Two Modifications of Nitrilotris(methylenephenylphosphinic) Acid: A Polymeric Network with Intermolecular (O=P–O–H)₃ vs. Monomeric Molecules with Intramolecular (O=P–O–H)₃ Hydrogen Bond Cyclotrimers

Abstract

Nitrilotris(methylenephenylphosphinic) acid (NTPAH₃) was silylated using hexamethyldisilazane to produce the tris(trimethylsilyl) derivative NTPA(SiMe₃)₃. From the latter, upon alcoholysis in chloroform, NTPAH₃ could be recovered. Thus, a new modification of that phosphinic acid formed. Meanwhile, NTPAH₃ synthesized in aqueous hydrochloric acid crystallized in the space group P3c1 with the formation of O-H⋅⋅⋅O H-bonded networks (NTPAH₃^P), in chloroform crystals in the space group R3c formed (NTPAH₃^M), the constituents of which are individual molecules with exclusively intramolecular O-H⋅⋅⋅O hydrogen bonds. Both solids, NTPAH₃^P and NTPAH₃^M, were characterized by single-crystal X-ray diffraction, multi-nuclear (¹H, ¹³C, ³¹P) solid-state NMR spectroscopy, and IR spectroscopy as well as quantum chemical calculations (both of their individual constituents as isolated molecules as well as in the periodic crystal environment). In spite of the different stabilities of their constituting molecular conformers, the different crystal packing interactions rendered the modifications of NTPAH₃^P and NTPAH₃^M similarly stable. In both solids, the protons of the acid are engaged in cyclic (O=P–O–H)₃ H-bond trimers. Thus, the trialkylamine N atom of this compound is not protonated. IR and ¹H NMR spectroscopy of these solids indicated stronger H-bonds in the (O=P–O–H)₃ H-bond trimers of NTPAH₃^M over those in NTPAH₃^P.

1. Introduction

Nitrilotris(methylenephosphinic) acids are essentially related to nitrilotris(acetic) acid I (Figure 1), which has been shown to crystallize as a zwitterionic compound [1,2]. The protonation of the α-amine N atom by one of the acidic groups is a common feature, which is very well known from biogenic α-amino acids [3]. Upon replacing carboxylic with phosphinic (compound II, [4]) or phosphonic acid groups (compound III [5]), the slightly more acidic P-containing acid groups (for example, PhMePOOH [6] and arylphosphinic acids of the type Aryl(H)POOH [7] were shown to be more acidic than benzoic acid) serve as protonators, and this is also true for nitrilotris(methylenephosphonic) acid IV [8] and compound V [9]. In contrast to the double-zwitterionic nature of V, cyclohexane-1,2-diamine derivative VI [10] has mono-zwitterionic features. The lone pair of its second N atom is involved in an N–H⋅⋅⋅N hydrogen bond and thus less susceptible to protonation.

In previous studies [11,12] nitrilotris(methylenephenylphosphinic) acid (NTPAH₃) was shown to be an interesting tripodal ligand. In the course of our ongoing investigations of Si-O-P compounds [13,14,15], phosphinic acids revealed interesting coordination features at silicon [16]. In this context, we prepared both NTPAH₃ and its trimethylsilyl derivative NTPA(SiMe₃)₃ as starting materials for further investigations. Thereby, we encountered highly interesting features of the molecular structure of NTPAH₃ itself in the solid state, which distinguish this acid from related aminoalkyl functionalized phosphinic and phosphonic (and carboxylic) acids. Both the preference of formation of O–H⋅⋅⋅O hydrogen bonds over the protonation of the intramolecular base (i.e., the alkylamine N atom) as well as the capability of solvent-dependent formation of intramolecular vs. intermolecular (O=P–O–H)₃ H-bond trimers render crystalline NTPAH₃ an interesting hydrogen bonding system. While the former stabilizes monomeric NTPAH₃, the latter renders this acid an interesting candidate as a building block for hydrogen-bonded organic frameworks (HOFs) [17,18,19]. As hydrogen bonding itself is of current interest (e.g., with respect to spectroscopic properties [20] and as a bond type that may complement other bond types in H-bond cross-linked materials [21]), we decided to have a closer look at this particular H-bond system of NTPAH₃.

2. Materials and Methods

2.1. General Considerations

Starting materials NH₄Cl (ORG-Laborchemie, Bunde, Germany, 99.7%), phenylphosphinic acid (Sigma-Aldrich, Steinheim, Germany, 99%), paraformaldehyde (thermo scientific, Kandel, Germany, 96%), hydrochloric acid (VWR chemicals, Fontenay-sous-bois, France, 37%), hexamethyldisilazane (abcr, Karlsruhe, Germany, 98.5%), chloroform, stabilized with amylenes (Fisher Scientific, Loughborough, UK, 99.8%), (CD₃)₂SO (deutero, Kastellaun, Germany, 99.8%) and CDCl₃ (deutero, Kastellaun, Germany, 99.8%) were obtained from commercial suppliers. The following solvents were dried prior to use: chloroform and (CD₃)₂SO were stored over 3 Å molecular sieves (Carl Roth, Karlsruhe, Germany); CDCl₃ was distilled from CaH₂ (Sigma-Aldrich, Steinheim, Germany, 95%) and kept under argon atmosphere.

Synthesis and characterization of NTPA(SiMe₃)₃ were carried out under an atmosphere of dry argon utilizing standard Schlenk techniques. Solution NMR spectra (¹H, ¹³C, ²⁹Si, ³¹P) (cf. Figures S1–S7 in the supporting information) were recorded on a Nanobay 400 MHz spectrometer (Bruker Biospin, Ettlingen, Germany). ¹H, ¹³C and ²⁹Si chemical shifts are reported relative to Me₄Si (0 ppm), either as internal reference or in accordance with the shift of the solvent signal ((CD₃)₂SO at δ 39.5 ppm in case of the ¹³C NMR spectrum of NTPAH₃), ³¹P chemical shifts are reported relative to 85% H₃PO₄ (from an external calibration) set at 0 ppm. Solid-state NMR experiments (for spectra cf. Figures S8–S13 in the supporting information) were performed on an Avance HD 400 MHz WB spectrometer (Bruker Biospin, Ettlingen, Germany) with ZrO₂ rotors. The ¹³C and ³¹P spectra were recorded using a 4 mm triple resonance CP MAS DVT probe with a spinning rate of 10 kHz. The ¹H spectra were measured on a 2.5 mm CP MAS VTN probe with a spinning rate of 30 kHz. The ³¹P NMR chemical shift was referenced to 85% H₃PO₄ using NH₄H₂PO₄ as external secondary standard. The ¹³C chemical shift was referenced to TMS using adamantane as external secondary standard. Cross-polarization NMR experiments were carried out with 1 ms contact time for ³¹P and 2 ms for ¹³C. An 80% ramp was used for ³¹P and a 70% ramp for ¹³C. The recycle delay of all ¹³C and ³¹P CP experiments was 30 s and for ¹H 10 s. IR spectra were measured at room temperature using a Nicolet 380 instrument (Thermo Fisher, Waltham, MA, USA). The solid material was ground with dry KBr in a mortar and pressed into a pellet. For each measurement, 32 scans were collected using wavenumbers from 400 cm⁻¹ to 4000 cm⁻¹. The spectra were corrected for background (prior to each measurement the background was collected using a freshly prepared KBr pellet).

For single-crystal X-ray diffraction analyses, crystals were selected under an inert oil and mounted on a glass capillary (which was coated with silicone grease). Diffraction data were collected on a Stoe IPDS-2/2T diffractometer (STOE, Darmstadt, Germany) using Mo Kα-radiation. Data integration was performed with the STOE software XArea version 2.3. The structures were solved using SHELXT and refined with the full-matrix least-squares methods of F² against all reflections with SHELXL-2019/3 [22,23,24]. All non-hydrogen atoms were anisotropically refined, C-bound hydrogen atoms were isotropically refined in idealized position (riding model), and hydrogen atoms of OH groups were located as residual electron density peaks and were refined without restraints. For details of data collection and refinement, see Appendix A,

Table A1. Crystallographic data from data collection and refinement for the nitrilotriphosphinic acid N(CH₂P(Ph,O,OH))₃ (NTPAH₃) in its modifications NTPAH₃^P and NTPAH₃^M, and for its tris(trimethylsilyl) ester N(CH₂P(Ph,O,OSiMe₃))₃ NTPA(SiMe₃)₃.

Parameter	NTPAH3P 1	NTPAH3M	NTPA(SiMe3)3 2
Formula	C21H24NO6P3	C21H24NO6P3	C30H48NO6P3Si3
Mr	479.32	479.32	695.87
T(K)	180(2)	180(2)	160(2)
λ(Å)	0.71073	0.71073	0.71073
Crystal system	trigonal	trigonal	trigonal
Space group	P3c1	R3c	$R\overline{3}$
a = b(Å)	9.3792(3)	18.2554(5)	12.1818(3)
c(Å)	15.6549(6)	11.4177(3)	92.431(3)
V(Å3)	1192.65(9)	3295.3(2)	11,878.8(7)
Z	2	6	12
ρcalc(g·cm−1)	1.34	1.45	1.17
μMoKα (mm−1)	0.3	0.3	0.3
F(000)	500	1500	4440
θmax(°), Rint	28.0, 0.0662	28.0, 0.0350	25.0, 0.0492
Completeness	100%	100%	99.9%
Reflns collected	14,281	18,163	255,15
Reflns unique	1932	1748	4686
Restraints	1	1	0
Parameters	99	98	265
GoF	1.036	1.055	1.106
χFlack	0.05(6)	−0.01(3)	n/a
R1, wR2 [I > 2σ(I)]	0.0317, 0.0711	0.0224, 0.0602	0.0376, 0.1017
R1, wR2 (all data)	0.0425, 0.0742	0.0243, 0.0610	0.0550, 0.1066
Largest peak/hole (e·Å−3)	0.21, −0.21	0.21, −0.15	0.30, −0.24

¹ The batch of crystals contained long trigonal prisms as well as pseudo-hexagonal plates. The data set was collected from a prism. The plates represent the same modification, but they are twins according to the twin law (1 1 0, 0–1 0, 0 0–1). ² The batch of crystals consisted of both thick plates and truncated bipyramids, both of which show trigonal symmetry. The data set was collected from a small cap of a truncated bipyramidal crystal. The central parts of the truncated bipyramids exhibited layering, which indicated stacking faults. Data collection from the larger trigonal plates revealed essentially the same unit cell parameters, but superstructure reflections along l indicated a primitive c axis of the same length. Those data sets allowed for structure solution in both space group types R$\overline{3}$ and P3, but both models refined to unsatisfactory R-values (above 0.20) and resulted in noticeable residual electron density peaks. We therefore interpret these crystals as two-phase twins, which result from large contributions of stacking faults along c and give rise to different local space group symmetries (i.e., space group types R$\overline{3}$ and P3). We are aware that these kinds of twins can be refined, for example it has been done for a 1:1 twin of space group types P2₁2₁2₁ and Pmnb [56], but such an analysis was beyond the scope of our study as crystal picking and collection of a data set in space group R$\overline{3}$ already provided satisfactory information of the molecular structure of compound NTPA(SiMe₃)₃.

. Graphics of molecular structures were generated with ORTEP-3 [25,26] and POV-Ray 3.7 [27]. CCDC 2365392 (NTPAH₃^P), 2365391 (NTPAH₃^M) and 2365393 (NTPA(SiMe₃)₃) contain the supplementary crystal data for this article. These data can be obtained free of charge from the Cambridge Crystallographic Data Centre via https://www.ccdc.cam.ac.uk/structures/ (accessed on 25 June 2024).

The geometry optimizations of isolated molecules were carried out with ORCA 5.0.3 [28] using the restricted PBE0 functional with a relativistically recontracted Karlsruhe basis set ZORA-def2-TZVPP [29,30] for all atoms, scalar relativistic ZORA Hamiltonian [31,32], atom-pairwise dispersion correction and with the Becke–Johnson damping scheme (D3BJ) [33,34]. Calculations were started from the molecular structures obtained by single-crystal X-ray diffraction analysis. Numerical frequency calculations were performed to prove convergence at the local minimum after geometry optimization and to obtain the Gibbs free energy (293.15 K). The crystal structures of NTPAH₃^P, and NTPAH₃^M were optimized with TURBOMOLE rev. V7-7-1 [35] using the restricted PBE functional with a basis set pob-TZVP and RI-J auxiliary basis set for all atoms. A grid size of m3 and atom-pairwise dispersion correction with a Becke–Johnson damping scheme (D3BJ) [33,34] were used. During the periodic boundary optimization, the redundant coordinates were optimized and cell parameters were held constant. At the final geometry, a single-point calculation using PBE0 functional was performed. QTAIM analyses were performed with Multiwfn 3.8 [36] starting from the results obtained with the PBE0 functional. Graphics were generated using Chemcraft version 1.8 (Build 164) [37] or TURBOMOLE rev. V7-7-1 [35]. The R-factor for the match of 3-fold symmetry of the individual optimized molecular structures was evaluated with Chemcraft [37].

2.2. Syntheses and Characterization

Compound NTPAH₃^P (C₂₁H₂₄NO₆P₃). This compound was synthesized by following a published protocol [38] using NH₄Cl (1.411 g, 26.37 mmol), paraformaldehyde (4.752 g, 158.2 mmol), phenylphosphinic acid (11.24 g, 4.232 mmol), 60 mL of water and 60 mL of hydrochloric acid. Yield: 5.98 g (12.5 mmol 47%) (cf. lit. 51.5% [38]) M.p. 238 °C (cf. lit. 233-236 °C [38]). As the original report [38] did not list NMR and IR spectroscopic data, we list the data obtained from our product. NTPAH₃ in solution: ¹H NMR (400 MHz, (CD₃)₂SO): δ (ppm) 13.07 (s, 3H, OH), 7.67-7.57 (m, 9H, Ph-o,p), 7.48 (m, 6H, Ph-m), 3.52 (d, ²J_PH = 4.4 Hz, 6H, CH₂); ¹³C NMR (101 MHz, (CD₃)₂SO): δ (ppm) 132.2 (Ph-p), 131.4 (d, ²J_PC = 10.6 Hz, Ph-o), 130.9 (d, ¹J_PC = 131.5 Hz, Ph-i), 128.4 (d, ³J_PC = 12.8 Hz, Ph-m), 58.9 (dt, ¹J_PC = 108.1 Hz, ³J_PC = 5.2 Hz, CH₂); ³¹P{¹H} NMR (162 MHz, (CD₃)₂SO): δ (ppm) 33.9. Solid NTPAH₃^P: IR: $\overline{\upsilon }$ (cm⁻¹) 419 (m), 476 (s), 537 (s), 694 (s), 752 (s), 763 (w), 796 (m), 886 (s), 955 (s), 987 (m), 1029 (w), 1071 (m), 1099 (m), 1099 (m), 1126 (m), 1157 (m), 1182 (s), 1246 (s), 1313 (w), 1326 (w), 1344 (m), 1394 (w), 1424 (s), 1440 (s), 1488 (w), 1592 (m), 1684 (b), 2126 (b), 2266 (b), 2560 (b), 2802 (m), 2846 (w), 2968 (m), 3056 (w).

Compound NTPA(SiMe₃)₃ (C₃₀H₄₈NO₆P₃Si₃). In a 100 mL Schlenk flask with magnetic stirrer and reflux condenser, 4.793 g (10.0 mmol) of NTPAH₃^P were suspended in 5 mL (ca. 3.85 g, 23.9 mmol) of hexamethyldisilazane and heated with stirring. The reaction mixture was refluxed under inert atmosphere until the evolution of ammonia had ceased and a clear colorless solution was obtained. This solution was allowed to cool to room temperature within few minutes to afford a white precipitate, whereupon the supernatant was removed with a syringe and the white solid was dried in vacuo. Yield: 6.80 g (9.77 mmol, 98%).

The P atoms of this compound are centers of chirality. In CDCl₃ solution, this compound forms a set of diastereomeric pairs of enantiomers, the RRR/SSS pair, which was encountered in the crystal structure, and the RRS/SSR pair. We observed an RRR/SSS: RRS/SSR ratio of 85:15. Because of the chemical identity of the three phosphinic acid moieties, the former pair gives rise to a single set of signals. The latter gives rise to a double set of signals in relative 2:1 intensity ratio (these additional signals are shown in the Supplementary Material in Figures S4–S7 and listed in Table S1). RRR/SSS pair: ¹H NMR (400 MHz, CDCl₃): δ (ppm) = 7.43 (appt. tq, ³J_HH = 7.5 Hz, ⁴J_HH + ⁵J_PH = 1.3 Hz, 3 H, Ph-p), 7.16 (appt. td, ⁴J_PH = 3.6 Hz, 6 H, Ph-m), 7.05 (ddd, ³J_PH = 12.2 Hz, ³J_HH = 8.1 Hz, ⁴J_HH = 1.4 Hz, 6 H, Ph-o), 4.41 (tt, ²J_HH + ²J_PH = 14.7, ⁴J_PH = 2.6 Hz, 1H, CH₂), 2.60 (dd, ²J_HH = 14.8, ²J_PH = 11.4 Hz, 1H, CH₂), 0.17 (s, 9H, Si(CH₃)₃); ¹³C{¹H} NMR (100 MHz, CDCl₃): δ (ppm) 132.0 (d, ²J_PC = 10.5 Hz, Ph-o), 131.6 (d, ¹J_PC = 132.3 Hz, Ph-i), 131.0 (d, ⁴J_PC = 2.8 Hz, Ph-p), 127.8 (d, ³J_PC = 13.1 Hz, Ph-m), 57.2 (dt, ¹J_PC = 122.6 Hz, ³J_PC = 9.8 Hz, CH₂), 1.27 (d, ³J_PC = 1.2 Hz, Si(CH₃)₃); ²⁹Si{¹H} NMR (79 MHz, CDCl₃): δ (ppm) = 22.2 (d, ³J_PSi = 9.3 Hz); ³¹P{¹H} NMR (162 MHz, CDCl₃): δ (ppm) = 29.4.

A coarse crystalline product (used for single-crystal X-ray diffraction analyses) was obtained according to a related protocol but using more hexamethyldisilazane as solvent, i.e., 2.08 g (4.34 mmol) of NTPAH₃^P in 20 mL (ca. 15.4 g, 95.6 mmol) of hexamethyldisilazane. The greater amount of solvent caused a lower yield of isolated crystalline product (2.24 g, 3.22 mmol, 74%).

Compound NTPAH₃^M (C₂₁H₂₄NO₆P₃). In a 10 mL Schlenk flask, NTPA(SiMe₃)₃ (0.569 g, 0.818 mmol) was dissolved in chloroform (5 mL) and anhydrous EtOH (1 mL) was added carefully (chloroform phase layered with ethanol), whereupon a small amount of white precipitate formed upon contact between the two phases within the first minute. Eventually, long crystalline needles of the product formed upon standing overnight. Thereafter, the solvent was decanted, and the white solid was dried in vacuo. Yield: 0.361 g (0.753 mmol, 92%). M.p. 238 °C. IR: $\overline{\upsilon }$ (cm⁻¹) 481 (m), 559 (s), 690(s), 712 (m), 722 (m), 734 (m), 762 (m), 957 (s), 1024 (w), 1035 (w), 1068 (w), 1097 (m), 1126 (s), 1154 (m), 1269 (w), 1310 (m), 1410 (m) 1439 (m), 1538 (b), 1595 (m), 2883 (m), 2937 (w), 2992 (w), 3051 (m).

3. Results and Discussion

3.1. Syntheses of NTPAH₃ and Its Trimethylsilyl Ester NTPA(SiMe₃)₃

The nitrilotriphosphinic acid NTPAH₃ in our study was prepared according to a previously published protocol [38] (Scheme 1), and the product obtained in this procedure (referred to as modification NTPAH₃^P; the superscript index ^P indicates its H-bonded polymeric nature in the solid state) was used for our study. Silylation of NTPAH₃ in excess hexamethyldisilazane (which served both as the silylating reagent and as the solvent) afforded the tris(trimethylsilyl) derivative NTPA(SiMe₃)₃, which crystallized from the reaction solution upon cooling to room temperature. This compound is highly soluble in chloroform and, in this solution, highly sensitive toward protolysis (e.g., alcoholysis). Thus, deliberate alcoholysis of NTPA(SiMe₃)₃ in chloroform afforded long needles of NTPAH₃ in a new modification (referred to as modification NTPAH₃^M, the superscript index ^M indicates its monomeric nature caused by exclusively intramolecular H-bonds).

3.2. Crystallographic Analysis of the Molecular Structures of NTPA(SiMe₃)₃, NTPAH₃^P and NTPAH₃^M

Compound NTPA(SiMe₃)₃ crystallized in the trigonal space group type R$\overline{3}$ with two independent thirds of molecules in the asymmetric unit (Figure 2, Appendix A Table A1). Their N atoms are located on crystallographic 3-fold rotation axes (Figure 2b,d). The Si–O bond lengths (1.6756(14) and 1.6760(14) Å) are similar to those encountered with trimethylsilyl phosphate O=P(OSiMe₃)₃ (1.674(3) Å, [39]) and a derivative of a cyclic phosphinic acid, (Me₃SiO)(O=)P[-C(SiMe₃)₂–CH₂–CH=C(SiMe₃)-] (1.672(2) Å, [40]). The latter was the only hit in a CSD search [41] for crystallographically characterized triorganosilyl derivatives of phosphinic acids. Thus, the crystal structure of NTPA(SiMe₃)₃ bears some novelty in this regard. Further metric data of these molecules (

Table 1. Selected corresponding bond lengths (Å) and angles (deg) in the crystal structures of NTPA(SiMe₃)₃, NTPAH₃^P and NTPAH₃^M.

	NTPA(SiMe3)3 1	NTPAH3P	NTPAH3M
N–C	1.473(2)/1.469(2)	1.469(3)	1.467(2)
P–C(CH2)	1.801(2)/1.802(2)	1.806(3)	1.817(2)
P–C(Ph)	1.788(2)/1.797(2)	1.796(3)	1.786(2)
P–O	1.575(2)/1.571(2)	1.547(2)	1.533(2)
P=O	1.469(2)/1.471(2)	1.494(2)	1.510(2)
C-N-C	110.37(12)/111.11(12)	109.56(17)	116.04(9)
N-C-P	113.21(14)/113.08(14)	115.74(18)	109.29(15)
C-P-C	107.12(10)/107.92(9)	109.81(13)	109.32(9)
O-P-O	115.57(9)/116.07(9)	115.79(15)	114.40(11)
C(CH2)-P-O(H/Si)	100.15(8)/99.73(8)	98.48(13)	107.29(1)
C(CH2)-P=O	114.93(9)/114.58(9)	113.47(12)	109.62(10)
C(Ph)-P-O(H/Si)	104.79(10)/104.74(9)	109.33(13)	106.67(10)
C(Ph)-P=O	112.98(11)/112.57(10)	109.46(14)	109.45(9)
ΣX-P=O 2	343.5/343.2	338.7	333.5

¹ Two individual sets of parameters are given in accordance with the asymmetric unit. ² Sum of angles about P=O (X = O, C).

) will be discussed in the context of the further analysis of the acid NTPAH₃.

Crystals of both modifications NTPAH₃^P and NTPAH₃^M were analyzed by single-crystal X-ray diffraction (Figure 3 and Figure 4, Table 1 and Table A1). Both modifications are trigonal (space group types P3c1 and R3c, respectively), and their asymmetric units consist of one third of a molecule of NTPAH₃ with its nitrogen atom located on a crystallographically imposed 3-fold rotation axis. Even though the three P-atoms within one molecule exhibit the same chirality, both crystal structures accommodate the respective (R,R,R)/(S,S,S)-pair of enantiomers because of the c glide in their space groups types. In both cases, the N atom is not protonated; i.e., the phosphinic acid moieties are retained. Hence, the molecules are not zwitterionic. This was an unexpected observation, as in both cases the trialkylamine N atoms’ lone pairs are essentially vacant (they are not involved in any other hydrogen bonds, the structures do not contain any other H-bond donors such as ammonium ions, water molecules etc., which could interfere with the N-located lone pair). This can be attributed to the involvement of all of the P(O)(OH) groups’ hydrogen atoms in cyclic hydrogen bond patterns, i.e., R${}_3{}^3\left(12\right)$ motifs [42] of the type (O=P–O–H)₃ (vide infra). The difference between the molecular conformations of NTPAH₃ contained in its two modifications is essentially based on an approximate 110° torsion about the CH₂–P bond (N–C–P–C torsion angles are 66.7(2)° in NTPAH₃^P and 176.3(1)° in NTPAH₃^M), as shown in Figure 3. This conformational difference gives rise to entirely different inter- and intramolecular interactions. While in modification NTPAH₃^P, the three phenyl groups form a calix shape and the P(O)(OH) moieties point away from the central N atom (Figure 4a,c), in modification NTPAH₃^M, the phenyl groups represent the periphery, and the three P(O)(OH) moieties are oriented toward the central 3-fold rotation axis (Figure 4b,d). In the latter case, this orientation results in the formation of a cyclic hydrogen bond system (indicated by the thin blue lines in Figure 4d). In modification NTPAH₃^P, a related R${}_3{}^3\left(12\right)$ hydrogen bond motif is formed, also about a crystallographically imposed 3-fold rotation axis, but in an intermolecular manner (cf. Figure 4e). The molecular conformation of NTPAH₃ in NTPAH₃^P resembles the conformation of NTPA(SiMe₃)₃.

For a comparison of the individual molecules, a set of selected corresponding bond lengths and angles of compounds NTPA(SiMe₃)₃, NTPAH₃^P and NTPAH₃^M is listed in Table 1. The N-C-P angles in NTPA(SiMe₃)₃ and NTPAH₃^P are noticeably wider than the tetrahedral angle, and we interpret this widening as originating from the steric congestion about the CH₂ groups. This angle is much closer to the tetrahedral angle in the cage structure of NTPAH₃^M, which lacks that kind of steric repulsion of the N- and P-bound substituents. In the compounds studied here, the coordination spheres about the P atoms are distorted tetrahedrally, which is to be expected from the VSEPR effect of the P=O bond. Thus, the widest angles are associated with X-P=O (X = C, O) moieties. Interestingly, the O-P=O angles are slightly wider than the C-P=O angles, and two reasons can be considered for this phenomenon: The C(Ph)-P=O angle is particularly small (smaller than the C(CH₂)-P=O angle) because of C–H⋅⋅⋅O=P attraction, which involves the P=O moiety and an ortho-H atom of the phenyl group. The C(CH₂)-P=O angle is still smaller than the O-P=O angle. The P–O bond may feature partial multi-bond character (in case of the P–O–Si moiety because of the electropositive silyl substituent, which renders the P–O-bound oxygen electron-rich, and in the case of the P–O–H moiety, because of the hydrogen bridge, which renders the P–O-bound oxygen electron-rich and also lowers the P=O double bond character). That would explain the enhanced O–P=O repulsion vs. relatively lowered C–P=O repulsion.

The inter- vs. intramolecular O–H⋅⋅⋅O hydrogen bonding in NTPAH₃^P and NTPAH₃^M, respectively, results in further differences between the two crystal structures: In NTPAH₃^P the C-N-C angles are 109.6(2)°. Thus, they are similar to the C-N-C angles in NTPA(SiMe₃)₃ (110.4(1) and 111.1(1)°). In NTPA₃^M, the corresponding angles are significantly wider (116.0(1)°). We attribute this angle widening to the strain associated with the cage formation by the intramolecular H-bonds. The metrics of the R${}_3{}^3\left(12\right)$ motifs are different, too. With O-O-distances of 2.487(3) Å in the O–H⋅⋅⋅O motifs of NTPAH₃^P its intermolecular hydrogen bonds appear to be weaker than the intramolecular H-bonds in NTPAH₃^M with O-O-distances of 2.450(2) Å. The P=O and P–O bonds seem to exhibit notable differences, too (1.494(2) and 1.547(2) Å, respectively, in NTPAH₃^P vs. 1.510(2) and 1.533(2) Å, respectively, in NTPAH₃^M). As these features may suffer from unresolved disorder effects (simultaneous cyclic disordering of O–H⋅⋅⋅O vs. O⋅⋅⋅H–O), we employed computational analyses for closer inspection of these hydrogen bond systems (cf. Section 3.3.).

The different molecular conformations of NTPAH₃ in its modifications result in different intra- and intermolecular interactions in the solid state. In particular, in NTPAH₃^M the O–H⋅⋅⋅O hydrogen bonds are an intramolecular feature, whereas in NTPAH₃^P they contribute to the intermolecular packing. To account for related kinds of interactions in a comparable manner, a Hirshfeld surface analysis was performed for the asymmetric units (which are one third molecule in each case), and therefore the O–H⋅⋅⋅O hydrogen bonds were accounted for in the same manner (Selected d_norm isosurface plots of the Hirshfeld surface analyses are contained in the Supplementary Material Figures S14 and S15). The d_norm ranges (−1.092…+2.229 for NTPAH₃^P, −1.094…+1.167 for NTPAH₃^M) vary noticeably, and the pronounced upper value of +2.229 in NTPAH₃^P indicates less efficient packing in this modification (well in accordance with the volume per molecule, which is much larger for NTPAH₃^P than for NTPAH₃^M with 596 vs. 549 ų, respectively). Figure 5 shows the fingerprint plots (d_e vs. d_i) of these analyses. Note: Because of the analysis of the asymmetric unit, the N–C bonds (and adjacent intramolecular N-H interactions) are also shown in the fingerprint plots, even though they are intramolecular features and determine the lowest d_norm values in both analyses. The O-H interactions are dominated by the O–H⋅⋅⋅O hydrogen bonds. Therefore, these contributions are representative of the (intermolecular and intramolecular, respectively) O–H⋅⋅⋅O hydrogen bond motifs in NTPAH₃^P and NTPAH₃^M, respectively. (C–H⋅⋅⋅O contacts are also present in both modifications; they contribute to the longer distance O-H interactions. For comparison of fingerprint plots (d_e vs. d_i) of all O-H interactions see Figure S16). Whereas these parts of the fingerprint plots are similar for the two modifications, great differences are evident for the long-range interactions part of the maps, where C-C, C-H and H-H interactions can be found, and C-C interactions are responsible for the greatest difference. Figure 6 shows that the long-range C-C interactions in NTPAH₃^P are caused by the calix shape created by the three phenyl groups about the 3-fold rotation axis. The faces of the phenyl rings inside the calix are thus protected from closer packing. In contrast, in NTPAH₃^M, the phenyl groups are more accessible for intermolecular packing from all sides. In addition to the closer C-C interactions, a larger fraction of C-C contacts on the packing of NTPAH₃^M results therefrom (5.8% in NTPAH₃^M, 2.8% in NTPAH₃^P) vs. a larger fraction of C-H contacts in NTPAH₃^P (16.8% in NTPAH₃^P, 12.4% in NTPAH₃^M), see diagram in Figure 7. In contrast, the O-H interactions (and thus the O–H⋅⋅⋅O hydrogen bonds in particular) contribute to surprisingly similar extent (26.2% in NTPAH₃^P, 27.8% in NTPAH₃^M).

Note: As the O-bound hydrogen atoms in the crystal structures of NTPAH₃^P and NTPAH₃^M were refined without restraints, the O–H bond lengths obtained from crystal structure refinement varied noticeably (0.62(4) Å in NTPAH₃^P, 0.93(5) Å in NTPAH₃^M). It is well known that there are problems associated with locating H atoms from X-ray diffraction data. In order to rule out the possibility that this deviation of different refined O–H bond lengths has greater influence on our Hirshfeld surface analysis, we performed the same analysis with structure models that were obtained by a refinement with idealized OH groups (riding model, H atom attached with the ShelX-code AFIX 147). In principle, it reflects the same findings (same appearance of fingerprint maps, same d_norm ranges). A representation of the contributions of contacts (corresponding to Figure 7, including the models with idealized OH groups) is included in the Supplementary Material, Figure S17.

3.3. Computational Analyses of NTPAH₃^P and NTPAH₃^M

For energetic comparison of the two modifications of NTPAH₃ and for closer inspection of the H-bond situation and its influence on bond lengths, we have optimized the molecular structures of NTPAH₃^P and NTPAH₃^M both as isolated molecules (NTPAH₃^P_OPT-1 and NTPAH₃^M_OPT-1) and in their periodic crystal environment (NTPAH₃^P_OPT-CRYST and NTPAH₃^M_OPT-CRYST). For details of the computational analyses, please see Section 2.1. and the Supplementary Material (Figures S18–S21, Tables S2–S5).

For both approaches of optimization of the molecular structures, the atomic coordinates obtained from crystallographic analyses were used as starting point, and optimizations were performed without constraints of symmetry operations. With respect to the latter, in all cases the 3-fold molecular symmetry was essentially retained in the local minima molecular conformations (with R-factors for the 3-fold symmetry match of 0.082 (NTPAH₃^M_OPT-1), 0.008 (NTPAH₃^P_OPT-1), 0.028 (NTPAH₃^M_OPT-CRYST) and 0.013 (NTPAH₃^P_OPT-CRYST), with an R-factor of 0.000 representing a perfect C₃ symmetry).

Comparison of the total energies revealed that the two modifications exhibit very similar stability with NTPAH₃^M_OPT-CRYST being only marginally less stable than NTPAH₃^P_OPT-CRYST by 0.34 kcal⋅mol⁻¹ (PBE) or 1.66 kcal·mol⁻¹ (single-point calculation with PBE0). As an isolated molecule, NTPAH₃^M_OPT-1 is noticeably more stable than NTPAH₃^P_OPT-1 by −20.6 kcal·mol⁻¹ (PBE0). This difference is mainly attributed to the presence vs. absence of the O–H···O hydrogen bonds in NTPAH₃^M_OPT-1 and NTPAH₃^P_OPT-1, respectively. Nonetheless, the energy involved in formation of the R${}_3{}^3\left(12\right)$ H-bond trimer can be expected to exceed that value. Dimerization energies of phosphinic acids (formation of R${}_2{}^2\left(8\right)$ motifs by two O–H⋅⋅⋅O hydrogen bonds) were experimentally determined at 21 ± 6 to 60 ± 10 kcal·mol⁻¹ [44,45], which would imply H-bond energies around 10.5 to 30 kcal·mol⁻¹ per H-bond and thus energies around 31.5 to 90 kcal·mol⁻¹ for related H-bond cyclo-trimers. However, the energy of the molecular conformation of NTPAH₃^P_OPT-1 also gains contributions from intramolecular interactions, which are associated with the formation of the (P-Ph)₃ calix and which are absent in NTPAH₃^M_OPT-1. These contributions serve as some compensation to the expected energy difference. Therefore, we analyzed the wave function features of the H-bonds to evaluate their energetic properties. Using quantum theory of atoms in molecules (QTAIM) analyses, we observed bond-critical points at the O–H···O hydrogen bonds. Emamian et al. [46] reported that via the electron density at the bond critical point, the hydrogen bond strength can be estimated. For NTPAH₃^M_OPT-1, an average H-bond energy of 17 kcal·mol⁻¹ per H-bond is observed, and for each of the sets, NTPAH₃^M_OPT-CRYST and NTPAH₃^P_OPT-CRYST, it is ca. 20 kcal·mol⁻¹. These values are in the above-mentioned expected range of 10.5 to 30 kcal·mol⁻¹ per H-bond. The energetic difference between the H-bonds of 0.4 kcal·mol⁻¹ (20.3 kcal·mol⁻¹ for NTPAH₃^M_OPT-CRYST and 19.9 kcal·mol⁻¹ for NTPAH₃^P_OPT-CRYST) vaguely indicates stronger H-bonds in NTPAH₃^M but is too marginal for a reliable conclusion.

Closer inspection of the bonds involved in the H-bond systems in NTPAH₃^P and NTPAH₃^M (

Table 2. Selected interatomic distances (Å) and angles (deg) in NTPA(SiMe₃)₃ (from XRD) as well as in NTPAH₃^P and NTPAH₃^M (data from XRD and optimized molecular structures).

Compound	P=O	P–O	O–H	H⋅⋅⋅O	O⋅⋅⋅O	C-N-C
NTPA(SiMe3)3 (XRD)	1.469(2)	1.574(2)	-	-	-	110.4(1)
	1.471(2)	1.571(2)	-	-	-	111.1(1)
NTPAH3P (XRD)	1.494(2)	1.547(2)	- 1	- 1	2.487(3)	109.6(2)
NTPAH3M (XRD)	1.510(2)	1.533(2)	- 1	- 1	2.450(2)	116.0(1)
NTPAH3POPT-1 2	1.476	1.608	0.96	-	-	112.5
NTPAH3MOPT-1 2	1.496	1.553	1.03	1.47	2.495	116.8
NTPAH3POPT-CRYST 2	1.559	1.607	1.06	1.41	2.469	109.3
NTPAH3MOPT-CRYST 2	1.558	1.610	1.07	1.41	2.480	116.9

¹ The O–H bond lengths from XRD analyses are not taken into consideration for comparison because of the inherent problem of uncertainty of determination of H-atom positions by X-ray diffraction. ² For molecules or structures with multiple independent bonds or atom distances of the same kind, the average value is reported here.

) indicates differences between the strength of the R${}_3{}^3\left(12\right)$ H-bond trimers in these two modifications. While in the silyl derivative NTPA(SiMe₃)₃ the P–O bond is ca. 0.10 Å longer than the P=O bond, this bond length difference is less pronounced in the crystal structures of the acids NTPAH₃^P and NTPAH₃^M (with differences of 0.05 and 0.02 Å, respectively). The enhanced averaging of P–O and P=O bond lengths in NTPAH₃^M could be an indication of a stronger H-bond system with respect to NTPAH₃^P, and the shorter O···O distance (by 0.04 Å) in NTPAH₃^M provides further support. These observations may (at least in part) also originate from unresolved O–H···O vs. O⋅⋅⋅H–O disorder in the crystal structures, and therefore further support is essential to confirm this hypothesis. Comparison of NTPAH₃^P_OPT-1 (single molecule devoid of H-bonds) and NTPAH₃^M_OPT-1 (single molecule with intramolecular H-bond system) already shows that P=O and P–O bonds are elongated and shortened, respectively, upon H-bond formation, and elongation of the O–H bond is observed as well. However, the additional interactions in the crystal packing cause further changes, which is evident from the comparison of NTPAH₃^M_OPT-1 and NTPAH₃^M_OPT-CRYST. Longer P=O and P–O bonds (with respect to the related bonds in the isolated molecule) are observed, and features of stronger H-bond systems are revealed. In NTPAH₃^M_OPT-CRYST, a longer O–H bond, shorter H⋅⋅⋅O distance and an overall shorter O⋅⋅⋅O distance is observed. Comparison of the two optimized molecular structures in the crystal environment (NTPAH₃^P_OPT-CRYST and NTPAH₃^M_OPT-CRYST) reveals that these modifications exhibit very similar H-bond features. Whereas in NTPAH₃^M the slightly longer O–H bond may indicate a stronger H-bond system, the slightly longer O⋅⋅⋅O distance is indicative of the opposite. Hence, the structural differences are less pronounced than expected from the crystallographic analysis. Therefore, further support will be delivered by solid-state ¹H NMR and IR spectroscopy (cf. Section 3.4). Last but not least, the different C-N-C bond angles in NTPAH₃^P and NTPAH₃^M were also confirmed by the computational analyses. Hence, the wide C-N-C angles are a characteristic feature of the cage structure of the molecule in NTPAH₃^M.

3.4. IR and Solid-State NMR Spectroscopic Analyses of NTPAH₃^P and NTPAH₃^M

The IR spectra of NTPAH₃^P and NTPAH₃^M (Figure 8) are characteristic fingerprints, which clearly distinguish between the two modifications. The most prominent difference is associated with the OH groups of these compounds. As reported earlier [47], the vibrational modes of phosphinic acids´ OH groups in cyclic hydrogen bond situations may give rise to an “A-B-C”-pattern of broad bands of high intensity. These bands are located around 2560, 2200 and 1680 cm⁻¹ for NTPAH₃^P and around 2630, 2190 and 1540 cm⁻¹ for NTPAH₃^M. Notably, the “C” band is shifted to lower wavenumbers and gains in relative intensity over “A” and “B” bands for NTPAH₃^M. In the literature [47] it is interpreted as a sign of the increasing strength of the set of hydrogen bonds involved. Furthermore, a splitting of the “B” band is particularly obvious for NTPAH₃^P, and in the literature that splitting has been mentioned in the context of weaker H-bonds.

¹H MAS NMR spectroscopy of these solids also supports the presence of stronger H-bonds in NTPAH₃^M. As shown in Figure 9, the signal of the OH protons (at δ_iso 13.6 ppm for NTPAH₃^P and at δ_iso 16.6 ppm for NTPAH₃^M) is sufficiently well resolved from the shift range of the aryl protons (the broad signal in the range 4–10 ppm) and thus clearly shows that the OH signal of NTPAH₃^M is shifted by Δδ ca. +3 ppm. It is well known that ¹H NMR signals of protons in H-bond situations are shifted to higher frequencies and that this shift is more pronounced for nuclei in stronger H-bonds [48,49]. This has been underlined by correlations of decreasing OH groups’ ¹H NMR shielding with shortening in O⋅⋅⋅O distances in series of carboxylic acids and phosphonic acids [48]. Moreover, this shift difference of the OH signal represents the greatest difference in the ¹H MAS NMR spectra of these two compounds.

The ¹³C MAS NMR (Figure 10) and ³¹P MAS NMR spectra (Figure 11) reveal further spectroscopic differences between NTPAH₃^P and NTPAH₃^M. In the former, the notable shift difference Δδ of ca. 10 ppm between the signals of the CH₂ carbon atoms (at δ_iso 54 ppm for NTPAH₃^P and at δ_iso 64 ppm for NTPAH₃^M) indicates the involvement of these groups in different bonding situations in the two modifications (i.e., as part of a cage motif in NTPAH₃^M vs. absence of this motif in NTPAH₃^P). In contrast, the shift differences of the aromatic ¹³C signals (in the range 125–135 ppm) are less pronounced. The ³¹P MAS NMR signals (at δ_iso 34.2 ppm for NTPAH₃^P and at δ_iso 37.7 ppm for NTPAH₃^M) also allow us to assign the modification to the respective solid. Both fingerprint features of the two modifications, the ¹³C signal of the CH₂ groups and the ³¹P signals, are sufficiently well separated to rule out the presence of significant amounts of the other modification in the respective sample.

4. Conclusions

In general, nitrilotris(methylenephenylphosphinic) acid (NTPAH₃) can be silylated using hexamethyldisilazane to afford the tris(trimethylsilyl) derivative, NTPA(SiMe₃)₃. The latter is sensitive toward protolysis, and alcoholysis in chloroform liberates the acid NTPAH₃. The different solvents used for synthesis of NTPAH₃ (aqueous hydrochloric acid vs. chloroform) give rise to the crystallization of two different modifications of this acid, i.e., NTPAH₃^P and NTPAH₃^M, respectively. As both modifications exhibit similar stability (shown by computational analysis), the formation of these modifications can be attributed to the different crystallization kinetics in these different solvents, which may depend on the predominant conformer in solution. Whereas aqueous environment may foster H-bond interactions of NTPAH₃ with the solvent, thus supporting the conformation encountered in NTPAH₃^P crystallized therefrom, chloroform cannot be involved into intermolecular H-bonds to similar extent, thus fostering the formation of intramolecular H-bonds in NTPAH₃ and supporting the conformation encountered in NTPAH₃^M crystallized therefrom. The absence of zwitterionic species in the crystals of NTPAH₃ can be attributed to the special feature of cyclic (O=P–O–H)₃ H-bond trimers (i.e., R${}_3{}^3\left(12\right)$ motifs), which engage all H-bond donor and acceptor sites of the phosphinic acid moieties, thus leaving no acid for protonation of the trialkylamine N atom. Therefore, we conclude that the 3-fold symmetry of the NTPAH₃ molecule, which supports crystallization in trigonal space groups and thus also supports formation of (O=P–O–H)₃ H-bond trimers (driven by crystal symmetry) represents a key to this rather unexpected appearance of a non-zwitterionic aminomethylphosphinic acid in its solid-state modifications. The crystal structure of NTPA(SiMe₃)₃, which is devoid of related H-bond systems, underlines this trend of crystallization of the NTPA motif in a trigonal crystal symmetry.

Molecular networks with a high density of hydrogen bonds (e.g., in layered silicon hydrogen phosphates [14] and various other H-bonded networks [50,51,52,53,54,55]) may exhibit good proton conductivity. Thus, further modification of the NTPA motif toward a covalently or coordinatively bound network with retention of O–H⋅⋅⋅O H-bond motifs and the additional trialkylamine N atoms may give rise to H⁺-conductive materials. Further investigations in this direction are currently under way.