1 Introduction

The properties and reactions of molecular metal complexes are highly dependent on the ligand and the metal [1,2,3]. For instance, early transition metals (ETM) are more earth-abundant than late transition metals, such as Ru and Re. Therefore, their mediated reactions tend to be a better choice for sustainable transformations [4, 5]. Besides, the reactivity of ETMs differs from late metals, which opens new routes for a variety of novel reactions [6].

Among the early metal atoms (groups 3–7), scandium is seen as an environmentally benign Lewis acid, and it is compatible with Lewis bases and water [6,7,8]. For instance, some distinct scandium complexes have been explored in \(\hbox {CO}_2\) activations [6, 9,10,11]. In particular, the \(\hbox {CO}_{2}\,+ \hbox { M}^{+}\,\rightarrow \hbox { MO}^{+}\,+\) CO conversion, with \(\hbox {M}=\hbox {TM}\), to CO by metal ions is an important channel in gas-phase, and it has been accomplished by the \(\hbox {Sc}^{+}\) species [12,13,14,15]. The CO product can further be employed for introducing carbonyl functionalities into a plethora of molecular systems, and it is a key component in crucial industrial processes such as the Monsanto/Cativa acetic acid synthesis and the Fischer–Tropsch synthesis of hydrocarbons [16, 17]. Chemical reactions of \(\hbox {CO}_2\) that result in more reactive and value-added chemical species become a crucial way of reducing environmental impact [18,19,20,21,22,23,24]. However, due to experimental complications of working with electropositive and small elements, such as scandium ions, the study of \(\hbox {CO}_2\) reactions mediated by their complexes is fairly recent compared to the other ETM systems [6].

To harness the high reactivity of low-valent ETMs toward the activation of small molecules is of great importance [25,26,27,28,29,30]. Although not usual, stable low-valent scandium complexes, such as the \(\hbox {Sc}^{+}\) complexes, \([\{\eta ^5\hbox {-}\hbox {P}_3 \hbox {C}_2 \hbox {tBu}_2)\hbox {Sc}\}_2(\mu \hbox {-}\eta ^{6}:\eta ^{6}\hbox {-}\hbox {P}_3\hbox {C}_3\hbox {tBu}_3)]\) [31],\([\hbox {Sc}(\eta ^{5}\hbox {-}\hbox {P}_3\hbox {C}_2\hbox {tBu}_2)(\mu \hbox {-}\eta ^2:\eta ^5\hbox {-}\hbox {P}_3\hbox {C}_2\hbox {tBu}_2)\hbox {Sc}(\eta ^5\hbox {-}\hbox {P}_3\hbox {C}_2\hbox {tBu}_2)]\) [32] and \((\hbox {LMgBr})_2\hbox {ScBr}\,(\hbox {L}=\hbox {Et}_2\hbox {NCH}_2\hbox {CH}_2\hbox {NC(Me)CHC(Me)}\hbox {NCH}_2\hbox {CH}_2\hbox {NEt}_2)\) [33], have been reported. Others examples of \(\hbox {Sc}^{0}\) complex and \(\hbox {Sc}^{2+}\) complexes [34, 35] are also found. For instance, the \({\hbox {Sc}[\hbox {N}(\hbox {SiMe}_3)_2]_3}^{-}\) was described to react with the \(\hbox {CO}_2\) molecule [11].

Neutral N-heterocyclic olefins (NHOs), in special, have played a growing role as ligands in transition metal catalysis, mainly because of their pronounced ylidic character [36,37,38,39,40,41,42]. Hence, it is expected that NHOs provide extra stabilization to low-valent metal centers such as \(\hbox {Sc}^{+}\) [43,44,45]. Although other types of N-heterocyclic scandium complexes have been described [6, 9, 10, 46], at the best of our knowledge, none has explored the potential of NHO-scandium (I) compounds toward the \(\hbox {CO}_2\) to CO reaction. In particular, the direct influence of NHO ligands on the \(\hbox {CO}_2\) to CO conversion mediated by the \(\hbox {Sc}^{+}\). A study that allows us to investigate the sole importance of such ligands, simulating a controlled environment cleared with perturbing factors, provides a direct insight into the molecular or ionic properties of chemically interesting species. Therefore, it offers an ideal way to shed light on mechanistic details of corresponding and more complex condensed-phase systems, i.e., it might reveal the potential of Sc (I) related molecular complexes and their use in \(\hbox {CO}_2\) reactions [47,48,49,50].

In this context, the present work explores, by means of computational methods, the potential of NHOs as ligand for the scandium(I) mediated reduction of \(\hbox {CO}_2\) to CO. Besides, as their structure offers elevated degree of electronic control through the choice of different substituents (R) [2, 39, 43,44,45, 51,52,53,54], we also explore this effect. Figure 1 presents the R groups attached to the current scandium NHOs systems, as well as the nomenclature adopted along with this work.

Scheme 1
scheme 1

a Parent NHO ligand. b General structure of the NHO ligand. c General N-heterocyclic olefin scandium(I) complex. d Different substituents (R) explored in the \(\hbox {CO}_2\) to CO reduction mediated by different NHO-Sc(I) complexes

Full size image

2 Computational methods

All electronic structure calculations were performed using the Gaussian 09 suite of quantum chemical programs (Revision D0.1) [55]. Geometry optimizations and frequency calculations of all chemical species were performed using the DFT/M06 method [56,57,58] along with the def2-TZVP basis set [59]. This level of theory is referred as DFT/M06/def2-TZVP along with the work. The M06 density functional showed to be an efficient approach to explore scandium-containing systems [60]. Besides, among other tested density functional methods, it was also able to better reproduce the experimental relative energy of \(^3 \hbox {Sc}^{+}\) and \(^1 \hbox {Sc}^{+}\) [61, 62]. See the Electronic Supplementary Material for more details. The nature of all structures, as minima or transition states (TSs), was confirmed by the vibrational analysis performed within the harmonic approximation at 298 K and 1 atm. The TSs were identified by only one imaginary frequency mode, which was further followed in order to check the connectivity between the transition state and its corresponding minima.

To analyze the electronic structure and stabilizing effects on the stationary points of the explored potential energy surface (PES), the intrinsic bond orbital (IBO) analyses were carried out using the DFT/PBE/def2-TZVP [59, 63, 64] level of theory, with the univ-JFIT [63] fitting basis set, as implemented in the IboView program [65,66,67]. Because this program does not have the M06 functional available, the PBE functional was chosen. However, in a previous study [30], we did not find a significant difference between the IBOs computed using PBE and M06 density functionals.

3 Results and discussion

This section is organized as follows: first, a detailed comparison between the \(^{1,3}(\hbox {Sc}^{+}) + \hbox {CO}_2\) PESs is presented. Next, we use the parent NHO system (Scheme 1a) to explore the NHO ligand effect on this conversion. Finally, the influence of selected electron-withdrawing (–\(\hbox {CF}_3\) and –CN) and electron-donating substituents (–OMe and –tBu) on the NHO system is also investigated.

3.1 \(\hbox {CO}_2\) to CO mechanism mediated by \(\hbox {Sc}^{+}\)

As the main point of the current work is to evaluate the influence of the NHO ligand on scandium (I) mediated \(\hbox {CO}_2\) to CO reaction, an essential and prior step consists of the analysis of the isolated \(\hbox {Sc}^{+}\) ion mediated conversion. Figure 1 shows the energy profile for this reaction on the singlet (S) and triplet (T) potential energy surfaces.

Fig. 1
figure 1

Singlet (red) and triplet (black) relative energy profiles of the \(\hbox {CO}_2\) to CO mechanism mediated by the \(\hbox {Sc}^{+}\) ion. All energies are given in \(\hbox {kcal mol}^{-1}\) with respect to the separated reactants in their singlet electronic state. Relative energies that do not and do include zero-point energy (ZPE) corrections are written in regular and bold, respectively. Bond distances are in Å

Full size image

First of all, we note that the inclusion of the zero-point energy (ZPE) does not change the overall trend along with the reaction. Therefore, in the discussion that follows, we retain ourselves to the electronic energy only. Although the \(^1 \hbox {Sc}^{+}\) ion is more energetic than its respective triplet state, the singlet PES offers a way to a more feasible \(\hbox {CO}_2\) to CO reduction mechanism. For instance, at the DFT/M06/def2-TZVP level of theory, the \(^1\hbox {ScO}^{+}\,+\) CO channel lies at \(-40.6\hbox { kcal mol}^{-1}\), while the \(^3\hbox {ScO}^{+}\,+\) CO one is at \(50.7\hbox { kcal mol}^{-1}\). This result is in agreement with previous studies on the gas-phase \(\hbox {CO}_2 \rightarrow \hbox { CO}\) conversion mediated by the \(\hbox {Sc}^{+}\) ion [12, 13].

The reaction initiates with the formation of the linear \(^{1,3}[\hbox {ScOCO}]^+\) form, with the triplet structure only \(1.6\hbox { kcal mol}^{-1}\) more stable than the singlet one. The singlet and triplet transition states, \(^{1,3}\)[TS1], that yield to the \(^{1,3}[\hbox {OScCO}]^{+}\) systems, are \(-20.9\) and 13.4 \(\hbox {kcal mol}^{-1}\), respectively, in the explored energy profiles. Their respective imaginary frequencies are \(-329.5\) and \(-370.9\hbox { cm}^{-1}\), respectively. The energy required to \(^{1}[\hbox {ScOCO}]^{+}\) to convert into \(^{1}\)[OScCO]\(^+\) is only 6.1 \(\hbox {kcal mol}^{-1}\), while the respective conversion in the triplet PES requires \(42.0\hbox { kcal mol}^{-1}\). The singlet \([\hbox {OScCO}]^{+}\) structure is over \(30\hbox { kcal mol}^{-1}\) more stable than its related linear one. In the singlet [OScCO]\(^+\) intermediate, the ScC bond distance is 2.47 Å, comparatively to 2.27 Å on the triplet \([\hbox {OScCO}]^{+}\) structure. Therefore, the energy required to dissociate the singlet system into \(\hbox {ScO}^{+}\,+\) CO is lower, as viewed in Fig. 1.

Since the singlet PES is the most feasible way to convert \(\hbox {CO}_2\) to CO using \(\hbox {Sc}^{+}\), all the remaining discussion that follows focus on this multiplicity.

3.2 \(\hbox {CO}_2\) conversion to CO by the parent NHO-Sc(I) complex

Figure 2 shows the minimum energy path (MEP) for the \(\hbox {CO}_2\) to CO reduction mediated by the parent NHO-Sc (I) system. The main point here is to evaluate whether the NHO system has the potential to be used as ligand in larger Sc(I) systems; for instance, if NHO could be a potential ligand in Sc complexes for \(\hbox {CO}_2\) conversions.

As revealed by the energetics of the reaction, the described NHO-Sc (I) system is able to convert \(\hbox {CO}_2\) to CO in an exoergic way. The \([(\hbox {Me})_2\hbox {NHO-ScO}]^{+}\,+\) CO channel lies at \(-30.0\)\(\hbox {kcal mol}^{-1}\), i.e., \(10.6\hbox { kcal mol}^{-1}\) above the corresponding ScO\(^{+}\,+\) CO one (Fig. 2, gray line). Therefore, this result raises the potential of NHO systems as ligands for Sc complexes to be employed on the \(\hbox {CO}_2\) to CO reduction. In addition, both intermediates, the linear \([(\hbox {Me})_{2}\hbox {NHO-ScOCO})]^+\) and the oxidative product \([(\hbox {Me})_2\hbox {NHO-OScCO}]^{+}\) forms, are stable ones (Fig. 2). In the \([(\hbox {Me})_2\hbox {NHO-ScOCO}]^{+}\) system, the \(\hbox {Sc-O}^2\) bond distance is 0.13 Å larger than the corresponding bond in \(^1[\hbox {ScOCO}]^+\), which could help to explain the better stabilization of this latter. The transition state \([(\hbox {Me})_2\hbox {NHO-TS1}]^+\) (see Fig. 2) that leads to the \([(\hbox {Me})_2\hbox {NHO-OScCO}]^{+}\) intermediate has an imaginary mode of \(-113.1\hbox { cm}^{-1}\) and it is \(-13.3\hbox { kcal mol}^{-1}\) on this potential energy surface. Moreover, since its associated activation energy is only 1.9 \(\hbox {kcal mol}^{-1}\), the process can occur at room temperature.

Fig. 2
figure 2

MEP of the \(\hbox {CO}_2\) to CO reduction mediated by the parent NHO, the \([(\hbox {Me})_2\hbox {NHO-Sc}]^+\) system (red line). The corresponding mechanism mediated by the single \(\hbox {Sc}^{+}\) ion is also shown for comparison (gray line and gray values). All energies are given in \(\hbox {kcal mol}^{-1}\) with respect to the separated reactants. Relative energies that do not and do include zero-point energy (ZPE) corrections are written in regular and bold, respectively

Full size image

To add a quantitative interpretation to the bond breaking/forming processes and stabilizing effects along with the mechanism, the main active IBOs along with the reaction are also viewed, Fig. 3. IBOs consist of a set of occupied molecular orbitals, which form an exact representation of a given Kohn–Sham wave function.

Fig. 3
figure 3

Relevant IBOs along the reaction mechanism mediated by a the \(\hbox {Sc}^{+}\) ion and b the parent \([(\hbox {Me})_2\hbox {NHO-Sc}]^+\) complex. Each IBO has a different color. The numbers after the colon are the fraction of electrons assigned to individual atoms of each IBO. The fraction of electrons adds up to two. Some IBOs are plotted together just for convenience

Full size image

The analysis over the transition states structures reveals that all five relevant IBOs have contributions upon the formation of the \(\hbox {Sc-C}^{1}\) bond on the \(\hbox {Sc}^{+}\) ion mediated mechanism (Fig. 3a), while in the \([(\hbox {Me})_2\hbox {NHO-Sc}]^+\) one, this observation holds true for three of them, IBOs (3), (4) and (5) (Fig. 3b). This result sheds light on the extra stabilization seen on the [TS1]\(^+\) comparatively to the \([(\hbox {Me})_2\hbox {NHO-TS1}]^+\). The major contributor to this forming process is the IBO (5). IBO (5) starts centered at scandium—it has a d character essentially—on the \([(\hbox {Me})_2\hbox {NHO-ScOCO}]^{+}\) and ScOCO\(^+\) intermediates, and it spreads through the Sc, \(\hbox {C}^{1}\), \(\hbox {O}^{2}\) and \(\hbox {O}^{3}\) atoms on their respective transition states. For more details on the contributions of each atom, see Fig. 3. On the [OScCO]\(^+\) and \([(\hbox {Me})_2\hbox {NHO-OScCO}]^{+}\) intermediates, this IBO remains basically over Sc and O\(^2\) to form a \(\pi\) bond character between them. In addition to this \(\pi\) bond, the Sc and O\(^2\) interaction on the oxidative addition intermediates has great contributions from the IBOs (3) and (4), which show a \(\pi\) and \(\sigma\) character, respectively.

3.3 Substituents effects on the \(\hbox {CO}_2\) conversion

Next, we explore the effect of selected electron-withdrawing (–\(\hbox {CF}_3\) and –CN) and electron-donating substituents (–OMe and –tBu) on the \(\hbox {CO}_2\,\rightarrow\) CO reduction mediated by the NHO-Sc (I) systems. Figures 4 and 5 present all the corresponding relative energy profiles. The general [(R)\(_2\hbox {NHO-Sc}]^+\) structures are seen in Scheme 1. The optimized structures of the stationary points discussed below are presented in Electronic Supplementary Material.

Clearly, the \(\hbox {CO}_2\) reduction to CO is a feasible pathway using the NHO-Sc systems, regardless the substituent group. As seen in Figs. 4 and 5, all reactions are exoergic. The linear intermediates, \([(\hbox {tBu})_2\hbox {NHO-ScOCO}]^{+}\), \([(\hbox {OMe})_2\hbox {NHO-ScOCO}]^{+}\), \([(\hbox {CF}_3)_2\hbox {NHO-ScOCO}]^{+}\) and \([(\hbox {CN})_2\hbox {NHO-ScOCO}]^{+}\) are well stabilized, \(-13.7\), \(-19.5\), \(-17.0\) and \(-18.7\hbox { kcal mol}^{-1}\), respectively, in the explored PES. The –tBu substituent provides a \([(\hbox {tBu})_2\hbox {NHO-Sc}]^+\) mediated mechanism energetically similar to its –\(\hbox {CH}_3\) counterpart (Fig. 4a, gray line). This is a reasonable behavior since both are alkyl groups; for instance, the transition states \([(\hbox {tBu})_2\hbox {NHO-TS1}]^+\) and \([(\hbox {Me})_2\hbox {NHO-TS1}]^+\) are very close in energy, as well as the intermediates \([(\hbox {tBu})_2\hbox {NHO-OScCO}]^{+}\) and \([(\hbox {Me})_2\hbox {NHO-OScCO}]^{+}\). As the electron-donating substituent type is changed, the –OMe group in our case (Fig. 5b), the stabilization along the mediated process is more evident. The transition state \([(\hbox {OMe})_2\hbox {NHO-TS1}]^+\) lies at \(-17.8\hbox { kcal mol}^{-1}\) in the current profile, while the oxidative addition intermediate \([(\hbox {OMe})_2\hbox {NHO-OScCO}]^{+}\) is at \(-50.9\hbox { kcal mol}^{-1}\). The imaginary frequencies associated with the \([(\hbox {tBu})_2\hbox {NHO-TS1}]^+\) and \([(\hbox {OMe})_2\hbox {NHO-TS1}]^+\) transition states are relatively low, \(-44.7\) and \(-66.8\hbox { cm}^{-1}\), respectively, as well as their respective activation barriers, 1.0 and 1.7 \(\hbox {kcal mol}^{-1}\).

Fig. 4
figure 4

Relative energy profiles of the \(\hbox {CO}_2\) to CO mechanism mediated by different NHO-\(\hbox {Sc}^{+}\) complexes: a\([(\hbox {tBu})_2\hbox {NHO-Sc}]^+\) and b\([(\hbox {OMe})_2\hbox {NHO-Sc}]^+\). All energies are given in \(\hbox {kcal mol}^{-1}\) with respect to the separated reactants. For comparison, the corresponding mechanism mediated by the parent \([(\hbox {Me})_2\hbox {NHO-Sc}]^+\) is also indicated by the gray line and the gray values. Relative energies that do not include and account zero-point energy (ZPE) corrections are written in regular and bold, respectively

Full size image

Within the electron-withdrawing (–\(\hbox {CF}_3\) and –CN) class (Fig. 5), although the transition states \([(\hbox {CF}_3)_2\hbox {NHO-TS1}]^+\) and \([(\hbox {CN})_2\hbox {NHO-TS1}]^+\) have similar energies along their respective mechanism, the –CN substituent offers the largest stabilization to the oxidative addition products, \(-45.5\hbox { kcal mol}^{-1}\) and \(-53.4\hbox { kcal mol}^{-1}\), respectively, for \([(\hbox {CF}_3)_2\hbox {NHO-OScCO}]^{+}\) and \([(\hbox {CN})_2\hbox {NHO-OScCO}]^{+}\). The transition states are characterized by an imaginary frequency of \(-117.4\hbox { cm}^{-1}\) for \([(\hbox {CF}_3)_2\hbox {NHO-TS1}]^+\) and, \(-165.9\hbox { cm}^{-1}\) for \([(\hbox {CN})_2\hbox {NHO-TS1}]^+\); in addition, their activation energies are \(-2.0\) and \(-3.1\hbox { kcal mol}^{-1}\), respectively. Overall, the electron-withdrawing substituents also make the respective oxidative addition intermediates require less energy to release the CO molecule; for instance, \([(\hbox {CF}_3)_2\hbox {NHO-OScCO}]^{+}\) and \([(\hbox {CN})_2\hbox {NHO-OScCO}]^{+}\) require \(7.6\hbox { kcal mol}^{-1}\) and \(12.6\hbox { kcal mol}^{-1}\), respectively, while the \([(\hbox {OMe})_2\hbox {NHO-OScCO}]^{+}\) needs \(14.9\hbox { kcal mol}^{-1}\).

Fig. 5
figure 5

Relative energy profiles of the \(\hbox {CO}_2\) to CO mechanism mediated by different NHO-\(\hbox {Sc}^{+}\) complexes: a\([(\hbox {CF}_3)_2\hbox {NHO-Sc}]^+\) and b\([(\hbox {CN})_2\hbox {NHO-Sc}]^+\). All energies are given in \(\hbox {kcal mol}^{-1}\) with respect to the separated reactants. For comparison, the corresponding mechanism mediated by the parent \([(\hbox {Me})_2\hbox {NHO-Sc}]^+\) is also indicated by the gray line and the gray values. Relative energies that do not include and account zero-point energy (ZPE) corrections are written in regular and bold, respectively

Full size image

4 Conclusions

In the present work, DFT computations were applied to explore the potential of NHOs as a ligand for scandium(I) mediated reduction of \(\hbox {CO}_2\) to CO. The \(\hbox {CO}_2\) to CO mechanism mediated by the \(^1\hbox {Sc}^{+}\) ion was found to be the most feasible way for such conversion. Based on that, the potential of different \(^1\)[NHOs-Sc]\(^+\) mediators was also investigated.

As revealed by the energetics of the reaction, the parent NHO-Sc (I) system, the \([(\hbox {Me})_2\hbox {NHO-Sc}]^+\) complex, was also able to convert \(\hbox {CO}_2\) to CO in an exoergic way. Therefore, it points out the potential of NHOs ligands for scandium complexes in the \(\hbox {CO}_2\) to CO reduction. Besides, the effect of selected electron-withdrawing (–\(\hbox {CF}_3\) and –CN) and electron-donating (–OMe and –tBu) substituents on the \(\hbox {CO}_2\,\rightarrow\) CO reduction mediated by the NHOs-Sc (I) systems was also investigated. Regardless of the substituent group, the \(\hbox {CO}_2\) reduction to CO was seen as a viable process by using these NHOs-Sc systems. Among the selected substituents, the –CN group showed to be a better choice for releasing CO more easily. The results discussed along this work raise the potential of the newly NHOs-Sc complexes in \(\hbox {CO}_2\) transformations.