Ammonothermal Synthesis of Luminescent Imidonitridophosphate Ba₄P₄N₈(NH)₂:Eu²⁺

Abstract

The structural variability of a compound class is an important criterion for the research into phosphor host lattices for phosphor‐converted light‐emitting diodes (pc‐LEDs). Especially, nitridophosphates and the related class of imidonitridophosphates are promising candidates. Recently, the ammonothermal approach has opened a systematic access to this substance class with larger sample quantities. We present the successful ammonothermal synthesis of the imidonitridophosphate Ba₄P₄N₈(NH)₂:Eu²⁺. Its crystal structure is solved by X‐ray diffraction and it crystallizes in space group Cc (no. 9) with lattice parameters a=12.5250(3), b=12.5566(4), c=7.3882(2) Å and β=102.9793(10)°. For the first time, adamantane‐type (imido)nitridophosphate anions [P₄N₈(NH)₂]⁸⁻ are observed next to metal ions other than alkali metals in a compound. The presence of imide groups in the structure and the identification of preferred positions for the hydrogen atoms are performed using a combination of quantum chemical calculations, Fourier‐transform infrared, and solid‐state NMR spectroscopy. Eu²⁺ doped samples exhibit cyan emission (λ_max=498 nm, fwhm=50 nm/1981 cm⁻¹) when excited with ultraviolet light with an impressive internal quantum efficiency (IQE) of 41 %, which represents the first benchmark for imidonitridophosphates and is promising for potential industrial application of this compound class.

The imidonitridophosphate Ba₄P₄N₈(NH)₂ was synthesized from ammonothermal synthesis. The structure and presence of the imide groups could be determined in a multi‐step characterization process using X‐ray diffraction, infrared and solid‐state NMR spectroscopic methods. These investigations allowed to determine a preferred position of the hydrogen atoms among the possible imide positions. Doped samples show an efficient cyan luminescence under ultraviolet light.

Introduction

In the search for suitable compound classes to be used as phosphor host lattices for application in phosphor‐converted light‐emitting diodes (pcLEDs), the importance of multifarious structural possibilities is often emphasized. With respect to this key feature, nitridophosphates have gained increasing interest in recent years.[ ¹ , ² , ³ , ⁴ , ⁵ ] Here, the high structural variability is expected corresponding to the variety observed in oxosilicates due to the isoelectronic relation of P/N to Si/O. The challenge to gain synthetic access to this compound class arises from the combination of low thermal stability of P₃N₅ together with the necessity of high reaction temperatures for the crystallization of nitridophosphates. Therefore, only a limited number of nitridophosphates could be prepared at ambient pressures. As appropriate solution pathways to this challenges, based on Le Chateliers’ principle, various high‐pressure synthesis routes in large volume presses have been developed in the past to synthesize a large number of nitridophosphates. ^[6]

Furthermore, additional medium pressure methods have been reported as suitable tools for the synthesis of nitridophosphates. These overcome the problem of small sample quantities obtained from the large volume press and allow a systematic access to larger amounts of the compounds, making an industrial application attractive.[ ³ , ⁷ ]

In addition to hot isostatic press approaches, where nitrogen gas pressures are applied during the synthesis, the ammonothermal approach is one of these medium pressure methods. Here, supercritical ammonia is generated during the reaction in specially designed autoclaves and used as a solvent, mainly for the synthesis of amides, imides or nitrides.[ ⁸ , ⁹ ] To increase the solubility of the starting materials in this solution‐based process, mineralizers are usually used generating ammonobasic, ammononeutral, or ammonoacidic environments during synthesis. ^[10] After some first reports on ammonothermal syntheses of nitridophosphates, a more universal approach was presented by Mallmann et al.[ ⁷ , ¹¹ , ¹² ] Subsequently, also first luminescent samples were obtained successfully from ammonothermal approaches.[ ⁴ , ¹³ ]

Next to nitridophosphates, N−H functionalized phosphorus nitride imides and imidonitridophosphates have attracted considerable interest in recent years as well. Two stoichiometric compositions of phosphorus nitride imides are known: HPN₂ and HP₄N₇, the former of which was also prepared ammonothermally. They consist of three‐dimensional networks built up from PN₄ tetrahedra.[ ¹⁴ , ¹⁵ , ¹⁶ , ¹⁷ ] For both compounds, high‐pressure polymorphs are reported, namely β‐HPN₂, β‐HP₄N₇ and γ‐HP₄N₇.[ ¹⁸ , ¹⁹ , ²⁰ ] Furthermore, the incorporation of ammonia gas molecules into pores of a tetrahedra network is reported for the nitridic clathrate P₄N₄(NH)₄(NH₃). ^[21]

For the imidonitridophosphates, syntheses in the large volume press allowed to incorporate negatively charged three‐dimensional imidonitridophosphate networks next to alkaline earth cations as observed in SrP₃N₅NH, Ba₄P₆N₁₀NH:Eu²⁺ and EA ₂AlP₈N₁₅(NH):Eu²⁺ (EA=Ca, Sr, Ba) as well as in imide‐doped samples of Sr₃SiP₃O₂N₇:(NH)²⁻ and Sr₅Si₂P₄ON₁₂:(NH)²⁻.[ ²² , ²³ , ²⁴ , ²⁵ ] Additionally, imidonitridophosphate layers could be realized in EAP₆N₈(NH)₄ (EA=Mg, Ca, Sr) using this method.[ ²⁶ , ²⁷ ]

From the ammonothermal approach with alkali metal amides and P₃N₅ as starting materials, on the other hand, several imido(nitrido)phosphate amides with isolated imidonitridophosphate anions could be prepared. In Cs₅[P(NH)₄](NH₂)₂, isolated P(NH)₄ tetrahedra are formed while in Na₁₀[P₄(NH)₆N₄](NH₂)₆(NH₃)_0.5 and Rb₈[P₄N₆(NH)₄](NH₂)₂ adamantane‐type T2 supertetrahedra anions are incorporated next to free amide ions.[ ²⁸ , ²⁹ , ³⁰ ] These structural units have also been observed in lithium‐containing nitridophosphates.[ ³¹ , ³² , ³³ ]

Therefore, with only few reported representatives of the imidonitridophosphate compound class, already a wide range of structural features could be realized.

While imidonitridophosphates were initially only discussed as possible intermediate products during the formation of nitridophosphates, they already showed interesting luminescent properties themselves. ^[27] Despite the expectation that oscillators such as N−H would quench the emission, luminescent samples of Ba₄P₆N₁₀NH:Eu²⁺ and EA ₂AlP₈N₁₅(NH):Eu²⁺ (EA=Ca, Sr, Ba) could be prepared and show bright emission in the blue to green spectral region. Here, a low ratio of the NH groups among the anions (NH : N=1 : 10 and 1 : 15, respectively) was discussed as a possible reason for the observed luminescence.[ ²³ , ²⁴ ]

In the following, we present the successful ammonothermal synthesis of the alkaline earth imidonitridophosphate Ba₄P₄N₈(NH)₂. X‐ray diffraction analysis revealed for the first time a structure type containing [P₄N₈(NH)₂]⁸⁻ anions next to alkaline earth metal ions. The presence of N−H functionality is confirmed by Fourier‐transform infrared (FTIR) and magic angle spinning (MAS) NMR spectroscopy, which allow the determination of the nitrogen atoms that are preferentially bound to hydrogen atoms. Finally, the optical properties and the luminescent behavior of Eu²⁺‐doped samples of Ba₄P₄N₈(NH)₂ are investigated showing promising results for future research on luminescent imidonitridophosphates.

Results and Discussion

Synthesis

The ammonothermal synthesis of the barium imidonitridophosphate Ba₄P₄N₈(NH)₂ was performed at ammonobasic conditions in custom‐built high‐pressure autoclaves made from a nickel‐based superalloy. The colorless product crystallizes in block‐like crystals (Figure 1). The product decomposes at ambient conditions after a few days and is therefore sensitive towards moisture. To prevent decomposition, the product was handled under inert gas conditions.

Figure 1.

SEM image of a Ba₄P₄N₈(NH₂) crystallite.

First synthesis attempts contained BaH₂ and P₃N₅ in the molar ratio 9 : 5 as well as an excess of the mineralizer NaN₃ to increase the solubility of the starting materials in supercritical ammonia. After the correct stoichiometry of the compound was determined from X‐ray diffraction data, samples were obtained from stoichiometric ratios of the starting materials BaH₂ and red phosphorus with an excess of NaN₃ as mineralizer.

From the so‐prepared samples, an unidentified side phase could be removed by washing the sample with dry ethanol. Luminescent samples were obtained using the dopant Eu(NH₂)₂ (1 mol % with respect to the barium content). Doped samples show a yellow body color and cyan luminescence upon irradiation with ultraviolet (UV) light which is further discussed in the luminescence part.

Crystal Structure Description

As the structure elucidation of Ba₄P₄N₈(NH)₂ comprised a multi‐step process involving several analytical methods which can be followed more easily with the knowledge of the final structure model, a description of this model is given here in advance to the structure determination.

The structure of Ba₄P₄N₈(NH)₂ is composed of adamantane‐type [P₄N₈(NH)₂]⁸⁻ anions which can be described as T2 supertetrahedra consisting of four P(N/NH)₄ tetrahedra. These anions occur in two orientations (green and pink in Figure 2a). Along [001], columns of adjacent supertetrahedra ions are formed, all oriented in the same direction. Along [100] and [010], the orientation of the adjacent supertetrahedra alternates (Figure 2b). The P(N/NH)₄ tetrahedra exhibit minimal distortion as evidenced by the P−N distances and bonding angles, which fall within the typical range reported in the literature (Figure 2c, Table S3).[ ²⁹ , ³⁰ , ³¹ , ³² ] As the following structure determination will show, it is most probable that the hydrogen atoms are connected to terminal nitrogen positions of the supertetrahedra with a higher probability of a connection to nitrogen atoms N1 and N3.

Figure 2.

Representation of the crystal structure of Ba₄P₄N₈(NH₂) showing columns of [P₄N₈(NH)₂]⁸⁻ anions (a) with view along [001] (b). The barium atoms are shown in orange, the phosphorus atoms in green and the nitrogen atoms in blue. The two orientations of the adamantane‐type [P₄N₈(NH)₂]⁸⁻ anions (c) are represented in green and pink, respectively. The coordination spheres of the barium atoms are shown below (d).

The space between the [P₄N₈(NH)₂]⁸⁻ anions is occupied by barium atoms. They are coordinated by six to eight nitrogen atoms (Figure 2d), respectively. The coordination polyhedra of the barium atoms can be described as distorted polyhedra, namely as monocapped trigonal frustum (Ba1), monocapped trigonal prism (Ba2), biaugmented isosceles wedge (Ba3) and isosceles wedge (Ba4). ^[34] The Ba−N distances range from 2.767(6) to 3.162(7) Å. This is consistent with other reported barium nitridophosphate compounds.[ ¹³ , ²³ , ³⁵ , ³⁶ , ³⁷ ]

Regarding the compositions of (imido)nitridophosphate compounds comprising T2 supertetrahedra anions, they all show the same condensation degree κ=0.4 which is defined as the ratio of tetrahedron centers to tetrahedron corners.[ ²⁹ , ³⁰ , ³¹ , ³² ] Ba₄P₄N₈(NH₂)₂ exhibits an unprecedented arrangement of the supertetrahedra anions resulting in a new structure type. While two orientations of the anions are observed frequently in this structural family, only Rb₈[P₄N₆(NH)₄](NH₂)₂ exhibits columns of supertetrahedra with the same orientation along [001] in the structure. The orientation of the anion columns alternates along [100] and [010] as well. ^[29] In contrast to Rb₈[P₄N₆(NH)₄](NH₂)₂, where one corner of the supertetrahedra points in the direction of the columns, the supertetrahedra in Ba₄P₄N₈(NH₂)₂ point with edges in the direction of the columns.

However, next to the structural similarity of the supertetrahedra anions, the compositions of the imidonitridophosphate (amides) Na₁₀[P₄(NH)₆N₄](NH₂)₆(NH₃)_0.5, Rb₈[P₄N₆(NH)₄](NH₂)₂ and Ba₄P₄N₈(NH)₂ vary significantly. This phenomenon can be attributed to a multitude of factors, including the relative quantity of metal cations, the ratio of additional amide anions, and the proportion of imide groups present in the T2 supertetrahedra. Among these compounds, Ba₄P₄N₈(NH)₂ shows the lowest hydrogen content. This may be achieved by the incorporation of the large divalent alkaline‐earth metal ion Ba²⁺ next to the T2 supertetrahedra structural motif for the first time which could render the incorporation of additional amide anions unnecessary to form a stable structure.

Crystal Structure Determination

From single‐crystal X‐ray diffraction data, a first structure model was obtained without hydrogen atoms. Ba₄P₄N₈(NH)₂ crystallizes in the monoclinic space group Cc (no. 9) with the lattice parameters a=12.5250(3), b=12.5566(4), c=7.3882(2) Å and β=102.9793(10)°. Information on the structure refinement is summarized in Table 1. The atomic coordinates together with the respective Wyckoff positions and displacement parameters are reported in the Supporting Information in Tables S1–S3. ^[38]

Table 1.

Crystallographic data for Ba₄P₄N₈(NH)₂ obtained from single‐crystal X‐ray diffraction, standard deviations are given in parentheses.

Formula	Ba4P4N8(NH)2
Crystal system	monoclinic
Space group	Cc (no. 9)
a/Å b/Å c/Å β/°	12.5250(3) 12.5566(4) 7.3882(2) 102.9793(10)
Cell volume/Å3	1132.27(5)
Formula units Z/cell	4
Density/g cm−3	4.783
μ/mm−1	14.285
T/K	298(2)
Diffractometer	Bruker D8 Venture
Radiation (λ/Å)	Mo‐K α (0.71073)
F(000)	1424
θ range/°	3.245–36.965
Total no. of reflections	8923
No. of independent reflections	4321
Observed reflections [F 2>2σ(F 2) ]	4077
R int, R σ	0.0328, 0.0514
Refined parameters/restraints	163/2
Flack parameter	0.044(17)
Goodness of fit (χ 2)	1.051
R1 (all data)/R1 [F 2>2σ(F 2)]	0.0336/0.0297
wR2 (all data)/wR2 indices [F 2>2σ(F 2)]	0.0555/0.0540
Δρ max/Δρ min [eÅ−3]	1.62/−1.95

The refined structure model is supported by CHARDI and BVS calculations (Tables S9–S10). Additionally, Rietveld refinement of powder X‐ray diffraction (PXRD) data confirmed that Ba₄P₄N₈(NH)₂ is formed without crystalline side phases as bulk material after the ethanol washing (Figure 3, Tables S7–S8).

Figure 3.

Rietveld refinement based on PXRD data of Ba₄P₄N₈(NH₂) with experimental data (black dots, Ag‐Kα ₁ radiation), calculated diffraction pattern (red line), difference profile (gray line) and reflection positions of Ba₄P₄N₈(NH₂) (blue bars).

From the refined atomic positions, two charge‐neutral sum formulas Ba₄P₄N₈ X ₂ where X is either O²⁻ or (NH)²⁻ were reasonable, since oxygen incorporation cannot be excluded due to contamination from the autoclave wall. ^[13] From the residual electron density obtained from the difference Fourier map, maxima in reasonable distances to the nitrogen atoms for an imide functionality were found. Therefore, further spectroscopic investigations were performed to clarify the structural composition.

The results of the EDX measurements support the Ba : P ratio obtained from the XRD structure model (Table S11). However, the differentiation of the lighter elements oxygen and nitrogen by this method is not reliable for the investigated compound due to its moisture‐sensitivity. It is possible that the measured oxygen content results from surface hydrolysis of the sample, as the sample is in contact with air during preparation for the EDX measurement, as well as from the washing step with dry ethanol.

In the FTIR spectrum (Figure 4), typical N−H stretching vibrations in the region of 3100–3300 cm⁻¹ were observed. The absorption bands at 612, 3208 and 3258 cm⁻¹ (literature: 591, 3208 and 3257 cm⁻¹) can be assigned to NaNH₂ which is formed during the reaction from the excess of mineralizer employed. ^[39] In general, the measured spectrum agrees well with a theoretical spectrum obtained from quantum chemical density functional theory calculations based on a structure model containing hydrogen atoms, as described later. The signal at 3132 cm⁻¹ can be assigned to the stretching vibration ν(NH²⁻) of Ba₄P₄N₈(NH)₂ with the help of the theoretical spectrum (Table S12). This indicates the presence of NH groups in the sample in general. In contrast to Na₁₀[P₄(NH)₆N₄](NH₂)₆(NH₃)_0.5, Ba₄P₄N₈(NH)₂ shows a sharp band at 3132 cm⁻¹ and therefore a structure without disorder of the imide ions is expected. The signals in the region of 500–1200 cm⁻¹ can be assigned to various lattice vibrations of the anionic [P₄N₈(NH)₂]⁸⁻ units.

Figure 4.

Comparison of the experimental FTIR spectrum of a sample containing Ba₄P₄N₈(NH₂)₂ (black line) with the theoretical spectrum from DFT‐PBE calculations (green line) and the positions of the maximum absorption of NaNH₂ (orange bars). The area showing the N−H stretching vibrations (3100–3300 cm⁻¹) is shown as enlargement.

For a further assignment of the imide groups among the nitrogen positions, ¹H, ³¹P and cross polarization ³¹P{¹H} solid‐state MAS NMR experiments were performed.

The ¹H spectrum shows a strong signal with a maximum at 2.1 ppm and a smaller broad shoulder around 2.4 ppm (Figure 5a). The observation of a ¹H NMR signal confirms the presence of hydrogen‐containing groups in the sample. In combination with the FTIR spectrum that shows no O−H vibrations, this allows the presumption that the compound contains imide ions. While the strong signal can be assigned to the main phase, the shoulder may originate from a disorder of the hydrogen atoms which could be statistically bound to different nitrogen atoms of the [P₄N₈(NH)₂]⁸⁻ units next to preferred positions which are responsible for the main signal. Another possible explanation for the shoulder of the signal could be unidentified minor side phases or hydrolysis products from the decomposition of Ba₄P₄N₈(NH)₂. In comparison to other alkaline earth metal imidonitridophosphates, for which ranges of the ¹H chemical shift from 9–5 ppm are reported, an upfield shift for the signal of Ba₄P₄N₈(NH)₂ is observed.[ ²² , ²³ , ²⁴ , ²⁶ , ²⁷ ] This shift might be caused by a higher electron density in close proximity to the hydrogen atoms due to the higher relative content of the alkaline earth metal in Ba₄P₄N₈(NH)₂ compared to the compounds in the literature, such as BaP₆N₁₀(NH). ^[23]

Figure 5.

Solid‐state NMR spectra of the ¹H measurement (a) and normalized spectra of ³¹P and ³¹P{¹H} measurements (b), all at 19 kHz MAS rate. Rotational side bands are marked with an asterisk. The deconvolution of the ³¹P spectrum is shown next to a representation of a [P₄N₈(NH)₂]⁸⁻ unit from the structural model with the preferred positions of the hydrogen atoms (c).

The ³¹P NMR spectrum shows four separate signals at 25.2, 19.8, 9.2 and 6.2 ppm with small full widths at half maximum (fwhm) of 1.8–3 ppm (Figure 5b). These chemical shifts are in the range reported for nitridophosphates containing T2 supertetrahedra as structural feature.[ ³² , ³³ ] In the cross‐polarized ³¹P{¹H} spectrum, it is visible that all four signals occur as well under indirect polarization. Normalized to the signal at 25.2 ppm, however, the other three signals lose intensity under indirect polarization compared to the ones under direct polarization. The observation of the signals in the cross‐polarized spectrum proofs the presence of hydrogen in a phosphor‐containing compound and, therefore, confirms the presence of the imide groups in Ba₄P₄N₈(NH)₂. For an assignment of the signals to the crystallographic positions, a deconvolution of the ³¹P spectrum was conducted (Figure 5c). From this, it can be derived that the signals at 25.2 ppm (signal A), 9.2 ppm (signal B) and 6.2 ppm (signal C) show an integral ratio of 1.8 : 1.0 : 0.8. This fits well to the four independent crystallographic phosphorus positions from the structure model (with an equal site multiplicity of Wyckoff positions 4a). It seems that the chemical environments of two of the phosphorus nuclei are very similar so that no separation into two signals is observed for signal A. Furthermore, the signal at 19.8 ppm is smaller than the others and its relative intensity loss in the ³¹P{¹H} spectrum is significantly higher. Therefore, it can be assigned to an unidentified minor side phase which also contains hydrogen in the spatial proximity of phosphorus and which might be the same phase that contributes to the shoulder in the ¹H spectrum.

A possible assignment of the signals A to C to crystallographic phosphorus sites can be attempted regarding their respective chemical shift and the distances of the phosphorus atoms to their neighboring atoms (Table S3). In their first coordination sphere, all phosphorus atoms are surrounded by four nitrogen atoms. The respective P−N distances deviate only slightly among the different phosphorus positions (d _P−N=1.572(7)–1.702(6) Å). Therefore, no significant variations in the chemical shifts are expected due to this coordination. In the second coordination sphere, the distance to the barium atoms is regarded. Here, differences in the distances to the closest barium atom are obtained from the crystal structure refinement. The atoms P2 and P4 show significantly shorter P−Ba distances of 3.147(2) and 3.129(2) Å compared to the positions P1 (3.421(2) Å) and P3 (3.446(2) Å). Therefore, for the nuclei P2 and P4, an upfield shift could be expected compared to the nuclei P1 and P3 due to the electron density of the barium atoms in closer proximity to their nuclei. Subsequently, an assignment of the signals B and C to atom positions P2 and P4 is reasonable, although a further distinction of the two sites is not feasible. Accordingly, signal A may be assigned to positions P1 and P3 which show less electron density close to their nuclei.

Comparing the ³¹P and ³¹P{¹H} spectra, the proximity of the hydrogen atoms to the phosphorus atoms may be derived according to the relative changes in signal intensity. The magnetization transfer under cross polarization is mediated by dipolar couplings through space, which scale with the inverse cube of the distance. Therefore, the further away a phosphorus atom is from the next hydrogen atoms, the larger is the loss of signal intensity in the cross polarization experiment. ^[40] From the spectra in Figure 5b, it is consistent to say that the probability of hydrogen atoms is highest close to the positions belonging to signal A, most likely P1 and P3. As the signals B and C are still visible in the cross polarization spectrum, a local proximity to hydrogen atoms for these positions may still be assumed, which could cause the observed shoulder in the ¹H spectrum as well.

To determine the imide positions among the ten possible crystallographic nitrogen positions, the results of the NMR spectra were used. From the small fwhm of signal A, a regular environment for these ³¹P nuclei is expected which makes mixed occupations of several positions and the simultaneous presence of a terminal and a bridging imide group close to P1 and P3 improbable as both of these possibilities would cause a broadening of the signal. Additionally, the simultaneous occupation of two bridging nitrogen atoms can be ruled out, as the intensity loss for the ³¹P{¹H} signals then should be in the same dimension for at least three signals. Therefore, it is expected that the hydrogen positions are bound to the terminal nitrogen positions N1, N2, N3 and N4 of the supertetrahedra which is in accordance to Pauling's second rule. ^[41]

From the NMR data, it is reasonable to say that the hydrogen atoms are partly statistically bound to all four terminal nitrogen atoms since all ³¹P signals of Ba₄P₄N₈(NH)₂ are present in the cross polarization spectrum as well. Additionally, a preferred occupation of positions close to P1 and P3 can be derived from the data and therefore, the hydrogen is preferably bound to the imide positions N1 and N3. This assignment is supported by the results from CHARDI and BVS calculations (Tables S9 and S10).

Based on these considerations, a second structure model for the quantum chemical calculations of the theoretical IR spectrum was generated by adding two hydrogen positions to residual electron density peaks from the Fourier map close to N1 and N3 and restricting the N−H distance to 0.90(2) Å. Information on this structure model is summarized in Tables S4–S6.

UV/Vis Reflectance Spectroscopy

To allow an estimation of the optical band gap of Ba₄P₄N₈(NH)₂, a diffuse reflectance spectrum of an undoped sample was collected (Figure S1). To determine the optical band gap from this spectrum, a pseudo‐absorption spectrum was generated using the Kubelka‐Munk function F(R)=(1−R)²/2R, with the measured reflectance R. ^[42] From a Tauc plot (Figure 6), a linear region is evident assuming a direct band gap. ^[43] By applying a tangent to the inflection point, an optical band gap of ≈3.5(2) eV was estimated.

Figure 6.

Tauc Plot of an undoped sample of Ba₄P₄N₈(NH)₂ (black) with a tangent to the inflection point (red line).

Luminescence

Due to the observation of promising luminescence properties in other compound classes with tetrahedra‐based, isolated anions, such as orthosilicates and orthophosphates, doping experiments were performed.[ ⁴⁴ , ⁴⁵ ]

When doped with approximately 1 mol % Eu²⁺ with respect to the barium content, Ba₄P₄N₈(NH)₂:Eu²⁺ shows cyan luminescence upon irradiation with near UV to blue light. The room temperature photoluminescence (PL) and photoluminescence excitation (PLE) spectra and a microscope image of the probed microcrystal under 400 nm light excitation are shown in Figure 7. Ba₄P₄N₈(NH)₂:Eu²⁺ shows a narrow‐band emission (fwhm: 50 nm/1981 cm⁻¹) with a maximum at 498 nm.

Figure 7.

Spectra of the photoluminescence excitation (PLE, dashed black line) and photoluminescence (PL, green line) of Ba₄P₄N₈(NH)₂ upon excitation at 400 nm at room temperature.

Due to the different coordination numbers and bond lengths (Figure 2d, Table S3) for the different barium sites, differences in absorption band positions and the Stokes shifts are expected for the four Ba sites which could explain the broadening of the PL emission band. Additionally, it is possible that not all of the four possible cations are occupied with Eu²⁺ which could explain the narrow‐band emission. Future investigations could include overdoping experiments to test this hypothesis. The observed luminescence of Ba₄P₄N₈(NH)₂:Eu²⁺ is comparable to other narrow‐band nitride‐based phosphors such as Sr₂AlP₈N₁₅(NH):Eu²⁺ (λ_max=496 nm, fwhm≈46 nm), β‐MgSrP₃N₅O₂ (λ_max=502 nm, fwhm≈42 nm) or Ba_3−x Sr _x [Mg₂P₁₀N₂₀]:Eu²⁺ (x=0–3, λ_max=492–515 nm, fwhm≈36–46 nm).[ ²⁴ , ³⁷ , ⁴⁶ , ⁴⁷ , ⁴⁸ ] Due to the maximum position and the fwhm of the emission, Ba₄P₄N₈(NH)₂:Eu²⁺ is a promising candidate phosphor to close the cyan‐gap occurring in state‐of‐the‐art pc‐LEDs. ^[49] Next to the PL emission, the internal quantum efficiency (IQE) and the thermal quenching behavior are important factors to assess the potential of a phosphor for a potential industrial application.

From the IQE measurements in an integrating sphere, an IQE of up to 41 % for as‐synthesized samples of for Ba₄P₄N₈(NH)₂:Eu²⁺ upon 443 nm laser excitation at room temperature was determined. This is a promising value for a possible application and can likely be further increased by synthesis and crystallite growth optimization in the future. To estimate the thermal behavior of Ba₄P₄N₈(NH)₂:Eu²⁺ at typical working temperatures of a LED, luminescence spectra at different temperatures were collected. Figure 8 shows the thermal quenching (TQ) of Ba₄P₄N₈(NH)₂:Eu²⁺ from room temperature to 200 °C derived from these spectra. The compound shows pronounced thermal quenching resulting in 40 % of the photoemission power at 150 °C compared to the emission power at room temperature.

Figure 8.

Thermal behavior of the normalized integrated emission power of Ba₄P₄N₈(NH)₂:Eu²⁺ under 415 nm LED excitation.

Compounds containing imide groups were considered as unsuitable for an industrial application due to possible absorption effects of the oscillator N−H. Additionally, luminescent imidonitridophosphates were until now only prepared using large volume press experiments which produce only small sample quantities that impede an industrial scale up. Therefore, no data on the efficiency and temperature behavior of luminescent imidonitridophosphates is available in the literature. ^[23] Hence, the data for Ba₄P₄N₈(NH)₂:Eu²⁺ present a first benchmark for the compound class of imidonitridophosphates.

The observation of a high IQE at room temperature in the imidonitridophosphate is surprising as Ba₄P₄N₈(NH)₂ shows the highest NH : N ratio of 1 : 4 compared to reported luminescent imidonitridophosphates which showed NH : N ratios of 1 : 10 and lower. For them, the low concentration of imide groups among the anions was discussed as possible reason for the observed luminescence despite the presence of imide groups in the structure.[ ²³ , ²⁴ ]

The thermal quenching observed for Ba₄P₄N₈(NH)₂:Eu²⁺ might be caused by the discussed absorption of the imide group oscillators. As there are other possible explanations, such as concentration quenching, which could cause the observed decrease in PL intensity at higher temperatures, further measurements, for example for different doping concentrations, should be conducted in the future.

Despite the unfavorable thermal behavior of Ba₄P₄N₈(NH)₂:Eu²⁺, the high IQE at room temperature in a structure with isolated anionic units shows that imidonitridophosphates are a promising compound class for host lattices of LED phosphors even for structures containing higher NH : N ratios.

Conclusions

We succeeded in the ammonothermal synthesis of the barium imidonitridophosphate Ba₄P₄N₈(NH)₂. The multi‐step structural characterization revealed the incorporation of the adamantane‐type [P₄N₁₀] structural unit next to metal cations other than alkali metal cations for the first time. Furthermore, the spectroscopic investigation enabled the determination of the presence of imide groups in the [P₄N₈(NH)₂]^8– anions as evidenced by FTIR, as well as the identification of the nitrogen atoms that are most likely to be bound to the hydrogen atoms, as indicated by solid‐state NMR. Doped samples of Ba₄P₄N₈(NH)₂:Eu²⁺ show a narrow‐band emission (fwhm: 50 nm/1981 cm⁻¹) with a maximum at 498 nm and a promising IQE of 41 %. The luminescence shows pronounced thermal quenching above room temperature.

Overall, this luminescent behavior is highly promising, as Ba₄P₄N₈(NH)₂ is built up from isolated anionic supertetrahedra. Additionally, it exhibits the highest NH : N ratio reported for luminescent nitridophosphates, as the latter was previously regarded as a potential impediment for luminescence in imide containing compounds. In light of the promising luminescence behavior observed for Ba₄P₄N₈(NH)₂, imidonitridophosphates should be considered in future research on suitable substance classes to be used as host structures in phosphor materials. Phosphors showing emission in the spectral region observed for Ba₄P₄N₈(NH)₂:Eu²⁺ have the potential to address the so‐called cyan gap in white‐light emitting pcLEDs. ^[49] For this possible application, the ammonothermal synthesis enables a medium pressure access to the imidonitridophosphate compound class which has – in contrast to the large volume press synthesis – the potential for industrial scale up.

Experimental Section

Since some of the starting materials as well as the product show sensitivity towards air and moisture, all handling of these compounds was conducted under inert gas conditions. Therefore, argon‐filled glove boxes (MBraun, O₂<1 ppm, H₂O<1 ppm) or flame‐dried Schlenk‐type glassware was used in combination with a vacuum line (ρ≤0.1 Pa) with supply of Ar and NH₃. To purify the two gases (Air liquide, 99.999 %), they passed through purification cartridges Micro Torr FT400‐902 for Ar and Micro Torr MC400‐702FV for NH₃ (both SAES Pure Gas Inc.) to reach a purity level of <1 ppbV H₂O, O₂ and CO₂ prior to their utilization. The respective amount of NH₃ that was condensed in the autoclave was estimated with the help of a mass flow meter D‐6320‐DR (Bronkhorst).

Synthesis of P₃N₅

Semi‐crystalline P₃N₅ was prepared according to the literature by ammonolysis of P₄S₁₀ (approx. 7 g, Sigma‐Aldrich, 99.99 %) with ammonia (purification described above) at 850 °C in a fused silica boat. ^[50] First, the insertion of the silica boat was performed in an Ar counter flow and the apparatus was floated with NH₃ for 4 h. Subsequent heating to 850 °C with a holding time of 4 h followed by cooling to room temperature yielded P₃N₅ as orange powder. The product was identified by PXRD, FTIR and CHNS analysis.

Synthesis of Eu(NH₂)₂

Eu(NH₂)₂ was synthesized from supercritical ammonia from elemental Eu (smart‐elements, 99.99 %) according to the literature. ^[51]

Ammonothermal Synthesis

For the synthesis of Ba₄P₄N₈(NH₂)₂, custom‐built Inconel^® 718 autoclaves (max. 900 K, 300 MPa, volume: 10 mL) were used. The autoclave was further equipped with a hand valve (SITEC), a bursting disk (Dieckers GmbH & Co. KG, pressure limit: 330 MPa) in a bursting disk holder (SITEC) and a pressure transmitter (HBM P2VA1/5000 bar). The starting materials BaH₂ (Materion, 99.7 %, 209.0 mg, 1.500 mmol) and P₃N₅ (135.7 mg, 0.8333 mmol) or red phosphorus (Sigma‐Aldrich, >97 %, 46.5 mg, 1.50 mmol) were ground together with the employed mineralizer NaN₃ (Sigma‐Aldrich, 99.5 %, 162.5 mg, 2.500 mmol). To synthesize doped samples, ≈1 mol % of the BaH₂ were replaced by Eu(NH₂)₂. The mixture was placed in a niobium or tantalum liner and subsequently transferred to the autoclave. To seal the autoclave, a silver‐coated Inconel^® 718 ring (GFD seals) was used by tightening the autoclave screws. Cooling of the autoclave with a liquid nitrogen/ethanol bath allowed to condense ammonia (ca. 8 mL, purification described above) in the autoclave. When the autoclave reached room temperature, it was placed in a tube furnace and heated to 670 K in 2 h, held at that temperature for 12 h and then heated to 870 K in 2 h. The temperature was held for 96–120 h, reaching a maximum pressure of 150 MPa, before the furnace was switched off and the autoclave was cooled to room temperature. After removing residual ammonia from the autoclave, the reaction product was separated from the liner wall in a glove box. The samples were subsequently washed with dry ethanol to remove the side phase.

Single‐Crystal X‐ray Diffraction

A single crystal (0.04×0.04×0.03 mm³) was isolated under oil using a microscope. The crystal was immediately transferred to a Bruker D8 Venture diffractometer to avoid hydrolysis and the data was collected using Mo‐K_α radiation (λ=0.71073 Å) and a combined Φ‐ω‐scan. To index and integrate the data, the APEX3 program was used as well as for the space group determination and semi‐empirical absorption correction (SADABS).[ ⁵² , ⁵³ ] The structure solution was performed using XPREP and SHELXT and for the refinement full‐matrix least‐squares methods (SHELXL) were employed in the program WINGX.[ ⁵⁴ , ⁵⁵ , ⁵⁶ , ⁵⁷ , ⁵⁸ ]

Powder X‐ray Diffraction

Powder X‐ray diffraction data were collected on a STOE STADI P diffractometer employing Ag‐K _α1 radiation (λ=0.5594217 Å), a Ge(111) monochromator as well as a Mythen 1 K detector in modified Debye‐Scherrer geometry. Prior to the measurement, the sample was prepared at argon atmosphere by grounding in an agate mortar and was transferred into a glass capillary (d=0.3 mm, wall thickness 0.01 mm, Hilgenberg GmbH). The TOPAS software package was used to perform a Rietveld refinement of the collected diffraction data. ^[59]

Fourier‐Transform Infrared (FTIR) Spectroscopy

The FTIR spectroscopy data were collected in an argon‐filled glove box on an Alpha II FTIR spectrometer (Bruker) equipped with a diamond attenuated total reflectance (ATR) unit. Using the program OPUS 8.7, a spectrum in the range of 400–4000 cm⁻¹ with a resolution of 2 cm⁻¹ was collected at ambient temperature. ^[60]

DFT Calculations

To simulate the theoretical IR spectrum, ab initio electronic structure calculations were performed to determine the vibrational frequencies. Periodic DFT calculations were conducted using the Vienna ab intio Simulation Package (VASP).[ ⁶¹ , ⁶² , ⁶³ , ⁶⁴ ] VASP separates core and valence electrons using projector‐augmented waves (PAW).[ ⁶⁵ , ⁶⁶ ] The generalized gradient approximation (GGA) was used to calculate exchange‐ and correlation‐energy, as described by Perdew, Burke and Ernzerhof (PBE). ^[67] A dense k‐point sampling with a 4×4×4 Γ‐centered grid (~0.2/Å) and a plane wave energy cutoff of 520 eV ensured a well‐converged structure. An optimization with full ionic degrees of freedom, i. e. atomic positions, cell shape and cell volume was performed using the conjugate gradient algorithm. ^[68] The extraction of vibrational frequencies from a density‐functional perturbation theory (DFPT) linear response calculation was performed using the plotIR script provided by Dr. Karhánek. The energy convergence criterion was set to 10⁻⁵ eV and the Hellmann‐Feynmann forces and stresses were relaxed until the convergence criterion of 10⁻³ eV/Å was reached. The same convergence criteria were used for the DFPT calculation.

Solid‐State MAS NMR Spectroscopy

The ¹H, ³¹P and ³¹P{¹H} spectra were recorded employing an Avance III 500 spectrometer (Bruker) operating at a ¹H frequency of 500.25 MHz (magnetic field strength: 11.7 T). The sample was ground in a glove box and transferred in a ZrO₂ rotor with an outer diameter of 2.5 mm which was rotated with a frequency of 19–20 kHz. Device‐specific software was used for the evaluation of the spectra. As a secondary reference, the ¹H resonance of 1 % Si(CH₃)₄ in CDCl₃ was used, adapting the Ξ values for ³¹P relative to H₃PO₄ as reported by the IUPAC. ^[69]

Scanning Electron Microscopy

For the generation of the electron microscope images of the crystallites and the energy dispersive X‐ray (EDX) spectroscopy data, a Dualbeam Helios Nanolab G3UC (FEI) equipped with an X–Max80 SDD detector (Oxford instruments) was used. The sample was placed on an aluminium holder using carbon foil. Additionally, to prevent electrostatic charging of the sample, a high‐vacuum sputter coater (CCU‐010, Safematic GmbH) was used to carbon‐coat the sample.

UV/Vis Spectroscopy

Diffuse reflectance spectra were collected using a Jasco V‐650 UV/vis spectrophotometer (JASCO) equipped with a deuterium and a halogen lamp, a CzernyTurner monochromator with 1200 lines/mm, concave grating and a photomultiplier tube detector.

Luminescence Spectroscopy

Particles of Eu²⁺‐doped samples of Ba₄P₄N₈(NH)₂ were analyzed by luminescence spectroscopy employing a HORIBA Fluoromax4 spectrofluorimeter system connected to an Olympus BX51 microscope. Recording of the respective PL and PLE spectra was performed at room temperature with emission and excitation wavelengths of λ _emi =498 nm and λ _exc =420 nm in a range from 400–650 nm with a step size of 2 nm. For the determination of the internal quantum efficiency, a powder sample of Ba₄P₄N₈(NH)₂ was placed in a PTFE sample holder with polymeric lid in a glove box before it was transferred to the spectrometer. The measurements were performed on an in‐house built system which is based on an integrating sphere with attached spectrofluorimeter (Instrument Systems CAS 140D). The internal quantum efficiency measurement was performed using the two measurement method and an excitation wavelength of 443 nm (laser diode). ^[70] For the temperature‐dependent measurements of emission spectra, a very thin powder layer was produced by curing a silicone suspension containing the sample between an alumina substrate and a cover glass at 150 °C in a glove box. The so‐prepared sample was heated on a Linkam THMS600 stage and the emission spectra at 415 nm excitation were measured after thermal equilibration using a calibrated Ocean Insight HR2000Plus ES spectrometer controlled by the measurement software SweepMe! (Axel Fischer and Felix Kaschura, SweepMe! – A multi‐tool measurement software (sweep‐me.net)).

Supporting Information Summary

The authors have cited additional references within the Supporting Information.[ ⁷¹ , ⁷² , ⁷³ , ⁷⁴ ]

Conflict of Interests

The authors declare no conflict of interest.

Supporting information

As a service to our authors and readers, this journal provides supporting information supplied by the authors. Such materials are peer reviewed and may be re‐organized for online delivery, but are not copy‐edited or typeset. Technical support issues arising from supporting information (other than missing files) should be addressed to the authors.

Supporting Information

Engelsberger F. M., Pritzl R. M., Steinadler J., Witthaut K., Bräuniger T., Schmidt P. J., Schnick W., Chem. Eur. J. 2024, 30, e202402743. 10.1002/chem.202402743

Data Availability Statement

The data that support the findings of this study are available in the supplementary material of this article.