Reactive ammonia in the solar protoplanetary disk and the origin of Earth’s nitrogen

Abstract

Terrestrial nitrogen isotopic compositions are distinct from solar and cometary values and similar to those of primitive meteorites, suggesting that Earth’s atmospheric nitrogen originates from a primordial cosmochemical source^1,2. Prebiotic organic compounds containing nitrogen that formed in the solar protoplanetary disk, such as amino acids, may have contributed to the emergence of life on Earth^3,4. However, the original reservoirs of these volatile compounds and the processes involved in their distribution and chemical modification before accretion remain unclear. Here we report the occurrence of the mineral carlsbergite (chromium nitride) within nanocrystalline sulphide inclusions of primitive chondritic meteorites using transmission electron microscopy and secondary ion mass spectrometry. The characteristics and occurrence of carlsbergite are consistent with precipitation from a chromium-bearing metal in the presence of reactive ammonia. The carlsbergite crystals have nitrogen isotopic compositions that differ from ammonia in cometary ices, but are similar to Earth’s atmospheric nitrogen. We suggest that the reactive ammonia proposed to have initiated formation of the carlsbergite came from ices within regions of the protoplanetary disk that were affected by the distal wakes of shock waves. Our findings imply that these primordial ammonia-bearing ices were a nitrogen reservoir within the formation region of the chondritic meteorite parent bodies and could have been a source of volatiles for the early Earth.

Main

Phosphorus- and chromium-bearing Fe, Ni sulphide (PCS, Fig. 1a) is a characteristic minor constituent of the volatile-rich CM2 chondrites⁵, but its petrogenetic significance remained unclear. To address this issue, we used focused ion-beam (FIB) preparation and transmission electron microscopy (TEM) to study PCS from the CM2 chondrites Yamato (Y-) 791198 and Y-793321. Earlier studies by TEM and selected area electron diffraction (SAED) had indicated that PCS is a nanocrystalline, polyphase material, but remained inconclusive about the mineralogical identities owing to broadened electron diffraction rings⁶. On the basis of our TEM observations, we have identified phosphorus-bearing, nanocrystalline pentlandite [n-Pn; (Fe, Ni)₉(S, P)₈] as the major component of PCS (Fig. 1b and Supplementary Figs 1 and 2).

Figure 1: Mineralogy of PCS grain 98F03 from Y-791198.

a, Backscattered electron image of the PCS grain before FIB sectioning. b, SAED pattern (∼12 μm² area) of the FIB sample showing the main lattice spacings of nanocrystalline pentlandite (n-Pn) as two broad rings accompanied by ring segments of carlsbergite (CrN) reflections, which indicate a non-random crystal orientation. c, Bright-field TEM image of the FIB sample showing carlsbergite platelets embedded in n-Pn. d, High-resolution image of a single carlsbergite platelet showing the dominant {100} face with measured lattice spacings (95% confidence of the mean, n = 4).

The second and surprising major component of PCS in both meteorites is carlsbergite, a cubic chromium nitride (CrN; Fig. 1b, c). Carlsbergite forms platelet-shaped crystals of typically less than 100 nm width and 3–4 nm thickness. In two PCS grains we found a few exceptional large platelets of 0.4–1.2 μm width and 40–120 nm thickness. The dominant crystal forms are pairs of parallel {100} faces (Fig. 1d). TEM imaging and SAED patterns show that small carlsbergite platelets occur abundantly distributed within PCS (∼500 μm⁻²) and have preferred orientations (Fig. 1b). The composition of carlsbergite was verified by electron energy-loss spectrometry (EELS, Supplementary Fig. 3). The bulk PCS contains 1–2 weight% (wt%) nitrogen, which was estimated from the bulk chromium contents measured by energy-dispersive X-ray spectrometry (EDXS) on regions containing only carlsbergite and Cr-free n-Pn (Supplementary Table 1).

Isotopic imaging of PCS by secondary ion mass spectrometry (NanoSIMS) confirmed high nitrogen concentrations not associated with carbon-rich regions but with internal carlsbergite crystals (Fig. 2 and Supplementary Fig. 4). Also the surrounding PCS was shown to contain nitrogen, but at lower concentrations. The nitrogen isotopic compositions of carlsbergite crystals are indistinguishable from those of PCS. The average ¹⁵N/¹⁴N ratio of four PCS grains in Y-791198 is 3.86 ± 0.06 × 10⁻³ (95% confidence, Supplementary Table 2) and overlaps with typical bulk values of CM chondrites⁷. The value is comparable to modest enrichment of ¹⁵N in ammonia (NH₃) and soluble organic matter in CM chondrites⁸ and their amino acids^4,8. It is close to the isotopic composition of Earth’s atmosphere⁹ (¹⁵N/¹⁴N = 3.676 × 10⁻³), which falls between the extremes of the solar wind⁹ (¹⁵N/¹⁴N = 2.18 ± 0.02 × 10⁻³), nitrogen in cometary comae¹⁰ (¹⁵N/¹⁴N = 6.8 ± 0.5 × 10⁻³) and nitrogen in the CB–CH chondrite Isheyevo¹¹ (¹⁵N/¹⁴N up to 2.2 × 10⁻²).

Figure 2: Localization of nitrogen by NanoSIMS.

a, Secondary electron SEM image of the surface after NanoSIMS analysis. The nanocrystalline pentlandite (n-Pn) has been sputtered preferentially and exposed internal grains. b, The same image overlaid with the distribution of ¹²C¹⁴N secondary ions obtained by NanoSIMS. Exposed carlsbergite grains correlate with local maxima (blue–violet) of the ion intensity. Grains not associated with intensity maxima are probably schreibersite. Organic material is present in the periphery.

The mineralogical context of PCS is often elusive, because PCS occurs mostly isolated within matrix material or fine-grained chondrule rims. In Y-791198 we encountered 50–60 μm large aggregates containing cores of PCS surrounded by shells of polycrystalline sulphides (Supplementary Fig. 5). TEM revealed contrasting oxidation states of Fe, Cr and P among the two zones: the inner PCS consists of a reduced assemblage of n-Pn, carlsbergite, schreibersite, daubréelite, and Fe, Ni metal, whereas the outer shell contains an oxidized assemblage of normal pentlandite, Fe-deficient 4C-pyrrhotite (Fe_0.875S), chromite, magnetite and a Ca–Na-phosphate (Supplementary Fig. 6).

The distinct platelet morphology, size range and non-random orientation of carlsbergite crystals strongly resemble CrN precipitates obtained by ammonia nitriding of Cr-bearing steels¹², where these properties are controlled by the Baker–Nutting orientation relationship between the crystal structures of CrN and body-centred cubic iron (Supplementary Note). The characteristics of natural carlsbergite crystals in PCS therefore strongly indicate that the nitride originally formed by precipitation from a Cr-bearing Fe alloy under conditions similar to ammonia-based nitriding.

Cr-bearing Fe, Ni alloys are indeed found in CM chondrites and typically contain 0.1–1 wt% Cr and P. However, we did not detect carlsbergite in five Cr-rich metal grains from Y-791198 and Y-793321. The partial pressure of N₂ in the protoplanetary disk was several orders of magnitude lower than the pressure required to dissolve the amount of nitrogen found in PCS in the precursor metal (Supplementary Discussion). Therefore, the sustained growth of carlsbergite itself must have been the mechanism for the nitrogen enrichment.

Although carlsbergite is estimated to carry less than 1% of the total nitrogen in CM chondrites, it gives an insight into the processing of nitrogen in the early Solar System. The sustained growth of carlsbergite requires the activity of dissolved monatomic nitrogen in the metal to be maintained in exchange with the surrounding gas. Molecular nitrogen (N₂) is a poor nitriding agent owing to its low solubility and strong N ≡ N bonding. However, mixtures of H₂ and NH₃ can provide high activities of monatomic nitrogen (N) through metal-catalysed dissociation of metastable ammonia according to the reaction NH₃ → N + 3/2 H₂. Under such conditions, the activity of atomic nitrogen, expressed as the nitriding potential $P_{{\text{NH}}_3}/P_{{\text{H}}_2^{3/2}}$, can be several orders of magnitude higher than that of N₂ gas¹³.

We suggest that the formation of n-Pn occurred through the reaction of the meteoritic metal with hydrogen sulphide (H₂S) at temperatures of 700–800 K shortly after the precipitation of carlsbergite had taken place. The sulphidation left the chemically resistant nitride unaffected, except for partially disturbing its preferred orientations. The formation of PCS by sulphidation is supported by experimental observations showing that P-bearing Fe, Ni metal reacts with H₂S to form P-enriched sulphide layers associated with normal pyrrhotite/troilite and pentlandite¹⁴.

The coexistence of Fe-deficient pyrrhotite and magnetite in the outer zone of shelled PCS grains places further constraints on the formation environment, requiring elevated sulphur and oxygen fugacities and correspondingly high H₂S/H₂ and H₂O/H₂ ratios. We interpret the zoning as a result of diffusive and reactive fractionation of gas species in the outer, oxidized zone. This led to a steep gradient in the redox conditions towards more reducing conditions close to the reacting metal interface. Hence, nitridation and sulphidation must have been rapid processes capable of maintaining the disequilibrium mineral assemblage of the zoned PCS/sulphide aggregates.

The alternative formation of carlsbergite by direct condensation from a solar gas was previously ruled out¹⁵. Furthermore, it can be expected that the unobstructed growth from a gas phase would result in isometric carlsbergite crystals that mirror its cubic symmetry. The platelet shapes observed in our samples clearly point to growth controlled by crystallographic interface energies. This argument also effectively rules out the formation of carlsbergite by low-temperature hydrothermal activity on the CM parent body below 350 K (ref. 16), because nitridation would have been extremely slow and basically impossible owing to the stability of ammonia at these temperatures (Supplementary Discussion). Moreover, the formation of highly Fe-deficient 4C-pyrrhotite instead of pyrite (FeS₂) in the oxidized outer sulphide zone requires temperatures above 750 K (Supplementary Fig. 7) that are incompatible with parent-body alteration.

Although the formation of ammonia is thought to have been kinetically unfavourable in canonical solar gas¹⁷, enrichment of ammonia is expected to occur in ices, as indicated by condensation calculations^17,18 and shown by spectroscopic studies of cometary comae¹⁹ and pre-stellar molecular clouds, where ammonia is probably the main nitrogen reservoir²⁰. By using the production rates of gaseous species of comet Hale–Bopp¹⁹ as proxies for the elemental compositions of primordial ice, we have calculated the equilibrium speciation of mixtures of solar gas and ice at variable mixing ratios and temperatures above 750 K (Supplementary Fig. 8). At these temperatures ammonia is not a major equilibrium species and the calculated nitriding potentials are distinctly smaller than the 3 × 10⁻⁴ Pa^−1/2 potential used in rapid, laboratory-scale nitriding of low-Cr Fe alloy²¹, which we consider a minimum requirement.

In contrast, kinetic calculations show that between 750 K and 1,260 K (the Fe–FeS eutectic) ammonia is retained metastably and the nitriding potential remains above the critical value at least for days (Fig. 3 and Supplementary Fig. 9). This suggests that nitridation in ice-enriched environments would have been possible on the timescale of hours to days even if only a fraction of the ammonia remained undissociated after the release. At 750 K, the mixing trajectories of H₂O/H₂ and H₂S/H₂ trend towards the pyrite–pyrrhotite–magnetite triple point (Fig. 4), suggesting that the formation of Fe-poor 4C-pyrrhotite and magnetite was possible at a C/O ratio of about 0.16 in the ice, which is lower than the calculated C/O ratio of 0.29 for Hale–Bopp ice.

Figure 3: Kinetic evolution of the nitriding potential in gas mixtures between solar gas and an ice-derived gas with Hale–Bopp-like species composition.

Shown are mixtures with 99 mol% (circles) and 90 mol% (squares) contributions of ice at 750 K/10 Pa total pressure (red), 750 K/1,000 Pa (green), 1,260 K/10 Pa (blue) and 1,260 K/1,000 Pa (orange). Coloured dashed lines indicate the corresponding equilibrium values. In a homogeneous gas the nitriding potential remains metastably above 3 × 10⁻⁴ Pa^−1/2 at least for days. At this value rapid CrN formation has been demonstrated²¹.

Figure 4: Gas mixing at 750 K between solar gas (diamonds) and an ice-derived gas of Hale–Bopp-like elemental composition (circles).

Elemental C/O = 0.29 (red) and C/O = 0.16 (blue, orange). Mixing trajectories at 10 Pa (solid lines) and 10³ Pa (dashed lines) total pressure are calculated for full equilibration (blue, immediate reaction of NH₃) and suppressed N₂ (orange, maximum retention of NH₃). The ratios of H₂O/H₂ and H₂S/H₂ trend towards the phase boundary of Fe-deficient pyrrhotite and magnetite for contributions of more than 90 mol% ice (squares).

Our observations provide evidence for the involvement of reactive ammonia in the formation of the nitride–sulphide–oxide assemblages and, hence, in protoplanetary processes. The most plausible environments for the release of ammonia from multi-component ices are ice-bearing disk regions affected by relatively gentle, yet rapid, heating in the distal wakes of shock waves. Proximal shock waves of moving bodies and gaseous impact plumes created by the collision of icy planetesimals seem less likely, because in these cases large fractions of the ices would have been subjected to high temperatures and therefore more rapid dissociation of ammonia.

On theoretical grounds, the shock-wave model has been found feasible for the pre-accretionary hydration of silicates²². Our results provide natural evidence that similar gas–solid interactions indeed took place. This has important implications for the evolution and interaction of volatile reservoirs in the early Solar System: the involvement of multi-component, ammonia-bearing ices implies that generally favourable conditions for complex chemical reactions (for example, Fischer–Tropsch-type reactions²³) towards the formation and modification of organic matter existed. Moreover, the only modest enrichment of ¹⁵N in carlsbergite suggests that the involved ammonia was part of a primordial icy reservoir with a nitrogen isotopic ratio close to that of Earth’s atmosphere, but distinct from the solar isotopic composition and cometary nitrogen. Hence, this reservoir may have substantially contributed to Earth’s early atmosphere, either as ices or modified organics.

Moderate enrichment of ¹⁵N occurs in prebiotic amino acids of CM chondrites^4,8 and is thought to have been incorporated during their synthesis within the parent asteroid(s). This process was coeval with their alteration by aqueous fluids and required dissolved ammonia of an as yet unclear origin²⁴. The incorporation of ammonia-bearing ‘chondritic ice’ into the chondrite-forming region could therefore have played an important role in the synthesis and modification of organics and their subsequent delivery to the early Earth.

The origin of this icy reservoir is, however, unknown. Ices may have been delivered into the inner regions of the emerging Solar System by the scattering of trans-Jovian, icy planetesimals during the ‘Grand Tack’²⁵. The nitrogen may have been mixed from isotopically light solar nitrogen and heavy, comet-like nitrogen of the outer protoplanetary disk, or it may have been derived from a region in the disk where originally ¹⁵N-poor interstellar ammonia²⁶ underwent only moderate ¹⁵N-enrichment^2,27. Although at least one comet with close-to-terrestrial D/H ratios is known²⁸, its canonical, ¹⁵N-rich nitrogen isotopic composition is at odds with such comets having delivered nitrogen to Earth¹. Although many of the ice-rich bodies probably lost most volatiles after the dissipation of the protective, gas-rich disk, it remains a possibility that other short-period comets, in particular main-belt comets², Kuiper belt objects²⁹, or even the dwarf planet Ceres³⁰ retained ammonia, which may be of Earth-like isotopic composition.

Methods

The samples of the CM2 chondrites Yamato 791198 and Yamato 793321 were obtained as clean interior fragments from NIPR (Japan). Polished samples were produced using standard petrographic methods. Care was taken to avoid excessive heating of the samples beyond 50 °C to avoid alterations. The fragments of Y-791198 intended for NanoSIMS analysis were initially impregnated with a solution of research grade poly(methyl methacrylate; PMMA) in dichloromethane. This preparation step was necessary, because epoxy resins contain nitrogen which could have interfered with the measurements of the nitrogen isotopic ratios. Furthermore, the impregnation provided evenly distributed carbon in the sample for the measurement of CN⁻ ions by SIMS, for which we used a Cameca NanoSIMS 50, measuring the ions ¹²C⁻, ¹²C¹⁴N⁻, ¹²C¹⁵N⁻ and ³²S⁻. The instrument was operated with a Cs⁺ primary ion beam of ∼1 pA, which was focused into a spot size of ∼100 nm. Ion intensities were recorded in multi-collection using four electron multipliers on movable trolleys. All measured species were free of isobaric interference. This is particularly important for ¹²C¹⁵N⁻, which was fully separated from ¹³C¹⁴N⁻. As reference, and for the control of instrumental mass fractionation, we used a synthetic mixture of chromium nitride and natural pentlandite (Sudbury, Canada). We used a FEI Quanta3D cross beam workstation for FIB preparation of TEM samples from selected PCS grains at Ga⁺ ion energies, initially of 30 keV and finally of 5 keV. For the microscopic study we used a Philips CM20 field-emission TEM with a post-column electron energy-loss spectrometer and a LEO 922 energy-filtered TEM, both operated at 200 kV and equipped with Thermo Scientific NORAN energy-dispersive X-ray detectors. Gas mixing and equilibration calculations were carried out using the Cantera software suite and included the nitrogen-bearing species N, N₂, NH₃, NH₂, NH, NO, NO₂, N₂O and HCN, among others. Further details are outlined in the Supplementary Methods.