Interplay between Gold(I)-Ligand Bond Components and Hydrogen Bonding: A Combined Experimental/Computational Study.

The influence of weak interactions on the donation/back-donation bond components in the complex [(NHC)Au(SeU)]+ (NHC = N-heterocyclic carbene; SeU = selenourea) has been studied by coupling experimental and theoretical techniques. In particular, NMR 1H and pulsed-field gradient spin-echo titrations allowed us to characterize the hydrogen bond (HB) between the -NH2 moieties of SeU and the anions PF6 - and ClO4 -, whereas 77Se NMR spectroscopy allowed us to characterize the Au-Se bond. Theoretically, the Au-Se and Au-C orbital interactions have been decomposed using the natural orbital for the chemical valence framework and the bond components quantified through the charge displacement analysis. This methodology provides the quantification of the Dewar-Chatt-Duncanson (DCD) components for the Au-C and Au-Se bonds in the absence and presence of the second-sphere HB. The results presented here show that the anion has a dual mode action: it modifies the conformation of the cation by ion pairing (and this already influences the DCD components) and it induces new polarization effects that depend on the relative anion/cation relative orientation. The perchlorate polarizes SeU, enhancing the Se → Au σ donation and the Au → C back-donation and depressing the C → Au σ donation. On the contrary, the hexafluorophosphate depresses both the Se → Au and C → Au σ donations.

Introduction

The Dewar–Chatt–Duncanson (DCD) framework for the characterization of the coordination metal–ligand bond[1,2] certainly has a central role in inorganic chemistry. It was proposed to rationalize the interaction between an olefin with a transition coinage metal (M), but it has been rapidly adopted as a general scheme for the coordination of other ligands (L),[3] including phosphanes,[4−6] carbon monoxide,[7] and carbenes.[8,9]

In short, according to the DCD theory, contributions of the M–L coordinative bond can be mainly divided into two terms: M ← L σ donation, from the filled orbitals of the ligand to the empty orbitals of the metal with axial symmetry, and M → L π back-donation, from the filled d orbitals of the metal to empty orbitals of the ligand with planar symmetry.

During the years, this framework gained an enormous success, so much so that it has been recently adopted also for molecules without any transition metal, as selenones,[10] phosphinidenes,[11,12] and chalcogeniranium cations,[13,14] among others. On the other hand, it also led to some misconceptions: for example, according to an old classification of the metals, the fact that the stretching frequency of the carbonyl moiety in [LAu(CO)] is generally higher than that of the free CO[15−17] (non-classical carbonyl[18]) was considered as a proof that gold is unable to give back-donation. Nowadays, it is known that this conclusion is wrong, as the carbonyl frequency is given by the large gold-induced C ← O polarization that outweighs the effect of back-donation.[7]

Also regarding the N-heterocyclic carbene (NHC) ligands,[19] initially they were classified as strong σ donors and very poor π acceptors,[20,21] but the π-acid properties of NHCs have been recently re-evaluated and extensively studied, both experimentally[8,9,22,23] and theoretically.[24−28]

Therefore, the correct evaluation of the DCD components of an M–L bond is extremely important, especially now that the rational design of “new” catalysts is a real possibility. For this reason, in the recent past, the group of Tarantelli and Belpassi carried out a systematic combined experimental–theoretical work on the characterization of the Au–L bond,[7,26,29−31] in many cases by using the charge displacement (CD) analysis.[32] The latter is a computational tool that decomposes the orbital interaction between two fragments according to irreducible representations of the group symmetry of the system[31] or, in absence of an adequate symmetry, to the natural orbital for the chemical valence (NOCV) framework.[33]

In the recent years, the same group and others demonstrated (and, in some cases, explained) that also weak interactions can influence the performances of gold(I)-based catalysts. For instance, π stacking can enhance the catalytic activity of complexes containing a pyrene moiety;[34] hydrogen bonding acceptors can direct and facilitate the nucleophilic attack by a rational ligand design[35,36] or by a wise choice of the counterion.[37−43] Indeed, the interplay between covalent and noncovalent interactions recently led to interesting results in dynamic combinatorial library-driven synthesis[44−46] (also applied to biomolecules[47,48]), surface chemistry,[49,50] and reactivity of halogen-containing moieties.[51]

More specifically, the presence of a noncovalent interaction can influence the covalency of a chemical bond. For example, when I2 establishes a halogen bond,[52] the I–I bond lengthens until, in the extreme cases, it breaks and becomes a halogen bond itself.[53,54] In a similar way, the C–Se bond order in selenourea is influenced by the presence of the hydrogen bond (HB) acceptors.[55]

We therefore asked ourselves whether and how second-sphere weak interactions can influence the DCD components of an M–L bond.[56] In order to do this, we decided to use the organometallic salt [(IPr)Au(SeU)]+ PF6– (1X, IPr = 1,3-bis(2,6-diisopropylphenyl)imidazol-2-yliden, SeU = selenourea, X– = PF6–, Scheme ).[57] The carbene ligand has been chosen for giving enough stability to the complex, while the PF6– anion for being a poor Lewis base that can be easily replaced by other stronger HB acceptors. Finally, SeU has been chosen for being a HB donor and for possessing an NMR-active nucleus, the 77Se, that can effectively probe the variations of Au–Se and Se–C bonds. Furthermore, it has been demonstrated recently that an anion interacting with the amine protons of SeU makes the halogen bond between selenium and a polarized iodine (as in IC6F13) stronger.[55] This is because the interaction NH···X– makes the zwitterionic resonance structure more important, inducing a larger negative charge on selenium, which in its turn becomes more basic.

Scheme 1

Two Main Resonance Structures for 1

In this paper, we characterize the HB adducts between 1 and two anions through experimental techniques, such as NMR titrations, pulsed-field gradient spin-echo (PGSE), and 77Se NMR spectroscopy, and theoretically by means of density functional theory (DFT) (B2PLYP/def2-TZVP//BP86-D3/def2-TZVP level[58]) and NOCV-CD analysis.[59,60] Here the Lewis acid (the cationic gold moiety) is much stronger than a polarized iodine, but the coordination bond is also more complex, in terms of active components, than the XB. In this case, the coordinated SeU can have many resonance structures and the two most important ones are depicted in Scheme .

The results indicate that the anion has a dual mode action: it modifies the conformation of the cation by ion pairing (and this already influences the DCD components) and it induces new polarization effects that depend on the relative anion/cation relative orientation. The perchlorate polarizes SeU, enhancing the Se → Au σ donation and the Au → C back-donation and depressing the C → Au σ donation. On the contrary, the hexafluorophosphate depresses both the Se → Au and C → Au σ donations.

Results and Discussion

NMR Studies

The characterization of the HB between 1 and an anion requires some consideration on the experimental conditions to be used. Indeed, because 1 is a salt, it is important to minimize the ion-pairing effect[61] to avoid any interference between the anion under examination and the anion already present in the salt as a counterion. An effective strategy can be the use of an easily replaceable, poorly coordinating, and poorly hydrogen-bonded acceptor anion, such as PF6–, a polar solvent, such as acetone-d6, and a low salt concentration, say around 10 mM. Considering the available information about the aggregation of gold(I)-[62] and tetraalkylammonium salts,[63] the illustrated strategy seems adequate to study “isolated” cations of 1 in solution. The addition of an excess of tetraalkylammonium salt NR4X will lead to the formation of the HB adduct/ion pair 1···X–.

The interaction between 1 and PF6–, for instance, has been studied adding tetraethylammonium hexafluorophosphate (TEAPF) to a 8.5 mM solution of 1PF6 and monitoring the 1H NMR chemical shift of the protons on the carbene backbone (CH) and the two amino protons (NHa and NHb) (Figure ). All of them indeed are acidic sites[64] and, therefore, potential sites for the HB interaction with the anion.

Figure 1

1H NMR chemical shift variation of (a) 1PF6 (0.0119 M in acetone-d6) and (b) 1PF6 (0.0085 M in acetone-d6) as a function of (a) [TBAClO] and (b) [TEAPF]. Solid lines represent the best-fit equations.

Experimental fitting[65] of experimental data led to three different HB adduct formation constants (KPF, KPF, and KPF). Their values are 1.1 ± 0.3, 3.5 ± 0.5, and 1.5 ± 0.6 M–1. All the association constants are low, coherently with the low basicity of PF6–. Anyway, a difference between the acidic sites can be noted, and the interaction of the amine protons in pseudo-cis with respect to selenium is stronger than the others, likely for the possibility of having two cooperative HBs, as it generally happens with the urea-like moiety.[66]

The same experiment has been carried out in the absence of the organometallic fragment, titrating free SeU with TEAPF. Unfortunately, the broad signal of the amino protons does not shift considerably after the addition of an excess of TEAPF. For this reason, a standard 1H NMR titration is not feasible and a 1H PGSE NMR titration is needed.[67] In fact, by means of the latter, the diffusion coefficient (Dt) of the species under examination can be measured, and aggregation processes can be evaluated, even if very weak or involving many species.[68,69]

The hydrodynamic volume of SeU (VH) which can be derived from the value of Dt(70) goes from 176 to 205 Å3 in the absence and presence of an excess of TEAPF (Table ). The corresponding association constant (KPF) is 2.0 ± 0.3 M–1. It can be noted that the latter is smaller than KPF, demonstrating that the metal center polarizes the coordinated SeU, making it more prone to establish hydrogen bonding with the anion (structure b in Scheme ).

Table 1

Diffusion Coefficients (Dt, 10–10 m2 s–1), Hydrodynamic Radii (rH, Å), and Hydrodynamic Volumes (VH, Å3) of SeU (24 mM) in the Absence and Presence of Tetralkylammonium Salts

additive	Dt	rH	VH
	24.8	3.48	176
[TEAPF6] = 8 mM	23.6	3.60	195
[TEAPF6] = 16 mM	23.5	3.61	197
[TEAPF6] = 38 mM	23.4	3.62	199
[TEAPF6] = 65 mM	23.3	3.63	200
[TEAPF6] = 83 mM	23.0	3.66	205
[TBAClO4] = 4.2 mM	24.4	3.53	184
[TBAClO4] = 17 mM	23.0	3.66	205
[TBAClO4] = 24 mM	22.5	3.71	214
[TBAClO4] = 84 mM	21.5	3.83	235

Using tetrabutylammonium perchlorate (TBAClO) instead of TEAPF, the 1H NMR titration of 1 in acetone leads to the following results: KClO,KClO, and KClO are 19 ± 2, 278 ± 30, and 211 ± 40 M–1, respectively. All the values are considerably higher than in the case of 1/PF6–, coherently with the higher basicity of the perchlorate anion. In this case, the CH/NHa selectivity is much higher, likely because of the larger stability of the supramolecular cycle between the two N–H moieties and the two Cl–O HB acceptors.

As given above, the titration in the absence of the gold fragment has to be carried out by using the PGSE NMR technique, because of the broadness of the NH NMR peak. Moreover, in this case, diffusional data have been fitted with a biexponential decay instead of the standard monoexponential one (see Experimental Section for details). The same thing happens when a benzoate anion is used,[55] likely for a chemical exchange between the amino protons and water, which is facilitated by basic anions. In fact, of the Dt values obtained by the biexponential fitting, the smallest one is compatible with SeU (around 23 × 10–10 m2 s–1), whereas the largest one is relative to a very small molecule (around 50 × 10–10 m2 s–1, presumably water).

Focusing only on the smallest value of Dt obtained by the fitting procedure (Table and Supporting Information), the VH of SeU goes from 176 Å3 in the absence of TBAClO to 235 Å3 in the presence of a large excess of TBAClO. Considering that the van der Waals volume of the perchlorate anion is 50 Å3, we can say that practically all selenourea in solution interact with the perchlorate. Fitting the data listed in Table , the association constant between SeU and ClO4– (KClO) results to be 101 ± 9 M–1. Comparing the latter with KClO (278 M–1), we can say that the HB is greatly enhanced by the coordination of SeU on the metal fragment, again because of the polarization of the ligand.

The systems described above have been characterized also by 77Se NMR spectroscopy. The selenium nucleus of the isolated SeU in acetone-d6 resonates at 220 ppm, whereas when it interacts with the gold fragment, the frequency is lowered down to 174 ppm, indicating a larger amount of electronic density around the nucleus (resonance b in Scheme ). This value is slightly lowered even more by the presence of an excess of TBAClO (173 ppm, [1PF6] = 10 mM, [TBAClO] = 80 mM), whereas an excess of TEAPF induces a deshielding effect (176 ppm, [1PF6] = 10 mM, [TEAPF] = 150 mM).

The titration of 1 with tetraetylammonium chloride or tetrabutylammonium benzoate did not lead to the determination of the corresponding association constant, as in their presence, the NH NMR peaks broaden and disappear. This indicates that the amine protons of selenourea are involved in a dynamic process. A likely hypothesis can be the equilibrium depicted in Scheme .

Scheme 2

Possible Reaction between 1 and Basic Anions

Finally, tetrabutylammonium sulfate resulted to be too hygroscopic to be dried and used in a titration like this.

Computational Studies

DFT studies have been carried out to quantitatively analyze the bonds between gold and SeU or gold and carbene in the presence and absence of HB. To save computational resources, the IPr ligand has been simplified substituting the aromatic substituents with methyl groups ([(NHC)Au(SeU)]+, 1s). The geometry of the following adducts has been optimized: 1s, 1sPF, and 1sClO. For all of them, the NOCV-CD analysis has been performed with the following fragmentation schemes: [(NHC)Au]+···[SeU] and [NHC]···[Au(SeU)]+ for 1s, [(NHC)Au(SeU)]+···[X]−, [(NHC)Au]+···[(SeU)X]−, and [NHC]···[Au(SeU)X] for 1sX.

From the DFT-optimized geometries, the C–Se bond in isolated SeU results to be 1.831 Å (Mayer bond order[71] = 1.63). The interaction with the gold moiety (1s structure) elongates the Se–C bond up to 1.887 Å and lowers the Mayer bond order to 1.24. When PF6– interacts with the two amino groups, the Se–C bond becomes 1.925 Å and the bond order is 1.16, whereas in the presence of ClO4–, the Se–C bond becomes 1.933 Å and the bond order is 1.11. A qualitative difference can be noted between the geometries of 1sPF and 1sClO (Figure ). In the latter, the anion interacts with the amino groups, forming a strong HB (average O···H distance of 1.75 Å), and with the hydrogens of one methyl group on the carbene, forming a weaker HB (2.518 Å). In the former, the anion interacts with the amino groups, forming a weak HB (1.995 Å) and with both the methyl groups of the carbene through weak HBs (2.57 Å). The distance anion···gold is much lower in 1sPF (around 3.3 Å) than in 1sClO (around 4.9 Å).

Figure 2

DFT-optimized geometries (BP86-D3/def2-TZVP level) for (a) 1sPF and (b) 1sClO.

The Se–C bond length is already demonstrated to be a sensitive probe for the interactions in which SeU is involved, becoming longer in the presence of a NH···benzoate HB or a Se···ICF3 XB and even longer in the presence of both.[55]

First, the Au–Se and Au–C bonds will be analyzed in 1s. The energy decomposition analysis (EDA)[72] shows that the total interaction energy (Eint) between the fragments [(NHC)Au]+ and SeU is −73.1 kcal/mol (Supporting Information), of which −5.6 are the steric contribution (Est, Pauli + electrostatic), −5.6 are the dispersion contribution (Edisp), and −61.8 kcal/mol is due to the orbital interaction (Eoi).

By using the NOCV methodology, Eoi can be further decomposed into chemically relevant contributions, numbered with an integer number k (Δρ), each of which describes a particular charge rearrangement upon the adduct formation with respect to the fragments [(NHC)Au]+ and SeU (see Computational Details). Further, each contribution is also associated with a portion ΔE of the total orbital energy of the adduct.

In the case of the Au–Se bond in 1s, for Δρ0 (ΔE0 = −44.0 kcal/mol), the 3D charge rearrangement plot shows a charge depletion (red-colored in Figure ) at selenium, whose shape resembles its p orbital, and a charge accumulation at the gold and carbene (blue-colored), all of which with an approximately axial symmetry. Therefore, Δρ0 is clearly related to the σ Se → Au donation.

Figure 3

Isodensity surfaces (±0.003 e/au) for the most relevant Δρ (k = 0–3) of the [(NHC)Au]+···[SeU] bond for complex 1s.

Δρ1 (ΔE1 = −5.7 kcal/mol) shows a charge accumulation at selenium and a depletion at the carbon and nitrogen atoms of selenourea, whereas no contribution is present at the gold–carbene moiety. Therefore, this contribution describes the C → Se polarization because of the formation of the complex, with selenium that becomes more negatively charged, and the C=Se double bond order decreases (structure b in Scheme ), as confirmed from the abovementioned DFT geometries.

Δρ2 and Δρ3 (ΔE = −4.1 and −2.9 kcal/mol, respectively) show charge depletion regions at gold, the shape of which resemble the d and d orbitals of the metal, and accumulation regions at selenium. These contributions can be related to the Au → Se back-donation of π and σ symmetry. Even if the back-donation is generally thought to be with π symmetry, there are examples of σ back-donation.[26]

These components can be integrated to obtain a quantitative estimation of the electronic fluxes between the fragments upon the formation of the adduct through the CD analysis: the CD0 curve, obtained by integrating Δρ0 along the z axis that passes through the gold and selenium nuclei, is always positive, indicating a net Se → Au charge transfer (Figure ). CD1 is mainly localized in the region of selenium, coherently with Figure . CD2 and CD3 are negative in the boundary region, indicating a Au → Se charge transfer (Figure ). At the boundary, the curves assume the values (CT) of 0.384, 0.027, −0.022, and −0.016 e for k going from 0 to 3, respectively.

Figure 4

NOCV-CD curves for the most relevant components of the Au–Se bond in the complex 1s. Black dots indicate the z position of the atomic nuclei. A yellow vertical band indicates the boundary between the [(NHC)Au]+ and SeU fragments.

The Au–C bond can be similarly analyzed (Figure S5, Supporting Information), using [NHC] and [Au(SeU)]+ as fragments (Eoi = −80.8 kcal/mol). Δρ0 (ΔE0 = −50.1 kcal/mol) is associated with the C → Au σ donation, Δρ1 (ΔE1 = −10.5 kcal/mol) and Δρ3 (ΔE3 = −5.1 kcal/mol) are associated with the Au → C π back-donation, and Δρ2 (ΔE2 = −8.3 kcal/mol) with the Au → C σ back-donation. Integrating the four main contributions leads to CT = 0.337, −0.041, −0.038, and −0.037 e for k going from 0 to 3, respectively. The NHC moiety has a slightly lower σ donation than SeU, but it accepts more back-donation (−0.116 e) than the latter (−0.038 e).

Now that the coordinative bonds in the isolated cation are completely characterized in terms of DCD components, it is possible to study the effect of a second-sphere HB.

First, 1sClO can be fragmented into [(NHC)Au(SeU)]+ and [ClO4–], in order to study the effect of the anion on the electron density of the cation. In this case, Eoi is only −27.4 kcal/mol, coherently with the fact that the HB is not a covalent bond but mostly electrostatic in nature (Eint = −99.9 kcal/mol). The decomposition of Eoi leads to two main contributions Δρ0 and Δρ1 associated with ΔEk = −8.7 and −7.9 kcal/mol, respectively. All the other contributions count for less than 1.5 kcal/mol each. Δρ0 is associated with a polarization of SeU, with the electron density of the anion that moves toward the amino protons of the cation and the electron density of selenourea that moves from the amino protons to the nitrogen atoms (Figure ). Moreover, also the C=Se bond is polarized, with a depletion region at carbon and an accumulation at the selenium atom.

Figure 5

Isodensity surfaces (±0.001 e/au) for Δρ0 of the [(NHC)Au(SeU)]+···[ClO4]− bond for complex 1sClO.

In other words, as already reported for the benzoate···SeU···ICF3 system,[55] the anion enhances the anionic character of selenium. Δρ1 describes a similar contribution and it is reported in the Supporting Information (Figure S6). The effect of PF6– is similar but smaller (Eoi = −15.7 kcal/mol, Figure S7, Supporting Information), and the accumulation region at selenium is not visible. A lower threshold is necessary to make it emerge.

Before analyzing the effect of the anion on the Au–Se and Au–C DCD components, it is important to take into account that the anion deeply modifies the conformation of the ligands, in different ways if different anions are introduced, as already discussed (Figure ). To do that, the NOCV-CD analysis will be performed using the geometries of 1s in 1sPF and 1sClO but obviously without the anion.

Deleting the anion from the geometry of 1sClO, the analysis of Δρ (k = 0–3) reveals that the change of geometry has a noticeable effect on the qualitative nature of the NOCV pairs. In fact, for the Au–Se bond, Δρ0 describes again the Se → Au σ donation, but the π electron of the C–Se bond is also involved (Figure ). This is likely because of the different relative orientations between gold and SeU: indeed, the Au–Se–C–N dihedral angle is 1.18° and 114.9° in 1s and 1sClO, respectively. In other words, in the latter, the π system of the Se–C double bond directly faces the gold atom, facilitating the σ donation from the π(SeC) orbital to the empty orbital of gold. This contribution does not substitute the p(Se) → Au donation, but they coexist, as the shape of the negative region at SeU demonstrates. This change is evidenced also by the NBO second-order perturbation theory interaction energy analysis, according to which the π(SeC) → Au or σ*(Au–C) orbital interactions pass from 1.5 to 3.7 kcal/mol passing from 1s to 1sClO.

Figure 6

Isodensity surfaces (±0.0015 e/au) for the most relevant Δρ (k = 0–3) of the [(NHC)Au]+···[SeU] bond for complex 1s (at the geometry that the cation adopts in 1sClO).

Δρ1 does not describe anymore the SeU polarization but the polarization of the whole molecular system, with both positive and negative regions at selenium and gold. A more detailed view of Δρ1 is reported in the Supporting Information (Figure S8), with a lower threshold.

Δρ2 describes the Au → Se π back-donation, from the d orbital of the metal associated to a polarization contribution; finally, Δρ3 describes the Au → Se σ back-donation, from the d orbital of the metal.

The Δρ functions corresponding to the analysis of the Au–C bond do not change (from the qualitative point of view) by using the geometry of the cation in 1sClO or 1s.

As above, the integration of Δρ leads to the quantification of the different contributions (Table ), allowing a comparison among the different situations: for the Au–Se bond, the CT values change only marginally and the largest deviation is 0.017 e for CT1, whereas for the Au–C bond, CT0 and CT1 increase in the absolute value with respect to the original geometry and all the other values undergo small variations.

Table 2

Bond Analysis Results (CT, in Electrons and, in Parenthesis, ΔE, in kcal/mol) for all the Cationic Systems

fragments	CT0 (ΔE0)	CT1 (ΔE1)	CT2 (ΔE2)	CT3 (ΔE3)	CTtot (Eoi)
[(NHC)Au]+···SeU	0.384 (−44.0)	0.027 (−5.7)	–0.022 (−4.2)	–0.016 (−2.9)	0.373 (−61.8)
[(NHC)Au]+···[SeU]a	0.381 (−44.1)	0.044 (−4.7)	–0.018 (−3.0)	–0.020 (−3.3)	0.387 (−61.3)
[(NHC)Au]+···[SeU]b	0.369 (−44.0)	0.025 (−4.3)	0.032 (−3.7)	–0.019 (−3.1)	0.378 (−60.8)
NHC···[Au(SeU)]+	0.374 (−50.1)	–0.033 (−10.5)	–0.034 (−8.3)	–0.030 (−5.1)	0.278 (−80.8)
NHC···[Au(SeU)]+a	0.381 (−49.9)	–0.059 (−11.1)	–0.033 (−9.0)	–0.025 (−5.0)	0.265 (−81.3)
NHC···[Au(SeU)]+b	0.381 (−49.9)	–0.043 (−11.2)	–0.032 (−8.5)	–0.030 (−5.3)	0.274 (−81.4)

At the geometry adopted in 1sClO but not considering the anion.

At the geometry adopted in 1sPF but not considering the anion.

Similar considerations can be made by using the geometry of the cation in 1sPF. In this case, the π(SeC) → Au or σ*(Au–C) orbital interactions count for 4.5 kcal/mol. In this case, comparing the values of CT at the two different geometries, the largest deviations are for CT0 that decreases and CT2 that passes from negative to positive. Likely, in this conformation, the polarization outweighs the back-donation, making the net flux from selenium to gold.

Now the electronic effect of the anion can be separately analyzed using the fragmentation scheme [(NHC)Au]+···[(SeU)ClO4]−. The analysis of Δρ (k = 0–3) does not qualitatively differ from that discussed (compare Figures and S9, Supporting Information). It is important to add that no noticeable accumulation/depletion regions are present at the anion; therefore, also in this case, the integration of Δρ can be directly related to the DCD and polarization components of the Au–Se bond. In this case, CT is 0.395, 0.051, −0.021, and −0.016 for k going from 0 to 3, respectively (Table ). Comparing these values with those previously obtained (Table ), it can be noted that the Se → Au is larger than previously obtained (0.384 e). Likely, because the perchlorate interacts with SeU, it makes the selenium atom more negative and more σ donor. The other values do not change much.

Table 3

Bond Analysis Results (CT, in Electrons and, in Parenthesis, ΔE, in kcal/mol) for all the Neutral Systems

fragments	CT0 (ΔE0)	CT1 (ΔE1)	CT2 (ΔE2)	CT3 (ΔE3)	CTtot (Eoi)
[(NHC)Au]+···[SeUPF6]−	0.352 (−42.4)	0.005 (−3.6)	0.011 (−3.7)	–0.024 (−3.6)	0.368 (−64.6)
[(NHC)Au]+···[SeUClO4]−	0.395 (−46.1)	0.051 (−5.3)	–0.021 (−3.3)	–0.016 (−2.8)	0.409 (−67.5)
NHC···[Au(SeU)PF6]	0.368 (−49.2)	–0.049 (−10.2)	–0.033 (−8.2)	–0.031 (−5.1)	0.255 (−79.4)
NHC···[Au(SeU)ClO4]	0.372 (−47.2)	–0.071 (−10.7)	–0.033 (−8.9)	–0.027 (−5.0)	0.241 (−78.9)

Another interesting consideration arises comparing the DCD components of the Au–C bond in the presence and absence of the anion. CT0 (0.372 e) is lower than previous value (0.382 e), whereas the back-donation contributions (CT1–3) are larger. Also, these variations can be ascribed to the polarization of SeU. In fact, the enhancement of the Se → Au donation makes gold less σ acceptor and more π donor.

In the case of the weakly coordinating PF6–, the CT0 value for the Au–Se bond is 0.352 e. It decreases with respect to the previous value (0.369 e), contrarily to what happens for the perchlorate. This can be because of the different relative anion/cation orientations (Figure ). In fact, the hexafluorophosphate is located over gold, polarizing SeU to a lesser extent, likely repelling the Au–Se electrons and therefore depressing the Se → Au donation. Similarly, all the other values of CT also become smaller or more negative (Table ).

For the Au–C bond in the presence of PF6–, all the CT values decrease, again because of the central position of the anion that depresses all the donations toward gold.

The effect of the anion on the Au–C and Au–Se DCD components is only apparently small, as it has been recently demonstrated that the gold–carbene bond is exceptionally stable and only few structural modifications are able to effectively modulate the σ donation and π back-donation.[26]

It is now possible to compare the experimental and theoretical results. From one point of view, the increase of KHB between the pairs SeU···X– and [(NHC)Au(SeU)]+···X– is certainly due to more favorable electrostatic contributions, but the polarization of SeU by the metal fragment (Δρ1 in Figure ) likely also plays a role, especially in the case of perchlorate, for which the difference is more pronounced than in the case of PF6–.

The other experimental probe, the 77Se NMR chemical shift, is certainly linked to all the theoretical parameters illustrated above (donation/back-donation variations, bond order...) but a deep understanding is not easy because the theoretical parameters influence the electron density at selenium in contrasting ways. For instance, the HB between selenourea and the perchlorate certainly polarizes the coordinated selenourea (Figure ), making the electron density around selenium to increase (a shielding effect can be predicted) with consequent decrease of the Se–C bond order. On the other hand, the NOCV-CD results demonstrate that the Se → Au σ donation increases in the presence of ClO4–, which decreases the electron density at selenium (a deshielding effect can be predicted). The two things are coherent with each other, as a more negatively charged selenium tends to be a better σ donor, but it is impossible to say a priori which parameter has the larger effect on the 77Se NMR chemical shift. As the experimental result is a small shielding effect (Δδ = −1 ppm), the former likely prevails on the latter. In the case of PF6–, the polarization is certainly less important, the Se → Au σ donation decreases (probably for polarization-induced effects), and the Au → Se π back-donation increases. The experimental result is a positive Δδ (+2 ppm). In order to better investigate this relationship, we recognize that more anions and cations should be investigated, but the strategy presented here is an adequate starting point and provides the general framework in which the future results can be discussed.

A more detailed comparison between the experimental and computational data is hampered by the fact that the correlation of an experimentally accessible parameter with a single DCD component is not easy, even if not impossible.[73−75] For this reason, new molecular systems are currently under study with the aim to better characterize the interplay between covalent and noncovalent interactions.

Conclusions

Here, we present a combined experimental/theoretical study on the relationship between a second-sphere HB and the DCD bond component for a gold(I) complex. The experimental results, obtained by 1H or diffusional NMR techniques, quantify the polarization of selenourea upon the coordination on the metal center and reveals that the selenourea affinity for the anion approximately doubles up in terms of association constant. The absolute values of K clearly depend on the basicity of the anion, but the effect is noticeable using either the noncoordinating PF6– or the more basic ClO4–. The use of more basic or coordinating anions has been prevented by the acidity of the selenourea protons which establish an acid–base equilibrium with the anion, which makes the corresponding NMR peak broaden and disappear. The 77Se NMR spectroscopy revealed that different anions induce a different sign for Δδ, positive for PF6– and negative for ClO4–.

From the theoretical point of view, the NOCV-CD analysis provided a quantitative analysis of the Au–Se and Au–C coordinative bonds, in the presence and absence of the anion, revealing that the latter influences the DCD bond components in many ways: it modifies the conformation of the ligands, in this case favoring the donation from the π orbitals of the Se–C bond to the metal and it polarizes selenourea, making selenium a better σ donor and, consequently, tends to decrease the Au → carbene donation and increase the back-donation.

Considering that the Au → carbene back-donation is poorly modulated by structural variations of the carbene,[26] the results presented here assume a unique relevance and pave the way for a different way of thinking. The use of external additives to modulate the DCD components can not only be as effective as the time-consuming, expensive “ad hoc synthesis” but also much easier and immediate.

Our laboratory is currently testing the potential of this strategy in catalysis and the results will be reported in due time.

Experimental Section

Materials and Instrumentation

Unless otherwise noted, all reagents were purchased from commercial sources and used without further purification. 1PF6 has been synthesized using the literature method.[57] Tetraalkylammonium salts were purchased, stored in an essicator, and used as received or synthesized by reaction of tetraalkylammonium hydroxide and the acid of the desired anion.

1H, 77Se and 13C NMR spectra were recorded with a Bruker “AVANCE DRX400” spectrometer equipped with a BBFO broadband probe. Chemical shifts were measured in ppm (δ) from TMS by residual solvent peaks for 1H and 13C.

PGSE NMR. 1H diffusion NMR measurements were performed by using the double-stimulated echo sequence with longitudinal eddy current delay at 298 K without spinning.[76] The dependence of the resonance intensity (I) on a constant waiting time and on a varied gradient strength G is described by the following equationwhere I is the intensity of the observed spin echo, I0 is the intensity of the spin echo in the absence of gradient, Dt is the self-diffusion coefficient, Δ is the delay between the midpoints of the gradients (0.2 s), δ is the length of the gradient pulse (4 ms), and γ is the magnetogyric ratio. The shape of the gradients was rectangular, and their strength G was varied during the experiments.

When necessary, the experimental data have been fitted through a two-component equationwhere Dt,1 and Dt,2 are the diffusion coefficient of the two species in mutual exchange and a1 and a2 are the two corresponding weighing factors.

The self-diffusion coefficient, Dt, was estimated by evaluating the proportionality constant for a sample of HDO (5%) in D2O (known diffusion coefficients in the range 274–318 K[77]) under the exact same conditions as the samples of interest. The solvent was taken as an internal standard. The hydrodynamic volume of the species has been calculated from the experimental value of Dt through the procedure previously described.[70]

Computational Details

Geometry Optimizations

If not otherwise specified, all the geometries were optimized with ORCA 4.0.1.2,[78,79] using the BP86 functional[80] in conjunction with a triple-ζ quality basis set (def2-TZVP). GGA functional demonstrated to be reliable for geometry optimizations,[58,81] whereas a single-point calculation using the double hybrid B2-PLYP has been used for the computation of the energies.[58] The dispersion corrections were taken into account during optimization cycles using the Grimme D3-parametrized correction with Becke–Jonhson damping to the DFT energy.[82] All the structures were confirmed to be local energy minima (no imaginary frequencies).

Energy Decomposition Analysis

The EDA[72] allows the decomposition of the bond energy into physically meaningful contributions. The interaction energy (Eint) is the difference of energy between the adduct and the unrelaxed fragments. It can be divided into contributions associated with the orbital, steric, and dispersion interactions, as shown in eq (72)Est is usually called the steric interaction energy and it is the sum of Eelst, the classical electrostatic interaction between the unperturbed charge distributions of the fragments (ρA and ρB) at their final positions in the adduct, and the Pauli repulsion (EPauli) that is the energy change associated with going from ρA + ρB to the antisymmetrized and renormalized wave function. The decomposition of Est is not possible with ORCA 4.0.1.2. Est comprises the destabilizing interactions between the occupied orbitals and is responsible for any steric repulsion. Eoi is the contribution arising from allowing the wave function to relax to the fully converged one, accounting for electron-pair bonding, charge transfer, and polarization, whereas Edisp is the contribution of the dispersion forces.

Charge Displacement Function Analysis

The CD function analysis is based on eq .[31] Δρ′(x,y,z) is the difference between the electron density of a complex and that of its noninteracting fragments placed in the same position as they occupy in the complex. In the present case, the fragmentation depends on the interaction under examination and is generally indicated in each case. The function Δq(z) defines, at each point z along a chosen axis, the amount of electron charge that, upon formation of the bond between the fragments, moves across a plane perpendicular to the axis through the point z. A positive (negative) value corresponds to electrons flowing in the direction of decreasing (increasing) z. Charge accumulates where the slope of Δq is positive and decreases where it is negative.

Because in many cases there is no molecular symmetry in the systems considered, we make use of the (NOCV):[60,83] Δρ′ is built from the occupied orbitals of A and B, suitably orthogonalized to each other and renormalized (promolecule), using the “valence operator” (eq ),[84−86]where ψi0 is the set of the occupied Kohn–Sham orbitals of fragments A and B, mutually orthonormalized, and ψi(AB) is the set of occupied orbitals of the adduct. The NOCVs can be grouped in pairs of complementary orbitals (φ, φ–) corresponding to eigenvalues with the same absolute value but opposite sign (eq ).where k numbers the NOCV pairs (k = 0 for the largest value of |ν|). In this framework, Δρ′ can be defined as in eq .

For each value of k, an energy contribution associated with the k-th NOCV pair is given. The most important ones are listed in the Supporting Information

Now the different Δρ′ can be separately integrated using eq .[33]

The NOCV orbital cubes have been manipulated through the software “Gabedit”,[87] whereas the electronic density matrices have been manipulated through the suite of tools “Cubes”.[88,89]

NBO analysis has been performed using the NBO6 suite of software.[90]