Atmospheric carbon dioxide is somehow being fixed by reaction with the water and Fe3P. This is by far the most interesting path, and could be plausible with the volume of CO2 trapped in the 250 mL round bottom flasks used. This area remains open to more research.

One P species conspicuously absent from the Fe3P reactions is the gas phosphine, PH3. However, it has been shown that small amounts of PH3 gas can be generated during the acidic corrosion of Fe3P at low (<2) pH. Currently unpublished work by Herschy (2013) in his Ph.D. thesis described the heating of Fe3P mineral in acidic fluids under anaerobic conditions to 90 °C for a period of 72 h afforded the production of small amounts of phosphine but mainly phosphite (Herschy 2013). Phosphine evolved during the reaction was collected and trapped by bubbling the gas stream though a 0.1 M solution of calcium hypochlorite [Ca(OCl)2] while the phosphite remained in the acidic fluids.

Concentrations of 7.0–7.5 mg/L were reported for phosphine and 6300–6800 mg/L for phosphite were measured in the experiments over a 72 h period by phosphmolybdate colorimetry that accounted for ~0.01 % and ~10 % of available phosphorus respectively. Further investigations into the rate of phosphine production showed that the majority of the phosphine observed in the experiment (~79.5 %) was produced in the first 24 h. This was speculated to be a result of the already corroded surface acting to inhibit the rate of corrosion.

Glindemann et al. (1998) also reported a 0.01 % production of phosphine from acidic corrosion of steel using a similar sample size and corrosion time. UV photolysis of PH3 in the presence of water vapor has been shown to generate phosphate and hypophosphite and has been proposed to be one method of cloud nucleation in the upper troposphere (Glindemann et al. 2003). The susceptibility of phosphine to react with water vapor under UV irradiation also accounts for the diurnal concentration and dissipation of phosphine in the atmosphere. Phosphine provides a highly reactive and highly mobile form of reduced P and the photo-oxidation of PH3 in the upper troposphere would be an excellent source of reduced P oxy-acids which would be dispersed in the atmosphere and then deposited back to the surface by rainfall.

For the researcher interested in working with this mineral in chemical reactions, schreibersite can be acquired by purchasing iron meteorites such as Canyon Diablo, Seymchan, and Sikhote Alin. In many of these meteorites, the schreibersite makes up ~5 % of the volume of the meteorite. The synthetic analog Fe3P can be purchased from chemical suppliers. This synthetic material varies from natural schreibersite in that it has no nickel, but is otherwise structurally identical, and reacts to produce identical phosphorus oxyanions as schreibersite (Pirim et al. 2014), albeit faster than the nickel-bearing variety.

Reacting Fe3P with water (usually about 1 g in 10–25 mL of water) releases phosphate and other P oxyanions to solution in detectable quantities over the course of about 1 week, but also rusts the iron to form Fe2+ and Fe3+. These ions interfere with NMR analyses and also bind phosphate and other anions to the surface of fine metal oxides. In this case, the addition of a strong base, typically 1 mL of 1 M NaOH or Na2S, forces the iron to precipitate and liberates P to solution. After centrifuging the solution and filtering it, the solution is ready for analysis by NMR, after adding some portion of D2O as an NMR lock. A typical NMR spectrum of a corroded schreibersite solution is shown as Fig. 4.

Fig. 4
figure 4

Typical 31P NMR spectrum of water after of Fe3P corrosion, pH 13. From left to right the peaks correspond to hypophosphate (P2O6 4−), phosphate (PO4 3−), phosphite (HPO3 2−), and pyrophosphate (P2O7 4−)

Full size image

The Context of Schreibersite on the Earth

Schreibersite completely corrodes primarily to phosphite in ocean water over about 10 years (Bryant et al. 2009), a short timescale in geologic terms. The discovery of phosphite in Archean limestones (3.52 Ga) by Pasek et al. (2013) may allow a determination of the quantity of phosphite, and thereby role of meteoritic material in the early oceans. The presumed carrier of this phosphite is CaHPO3 in these rocks. The simplest constraint on the primordial concentration of phosphite in the early oceans is the production of calcium phosphite:

$$ {{\mathrm{HPO}}_3}^{2-}+{\mathrm{Ca}}^{2+}\to {\mathrm{Ca}\mathrm{HPO}}_3\left(\mathrm{s}\right) $$

The K of this reaction is 106 (Van Wazer 1958), implying that, if the oceans had close to modern calcium concentrations (10−2 M), then the phosphite concentration of the Archean ocean would be about 10−4 M to form solid calcium phosphite (typically a monohydrate) and become trapped in the limestone. In contrast, the phosphate concentration of the modern ocean is about 10−6 M or less, suggesting that over 99 % of the P in early oceans bore a reduced oxidation state at this time.

It is important to note that very little is known about the geochemical behavior of the phosphite anion, as little justification existed for study of this compound. However, the recent discoveries of phosphite in natural samples (e.g., Pech et al. 2009, 2011; Pasek and Block 2009; Han et al. 2013; Pasek et al. 2014) suggest that studying this compound in the future may be beneficial. It is possible that phosphite may behave differently in solutions with iron oxides and iron hydroxides, or may form solid solutions with calcium-carbonate, or might precipitate at much lower concentrations with other cations. Any of these could alter the calculation of phosphite concentrations in the early oceans by orders of magnitude.

If all of the phosphite in the Archean oceans originated from meteorites, then the 10−4 M concentration above would have been released from 1015 to 1016 kg of schreibersite. This would correspond to a meteoritic mass of as much as 1017 to 1019 kg of material. We previously estimated the flux of phosphorus from schreibersite during peak bombardment as 108 kg/year (Pasek and Lauretta 2008). A heavy bombardment duration of 108 years would deliver the required phosphite to meet the phosphite concentration calculated above. This initial calculation was based on a heavy, rapid spike of bombardment at 3.8 Ga, consistent with the prevailing models of bombardment last decade. Recently, this bombardment spike has been called into question, as examination of orbital mechanics (Bottke et al. 2012), the geologic record (Johnson and Melosh 2012), and the lunar record (Zellner et al. 2009; McCaffrey et al. 2014) suggest a less intense pulse with a longer duration.

That schreibersite is the source of this phosphite is not proven (indeed, is unprovable) and there might be an alternative source for this material. One possibility could be volcanism. Yamagata et al. (1991) demonstrated polyphosphates in volcanic fumarole gas, but in addition, had 2–3 unknown compounds in their analyses. Some of these may have been reduced P compounds. If so, exhalation of P2O3 type gases (as opposed to P2O5 type) could have contributed to the reduced P budget of the early earth. Alternatively, the Archean crust may have provided some of this reduced P. Although it has recently fallen out of favor, the source material for the earth could have been the extremely reducing enstatite chondrites. Oxygen isotope systematics suggest that the source material for the earth are this class of meteorites characterized by extremely reduced mineralogy (Javoy et al. 2010). If the rocky portion of the earth indeed formed from enstatite chondrites, then a reducing ocean with phosphite as a significant component would be feasible, as would reduced oxidation state P from gases from early volcanoes.

One alternative source for phosphite on the early earth is impact reduction of crust material. During a meteorite impact, the target material and the impactor mix and oxygen is segregated into the plume (Melosh 1989). As a result, rocks and melt ejected by impact bear a lower redox state than the starting material (Fudali et al. 1987). Hypothetically this should hold true for P in the melt as well. Glindemann et al. (1999, 2004) and Pasek and Block (2009) demonstrated that a close analog process, lightning striking the ground, results in the reduction of phosphate oxidation state by lightning. This process could parallel that of the impact reduction of phosphate.

Independent of the source of reduced P on the early earth, if the phosphite was present in the Archean oceans at about 10−4 M, then ultimately this phosphite had to oxidize within a billion years or so (Pasek 2008). The oxidation of phosphite is best promoted by reaction with OH radicals or other oxidizing agents. If the oxidation of phosphite in the Archean proceeded by a reaction analogous to the Fenton-style reactor of Pasek et al. (2008), then the 10−4 M of phosphite would have oxidized to 6 × 10−5 M phosphate, 3 × 10−5 M pyrophosphate, and 5 × 10−6 M of both triphosphate and trimetaphosphate. Trimetaphosphate has recently been demonstrated to be a phosphorylating agent for ribozymes (Moretti and Müller 2014), and phosphite oxidation may be a plausible source for this early RNA synthesis.

A Brief Retrospective

In contrast to the some of the work in the origins of life field, experimental investigations of schreibersite are a recent development with about a decade of work. A retrospective may thus seem premature. However, the original idea that reduced oxidation state phosphorus could have been present on the early earth, possibly delivered by meteorites, was first espoused by Gulick over half a century ago (Gulick 1955, 1957). No experimental data was presented at the time to back the claim up, and Miller and Urey (1959), using thermodynamic arguments quickly dismissed this idea, as did much of the field at the time.

The possibility that meteorites could have acted as a source of phosphorus on the early earth was renewed again by George Cooper’s work on the Murchison meteorite phosphonates (Cooper et al. 1992). Cooper’s discovery demonstrated the presence of phosphonates—organic compounds with a single C-P linkage—at nanomolar concentrations. These compounds, which occurred as a homologous sequence of alkyl compounds from methylphosphonate to butylphosphonate, demonstrated that meteorites indeed carry soluble P compounds as part of their organic inventory. This discovery, although covering a somewhat obscure group of chemicals not typically associated with biology, brought renewed interest in the impact of meteorites on the inorganic prebiotic inventory of the early earth.

Several studies followed George Cooper’s initial report, including potential pathways for the formation of meteoritic phosphonates identified by De Graaf et al. (1995, 1997, 1998) and the possibility of potential C-P bearing nucleic acid analogs (Peyser and Ferris 2001). Alan Schwartz’s lab demonstrated a synthetic route to the phosphonates that involved reaction of organic compounds with phosphite, stimulated by UV radiation. The phosphite was proposed to come from the meteorite, possibly from reaction of schreibersite.

The first experimental work on schreibersite as a mineral capable of phosphorylation was arrived at independently by two groups. Matthew Pasek and Dante Lauretta first characterized the reaction of phosphides with water (Pasek and Lauretta 2005), and David Bryant and Terry Kee demonstrated that these reactions involve water and not trace oxygen (Bryant and Kee 2006). Pasek’s dissertation work at the time began as an investigation of the extracts of several carbonaceous meteorites, with the idea that extracts could be analyzed by 31P NMR to characterize bulk P speciation. Initial results were disappointing, and continued to be so, as the concentration of all ions aside from phosphate were too low to be measured by NMR. As a consequence, 31P NMR analyses of the extracts (using a solution of EDTA) of the major P-bearing minerals, including apatite and schreibersite (as synthetic Fe3P), were performed to determine P speciation. Intriguingly the extracts of Fe3P showed a lot more peak variety than the apatite extract. Identifying the species giving these peaks, the reaction processes necessary to make these species, and the prebiotic potential of these compounds has taken a full decade.

Conclusion

Invoking an extraterrestrial source for phosphorus in prebiotic chemistry may (and did) strike many as odd. However, the data suggest that schreibersite is a versatile source of P for the building of prebiotic molecules. In addition to enhanced P solubility, the enhanced reactivity of P coming from schreibersite would argue for its role in prebiotic chemistry. Additionally, some geochemical lines of evidence are pointing towards reduced P being present on the early earth. Although there may exist sources of reduced P that have not been identified, a meteoritic source could suffice.

Ultimately, the phosphorus in life had to come from a mineral source. The nature of this reaction has been long lost to the passage of geologic time. Indeed, whether P was even in the first biomolecules is unknown. Life has built on, and replaced its ancestral molecules (e.g., Hud et al. 2013), to the point where elucidating the exact chemical path for the origin of life is impossible. However, demonstration of a plausible prebiotic pathway to forming prebiotic molecules, including phosphorylated molecules, merits study as it identifies the fundamental chemistry that determines what life is. Tying these pathways to viable geochemical observations narrows down this work to a set of plausible routes that will eventually identify the suite of mechanisms that enable life to arise.