Andrea
Ienco
*,
Gabriele
Manca
*,
Maurizio
Peruzzini
and
Carlo
Mealli
Istituto di Chimica dei Composti Organometallici – Consiglio Nazionale delle Ricerche (CNR-ICCOM), Via Madonna del Piano 10, 50019, Sesto Fiorentino (FI), Italy. E-mail: andrea.ienco@iccom.cnr.it; gabriele.manca@iccom.cnr.it
First published on 12th November 2018
This paper is a comparative outline of the potential acid–base adducts formed by an unsaturated main group or transition metal species and P atoms of phosphorene (Pn), which derives from black phosphorus exfoliation. Various possibilities of attaining a realistic covalent functionalization of the 2D material have been examined via DFT solid state calculations. The distribution of neighbor P atoms at one side of the sheet and the reciprocal directionalities of their lone pairs must be clearly understood to foreshadow the best possible acceptor reactants. Amongst the latter, the main group BH3 or I2 species have been examined for their intrinsic acidity, which favors the periodic mono-hapto anchoring at Pn atoms. The corresponding adducts are systematically compared with other molecular P donors from a phosphine to white phosphorus, P4. Significant variations emerge from the comparison of the band gaps in the adducts and the naked phosphorene with a possible electronic interpretation being offered. Then, the Pn covalent functionalization has been analyzed in relation to unsaturated metal fragments, which, by carrying one, two or three vacant σ hybrids, may interact with a different number of adjacent P atoms. For the modelling, the concept of isolobal analogy is important for predicting the possible sets of external coligands at the metal, which may allow the anchoring at phosphorene with a variety of hapticities. Structural, electronic, spectroscopic and energy parameters underline the most relevant pros and cons of some new products at the 2D framework, which have never been experimentally characterized but appear to be reasonably stable.
In this paper, we will present systematic analyses, first addressing the chemical and structural aspects of the phosphorene Pn surface and, in particular, the distribution of the P lone pairs, which will interact with σ vacant orbitals at the acidic moieties to be added. The relevance of the donor power at P atoms has already been amply discussed by us relative to the I3P formation from the P4 + I2 reactants.12 It emerged that with respect to generic PX3 phosphines, the white phosphorus allotrope is definitely a much weaker donor; hence, at this point it becomes important to establish where the phosphorene's lone pairs are positioned compared to the previous limiting cases.
Fig. 1(a and b) starts illustrating the distribution of the Pn lone pairs on one side of the 2D material from two viewpoints.
The top view a clearly shows how all the lone pairs point out from a Pn plane although biased by about 60° away from orthogonality, while the lateral view b indicates how the two emerging from any P–P bond are reciprocally skewed also by about 60°. Importantly, the rugged Pn surface consists of parallel zigzag chains with the lone pairs projecting inside the ditches. In principle, any of them may overlap with a mono-functional σ acceptor and, in the case of a metal, contribute to a 2e− donation. The interactions with a multi-functional σ acceptor metal are instead more complex, in view of the simultaneous involvement of two or three neighbor lone pairs, which do not naturally converge into the same point. Still, multiple P–M bonds with one metal seem possible, although they cannot be necessarily equivalent for overlap reasons. Indeed, some geometric adaptation is necessary to allow the multiple-hapticity of phosphorene to a metal.
The mentioned stereochemical aspects will be important in our study of phosphorene's covalent functionalization, which is still substantially unexplored. In fact, many reports have concerned the van der Waals (vdW) functionalization of the Pn surface, examples being the interaction with either 7,7,8,8-tetra-cyano-p-quindodimethane (TCNQ) or a perylene diimide,13 which are characterized by charge transfer from the exfoliated black phosphorus to the organic moiety. Other experimental studies were related to the non-covalent adsorption of small molecules (e.g., CO) on Pn, with insufficient stereochemical details about the interactions.14 One of the few studies of Pn covalent functionalization with metals has concerned the adduct of the TiL4 tetrahedral fragment (L = sulphonic ester), whose stereochemistry has not been illustrated in detail. In fact, characterization of the product was performed by liquid NMR-techniques and the presence of a direct P–Ti bond assessed by Raman spectroscopy.15 In other cases, the absorption of some naked metal atoms on the surface has been computationally addressed with a focus on the number of coordinated P atoms and the effects on the band structures.16 Solid state calculations are instead reported to corroborate the absorption of a CrO3 unit on phosphorene with a unique P–Cr bond completing the tetrahedral coordination of the metal. The linkage of 2.45 Å is relatively large although it is still consistent with covalency, which is further corroborated by the −2.17 eV stabilization energy of the product. No comparison is proposed with any known molecular models of the metal fragment in question.17
A detailed experimental and in silico study of the Pn covalent functionalization concerns the micro-mechanically exfoliated black phosphorus over a Si/SiO2 substrate. By reaction with an aryl diazonium salt, which releases N2, formation of P–aryl covalent bonds is described. In our view, there are still open questions concerning the actual electronic nature of the system. For instance, the P–C bonds imply the formation of local phosphonium cations with a predictably associated anion, whose presence has not been addressed. Remarkably, an uncharged system without any counterion must imply unpaired spins, possibly dispersed throughout the Si/SiO2 substrate. In any case, none of the previous points have been focused on at either computational or experimental level.18
Based on the summarized state-of-the-art, the covalent functionalization of phosphorene is still bleary, especially from the experimental viewpoint. For this, we propose in this paper a series of potential reactants with suitable electronic features to be absorbed at the Pn surface through interactions with selected P lone pairs distributed as in Fig. 1. The detailed stereochemical, electronic and energy features of the products appear to be sound for both main group and transition metal reactants, all characterized by residual acceptor capabilities. Initially, our computational approach was based on limited portions of the rugged phosphorene's surface featuring terminal H atoms at the boundaries. However, it became rapidly evident that a model such as P38H16 could easily undergo geometric distortions that are impossible for an actually periodic Pn surface; hence actual solid state calculations were performed using the CRYSTAL17 software package.19 The increased reliability of the results was verified from structural, spectroscopic and energy viewpoints but also confirmed by the basic electronic properties grasped from the more qualitative molecular modelling.20 In summary, the two approaches ensured a continuity of interpretation, which did not only shed light on the covalent bonding enabled at the Pn surface but also afforded some predictability, which experimental chemists may be willing to try later.
The Pn slab was fixed by assuming a unit cell of four P atoms or sixteen valence orbitals. Accordingly, the overall band structure at the left side of Fig. 2 combines sixteen different bands, ten of which are populated by the twenty electrons per cell. In the lowest region, six bands stem from the P–P σ bonds, while the next higher four correspond to the lone pairs of the pyramidal P atoms. Although their orientations exclude convergence, the short contacts across the surface determine reciprocal repulsions, which cause a widening of the frontier band >1 eV, as estimated by the B3LYP functional.21 At the same time, a 2.26 eV gap is estimated between the valence and conduction regions, with the latter consisting of six bands having the prevailing P–P σ* character. As mentioned, it is sufficient to pile up only ten Pn sheets to approach the band gap proper of the bulk black phosphorus. The steep variation on stacking is likely a consequence of the cumulative bifacial repulsions between lone pairs with consequent widening of the frontier bands.
The covalent functionalization of phosphorene has also electronic and structural consequences on the band structure. For instance, some added reactants may have bands, which fall in between the Pn valence and conduction ones, an example being the case of the d orbital set of an anchored metal atom. In addition, acidic groups at the surface may withdraw electron density from the P lone pairs with a consequent decrease of their original repulsion effects. As another aspect, groups bound only at one side of a channel, but projecting over it, may become repulsive toward some opposite P atoms inducing a local physical broadening of the channel itself. Obviously, the more frequent is the repetition of the group along the same zigzag chain, the more constant is the widening of the channel itself. To evaluate some possible electronic consequences of these structural effects, we started from the optimized structure of the naked 2D phosphorene with the cell parameter b (see the inset of Fig. 3), which provides an indirect evaluation of the channel width. Then, a systematic stretching of b (in the range 4.5–4.9 Å) determined the band gap increase from 2.11 to 2.43 eV. The opposite decrease of the band gap occurs upon the stacking of the sheets, as a strategy to reconstruct black phosphorus. This point will be later important for discussing key aspects of phosphorene's functionalization.
The following discussion will be based on computed parameters, the first ones of geometric character. Thus, the strength of the donor–acceptor σ interactions in functionalized Pn adducts will be evaluated, for instance, from the length of the newly formed bonds and also other more or less evident structural variations occurring at either Pn or the added acidic molecule. The other important parameter is the estimated binding energy BE (BE = ΔEadduct − ΔEacid-unit − ΔEP-donor unit), which helps in guessing the propensity of a given molecule to form a stabilizing adduct with phosphorene.
Fig. 4 Structure of tris(di-t-butylphosphino)phosphane, P3P.24 The methyl groups of tBu substituents are omitted for clarity. |
The difference will be corroborated by a comparison of the P3P·BH3 adduct with the corresponding Pn·BH3 ones, optimized for different borane coverages at a given side of the surface (see Table 1 and the associated discussion). Finally, a correlation is made with the white phosphorus adduct P4·BH3, where, at variance with P3P, there are three additional P–P bonds between the basal atoms of the molecule. In this case, the six edges of the P4 tetrahedron form a highly compact σ bonding system, to which also the four axial P lone pairs are contributing, given that the lowest lying P4 bonding combination of a1 symmetry largely consists of pz in-pointing orbitals, and hence lies low in energy. As a consequence, the P4 basicity appears to be particularly low as previously pointed out by us.12
Model | P–B dist. | BE | Band gap |
---|---|---|---|
(CH3)3P·BH3 | 1.91 | −1.61 | |
P3P·BH3 | 1.96 | −1.15 | |
Pn·0.031(BH3) | 2.00 | −0.60 | 2.32 |
Pn·0.062(BH3) | 2.00 | −0.58 | 2.42 |
Pn·0.125(BH3) | 2.02 | −0.51 | 2.61 |
Pn·0.250(BH3) | 1.98/2.11 | −0.46 | 2.70 |
P4·BH3 | 2.08 | −0.32 |
In order to envisage some differences between the adducts, Table 1 reports optimized parameters such as the P–B distance, the BE binding energy and the band gap of the adduct. In particular, the (CH3)3P·BH3 one appears to be most stable in view of the shortest P–B distance of 1.91 Å (1.85 Å in some experimental structure)25 and the most negative Binding Energy (BE = −1.61 eV). P3P·BH3 is instead somewhat less stabilized, having a larger P–B distance of 1.96 Å and a less negative BE of −1.15 eV. The different basicity is not attributable to a lower energy of the pivotal P lone pair, given the similar destabilizing effects of either the C or P substituents. Rather, the reduced donor power toward the vacant B orbital is due to the minor contribution of the central P lone pair of P3P·(HOMO), which has instead about 20% larger mixing of the adjacent P lone pairs in spite of their somewhat unfavorable orientation.
Before addressing the behavior of phosphorene as a σ donor, it must be mentioned that the white phosphorus adduct P4·BH3 with P–B and BE values of 2.08 Å and −0.32 eV, respectively, represents the weakest adduct of the series. The reason for the very poor 2e− donor power of P4 has already been outlined in our previous study of its demolition with I2, and indeed the highest barrier in the entire process leading to the final I3P product was encountered in the formation of the initial P4·I2 adduct.12 On the other hand, P4 is known to have residual donor power as suggested by some known η1 metal complexes, a first example being the trigonal bipyramidal species (NP3)Ni(η1-P4) species, reported long ago by our institute.26
In the case of phosphorene itself, the P–B bonding strength determined for various Pn·xBH3 adducts was systematically examined for different x values that each time doubled the density of borane on the surface. Fig. 5 shows a top view of the three progressively more covered species, which correspond to one BH3 molecule for every 16 and 8 and 4 P atoms of Pn. Additionally, Table 1 reports the parameters also for the least dense adduct (ratio 1:32), where x corresponds to a coverage of about 3.125% vs. the other values of 6.25%, 12.50% and 25.0% in the series. Initially, the doubling of the borane coverage causes very small variations of the structural and energy parameters, because the added BH3 molecules do not significantly interfere with each other. Thus, the P–B distance of 2.0 Å remains unaffected and the BE is reduced by only 0.02 eV. In the subsequent passage from 16 to 8 P atoms for any BH3, the reduction of the Pn donor power is evident. In fact, the P–B distance elongates up to 2.02 Å and BE reduces to −0.51 eV.
Fig. 5 The three Pnx(BH3) adducts with x = 0.062, 0.125 and 0.250 (a, b and c, respectively), corresponding to one BH3 molecule for every 16, 8 and 4 P atoms. |
Finally, in reaching the maximum BH3 density (25%), the system is evidently affected, also in view of the two short contacts between neighboring boranes. This determines also an asymmetry of the adjacent P–B distances (1.98 and 2.11 Å) while the difference in BE per BH3 unit becomes as small as −0.46 eV, confirming the increasing repulsion between the absorbates, no more tightly bound.
Fig. 6, presenting a lateral view of the adduct with 6.2% BH3 coverage, highlights some Pn deformation. In particular, the channel on which the attached BH3 molecule projects is more broadened than the adjacent one by about 0.3 Å, thus allowing a larger cradle for the absorbed molecule(s). Recall that a similar broadening of the ditch was simulated for naked phosphorene by elongating the cell parameter b, as shown in Fig. 3. In that case, the gap between the frontier and conduction bands increased consistently with the results of Table 1 with a maximum 0.5 eV difference on increasing the BH3 density at the surface. From a qualitative electronic viewpoint, the larger separation of the facing lone pairs at the ditch determines decreased repulsions, the effect being further enhanced by the larger electron drifts into the added boranes. Finally, the narrowing of the lone pair band also implies a larger gap from the conduction band, which, being formed by the P–P σ* levels, is scarcely affected by these geometric perturbations.
Fig. 6 A portion of the Pn·BH3 adduct, where 1 borane molecule is added to a supercell consisting of 16 P atoms. |
Pdonor | Ppyramidal·I2 | ||
---|---|---|---|
P–I | I–I | BE | |
(CH3)3P | 2.78 | 3.07 | −0.98 |
P3P | 2.86 | 3.03 | −0.65 |
Pn | 3.17 | 2.94 | −0.18 |
P4 | 3.19 | 2.92 | −0.13 |
The calculations confirm a very weak interaction in P4·I2, as suggested by the almost unperturbed geometries of the two molecular components and the very small −0.13 eV exothermicity.
In a previous study of the multi-step demolition of P4 with I2 to give I3P, the evolution from the first adduct P4·I2 was the most difficult one for the entire process in view of the high and unique barrier (ΔG = +14.6 kcal mol−1). In actuality, a second I2 molecule had to be involved as well as in all the subsequent steps.12 In this manner, two distinct P–I linkages could be formed in place of the pre-existing P–P bond, while a new I2 molecule was regenerated in situ being ready for subsequent reactivity. With this picture in mind, we explored whether also phosphorene could undergo any similar P–P cleavage under the action of two diatomics. Again we started with the XB type adduct Pn·I2 where one diatomic interacts with a Pn cell of 16 P atoms with an activation that is only marginally larger than that in P4·I2 (differences in the P–I and I–I distances of about 0.2 Å and a slightly more exothermic energy balance of −0.18 vs. −0.13 eV). Upon the addition of a second I2 molecule, the optimized Pn·2I2 model, reported in Fig. S1,† shows that the di-iodine activation only barely increases. In fact, the first added I2 molecule has P–I1 and I1–I2 variations no larger than 0.05 Å and the same result applies also to the second I2 one (see Table S1†). In this case, the overall BE value indicates an extra energy gain of only −0.12 eV, but more important, a further evolution of the system seems to be prevented by an unsuitable stereochemistry. While in the P4 case, it was evident that a remote I atom of the 2I2 grouping can potentially perform a nucleophilic attack into the P–P σ* level and induce its cleavage,12 the same does not apply to the 2D Pn species in view of the unsuitably oriented P–P bonds. Perhaps, the mechanism could be pursued for the high temperature transformation of the red phosphorus allotrope into the black one thanks to the involvement of I2 molecules generated by SnI4.29 In this case, the stereochemistry of the P–P linkages of red phosphorus is not as flat as in phosphorene, possibly favoring the action of I2 similarly to that proposed for P4. Obviously, the problem has to be tackled in some depth in order to corroborate such a conclusion.
To choose the metal fragments which may be best suited for the phosphorene functionalization, it is first important to have a good understanding of the stereochemical and electronic properties of both the interacting phosphorene and transition metal fragment. In the choice of the latter, a leading concept is that of the isolobal analogy,30 which allows their tailoring to support single or multiple 2e− interactions with neighbor Pn lone pairs. A general strategy is to start from a formally saturated metal complex, which, on losing one or more coligands, acquires a variable number of vacant σ hybrids acting as the acceptors of some phosphorene P lone pairs. The donor power of the latter has already been found to be particularly weak; hence it is important that the original metal complex carries some even weaker ligands to be eventually replaced. As another problem, the generated metal fragments should not create peculiar steric problems in the approach to the 2D material but also in their reorientation on the surface to maximize the σ overlap(s). Another important limitation in the choice of the metal fragment is its possible charge derived from the combination of different metals and coligands. In this case, the necessary electroneutrality of the solid state system imposes the presence of a counterion in close proximity to the charge metal fragment anchored at the surface. Clearly, for an optimal metal selection in the P functionalization, one should preferentially work with an uncharged metal fragment.
Fig. 7 shows top views of one phosphorene's face with differently hanging metal fragments. The latter differ in the number of σ hybrids stemming from the metal and interacting with one, two or three Pn lone pairs, as shown in Fig. 7a–c. Since any P lone pair forms an angle of about ∼30° with the Pn sheet, the unique P1–M linkage of the η1 coordination (7a) is expected to be bent on the surface, although in some cases some reorientation is observed (see below).
In the case of the η2 and η3 coordinations (7b and 7c, respectively), not all the interactions imply an optimal orbital overlap. In the former case, for instance, the lone pairs at the P1 and P2 atoms, originally lying in parallel planes at the opposite sides of a ditch (see Fig. 1), need to interact simultaneously with a midway M atom. To maximize the overlap some local torsion should occur at the surface, which has however a very scarce degree of flexibility. A similar problem arises for the η3 coordination, which involves one atom at one bank of the ditch (P1), while the P2 and P3 ones are sequential at the opposite side. In this case, only the P1 lone pair can, in principle, attain an optimal overlap with one M σ vacant hybrid; conversely, those at the P2 and P3 ones should attempt a rather difficult reorientation. The problem will be later confirmed by the larger P2–M and P3–M bonding distances compared to the P1–M one.
The stabilization of the ClAu(η1-Pn) adduct is proved by the large BE value of −2.4 eV, which is about four times larger than the −0.58 eV value obtained for the non-metal species Pn BH3 in Table 1. In view of these results, it emerges that an opportune transition metal fragment can be one of the best candidates for the covalent functionalization of phosphorene. For an even more realistic energetic picture, the ClAu(η1-Pn) adduct has been obtained through two alternative paths. Eqn (1) involves the precursor ClAu(Me2S), which first loses Me2S with a cost of +2.2 eV; hence the final energy balance is exothermic by −0.2 eV. Remarkably, the adduct formation from the other mentioned complex ClAu(Me3P) is more difficult, since eqn (2) is endothermic by +0.44 eV. Thus, phosphorene functionalization can be generalized to proceed from reactants carrying a good leaving group.
ClAu(Me2S) + Pn → ClAu(η1-Pn) + Me2S | (1) |
ClAu(Me3P) + Pn → ClAu(η1-Pn) + Me3P | (2) |
It is worth mentioning that previous computational studies on gold-phosphorene were reported for a system with single Au(0) atoms dispersed on the surface and η3 locally bound to the cavity formed by the P1, P2, P3 atoms.16 The average optimized Au–P distance of about 2.37 Å and the −1.61 eV stabilization energy seemed to confirm the propensity of gold toward Pn functionalization but the peculiar electronic nature of the Au(0) carrying an unpaired spin each was not specifically illustrated.
At this point, it is worth mentioning that, based on the in silico predictions about the ClAu(η1-Pn) adduct, some experimental attempts were made by us to obtain a functionalized phosphorene surface, starting from the ClAu(Me2S) reactant. Unfortunately, we are still unable to present here any unquestionable result, because ClAu(Me2S) easily promotes the formation of gold nanoparticles.37 During the reaction, we indeed observed by-products, some of which have been spectroscopically characterized. The work, which is still in progress, will hopefully provide some reasonable explanation of the Au(I) → Au(0) reduction. To avoid the latter, we are trying to use some Au(I) linear complex other than ClAu(Me2S) for the Pn functionalization.
Concerning the anticipated effect on the Pn band structure induced by the functionalization with ClAu, we noticed a decrease of the band gap, rather than the increase expected for the widening of some ditch. Indeed, the 2.26 eV band gap of free phosphorene becomes 2.17 eV in ClAu(η1-Pn). A relatively easy explanation of the result emerges from the plots of the whole band and the corresponding Density of States (DOS) in Fig. 9.
Evidently, the highest portion of occupied bands (Fermi region) largely consists of a set of Au d orbitals (yellow projection), which are implicitly repulsive with P lone pairs. Importantly, the spread of the basic band structure determines a consequent reduction of the gap with respect to the conduction band. This point is further corroborated by a plot of the Crystal Overlap Orbital Population (COOP in Fig. S2†), which properly highlights the triggered d/P lone pair repulsions.
The simultaneous coordination of the two P atoms at the opposite sides of a channel imposes a shrinkage of the latter, which is indicated by the 0.2 Å shorter P1⋯P2 separation compared to naked phosphorene. Correspondingly, the adjacent channel is somewhat widened. The P1 and P2 lone pairs involved in the η2-Pn coordination lie, as known, in parallel planes, and hence do not naturally converge into the unique metal position with nonoptimal overlap with its vacant σ hybrids. For this, the metal fragment is rotated by about 25° with respect to the perfect square planar geometry. Scheme 2 helps in understanding the orbital underpinning of the distortion, needed to improve the σ overlaps.
Scheme 2 The suggested rearrangement of a L2Ni(II) fragment across one Pn channel to maximize σ overlap. |
Through some minor deformation of the phosphorene surface and the indicated metal rearrangement, the obtainment of the final Pn adduct is indeed exothermic by −0.32 eV, as estimated from the reaction in eqn (3).
(CH3)2Ni(H2O)2 + Pn → (CH3)2Ni(η2-Pn) + 2H2O | (3) |
The quasi square planar coordination of d8-Ni(II) on Pn determines an increase of the band gap with respect to that of naked phosphorene (from 2.26 to 2.31 eV, as shown in Fig. S3†). The trend is the opposite compared to that of the ClAu derivative. There must be a compromise between the reduced repulsion between the P1 and P2 lone pairs participating in metal binding and the larger repulsion between the non-involved ones due to the shrinking of the channel.
The η2 coordination of phosphorene may be potentially attained also by using alternative fragments. One of them can still be of the L2M type with a d10 rather than a d8 metal atom, which has as a precursor an 18e− tetrahedral complex (Td). In this case, the two vacant σ hybrids of the L2M fragment lie high in energy for having exclusive s/p character without any d contribution. In general, the known Ni, Pd, Pt tetrahedra are stabilized by strong σ donors such as four COs42 or 2COs + 2R3P ligands.43,44 Since some complex is known with a dinitrogen chelate,45 we used the species (CO)2Ni(NH3)2 as a model reactant for the reaction with Pn, reported in eqn (4). From the optimized adduct (CO)2Ni(η2-Pn), having a local tetrahedral geometry as shown in Fig. 12, the process is estimated to be exothermic by −0.17 eV.
(CO)2Ni(NH3)2 + Pn → (CO)2Ni(η2-Pn) + 2NH3 | (4) |
In the tetrahedral geometry of (CO)2Ni(η2-Pn), the Ni(0)–P distances of 2.24 Å are close to those of the Ni(II) square planar species of Fig. 11. Therefore, the channel width is again shortened by about 0.2 Å but the 2.22 eV band gap of the d10 adduct is now only slightly smaller than the 2.26 eV value of free phosphorene vs. the 2.31 eV value of the Ni(II) case. Fig. S4† shows trends similar to those of the ClAu adduct in view of the position of the metal d orbitals in the highest portion of the valence band. A final piece of information concerns the approximately 70 cm−1 blue shift of the CO ligands found in the computed IR spectrum of (CO)2Ni(η2-Pn) compared to that of the molecular model (CO)2Ni(PMe3)2. This feature is not surprising given the known weak donor power of the 2D material, with a consequently reduced back donation from the metal into CO ligands.
Another metal fragment for potential η2-Pn coordination is a d6-L4M butterfly, obtained upon removal of two cis equatorial ligands from a d6-L6M octahedron. A possible example is (CO)4M (M = Cr, Mo and W), obtainable for instance from an octahedron with two additional dialkyl sulphide ligands.46,47 Optimization of the (CO)4Mo(η2-Pn) adduct (see Fig. S5†) shows that the Mo–P interactions are relatively weak in view of the 2.56 Å distances and other evident distortions of steric origin. The scarce propensity of the butterfly (CO)4Mo fragment to support Pn di-hapto coordination is corroborated by the large +1.02 eV endothermicity of (CO)4Mo(η2-Pn), as estimated from eqn (5). Perhaps in general, phosphorene functionalization is unlikely with a butterfly fragment.
(CO)4Mo((CH3)2S)2 + Pn → (CO4)Mo(η2-Pn) + 2(CH3)2S | (5) |
From the orbital interaction viewpoint, the unsaturated trigonal pyramidal (TP) (CO)3Mo fragment has the three vacant and delocalized MOs of a1 + e (degenerate) symmetries, shown in Scheme 3. These can match with phosphorene's lone pair combinations to allow η3-Pn coordination.
However, the e symmetry matchings are not fully equivalent due to the natural isosceles shape of the P3 triangle on the surface. In fact, the optimized (CO)3Mo(η3-Pn) adduct of Fig. 13 has a P1–Mo bond of 2.46 Å, which is about 0.1 Å shorter than the P2–Mo and P3–Mo ones of 2.56 Å.
As indicated by Scheme 4, the P1 lone pair is properly oriented for optimal MO interactions with both a1 and the e orbitals of Scheme 3. Conversely, the P2 and P3 lone pairs are not equally well oriented and can hardly rearrange due to the Pn skeleton rigidity. Eventually, however, the divergences in bonding are not so dramatic; hence an essential tri-hapto coordination of phosphorene may be assumed.
Scheme 4 Disposition of the vacant metal σ hybrids and the P1, P2 and P3 lone pairs in a potential tri-hapto coordination. |
From the energy viewpoint, the η3-Pn formation is exothermic by about −0.33 eV, as estimated by the reaction in eqn (6), which proceeds from the precursor (CO)3Mo(η3-p-xylene).
(CO)3Mo(η3-p-xylene) + Pn → (CO)3Mo(η3-Pn) + p-xylene | (6) |
Incidentally, the scarce deformation of the coordinated Pnvs. the free 2D material has a cost of +0.28 eV, as it results from a single point calculation of the final adduct upon removal of the (CO)3Mo fragments. Evidently, the formed P–Mo bonds almost double the energy lost on the Pn deformation. In addition, the discrete complex (CO)3Mo(PMe3)3 was optimized to compare its computed IR spectrum with that of (CO)3Mo(η3-Pn). For consistency, the vibrational frequencies of all species were calculated by using the same CRYSTAL software package.19 Once again, the donor power of phosphorene is found to be scarce as indicated by the 60 cm−1 blue shift of the CO stretching in the adduct vs. those of the complex (CO)3Mo(PMe3)3. Consider that in the latter the carbo-substituted phosphines have high donor power, so that the larger electron density at the metal allows larger back-donation with different IR responses for CO stretching. As another point, adducts of the type L3Mo(η3-Pn) with three single bulky substituted phosphines49 or a common tripodal chelate such as CH3C(CH2)3PPh250 are unlikely because of the steric hindrance problems arising when the substituents point down on the Pn surface.
Finally, as already done for the Pn η1 coordination with the ClAu fragment, we examined how the electronic structure and the determining band gap may be influenced by the (CO)3Mo functionalization. The corresponding gap of 2.34 eV for (CO)3Mo(η3-Pn) is larger than that in free phosphorene and the discussed ClAu(η1-Pn) adduct (2.26 and 2.17 eV, respectively). At variance with the latter case, the P lone pairs, which participate in the P–Mo bonding, lose their reciprocal repulsive character, with a lowering of the valence band spread, and a consequent increase of the band gap. On the other hand, the right side of Fig. S6† shows that in both the valence and conduction bands, the π interactions between the t2g metal d orbitals and CO π* levels have a significant contribution. In particular, the stabilized Mo d orbitals represent the highest limit of the valence band, while the opposite occurs for ClAu(η1-Pn), where the repulsions between the low lying and filled Au d orbitals and the P lone pairs fix the band gap.
Next, we switched to various transition metal fragments featuring a variable number of vacant metal lobes, based on the isolobal analogy concept.30 Accordingly, we examined the possibility that some given metal acceptor interacts with one or more neighbor Pn lone pairs and affords its covalent functionalization. M–Pn derivatives of various hapticities were optimized, providing useful information on stereochemistry and energy. Evident steric hindrance problems emerge even when a metal fragment is electronically suited but has an unsuitable disposition of the coligands, in particular if bulky. Other derived properties of the adducts have been examined such as the vibrational spectra of the CO coligands, when present. Another tackled point concerns the band gap of the functionalized adduct vs. that of naked phosphorene as a consequence of the variable repulsion between P lone pairs. A first example of perturbation is given by the stacking of the Pn sheets to reform black phosphorus. Only ten Pn sheets are sufficient to reproduce the much smaller band gap of the 3D material, because the bilateral repulsions between the sheets quickly spread the frontier band and reduce its separation from the conduction band. On the other hand, the functionalization of a single Pn face has contrasting and discussed effects such as the widening of the ditches on associating acidic groups, which in turn reduces the repulsion between lone pairs. In the case of metal coordination, its d orbital set may fall at the top of the bands formed by P lone pairs with consequent modification of the ultimate band gap. Qualitative explanations have been proposed also for more complex cases.
While our analysis has indicated various possibilities of covalent functionalization of phosphorene with acidic groups, the corresponding experimental evidence remains scarce. Certainly, one reason is the difficult isolation and manipulation of the 2D material, given its known sensitivity to external agents such as oxygen or humidity. A preliminary elimination of the latter is fundamental to perform new useful chemistry. In any case, we trust that our models may be later exploited by some brave experimentalist to construct viable species and open new application gates in the chemistry of the important 2D material.
Footnote |
† Electronic supplementary information (ESI) available. See DOI: 10.1039/c8dt03628d |
This journal is © The Royal Society of Chemistry 2018 |