Cu 14 Cluster with Partial Cu(0) Character: Difference in Electronic Structure from Isostructural Silver Analog

Abstract

An atom‐precise Cu⁰‐containing copper cluster, Cu₁₄(C₂B₁₀H₁₀S₂)₆(CH₃CN)₈ (abbreviated as Cu₁₄‐8CH₃CN) is reported, which is synthesized via a simultaneous reduction strategy and fully characterized by single‐crystal X‐ray diffraction, ESI‐TOF‐MS, and X‐ray photoelectron spectroscopy. Cu₁₄‐8CH₃CN is the only copper cluster that has a virtually identical silver structural analog, i.e., Ag₁₄(C₂B₁₀H₁₀S₂)₆(CH₃CN)₈ (hereafter as Ag₁₄‐8CH₃CN). Nevertheless, density functional theory calculations reveal that the electronic structure of Cu₁₄‐8CH₃CN differs significantly from the superatom electronic configuration of Ag₁₄‐8CH₃CN. Moreover, Cu₁₄‐8CH₃CN shows room‐temperature luminescence and good electrocatalytic activities in the ethanol oxidation reaction and detection of H₂O₂. This pair of unprecedented analogous molecular nanoscale systems offer an ideal platform to investigate the fundamental differences between copper and silver in terms of catalytic activity and optical properties.

Atomically precise noble metal nanoclusters (NCs) are a relatively new and emerging class of materials linking atoms and nanoparticles that have gained increasing interest due to their aesthetically pleasing molecular structures and potential applications in nanodevices, catalysis, medicine, and imaging. To date, gold and silver NCs of various sizes are well documented, whereas comparable copper NCs remain scarce. In particular, copper NCs with metallic character are significantly rare probably due to the lower M^I/M⁰ half‐cell potential of Cu (0.52 V) versus those of Ag (0.80 V) and Au (1.68 V). To the best of our knowledge, only six such copper NCs, namely [Cu₂₀(CCPh)₁₂(OAc)₆)], [Cu₂₅H₂₂(PPh₃)₁₂]Cl, [Cu₂₉Cl₄H₂₂(Ph₂phen)₁₂]Cl (Ph₂phen = 4,7‐diphenyl‐1,10‐phenanthroline), [Cu₄₃Al₁₂](Cp*)₁₂ (Cp* = η⁵‐C₅Me₅), [Cu₅₃(CF₃COO)₁₀(CC^tBu)₂₀Cl₂H₁₈]⁺, and [Cu₁₃(S₂CNⁿBu₂)₆(CCR)₄](PF₆) (R = C(O)OMe, C₆H₄F) have been reported. Notably, most of them involved H⁻, which is believed to originate from the reductant NaBH₄ or Ph₂SiH₂; the Cu₂₅, Cu₂₉, and Cu₅₃ clusters contain coordinated H⁻, while the Cu₁₃ cluster was prepared from Cu hydride cluster precursor. It is well known that the use of excess reducing agents continually hinders the preparation of monodisperse products and makes their isolation more complex. The presence of H⁻ also usually causes problems regarding its precise location and decreases the stability of the NCs. Thus a synthetic method for copper NCs containing Cu⁰ that avoids the use of H⁻ is of vital consideration.

Despite the kinship of the coinage triad, Au, Ag, and Cu exhibit significantly different nanocluster assembly behavior due to their inherent chemical characteristics. Therefore, analogous nanoclusters of these metals are extremely difficult to prepare even under exactly the same synthetic conditions, although isostructural Au, Ag, and Cu coordination complexes are known in other low‐nuclearity systems, such as M₃(Pz)₃ (Pz = pyrazole) and M₄(SR)₄. Very recently, the Bakr group reported [Ag₂₅(SR)₁₈]⁻, the only silver nanoparticle identically analogous to [Au₂₅(SR)₁₈]⁻ in terms of the number of metal atoms, ligand count, superatom electronic configuration, and atomic arrangement, which provided the first model nanoparticle platform for understanding the fundamental differences between silver and gold in terms of nobility, catalytic activity, and optical properties. However, a detailed study of Cu nanoclusters in this regards remains unavailable.

Reducing‐cum‐protecting agents could be desirable in mixed‐valence cluster assembly. 1,2‐dithiolate‐o‐carborane has exhibited intense reducing capability and synchronously converted Ag^I to Ag⁰ during cluster formation. Herein we present the one‐pot self‐reduction synthesis and crystal structure of a fcc‐Cu₁₄ cluster with Cu⁰ character, Cu₁₄(C₂B₁₀H₁₀S₂)₆(CH₃CN)₈ (Cu₁₄‐8CH₃CN), an analog of the previously reported (Ag₁₄‐8CH₃CN). This cluster is the first copper nanoparticle with an exact silver analog in terms of size, composition, charge and crystal structure, and thus it provides a platform for direct comparison of the properties of copper and silver nanoparticles by both experimental and theoretical methods. Furthermore, the coordinated CH₃CN molecules on the cluster's surface could be site‐specifically replaced by 4‐dimethylamino‐benzonitrile (DMABN), affording [Cu₁₄(C₂B₁₀H₁₀S₂)₆(DMABN)₈]·2.5THF (Cu₁₄‐8DMABN). This finding further supports the universality of site‐specific surface modification strategy of metal clusters as proposed by us. Besides, Cu₁₄‐8CH₃CN shows good electrocatalytic activities in ethanol oxidation reaction and detection of H₂O₂.

The synthesis of Cu₁₄‐8CH₃CN cluster involves the reaction of Cu(CF₃COO)₂ with 1,2‐dithiol‐o‐carborane in CH₃CN‐THF (v/v = 1:1) at room temperature (Figure 1; see the Supporting Information for details). Fast self‐reduction of Cu²⁺ to Cu⁺/Cu⁰ can be clearly visualized by color fading of the CH₃CN solution of Cu(CF₃COO)₂ upon dropwise addition of 1,2‐dithiol‐o‐carborane in THF (Video S1, Supporting Information), which is rare in one‐step copper cluster synthesis. Block crystals suitable for single‐crystal X‐ray diffraction (SCXRD) analysis were obtained several days later (Figure S1, Supporting Information). To gain an insight into the possible redox reaction during the formation of the cluster, we analyzed the synthetic reaction solution of Cu₁₄‐8CH₃CN by ESI‐TOF‐MS, where a deboronated [(C₂B₁₀H₁₀S₂)(C₂B₉H₁₀S₂)]⁻ coupling species with disulfide bonds (Figure S2, Supporting Information) was found in the mass spectrum. Moreover, a byproduct including the disulfide species [Cu(CH₃CN)₄][(C₂B₁₀H₁₀S₂)(C₂B₉H₁₀S₂)] (abbreviated as Cu‐Disulfide) was fortunately isolated and well‐characterized by single‐crystal X‐ray structure analysis (Figure S3, Supporting Information). The dissociated B atoms from the carborane were probably converted to borates, as evidenced by the ¹¹B NMR spectrum of the reaction solution (Figure S4, Supporting Information). These observations indicate that the copper ions could be reduced by the thiol ligands as has been shown in the syntheses of Au and Ag thiolate nanoclusters or nanoparticles, and/or by the dissociated B atom, and form the resultant Cu(0)‐containing clusters.

Synthesis and structural dissection of desired Cu14‐8CH3CN cluster. a) Schematic representation of one‐pot synthesis. b–d) Cu coordination spheres and bonding environment of S atoms. e,f) Cu64+ core and Cu88+ shell of fcc‐Cu14 framework. Color codes: brown and pink = copper; yellow = sulfur; gray = carbon; blue = nitrogen; turquoise = boron.

The as‐synthesized Cu₁₄‐8CH₃CN was characterized by its ¹H NMR, ¹¹B NMR spectra, and energy dispersive X‐ray Spectroscopy (EDS) (Figures S5–S7, Supporting Information). The phase purity of as‐synthesized Cu₁₄‐8CH₃CN was confirmed by in situ powder X‐ray diffraction (PXRD) patterns recorded with milled crystals in the mother liquid (Figure S8, Supporting Information). However, the crystals decomposed very quickly once isolated from the solution, and the PXRD pattern of a dried sample displayed poor crystallinity. However, the retained red emission of the dry sample indicates that the framework of the cluster is robust, and the loss of crystallinity is probably caused by volatility of the lattice or coordinated solvent molecules. This assumption has been confirmed by the recovered PXRD pattern of a long‐term stored (1 month) dry sample soaked in a THF/CH₃CN mixture (Figure S8, Supporting Information). The stability of Cu₁₄‐8CH₃CN in solution was further studied by monitoring the complex in a CH₂Cl₂/CH₃OH (v/v = 1:1) solution with UV–vis spectroscopy at room temperature. The UV–vis spectra remained essentially unchanged over 24 h, indicating its high stability (Figure 2b). High‐resolution ESI‐TOF‐MS was also conducted to confirm the chemical formula and verify the stability. The excellent match of the experimental and simulated isotope patterns illustrated that the peaks at ≈2169.87, 2210.89, and 2251.91 Da correspond to the [M‐7CH₃CN+H]⁺, [M‐6CH₃CN+H]⁺, and [M‐5CH₃CN+H]⁺ species, respectively, which also indicates easy dissociation of the coordinated CH₃CN molecules (Figure a). The exposed Cu atoms generated by surface coordination bond cleavage might render this compound catalytically active in some reactions. The Cu Auger lines at 1848.55 and 1850.9 eV, obtained from the X‐ray photoelectron spectroscopy (XPS) spectrum as the sum of the Cu 2P_3/2 binding energy and the LMM Auger kinetic energy, indicate the coexistence of Cu(I) and Cu(0) (Figure S9, Supporting Information).

a) Positive mode ESI‐TOF‐MS spectrum of Cu14‐8CH3CN. Inset: Enlarged portion of the spectrum showing the measured (black) and simulated (red) isotopic distribution patterns (M = Cu14(C2B10H10S2)6(CH3CN)8). b) Time‐dependent UV–vis spectra of Cu14‐8CH3CN in CH2Cl2/CH3OH solution (v/v = 1:1, 2.5 × 10−5 mol L−1).

SCXRD analysis revealed that Cu₁₄‐8CH₃CN crystallizes in cubic space group Fmm (Table S1, Supporting Information), being unlike Ag₁₄‐8CH₃CN that crystallizes in triclinic space group P. Cu₁₄‐8CH₃CN contains a discrete cubic Cu₁₄ core with six face‐capping bidentate 1,2‐dithiolate‐o‐carborane ligands, and eight vertex‐capping CH₃CN ligands (Figure ). The structure is identical to that of the Ag₁₄‐8CH₃CN homolog, but the fcc metal framework is more compact (the edge of the cube is 4.098 Å in Cu₁₄‐8CH₃CN, and average 4.458 Å in Ag₁₄‐8CH₃CN) (Figure e,f). Each Cu atom in the Cu₆⁴⁺ core adopts ten‐coordinate geometry, bonding to four adjacent Cu atoms at the octahedral vertexes with a Cu⋅⋅⋅Cu distance of 2.4926 Å, four Cu atoms from the corresponding face of the cubic Cu₈⁸⁺ shell (Cu⋅⋅⋅Cu separation of 2.912 Å), and two S atoms from one 1,2‐dithiolate‐o‐carborane ligand (Figure b). In the cubic Cu₈⁸⁺ shell, each Cu atom adopts a seven‐coordinate geometry, bonding to three S atoms from three different 1,2‐dithiolate‐o‐carborane ligands, three Cu atoms of the Cu₆⁴⁺ unit, and one CH₃CN molecule located at a cubic corner (Figure c). Each thiolate group acts as a µ₃‐bridge between one Cu atom from the Cu₆⁴⁺ core and two Cu atoms from the Cu₈⁸⁺ shell (Figure d), with CuS bond distances of 2.2564 and 2.367 Å.

The jelliumatic electron count of Cu₁₄‐8CH₃CN is 2 (n = 14 − 6 × 2), being the same with that of Ag₁₄‐8CH₃CN. To gain insight into the electronic structure of Cu₁₄‐8CH₃CN and compare it to that of its Ag₁₄‐8CH₃CN analog, density functional theory (DFT) calculations were carried out. The degenerate lowest unoccupied molecular orbitals (LUMO to LUMO+2) indicate strong superatomic P character over the entire Cu₁₄ cube, which is highly similar to that of Ag₁₄‐8CH₃CN; however, unlike the S‐symmetric HOMO state of Ag₁₄‐8CH₃CN, the jellium 1S orbital of Cu₁₄‐8CH₃CN shifted down as a low‐lying HOMO‐6 state over the Cu₆⁴⁺ kernel (Figure 3 and Figure S10, Supporting Information). These findings indicate that the silver and copper cluster analogs with identical composition, molecular structure, and valence electron count do not necessarily have the same electronic structures.

Locations of the superatomic states for Cu14‐8CH3CN and Ag14‐8CH3CN. Ag14‐8CH3CN has the character of the 1S‐symmetric superatomic state in the HOMO level and the 1P‐symmetric state in the LUMO to LUMO+2 levels. Cu14‐8CH3CN has the character of the 1S‐symmetric superatomic state in the HOMO‐6 level and the 1P‐symmetric state in its LUMO to LUMO+2 level.

We simulated the optical absorption of Cu₁₄‐8CH₃CN with time‐dependent density functional theory (TD‐DFT) to identify the origin of the optical transitions. Compared to the experimental molecular‐like absorption spectrum of a CH₂Cl₂‐CH₃OH solution of dry Cu₁₄‐8CH₃CN crystals, the simulated spectrum displays an analogous profile but had a significant blueshift (Figure 4a). As mentioned above, the coordinated CH₃CN molecules could dissociate readily. The dry sample probably lost part or all the coordinated CH₃CN molecules before being used for the UV–vis absorption test, which might account for the observed shift. To prove this, one drop of CH₃CN was added to the test solution, and the resulting electronic absorption showed an obvious blueshift just as surmised (Figure S11, Supporting Information). Therefore, we simulated the optical absorption of a Cu₁₄ model with all CH₃CN molecules omitted (Figure S12, Supporting Information). In contrast, the calculated spectrum exhibits a dramatic redshift with respect to the experimental spectrum, being accompanied by a slight difference in profile. To devise a compromise, we created a Cu₁₄‐4CH₃CN model bearing only four CH₃CN ligand molecules (Figure S13, Supporting Information). The simulated optical absorption spectrum matches the experimental spectrum much better, with a major band at 287 nm (HOMO‐20 to LUMO) and a weaker band at 323 nm (HOMO‐6 to LUMO+1) corresponding to the experimental peaks at 294 and 339 nm, respectively. The acceptable deviations might be due to the inaccurate CH₃CN number and positions. A comparison of the absorption spectra of Cu₁₄‐8CH₃CN and its Ag analog Ag₁₄‐8py¹⁴ was shown in Figure S14 (Supporting Information), as Ag₁₄‐8CH₃CN exhibits compromised stability. The obvious difference, which is in accordance to their electronic structures, is attributed to the different nature of the metals.

a) Normalized computed optical absorption spectra of Cu14‐8CH3CN (red dot line), Cu14‐4CH3CN (blue solid line) and Cu14 (green dot line) compared to the experimental data of Cu14‐8CH3CN (black solid line). Gray bars show the individual transitions (delta‐function‐like peaks showing the relative oscillator strengths) (Table S2, Supporting Information). The continuous computational spectra are sums of Gaussian smoothed individual transitions (width of 15 nm). b) Excitation (black trace) and emission (red trace) spectra of Cu14‐8CH3CN. Inset: photograph of Cu14‐8CH3CN in CH2Cl2 irradiated with 365 nm UV light at room temperature.

Although Cu complexes have been investigated as luminescent materials for decades,1h the reported Cu NCs with Cu(0) character are all nonemissive. Interestingly, Cu₁₄‐8CH₃CN emits bright red light upon UV irradiation in both solid and solution states at room temperature. The luminescence spectrum (λ_ex = 400 nm) displays a weakly structured broad emission band with two peaks at 637 and 661 nm (Figure b), the profile of which is akin to that of Ag₁₄‐8CH₃CN. The quantum yield of the emission at room temperature is 0.31. The microsecond‐scale emissive lifetime (τ_{298 K} = 5.13 µs) indicates spin‐forbidden triplet phosphorescence. Based on DFT and TDDFT calculations (Figure a and Figure S10, Supporting Information), the emission of Cu₁₄‐8CH₃CN might mainly originate from the excited state that arose from S‐type HOMO‐6 and the ligand‐based HOMOs to P‐type LUMOs transitions. Upon photoexcitation, the excited electrons into superatomic 1P orbitals would lead to some distortions of excited states, which might be related to the large Stokes shift and the hump‐like peaks in emission spectra.

The surface modification with pyridyl ligands used for Ag₁₄‐8CH₃CN were not applicable to Cu₁₄‐8CH₃CN, highlighting the difference of silver and copper cluster analogs. To address this problem, 4‐(dimethylamino)benzonitrile was used to replace the CH₃CN molecules, and eight DMABN molecules occupied all fcc corners as expected, giving surface‐modified Cu₁₄‐8DMABN (Figure S15, Supporting Information). The product was characterized by ESI‐MS, ¹H and ¹¹B NMR spectroscopy (Figures S16–S18, Supporting Information). The UV–vis absorption spectrum of Cu₁₄‐8DMABN in DMF shows a similar profile to that of Cu₁₄‐8CH₃CN with two bands at 295 and 345 nm, which indicates that surface modification seldom affects the electronic structure of the cluster (Figure S19, Supporting Information). The enhanced absorption at around 295 nm is probably due to overlap of the n → π* or π → π* transitions of the DMABN ligands. These achievements further demonstrate that site‐specific replacement of the coordinated solvent molecules in the ligand shell to modify a metal cluster while retaining its metal‐core integrity is feasible and universal.

Development of low‐cost electrocatalysts capable of oxidizing ethanol with high efficiency holds great promise for resolving the impediments to developing practical direct ethanol fuel cells. The electro‐oxidation reaction of ethanol for Cu₁₄‐8CH₃CN was evaluated in a solution containing 0.1 m KOH and different concentrations of ethanol (Figures S20 and S21, Supporting Information). An oxidative peak at +0.84 V in the forward scan appeared in the presence of 0.5 m C₂H₅OH. And peak current rose and peak potential shifts to higher positive position with the increase of the ethanol concentration, verifying an excellent electrocatalytic activity of Cu₁₄‐8CH₃CN in this reaction. The linear relationship between anodic current density and ethanol concentration confirms that the ethanol oxidation process is controlled by the kinetic diffusion of ethanol. PXRD pattern of the sample collected after electrocatalytic reaction well reproduced the simulated one (Figure S22, Supporting Information), indicating that Cu₁₄‐8CH₃CN cluster remains intact.

Cu₁₄‐8CH₃CN clusters were also investigated for electrochemical detection of H₂O₂. The CVs of Cu₁₄‐8CH₃CN/GC electrode in 0.1 m phosphate buffer solution (PBS) with different concentrations of H₂O₂, which was bubbled with ultrahigh purity nitrogen for at least 15 min prior to the electrochemical measurements, shows that obvious reduction current appears after the introduction of H₂O₂ and the reduction current increases with the increase of H₂O₂ concentrations (Figure 5a). These findings indicate the high electrochemical activity of Cu₁₄‐8CH₃CN for H₂O₂ detection. The amperometric response of Cu₁₄‐8CH₃CN to the successive addition of H₂O₂ was investigated at −0.45 V. From the obtained i–t curve shown in Figure b, it can be clearly seen that the reduction current increases with successive addition of H₂O₂ and the current can reach a steady state rapidly, indicating the sensitive and fast response of Cu₁₄‐8CH₃CN to the concentration change of H₂O₂. The responding current and concentration of H₂O₂ display linear relationship (Figure a). The limit of detection (LOD) was estimated to be 2.3 × 10⁻⁶ m based on the three times the standard deviation for the average measurement of blank sample (LOD = 3σs⁻¹).

a) CVs of blank electrode (6 × 10−3 m H2O2) and Cu14‐8CH3CN in 0.1 m PBS with the absence and presence of H2O2 at different concentrations of 2, 4, and 6 × 10−3 m, scan rate: 100 mV s−1. Inset: The linear relationship between responding current and concentration of H2O2. b) The amperometric i–t curve of Cu14‐8CH3CN upon successive addition of H2O2 recorded at −0.45 V.

In summary, we have synthesized and fully characterized Cu₁₄(C₂B₁₀H₁₀S₂)₆(CH₃CN)₈ (Cu₁₄‐8CH₃CN), the first copper analog of known Ag₁₄‐8CH₃CN. The similarities and differences in electronic structure and optical property between these Group 11 cluster analogs have been thoroughly investigated by experimental and theoretical methods. In addition, the catalytic activities of Cu₁₄‐8CH₃CN in ethanol electro‐oxidation reaction and electrochemical detection of H₂O₂ were evaluated. This work provides a perspective for preparing Cu⁰‐containing clusters via a self‐reduction procedure, and it serves as a model platform for direct comparison of the physicochemical properties of copper and silver at the nanoscale level.

[CCDC 1861371, 1861372, 1886055 contains the supplementary crystallographic data for this paper. These data can be obtained free of charge from The Cambridge Crystallographic Data Centre via www.ccdc.cam.ac.uk/data_request/cif].

Acknowledgements

This work was supported by the National Science Fund for Distinguished Young Scholars (21825106), the National Natural Science Foundation of China (No. 21801228 and 21671175), the Program for Science & Technology Innovation Talents in Universities of Henan Province (164100510005), the Program for Innovative Research Team (in Science and Technology) in Universities of Henan Province (19IRTSTHN022) and Zhengzhou University.

Conflict of Interest

The authors declare no conflict of interest.