Halogenation of the Hexaphosphabenzene Complex [(Cp*Mo)2 (μ,η6 :η6 -P6 )]: Snapshots on the Reaction Progress.

The oxidation of [(Cp*Mo)2 (μ,η6 :η6 -P6 )] (1) with halogens or halogen sources was investigated. The iodination afforded the ionic complexes [(Cp*Mo)2 (μ,η3 :η3 -P3 )(μ,η1 :η1 :η1 :η1 -P3 I3 )][X] (X=I3 - , I- ) (2) and [(Cp*Mo)2 (μ,η4 :η4 -P4 )(μ-PI2 )][I3 ] (3), while the reaction with PBr5 led to the complexes [(Cp*Mo)2 (μ,η3 :η3 -P3 )(μ-Br)2 ][Cp*MoBr4 ] (4) [(Cp*MoBr)2 (μ,η3 :η3 -P3 )(μ,η1 -P2 Br3 )] (5) and [(Cp*Mo)2 (μ-PBr2 )(μ-PHBr)(μ-Br)2 ] (6). The reaction of 1 with the far stronger oxidizing agent PCl5 was followed via time- and temperature-dependent 31 P{1 H} NMR spectroscopy. One of the first intermediates detected at 193 K was [(Cp*Mo)2 (μ,η3 :η3 -P3 )(μ-PCl2 )2 ][PCl6 ] (8) which rearranges upon warming to [(Cp*Mo)2 (μ-PCl2 )2 (μ-Cl)2 ] (9), [(Cp*MoCl)2 (μ,η3 :η3 -P3 )(μ-PCl2 )] (10) and [(Cp*Mo)2 (μ,η4 :η4 -P4 )(μ-PCl2 )][Cp*MoCl4 ] (11), which could be isolated at room temperature. All complexes were characterized by single-crystal X-ray diffraction, NMR spectroscopy and their electronic structures were elucidated by DFT calculations.

Introduction

The halogenation of white phosphorus is the first step on an industrial scale to transform P4 to organophosphorus derivatives. First publications on the halogenation of white phosphorus date back more than 120 years.[ , ] In 1940, Wyllie et al. proved that the reaction of P4 with I2 leads to P2I4 or PI3, depending on the used stoichiometry, while the reaction of P4 with Br2 leads to PBr3. Based on these results, in 1994, Tattershall et al. reported four series of compounds resulting from the reaction of P4 with I2, Br2 or ICl that were identified by NMR spectroscopy. Later on, Stoppioni et al. presented the first example of a halogenation of coordinated white phosphorus in the coordination sphere of a transition metal by iodine, which resulted in the monocation [{CpRu(PPh3)2}2(μ,η1:η1‐P4H2I)]+). In 2019, Peruzzini et al. reported the ruthenium‐mediated halogenation of white phosphorus, resulting in the complex [Cp*Ru(PCy3)(μ,η2:η4‐P4Cl2)RuCp*] bearing the unprecedented P4Cl2 moiety. Recently, the stepwise degradation of white phosphorus coordinated to a Ru(II) complex upon the reaction with iodine was postulated by DFT computations. The key role of the metal was demonstrated since the proposed mechanism differs significantly from the concerted one that is valid for uncoordinated white phosphorus. As an alternative access, P4 moieties already converted to polyphosphorus complexes might be a valuable approach to functionalized polyphosphorus complexes.

Our recent study on the iodination of the cyclo‐E5 complexes [Cp*M(η5‐E5)] (M=Fe, Ru; E=P, As) has shown that this is a powerful method for the high‐yield synthesis of new types of functionalized polypnictogen compounds (Scheme 1b, red arrow). By examining different halogen sources, the halogenation of the tetrahedrane compound [{CpMo(CO)2}2(μ,η2:η2‐P2)] proved to lead to very diverse reaction products (Scheme 1a, red arrow).

Scheme 1

Selected examples of one‐electron oxidations versus halogenation reactions of Pn‐ligand complexes. Selected Pn ligand complexes are a) the tetrahedrane [{CpMo(CO)2}2(μ,η2:η2‐P2)]; b) [Cp*M(η5‐E5)] (M=Fe, Ru; E=P, As); c) [(Cp*Mo)2(μ,η6:η6‐P6)].

These studies clearly show the different reactivities of polyphosphorus complexes with oxidants such as halogens towards an alternative one‐electron oxidation. Thus, the oxidation of [{CpMo(CO)2}2(μ,η2:η2‐P2)] leads selectively to the dicationic complex [{CpMo(CO)2}4(μ,η2:η2:η2:η2‐P4)]2+[11] (Scheme 1a, B), while the halogenation yields, depending on the used stoichiometry and halogen, C–E (Scheme 1a). Similarly, a substantial difference in the one‐electron oxidation of [Cp*Fe(η5‐P5)] and its oxidation with halogens was realized. While the oxidation leads to the dicationic complex [(Cp*Fe)2(μ,η5:5‐P10)]2+ (Scheme 1b, G), the reaction with I2 yields the nortricylane derivative H (Scheme 1b). Since in triple‐decker complexes the middle deck is stabilized by two metal fragments, a higher stability of the products in the reaction with halogens is expected, possibly enabling the identification of intermediates and their isolation along the reaction pathway.

Therefore, the redox‐active compound [(Cp*Mo)2(μ,η6:η6‐P6)] (1) was chosen for a detailed study of its halogenation. The cyclo‐voltammogram of this 28 VE complex reveals a reversible one‐electron oxidation in which a 27 VE product results in a distorted cyclo‐P6 ligand in a bis‐allylic arrangement (Scheme 1c, I). Herein we report on the oxidation of the hexaphosphabenzene complex [(Cp*Mo)2(μ,η6:η6‐P6)] (1) by halogens and halogen sources such as I2, PBr5 and PCl5 as a novel synthetic approach to unprecedented halogen‐functionalized complexes of the type [(Cp*Mo)2PnXm]+/0 (n, m=1, 2, 3; X=I, Br, Cl) revealing PnXm middle decks.

Results and Discussion

The reaction of 1 with an excess of iodine (6 equiv.) in CH2Cl2, followed by the layering of the reaction solution with n‐pentane, leads to [(Cp*Mo)2(μ,η3:η3‐P3)(μ,η1:η1:η1:η1‐P3I3)][I3] (2‐I) as a red crystalline compound in 54 % isolated yield (Scheme 2). When a stoichiometric amount of I2 was used (3 equiv.), a few crystals of the similar complex 2‐I in which I3 − is replaced by I− as counterion were isolated, together with [(Cp*Mo)2(μ,η4:η4‐P4)(μ‐PI2)][I3] (3) as black blocks in a crystalline yield of 2 % (Scheme 2, cf. Supporting Information). The ESI‐MS spectrum of the latter reaction solution shows the molecular ion peak of 2 and 3, together with the one of the paramagnetic complex [(Cp*Mo)2(μ‐I)4][I3]. The latter compound was already described by Poli et al. Possible other products as for instance PI3 or P2I4 could not be detected in the 31P{1H} NMR spectrum of the reaction solution which shows only the signals of 2 (see below). The formation of 3 could not be ascertained by 31P NMR spectroscopy due it is paramagnetic nature. Since no signals of the starting material were detected, its full conversion can be assumed. The variable temperature (VT) 31P{1H} NMR spectrum of this reaction solution recorded from −80 °C to room temperature shows that the formation of 2 starts already at low temperatures, with 2 being the only P‐containing diamagnetic product that could be detected. The spectra at higher temperatures are silent because 2 precipitates completely from the solution already at low temperatures (cf. SI for VT‐31P{1H} NMR) and the remaining products are paramagnetic.

Scheme 2

Reaction of [(Cp*Mo)2(μ,η6:η6‐P6)] with I2, PBr5 and PCl5.

Our study of the halogenation of [{CpMo(CO)2}2(μ,η2:η2‐P2)] showed that the nature of the halogen (I2 vs. Br2) plays a decisive role regarding the type of the resulting compounds. Hence, the reaction of [(Cp*Mo)2(μ,η6:η6‐P6)] (1) with PBr5 as a bromine source was carried out. Due to the higher reactivity of bromine (as compared to iodine), the reaction of 1 with an excess of PBr5 (6 equiv.) in CH2Cl2 was carried out at −40 °C and the reaction solution was slowly warmed up to −20 °C. Precipitation of the concentrated reaction solution with cold n‐pentane led to a green precipitate which was dissolved in THF and layered with toluene affording a few crystals of [(Cp*Mo)2(μ,η3:η3‐P3)(μ‐Br)2][Cp*MoBr4] (4) (2 % yield). The supernatant mother liquor was decanted off and, while warming up to room temperature, crystals of the neutral species [(Cp*MoBr)2(μ,η3:η3‐P3)(μ‐P2Br3)] (5) were formed in 17 % yield. Performing the reaction under the same conditions but evaporating all volatiles at −10 °C, washing the residue with n‐pentane and toluene and finally recrystallizing from a mixture of CH2Cl2/n‐pentane, crystals of [(Cp*Mo)2(μ‐PBr2)(μ‐PHBr)(μ‐Br)2] (6) could be isolated (9 % yield; Scheme 2). Additionally, also a few crystals of the known side product [(Cp*MoBr2)2(μ‐Br)2] were formed. The 31P{1H} NMR spectrum of the reaction solution shows, among the signals corresponding to 4 and 6, a broad singlet at 230 ppm corresponding to PBr3 which overlaps with one of the two signals of 6. Signals corresponding to 5 could not be detected in the 31P NMR spectrum of the freshly prepared reaction solution, but were only detected after storing the sample at room temperature for five days. This indicates that 5 is not an initial product of the reaction of 1 with PBr5. Preparing an NMR sample by mixing a precooled solution of 1 and PBr5 and performing a VT 31P{1H} NMR experiment (starting at −80 °C) shows the formation of a very complex reaction mixture (cf. SI) from which no known complexes could be identified. By warming to room temperature, signals corresponding to 6 could be detected. Since the signal of PBr3 is always detected when using PBr5 as a reagent, the question arises which part of it results from the halogenation of 1. 31P NMR spectroscopic investigations showed that only about 30 % of the PBr3 originates from PBr5 while the remaining 70 % follows from the bromination of 1 (see Supporting Information).

Against the background that the reaction of 1 towards a bromine source was predictably more complex than the one with iodine, the question arises as to what would be the difference in the reactivity towards an even stronger halogenating agent as for instance chlorine. Therefore, a time‐dependent 31P{1H} NMR study of the reaction of 1 with an excess of PCl5 (6 equiv.) was carried out at 193 K (Figure 1a). The signal of 1 could not be detected, which proves that its conversion is complete already at 193 K after fifteen minutes. The first 31P{1H} NMR spectrum (t1= 15 minutes) shows four resonances centered at δ=420.8, 329.3, 145.1 and −361.7 ppm in a 1 : 2 : 2 : 1 integral ratio, corresponding to an AMM'OO'X spin system (Figure 1a,b) which can be assigned to 7, based on its similarity with 2 (see below). In addition, a singlet corresponding to PCl3 (δ=220.0 ppm) and a singlet at 6.5 ppm which could not be assigned were detected. 7 could not be isolated due to its high instability. Attempts to crystallize it led to the isolation of [(Cp*Mo)2(μ,η3:η3‐P3)(μ‐PCl2)2][PCl6] (8) instead (74 % yield, scheme 2). Based on the 31P NMR spectroscopic data, the first product of the reaction of 1 with PCl5 is 7 which then converts to 8, where its signals start to appear in the time‐dependent NMR study after one hour (t2=1 h; figure 1a). Solutions of 8 in CH2Cl2 at room temperature are stable for less than one hour, afterwards the color of the solution starts to change from bright red to dark brown. Its decomposition was monitored by NMR spectroscopy (see Supporting Information). The 31P{1H} NMR spectrum of the crystals of 8 dissolved in CD2Cl2 at room temperature reveals after seven hours its complete decomposition and formation of [(Cp*Mo)2(μ‐PCl2)2(μ‐Cl)2] (9), [(Cp*MoCl)2(μ,η3:η3‐P3)(μ‐PCl2)] (10) and PCl3 (Figure 2). Moreover, the intensity of the signal corresponding to PCl3 (220.2 ppm) increases with the temperature, while the one of [PCl6]− decreases until it disappears completely at room temperature (cf. SI for VT 31P{1H} NMR). By performing the reaction of 1 with PCl5 directly at room temperature, 9 and 10 can be isolated in 3 % and 4 % crystalline yield, respectively (Scheme 2). The 31P{1H} NMR spectrum of the reaction solution at 25 °C shows the characteristic signals of 9 (δ=317.2 ppm), 10 (see below) and PCl3 (δ=220.2 ppm) among other signals that could not be assigned (cf. SI). Roughly 78 % of the PCl3 originates from 1, which reflects the low yields of 9 and 10. Attempts to isolate other products of the chlorination led to the isolation of the 30 VE cationic triple‐decker compound [(Cp*Mo)2(μ,η4:η4‐P4)(μ‐PCl2)][Cp*MoCl4] (11) (Scheme 2).

Figure 1

a) Time‐dependent 31P{1H} NMR spectra of the reaction solution of 1 with PCl5 at 193 K. *=7, ♦=8; b) Selected signals of the 31P{1H} NMR spectrum of 7 (CD2Cl2, 193 K) c) Selected signals of the 31P{1H} NMR spectrum of 8 (CD2Cl2, 233 K).

Figure 2

31P{1H} NMR spectra of 8 at t0=20 minutes and t1=7 h (CD2Cl2, 300 K). Whereas, after 20 minutes, only signals of 8 (*) and PCl3 are visible, after 7 h, its complete decomposition into 9 (•) and 10 (▪) is detected.

11 is well soluble in CD2Cl2 but paramagnetic (cf. Supporting Information) and therefore could not be detected by NMR spectroscopy, but was identified by single crystal X‐ray diffraction analysis. In the 1H NMR spectrum of 11, the signal of the anion [Cp*MoCl4]− in 11 could be detected at −13.9 ppm as a broad singlet (ω1/2 = 170 Hz), which is in line with the reported chemical shift. When 1 was reacted with three equiv. of PCl5, the neutral analogue of 11, i. e. [(Cp*Mo)2(μ,η4:η4‐P4)(μ‐PCl2)] (12), could be isolated in 10 % yield. The 31VE triple‐decker complex 12 is paramagnetic, but no signals could be detected by EPR spectroscopy, probably due to its triplet spin state. Crystals of 11 were alternatively obtained by layering a solution of 8 with n‐pentane in CH2Cl2 at room temperature after few days. Therefore, together with 9 and 10, 11 represents another conversion product of 8 (Scheme 2). Additional proof of this is provided by the 1H NMR spectrum of crystals of 8 dissolved in CD2Cl2 which also shows, after 15 days at room temperature, the broad singlet of the counterion of 11, [Cp*MoCl4]−(cf. Supporting Information).

The 31P NMR spectra of 2 and 7 are very similar, both of them showing an AMM'OO'X spin system (cf. Supporting Information). The central phosphorus atom of the allylic‐like P3 unit (PX) resonates at high field (−327 ppm for 2 and −362 ppm for 7), while the peripheral P atoms (PM/PM’) resonate at lower field (349 ppm for 2 and 382 ppm for 7) and show the largest 1 J PP coupling constant (∼380 Hz). Similar chemical shifts and coupling constants were found for the allylic‐like P3 unit in the complexes 5, 10 and 8 (for details see Supporting Information), although 5 shows an ADHKX spin system (Figure 3, top) due to the chemically inequivalent peripheral P atoms of the allylic‐like P3 unit. The central P atom (PA) of the P3X3 subunit resonates at low field (380 ppm for 2 and 421 ppm for 7), while the peripheral P atoms (PO/PO’) resonate at moderately higher field (5 ppm for 2 and 145 ppm for 7) and show a smaller 1 J PP coupling constant (∼280 Hz; Figure 1b). Within the PBrPBr2 unit of 5, the 1 J PP coupling of 420 Hz is rather large and an additional large coupling (J PAPH=70 Hz) is observed within the P3 unit. Similarly, P−P couplings can be detected between the PCl2 and the P3 units (up to 35 Hz) in 8 (Figure 1c) and 10 (up to 41 Hz, Figure 3, bottom). Additionally, the resonance signal of the PCl6 − anion in 8 is observed at −296 ppm in the 31P NMR spectrum. Compound 8 shows rather broad signals at room temperature, which sharpen by lowering the temperature to −40 °C (Figure 1c).

Figure 3

Selected signals of the 31P{1H} NMR spectra of 5 (top) and 10 (bottom) (CD2Cl2, 300 K).

The 31P{1H} NMR spectrum of 6 shows two doublets with a coupling constant of 38 Hz, corresponding to the two nonequivalent phosphorus atoms. The signal at the highest field can be assigned to the PH ligand which splits in the 31P NMR spectrum into a doublet due to the coupling with the proton (1 J PH=444 Hz). For 9, only one singlet was detected (cf. Supporting Information). The 31P NMR chemical shifts and coupling constants for all complexes were determined by iterative simulation of the experimental spectra (see Supporting Information).

Apart from 7 which is extremely unstable even at low temperatures in solution and could not be isolated, the solid‐state structures of all other products were determined by single crystal X‐ray diffraction analysis (Figure 4), revealing that the halogenation of the 28VE cyclo‐P6‐containing triple‐decker complex [(Cp*Mo)2(μ,η6:η6‐P6)] (1) leads to the formation of dinuclear monocations and neutral species which, in most cases, retain their triple‐decker geometry. However, in some cases, degradation to [(Cp*MoBr2)2(μ‐Br)2], [(Cp*Mo)(I4)]− or [Cp*MoX4], the latter being observed as an anion in 4 and 11, was detected. For all complexes, the Mo−Mo bond lengths are below the sum of their covalent radii (3.08 Å) and longer than the corresponding distance in the neutral cyclo‐P6 complex 1 (2.647(1) Å) and in the related monocation [(Cp*Mo)2(μ,η6:η6‐P6)][FAI] ([FAI]=[FAI{OC6F10(C6F5)}3], 2.661(8) Å). They vary from 2.691(2) Å in 9 to 2.920(6) Å in 10. The only exception is observed within 4 with a Mo−Mo distance of 2.579(6) Å. The Cp* ligands in the compounds 2‐I, 3, 4, 6, 8, 9 and 11 are almost coplanar to each other, being only minimally tilted (tilt angle of 1° to 7°). In compounds 5 and 10, the coplanarity of the ligands is lost, and the Cp* ligands are tilted by 40° and 38°, respectively.

Figure 4

Molecular structures of 2‐I, 3, 4, 5, 6, 8, 9, 10 and 11 with thermal ellipsoids at 50 % probability level. Hydrogen atoms and the counterions of 2‐I, 3, 4, 8 and 11 are omitted for clarity. In case of disorder only the major parts are depicted. Selected bond lengths: 2I: Mo1−Mo2: 2.722(15) Å, P1−P2: 2.216(5) Å, P1−P6: 2.220(5) Å, P3−P4: 2.139(5), P4−P5: 2.138(6) Å; 3: Mo1−Mo2: 2.743(8) Å, P2A−P3A: 2.243(7) Å, P3−P3’: 2.162(4); 4: Mo1−Mo2: 2.579(6) Å, P1−P2: 2.126(2) Å, P2−P3: 2.118(19) Å; 5: Mo1−Mo2: 2.919(7) Å, P1−P2: 2.260(2) Å, P3−P4: 2.174(2) Å, P4−P5: 2.128(2) Å; 6: Mo1−Mo2: 2.709(3) Å; 8: Mo1−Mo2: 2.728(2) Å, P2−P3: 2.105(4) Å, P3−P4: 2.101(4) Å; 9: Mo1−Mo2: 2.691(2) Å; 10: Mo1−Mo1’: 2.920 (6) Å, P2−P3: 2.163 (2) Å, P3−P4 2.123 (3) Å; 11: Mo1−Mo2: 2.759(4) Å, P2−P3: 2.210(9) Å, P4−P5: 2.197(8), P3−P4: 2.155(9) Å.

The molecular structure of 2‐I shows an allylic‐P3 and a P3I3 ligand each of them bridging the two {Cp*Mo} fragments (Figure 4). In the P3I3 unit, the central P atom does not coordinate to molybdenum. Compared to the P−P bond lengths in 1 (average: 2.171(3) Å and in the range of P−P single bonds (2.22 Å) ), the P3‐P4 (2.139(5) Å) and P4‐P5 (2.138(6) Å) distances are shortened. Similar bond lengths were observed for the cationic complex [(Cp*Mo)2(μ,η6:η6‐P6)][FAI] whose cyclo‐P6 undergoes a bis‐allylic distortion upon one electron oxidation (average: 2.136(10) Å). The P1‐P2 (2.216(5) Å) and the P1‐P6 (2.220(5) Å) distances in the P3I3 ligand are longer than the one in 1, and correspond to P−P single bonds, while the distance between P2 and P6 with 2.459(5) Å is too long for a usual single bond, but considerably below the sum of the van der Waals radii (3.80 Å). DFT calculations for compound 2‐I reproduce this distance (2.434 Å) well, but no bonding interaction could be detected between these two P atoms (the Mayer bond order (BO) is lower than 0.1 and no orbital overlap occurs). The P⋅⋅⋅P distances between the two P3 ligands are rather long (P2⋅⋅⋅P3 2.652(5) Å and P5⋅⋅⋅P6 2.665(5) Å). Nevertheless, DFT calculations indicate the presence of a P⋅⋅⋅P interaction as shown by BOs of 0.16 and 0.18 for P2⋅⋅⋅P3 and P5⋅⋅⋅P6, respectively (see also Supporting Information). A similar cyclo‐P3I3 moiety was recently postulated by DFT computations as one of the intermediates involved in the last steps of the iodine‐induced stepwise degradation of the P4 ligand in [Cp*Ru(dppe)(ɳ1‐P4)], but without any experimental evidence. Therefore, 2‐I represents the first example of an isolated polyphosphorus complex bearing a P3I3 ligand. The presence of an allylic P3 ligand is recurrent among the products of the halogenation of 1, as in 4, 5, 8 and 10, which will be discussed together therefore. In the allylic P3 ligand of 4, the P1−P2 (2.126(2) Å) and the P2‐P3 (2.118(19) Å) bond lengths are shortened compared to the P−P bond lengths in 1 (average: 2.171(3) Å). In the case of 5, the P3 ligand is distorted (P3‐P4 2.174(2) Å; P4‐P5 2.128(2) Å) and additionally a P2Br3 ligand bridges the two {Cp*MoBr} fragments in an end‐on coordination mode with a P−P bond length that is in the range of a P−P single bond (P1−P2=2.260(2) Å). Several diphosphines of the type RP2Br3 were reported, for example with R=tBu by Baudler et al., R=CCl3 or R=CN, but no P2Br3 unit as a ligand attached to a metal fragment is known. Thus, the one in 5 is the first reported example. The P3 ligand in 8 is in line with what was observed for the analog P3 units in 2‐I, 4 and 5, with P2‐P3=2.105(4) Å and P3‐P4=2.101(4) Å. The average distance from this unit and the PCl2 groups (2.663(7) Å) suggests the presence of P⋅⋅⋅P interactions between them. The solid‐state structure of 10 resembles the one observed for 5, containing a distorted allylic P3 ligand (P2‐P3: 2.163(2) Å and P3‐P4: 2.123(2) Å) bridging between two {Cp*MoCl} units. The short distance between the P3 ligand and the PCl2 group (P1‐P2=2.577(19) Å) indicates an interaction which is also reflected in a rather large 2 J PP coupling constant between these nuclei. The short P1⋅⋅⋅P2 distance (2.551 Å) is well reproduced by DFT calculations and is accompanied by a BO of 0.22 indicating the presence of a bonding interaction. The Pn core of 10 is similar to the one observed in compound [{Cp*V}2(μ,η3:η3‐P3){μ‐P(NHCMe)}] resulting from the ring contraction of the vanadium analog of 1 ([(Cp*V)2(μ,η6:η6‐P6)]) induced by MeNHC.

The solid‐state structure of 3 reveals a cisoid‐P4 and a PI2 bridging ligand separated from each other by 2.596(9) Å. The P2‐P3 (= P2’‐P3’) (2.243(7) Å) and P3‐P3’ (2.162(4) Å) bond lengths in the cisoid‐P4 ligand all lie in the range of P−P single bonds. Therefore, it cannot be described as a tetraphosphabuta‐1,3‐diene‐like ligand, contrary to the analog cisoid‐P4 ligands in complexes ([(CpBIGFe)2(μ,η4:η4‐P4)], [(Cp’’Fe)2(μ,η4:η4‐P4)], [{CpRFe}2(μ,η4:η4‐P4)] (CpR=C5H3(SiMe3)2). The cisoid‐P4 ligand as middle deck in the 30VE species 11 is similar to the one in 3 with two longer P−P bonds (P2−P3=2.210(9) Å, P4−P5=2.197(8) Å) and a shorter one (P3−P4=2.155(9) Å). The P1−P2 and the P1−P5 distances in 11 are 2.664(8 Å) and 2.6591(7), respectively, indicating a P−P interaction (see below).

The solid‐state structure of 6 contains only two P atoms, in the form of a PBr2 and a PHBr bridging ligand, with a nonbonding distance between the two phosphorus atoms of 2.762(9) Å. A similar structure was observed for compound 9, which bears two equivalent bridging PCl2 ligands, separated from each other by 2.894(6) Å.

In order to investigate the bonding situation in complexes 2–12, DFT calculations at the D4‐TPSSh(CPCM)/def2‐TZVP level were conducted. The geometric parameters of the complexes are well reproduced, including the distances between the different Pn units in the molecules. In all complexes 2–12, a Mo−Mo bond was detected, being in line with the relatively short Mo−Mo distances. The Mayer bond order (BO) varies from 0.87 in 4 to 0.59 in 5 (Mo−Mo distances 2.574 and 2.892 Å in the optimized geometries, respectively). The intrinsic bonding orbitals representing the Mo−Mo bond in 2 and the bonding within the Mo2P3 unit in 4 are depicted in Figure 5.

Figure 5

Intrinsic bonding orbitals representing the Mo−Mo bond in 2 (top left) and 4 (top right) as well as the Interaction Region Indicator (IRI) plot in the plane defined by the phosphorus atoms (bottom left) and Electron Localization Function (ELF) plot (bottom right) in 8. Blue dots represent (3,−1) critical points.

The NBO analysis is in agreement with the IBO analysis, although only in 4 and 6 an Mo−Mo bond (Mo−Mo 2.579 Å (4) and 2.709 (6)) is predicted by the NBO analysis, while in the other complexes a nonbonding NBO on each Mo with an occupancy of approx. 1e is partitioned instead of a Mo−Mo bonding orbital. Although the P2−P6 distance in 2 is rather short, no bonding interaction could be detected by DFT calculations. In contrast, bonding interactions were detected between P2−P3 (2.649 Å) and P5‐P6 (2.628 Å) as shown by BOs of 0.16 and 0.18, respectively. Similar interactions were detected between P1−P2 A in 3 (P1‐P2 A 2.623 Å, BO 0.17; singlet spin state), P2−P3 in 5 (2.594 Å, BO 0.17), P1−P2 in 8 (2.517 Å, BO 0.24), P1−P2 in 10 (2.551 Å, BO 0.22), P2−P3 in 11 (2,626 Å, BO 0.18; singlet spin state) and P2−P3 in 12 (2.665, BO 0.17). The Interaction Region Indicator (IRI) clearly shows a bonding interaction between the P3 and the PCl2 units in 8, among the expected bonds. A plot of IRI in the plane defined by the phosphorus atoms is depicted in Figure 5 showing the regions with notable chemical bond interaction (orange) and areas where weak interactions occur (green). The Electron Localization Function (ELF) and Localized Orbital Locator (LOL) also support these interactions (see Supporting Information). Complexes 3 and 11 are paramagnetic in solutions at room temperature, however, DFT calculations show that the singlet spin state of the cation in 3 and 11 is with 90 and 102 kJ⋅mol−1, respectively, more stable than the triplet spin state. The overall paramagnetic behavior of 11 might be due to the paramagnetic nature of the counter anions [Cp*MoCl4]−, having a triplet spin state as determined experimentally.

Conclusions

In summary, we showed that the halogenation of [(Cp*Mo)2(μ,η6:η6‐P6)] (1) proceeds via a very complex pathway leading to a plethora of complexes containing different Pn units such as P4, P3, P2 and P1. An excess of halogens/halogen sources leads to phosphorus‐free complexes of the type [Cp*MoXn] (X=I, Br, Cl) as well as PX3 (X=Br, Cl) as final reaction products. Additionally, we showed that besides iodination, which is known to be a powerful tool for the synthesis of new Pn ligand complexes, bromination and chlorination can also be used for this purpose. Among the products of the iodination of the hexaphosphabenzene complex 1, the novel compound 2 bearing an unprecedented P3I3 ligand could be isolated. With a bromine source, complex 5 could be isolated, representing the first example of a compound bearing a P2Br3 unit as a bridging ligand between two Mo centers.

Whereas the products of the one‐electron oxidation of 1 showed only a distortion of the hexagonal geometry of the P6 middle deck, the use of halogens or halogen sources as oxidizing agents afforded a variety of new polyphosphorus compounds bearing synthetically useful novel PnXm units. As expected, the chemoselectivity of the reaction decreased with the enhancement of the oxidizing power of the halogen. However, with a strict control of the temperature it is possible to exert a good control of the reaction even with a stronger oxidant such as the chlorine source PCl5. First snapshots from the halogenation of the P6 ligand complex were found, and even though the complete pathway could not be clarified, they gave useful information concerning the progress of the reaction. Thus, first an allylic fragmentation of the cyclo‐P6 unit occurs followed by the monohalogenation of the P atoms of one of the allylic P3 units. Subsequently, a dihalogenation forming PX2 units occurs, which is followed by the removal of the P atoms as PX3 moieties. Only afterwards, the second allylic‐P3 unit appears to be halogenated.

This study adds triple‐decker complexes to the class of poly‐phosphorus compounds that can be successfully halogenated and thus functionalized. Therefore, future investigations will focus on the halogenation of heterobimetallic triple‐decker compounds as well as of En ligand derivatives that combine the features of triple‐decker complexes and separate En units. Moreover, since the halogenation reactions in general have shown to be a powerful tool for the generation of halogen‐functionalized polyphosphorus ligands, future studies will also focus on the use of these products for further derivatizations.

Deposition Numbers 2155218 ( contain the supplementary crystallographic data for this paper. These data are provided free of charge by the joint Cambridge Crystallographic Data Centre and Fachinformationszentrum Karlsruhe Access Structures service.

Conflict of interest

The authors declare no conflict of interest.

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

Click here for additional data file.