Introduction

Nitrogen (N) is an abundant element in the universe and forms various organic molecules by bonding with carbon (C), hydrogen (H), and oxygen (O). Ammonia (NH3), the primordial N, is the major N carrier in interstellar molecular clouds [1,2,3], as well as in the solar system bodies. Because of its high chemical reactivity, NH3 is considered as the starting molecule for various N-containing molecules. Interstellar dust surfaces are important reaction sites for the formation of organic molecules in photochemical reactions. The adsorption of NH3 on interstellar dust surfaces is the first step in their chemical reactions on interstellar dust surfaces.

N has two stable isotopes, 14N and 15N. The isotope ratios (δ15N, ‰: per mille deviation of the 15N/14N ratio from the Earth’s atmosphere) of molecules provides information on the environment and reaction processes that the molecules have experienced. Solar system bodies exhibit a wide range of the δ15N values. For instance, Sun and Jupiter have the δ15N values of − 350‰, whereas solid planets such as Earth, Venus, and Mars have the δ15N values of 0‰. Meanwhile, more primitive solar system materials, such as comets and chondrites, have the elevated δ15N values of >  + 1000‰ for comets [4], and <  + 200‰ for carbonaceous chondrites as bulk values [5]. In addition, there are large isotopic anomalies (< + 5000‰) in chondrites and cometary dust with significant local enrichment of heavy isotopes, which are called hotspots [6,7,8]. The 15N enrichment is considered to have originated in cold interstellar molecular clouds [9, 10] and star-forming regions [11, 12]. Several mechanisms of nitrogen isotopic fractionation in the interstellar environments have been proposed such as ion–molecule exchange reactions between N atoms and N-containing ions [13, 14] and isotope-selective UV photodissociation of N2 and formation of 15N-enriched NH3 ice [9, 12, 15]. However, the relevant mechanisms are not well understood, yet.

The isotopic fractionation of N during the adsorption and desorption of NH3 gas on solid particle surfaces has been previously reported by Sugahara et al. [16]. Laboratory experiments in which NH3 gas (δ15N =  + 27.0‰) was adsorbed on two Japan Clay Science Society (JCSS) standard siliceous clay minerals (montmorillonite and saponite) revealed a negative correlation between the N isotope ratios of the clay minerals and the adsorption rate of NH3, with a large N isotopic fractionation greater than + 50‰ [16]. The study further verified the δ15N changes during vacuum-induced desorption, revealing that the δ15N values of the N remaining on the clay sample decreased after a brief increase and finally settled at a certain value. These facts suggest that there are at least two chemical forms for the adsorption of NH3: one is easily desorbed under vacuum, and the other is strongly adsorbed on clay minerals. The former is in the form of an NH3 molecule, which are physically trapped in the interlayer space of clay minerals and replacements of the water molecules coordinated to interlayer cations, and the coordination to the Lewis acid sites and silanols. The latter is in the form of ammonium ion (NH4+), which were ionized by reacting with strong Brønsted acid sites such as surface-adsorbed water or interlayer water [17, 19].

X-ray absorption near-edge structure (XANES) is a molecular-scale analytical technique that determines the chemical environment from the electronic and structural information of a target element [20]. The chemical properties of specific elements can be studied, and information regarding the geometric arrangement of such elements of interest can be obtained. Because the chemical environment of N is fundamental to the biogenic cycle on Earth, N K-edge XANES has been applied to identify the form in which N is present in organic and inorganic composite materials [21], such as soils and sediments [22, 23]. The substitution of NH4+ into clay minerals and their XANES spectra have been reported by Leinweber et al. [22]; however, NH3 gas-adsorbed clay has not yet been verified.

In this study, XANES analysis was performed at the N-targeting K-edge to clarify the relationship between the N isotopic anomalies and the adsorption mechanism of NH3. Because the adsorption of NH3 was extremely small (0.1–1.3 wt%), determining the chemical form of N nondestructively and nonextractively in a laboratory experiment is impossible. In this regard, the partial fluorescence yield (PFY) method, a type of XANES, can be applied for high-sensitivity analyses on the order of ppm [24,25,26]. Thus, PFY-based XANES analysis was performed on four types of clay mineral samples (montmorillonite, saponite, dickite, and kaolinite) with adsorbed with NH3 molecules to determine the differences between their adsorption spectral characteristics and N chemical forms.

Materials and methods

Adsorption experiments on NH3 gas were conducted using reference clay samples of JCSS-3101b montmorillonite [(Na,Ca)0.33(Al,Mg)2Si4O10(OH)2·nH2O], JCSS-3501 saponite [Ca0.25(Mg,Fe)3((Si,Al)4O10)(OH)2·nH2O], JCSS-1301 dickite [Al2Si2O5(OH)4], and JCSS-1101b kaolinite [Al4Si4O10(OH)8]. The surface area data measured using the N-adsorption Brunauer–Emmett–Teller (BET) method were compiled with the previous results [16] (Table 1).

Table 1 Types of clay minerals used in the study
Full size table

The details of the adsorption experiments and N isotope analysis were described by Sugahara et al. [16]. Briefly, the individual clay minerals (50 mg each) were placed in 10-mL glass vials (Fig. 1). The clay minerals were preheated at 110 °C for more than a week to eliminate adsorbed water (Fig. 1). After the glass vials were sealed, atmospheric gas in the glass vials was evacuated. NH3 gas with a known δ15N value of + 27.0‰ was then introduced into the vials. The gas pressure was set at ~ 1 atm. The glass vials were kept at room temperature for one week. A sensitivity-improved elemental analyzer connected to an isotope-ratio mass spectrometer (nano-EA/IRMS) was used to determine the N content and their δ15N values [16, 27]. The nitrogen isotope values are expressed as conventional δ notation relative to atmospheric N2 (air). An authentic standard IAEA-N-2 (ammonium sulfate, + 20.3‰) and an interlaboratory calibrated standard BG-T (L-tyrosine, + 8.74‰, [28]) were used to calibrate the δ15N values. The δ15N analysis of standards and samples was performed in quantities of 18–250 nmol (0.3–3.4µg) nitrogen, and the analysis error estimated from repeated analysis of IAEA-N-2 was less than ± 0.5‰ (n = 5). After the initial analysis, part of the residual samples of montmorillonite and saponite were evacuated for 2 h and those of dickite and kaolinite for 1 h to examine the change in the adsorbed NH3 by this process. The evacuated samples were also analyzed using nano-EA/IRMS.

Fig. 1
figure 1

Schematic illustration of the NH3-adsorption experiments conducted in this study. Individual clay minerals (~ 50 mg each) were placed in 10-ml glass vials. The glass vials were evacuated to eliminate atmospheric air. NH3 gas was then introduced into the glass vials, which were adjusted to 1 atm. After stabilization for a week, the glass vials were opened, and portions of the NH3-adsorbed clay minerals were sampled for δ15N analysis through EA/IRMS and N-XANES analysis at SPring-8. Montmorillonite and saponite were further evacuated for 2 h and sampled for the same analysis

Full size image

N-targeting K-edge XANES analysis

The XANES measurements were performed at the c-branch of the Soft X-ray Photochemistry Beamline (BL27SU) at SPring-8 (Fig. 2). The light source was radiation from a figure-eight undulator that produced a linearly polarized photon beam [29]. The photon beam was dispersed using a soft X-ray monochromator with a plane grating with different line spacings [30]. XANES spectra were obtained by scanning the undulator gap to maximize the intensity of the incident soft X-ray and scanning the monochromator to keep the resolving power constant. Measurements were taken by scanning the widths of the entrance and exit slits. The photon energy resolution during measurements was set to 80 meV. The beam size at the focal point was circular with a diameter of 500 µm and a photon flux of 1 × 1011 Ph/s [31]. During the XANES measurements, the intensity of the incoming photon beam (I0) was monitored by measuring the drain current at the surface of the post-focusing mirror.

Fig. 2
figure 2

A Photograph of the soft X-ray beamline facility of BL27SU at SPring-8. B Schematic illustration of the analytical setup for the nitrogen K-edge XANES measurement. C Mineral structures of clay minerals and the possible adsorption states of ammonia. SDD silicon drift detector, CCD charge-coupled device, SR synchrotron radiation, PFY partial fluorescence yield; and TEY total electron yield

Full size image

For the XANES measurements, powdered clay mineral standards adsorbed with NH3 were fixed to a sample holder with conductive double-sided carbon tape. The sample holder was fixed to a linearly rotating manipulator and placed in a vacuum chamber at 1 × 10−5 Pa. The spectra were recorded by applying the PFY method using a silicon drift detector (SDD) [32]. The materials used in this study are listed in Table 1. The energy range for the N K-edge XANES measurements was 395–415 eV, with an energy step of 0.1 eV and an acquisition time of 4 s. As finer spectral structures have been reported only for N2 gas [33], an additional energy step of 0.01 eV was set in the 399.5–402.0-eV interval for additional analysis to confirm the presence of N2 gas. Data analysis for background removal and qualitative analysis of the XANES spectra were normalized to variations in the primary X-ray intensity (I0). Linear pre-edges were removed for each spectrum and the data were normalized using the edge jump height. Because of the low N concentration in the samples, PFY data were used owing to the high sensitivity of the measurement.

Results and discussion

Every sample showed the increased δ15N values compared to the initial δ15N value of the introduced NH3 gas (+27‰) (Figure 3 and Table 2). Dickite and kaolinite were plotted in the specific areas with lower adsorption ratios as shown in Figure 3. Dickite and kaolinite showed mean δ15N values of +47.7‰ and +41.8‰, with adsorption ratios of 0.03% and 0.29%, respectively. In contrast, our previous results showed that the nonevacuated samples of montmorillonite and saponite had variations in the δ15N values and their adsorption ratios indicating a negative correlation between them. Their δ15N values ranges from +28.9‰ to +71.4‰, whereas the adsorption ratios ranged from 0.18% to 1.15% [16]. These results demonstrated that the adsorption of NH3 caused the isotopic fractionation of N, and the adsorbed NH3 was enriched in 15N.

Fig. 3
figure 3

Variation in the nitrogen isotope ratios of ammonia gas adsorbed on clay minerals. The adsorption ratio of each clay mineral was normalized to the wt%. Filled circle, unfilled circle, filled square, unfilled square after Sugahara et al. [16]

Full size image
Table 2 Raw profiles of nitrogen adsorption (wt%) and nitrogen isotopic composition (δ15N, ‰ vs. Air) measured using nano-EA/IRMS
Full size table

After the evacuation of the NH3-adsorbed samples, the evacuated (1 h) dickite and kaolinite were plotted in a specific area with the elevated δ15N values of +52.1‰ and +55.7‰, respectively, with slightly decreased adsorption ratios. This result is consistent with the evacuated (2 h) montmorillonite and saponite, which were plotted in similar areas with uniform values. The adsorption ratios were 0.12% for the 2-h-evacuated montmorillonite and 0.09% for the 2-h-evacuated saponite, which were between those of dickite and kaolinite. The mean δ15N values were +62.2‰ and +54.9‰, respectively. These observations implied that the light isotope (14N) was preferentially removed by evacuation and that the remaining NH3 was enriched in 15N and the remaining NH3 existed in only one adsorption form.

Variations in the adsorption ratios and δ15N values, which were observed in nonevacuated montmorillonite and saponite, were due to the existence of exchangeable cations (e.g., Na+ and Ca2+) and interlayer water in the interlayer space of these clay minerals, whereas dickite and kaolinite did not. According to the Fourier Transform Infrared (FTIR) Spectroscopic study of montmorillonite and saponite, which were exposed to gaseous NH3, two different acidities of these clay minerals: Lewis acid and Brønsted acid, are keys to understand the adsorption mechanisms of NH3 [17, 19]. In the interaction with the Lewis acid site (e.g., aluminum ions), NH3 adsorbed as NH3 molecules [17, 19]. The hydrogen bonding with silanols also has a contribution in this adsorption mechanism [17, 19]. On the other hand, the reaction with the Brønsted acid site (e.g., interlayer water molecules) produce ammonium ions (NH4+) [17, 19]. This ionized NH4+ also further connect with NH3 molecules via hydrogen bonding [17, 34]. The acidity of the Lewis acid sites and silanols are week and the adsorbed NH3 are easily evacuated [17]. On the contrary, NH3 adsorbed on montmorillonite and saponite after evacuation, as well as on dickite and kaolinite, was strongly adsorbed on the mineral surfaces. They are protonated at the Brønsted acid site to form NH4+ and chemisorbed on the mineral surface. The elevated δ15N values of the chemisorbed NH4+ represented N isotopic fractionation caused by the protonation and chemisorption processes of NH3, and they were different among clay minerals.

The N K-edge XANES spectra obtained for NH3-adsorbed clay minerals are shown in Fig. 4. The N-XANES spectra were obtained for non-evacuated dickite and kaolinite and 2-h-evacuated montmorillonite and saponite. We also tried to measure the nonevacuated montmorillonite and saponite, but it was not succeeded. This was probably because gaseous NH3 easily escaped and/or decomposed by X-ray radiation. All NH3-adsorbed clay minerals exhibited similar N-XANES spectra. They had a relatively broad peak at 406 eV in common. Kaolinite also had a sharp peak at 401 eV. A weak 401-eV peak was also observed in dickite and 2-h-evacuated montmorillonite. In addition, 2-h-evacuated montmorillonites also had a broad feature around 410 eV.

Fig. 4
figure 4

A Nitrogen K-edge XANES spectra of clay minerals adsorbed with NH3 gas and B ammonium sulfate powder. C 401-eV resonance characteristics of N2 measured with a high-resolution scan at 0.01-eV steps

Full size image

Only a few N K-edge XANES studies have been conducted on solid soil samples [35, 36]. The spectra of bentonite and soil clay shared a prominent peak at 405.5 eV in the 1s → σ* region and a broad resonance feature of interlayer NH4+ near 410 eV. A previous study [22] reported the XANES spectra of montmorillonite (bentonite) substituted with NH4+ in the interlayer. This interlayer NH4+ has been proposed as a broad resonance at approximately 410 eV and is recognized in the spectra of fine and coarse soil samples [22]. Herein, the 410-eV interlayer NH4+-derived peak was more prominent in the 2-h-evacuated montmorillonite (Fig. 1). Smectite is generally a 2:1-type mineral with large amounts of exchangeable cations and water molecules in the interlayer and is a material with high viscosity, swelling, cation exchange capacity (CEC), and adsorption properties. The N K-edge XANES spectra of montmorillonite that reacted with NH3 gas showed that the adsorbed NH3 was chemisorbed as NH4+ between the layers after vacuum removal. In the kaolinite, dickite, and 2-h-evacuated saponite spectra, no distinct interlayer N peaks were observed, as in the 2-h-evacuated montmorillonite (Fig. 4).

Leinweber et al. [22] showed N K-edge XANES spectra for clay fractions in soil and identified three peaks between 398.8 and 401.0 eV. Based on the spectral characterization of reference compounds by Leinweber et al. [21], the peak at 398.8 eV and 399.8 eV were considered to be represented several nitrogeneous organic compounds such as nitrogen heterocycles and nitrile. The 401-eV could be also originated from some nitrogeneous organic compounds such as amino acids and nitrogen heterocycles. Ammonium salts also have 401-eV peak and could be the candidate. The peak at 405.5 eV, which was located on the higher energy side than these peaks, was derived from the 1s → σ* resonance of N-compounds and was considered to be a common feature in a wide variety of N-containing compounds. This 405.5-eV peak corresponds to the 407-eV peak in our NH3-adsorbed clay minerals. Meanwhile, the 401-eV peak is controversial because it is reported to the result of beam damage products for amino acids [37]. The distinct peak at 401 eV in kaolinite and the same small peak in the 2-h-evacuated montmorillonite and dickite are derived from N2. The presence of N2 can be easily distinguished that from of amine by measuring a high-resolution spectrum near 401 eV because a fine structure (N–N stretching vibration: ν = 235 meV) appears in the peak corresponding to the 1s → π* transition [33] (Fig. 4c). Our sample also confirmed the presence of N2 at ~ 401 eV using high-resolution measurements (Fig. 4c). This peak was not observed in the blank sample and thus was not due to residual N2. The appearance of the N2 signal in the soil can be explained by the photoinduced decomposition of NH4+ [38]. Moreover, the irradiation experiment of 150-eV photons on an NH3 ice film at 20 K demonstrated the formation of N2 by a series of photochemical reactions via intermediate radical species such as NH2, NH, N2H4, N2H3, and N2H2 [39]. Compared with their photon irradiation experiments [40], the photon intensity in our XANES measurements was very strong. In addition, the experimental temperature was higher (20 K vs. room temperature) in our experiments. Our experimental condition would work more for these photochemical reactions than their settings. Therefore, it is plausible that some NH3 and/or NH4+ adsorbed on clay minerals were decomposed through photoinduced decomposition by soft X-rays, and that this N2 derived from the decomposition of ammonium species (NH3, NH4+) was the origin of the 401-eV peak, which is indirect evidence for adsorbed NH3 and/or NH4. That N2 peak was very large for kaolinite, and the amount converted to N2 in the photochemical reaction seemed to vary depending on the structure of the clay mineral. In addition, the saponite peak exhibited a shoulder approximately 405 eV to the left of the main peak. This feature was also observed in kaolinite and dickite. NH3 molecules adsorbed on metal surfaces are known to have a broad peak around this region [40, 41]. Thus, this feature may indicate remaining NH3, which could exhibit been trapped in the interlayer space of the clay minerals.

The spectral similarity between bentonite loaded with interlayer NH4+ [22] and our NH3-adsorbed clay minerals implies that the N K-edge XANES spectra in our study show chemisorbed NH4+ in the interlayer space. This is consistent with the interpretation derived from the N isotopic results. Although a study has reported the N K-edge XANES spectra of interlayer NH4+ [22], there have been no reports on adsorbed NH3 in the gas phase. One of the reasons for this is the experimental difficulty of the XANES analysis, which requires high-vacuum conditions that desorb and evacuate the weakly adsorbed NH3. In addition, further photoinduced decomposition of NH3 eliminated much of the NH3 in its gaseous state. Therefore, the N K-edge XANES spectra reported here selectively capture the tightly bound form of N, which would exist as NH4+ in the interlayer space.

Conclusion and perspectives

The adsorption of NH3 on clay minerals caused large N isotopic fractionation. The adsorbed NH3 was enriched in 15N compared to the initial NH3 gas. Although montmorillonite and saponite showed variations in the adsorption ratios and the δ15N values due to their interlayer cations and space, dickite, kaolinite, and evacuated samples did not show such variations, and their values settled to certain values. The N K-edge XANES spectra of the NH3-adsorbed clay minerals revealed that the strongly adsorbed NH3 on clay minerals was in the form of ammonium ion (NH4+), which were chemisorbed on the mineral surface and/or interlayer space. Our study indicated that clay minerals hosted 15N-enriched NH3 and could have contributed to the formation of 15N-enriched organic molecules.

Primordial NH3 and N signatures (–NH) on phyllosilicates have been discovered in other solar system objects (e.g., Ceres [42]) and the inner asteroid belt (e.g., carbonaceous asteroid [162173] Ryugu [43, 44]). Recent findings from the analysis of Ryugu samples suggest the important role of N-bearing organic chemical synthesis and diverse molecular profiles on asteroids (e.g., amino acids [5, 45]; amines [6]; nucleobase [46]). Therefore, the successful identification of NH3 on the carbonaceous asteroid Ryugu [47] and cometary NH3 (67P/Churyumov-Gerasimenko [48]) is of great importance for the N-bearing chemical evolution and 15N/14N isotopic dynamics in the solar system.