Facet-Dependent Reduction Reaction of Diruthenium Metal–String Complexes by Face-to-Face Linked Gold Nanocrystals

Abstract

The facet-dependent redox reactions of diruthenium metal–string complexes by gold nanoparticles (AuNPs) are explored by using the surface-enhanced Raman scattering (SERS) technique. Gold nano-rhombic dodecahedrons (AuRDs), gold nanocubes (AuNCs), and gold octahedrons (AuOhs) with exclusive facets {110}, {100}, and {111}, respectively, were synthesized. These AuNPs linked face-to-face by metal–string complexes Ru₂M(dpa)₄Cl₂ (dpa = dipyridyl amino, M = Ni, Cu) with chloride axial ligands serve as both SERS substrates and reducing agents in the reactions. We employ the diruthenium core in these complexes with multiple redox states to study the reduction ability of varied AuNP facets upon plasmonic excitation. In Ru₂Ni(dpa)₄Cl₂, the Ru–Ru stretching mode ν_{Ru–Ru str.} lies at 327 cm^–1 on the SERS substrate AuOh, but this band shifts to 313 cm^–1 on the AuRD and AuNC. The diruthenium moiety was reduced to [Ru₂]⁴⁺ by the AuRD facet {110} and the AuNC {100}. The gold nanorods in the solution prepared with metal–string complexes bridging head-to-head on {111} facets were used for the SERS substrate. The SERS curves of the complexes in these self-assembled head-to-head rods display ν_{Ru–Ru str.} at 327 cm^–1, which is assigned to having an [Ru₂]⁵⁺ core. Hence, facets {110} and {100} have a reduction reactivity greater than that of {111}. In Ru₂Cu(dpa)₄Cl₂, the ν_{Ru–Ru str.} is observed to lie at 312 cm^–1 on AuRD, but shifts to 320 cm^–1 on the AuNC and AuOh. In the latter cases, the diruthenium moiety was reduced to having a charge of 4+ with electronic configuration π*²δ*², whereas the former case band at 312 cm^–1 with a weaker Ru–Ru bonding is also attributed to [Ru₂]⁴⁺ but with electron configuration π*⁴. π*⁴ lies at an energy greater than π*²δ*². The electrochemical SERS spectra of diruthenium complexes were recorded to verify their oxidation states. Conclusively, these results yield the reduction reactivity of the following facet: {110} > {100} > {111}. According to the results of the redox reactions, the valence states of the diruthenium metal–string complexes are verified. In the [Ru₂]ⁿ⁺ core, n = 4 π*⁴, 4 π*²δ*², 5 π*²δ*, and 6 π*δ*, and the ν_{Ru–Ru str.} is 312, 320, 327, and 337 cm^–1, respectively.

Introduction

Metal nanoparticles and various anisotropic nanoparticles are reported to have heterogeneous catalytic properties. The catalytic reactivity depends on the metal species, shape, size, and facets.¹⁻⁹ Observations on facet-dependent reactions are reported. Zhang and Wang described the high-index facets of gold nanoparticles having better catalytic reaction rates for hydrogenation of 4-nitrothiophenol.⁹ Zhang et al. found that the gold atoms preferentially deposited on Ag@Au cuboctahedrons with concave structures at the sites of the {111} facets.¹⁰ Chiu et al. synthesized the high-degree uniformity of cubic, octahedral, and rhombic dodecahedral gold nanocrystals and found the facet-dependent catalytic activity of gold nanaocrystals on reduction of p-nitroaniline.¹¹ The gold mediates the electron transfer to the aromatic nitro compounds to catalyze the reaction. They deduced that the order of reduction of catalytic activity of the facets was {110} > {100} > {111} and partially explained that this is because of the high binding energy of p-nitroaniline to the {110} facet.¹¹

Surface enhanced Raman scattering (SERS) is a sensitive technique and has unique vibrational fingerprinting capability to enable identification of transient intermediates of reactions.^9,12 Hence, SERS has been used to monitor the catalytic processes by nanostructures. Besides, the nanoparticles can act as redox reagents. Huang et al. observed oxidation of 4-aminothiophenol by silver nanoparticles during SERS measurements.¹³ Kim et al. explained that the hot electron released from the silver nanoparticles converted Fe³⁺ to Fe²⁺ in their SERS detection.¹⁴ Wu et al. were aware of the redox ability of varied nanoparticles and used it combining the SERS technique to obtain the bonding strength of Ru–Ru in diruthenium complexes after reacting with silver or gold substrates upon plasmonic excitation.¹⁵ Consequently, they determined the valence states of the diruthenium core in trinuclear metal–string complexes. These metal–string complexes are one kind of extended metal atom chains (EMACs), having structures of four polypyridyl amido ligands helically coordinating the metal ion line, for example [Ru₂Ni(dpa)₄Cl₂] (Hdpa = dipyridylamine), as displayed in Scheme 1.

Scheme 1. Structures of Ru₂Ni(dpa)₄Cl₂ and Ru₂Cu(dpa)₄Cl₂.

Wang et al. observed a direct plasmon-accelerated electrochemical reaction on gold nanoparticles.¹⁶ When the plasmonic band is excited, this produces more hot electrons/holes and accelerates redox reactions. Hence, in SERS measurements, redox reactions can occur and the vibrational structure of products can be measured in situ. In the present work, we report the facet-dependent reduction of the heteronuclear EMAC Ru₂M(dpa)₄Cl₂ (M = Ni, or Cu, structures shown in Scheme 1) by gold nanocrystals measured using SERS. These complexes have axial ligand chlorides as the anchoring group to bridge the surfaces of gold nanocrystals. Because of the multinuclear system these complexes have several redox potentials; hence, they serve as excellent candidates for evaluating the redox ability of nanoparticles. The diruthenium core can be [Ru₂]ⁿ⁺, n = 4 (electron configuration π*⁴ or π*²δ*²), 5 (π*²δ*¹), and 6 (π*¹ δ*¹).^15,17,18 We used two diruthenium complexes with different oxidation potentials to examine the reduction activity of varied facets of gold nanocrystals. Here, we synthesized octahedrons (Ohs), cubes, rhombic dodecahedrons (RDs) enclosed by {111}, {100}, and {110} facets, respectively, with a diameter of 50–60 nm to have hot electrons with energy near the Fermi level during plasmonic excitation.

Because of complications in this metal–string molecular system, electrochemical SERS (ECSERS), which combines SERS and an electrochemical technique, is used to obtain the Ru–Ru bonding strength and then to identify their valence states. The absorption during the redox processes can be recorded in situ to assist in assigning the electronic structures.^19,20 Diruthenium possesses high polarizability; thus, intense Raman signal is expected. This provides SERS detection with great sensitivity. In addition, understanding the redox states of diruthenium complexes is essential for future applications in their catalytic processes.

Results and Discussion

Physical Properties of Various Shapes of Au Nanocrystals

Gold nanocubes (AuNCs), Ohs, and RDs were synthesized using the seed-mediated method developed by Chiu et al.¹¹ The metal–string complexes dissolved in ethanol were dispersed in acetonitrile (ACN)-containing nanocrystals. All nanocrystals except spheres were surrounded with surfactant cetyltrimethylammonium chloride (CTAC) before adding complexes. Gold nanorods (AuNRs) were synthesized using the method of Vigderman and Zubarev²¹ covered with cetyltrimethylammonium bromide (CTAB), which was then replaced with CTAC; hence some CTAB might remain around the sides of the rods but the heads were mostly enclosed with CTAC. All the facets used for comparing their reduction activity here were covered with the same kind of surfactant to avoid possible complications. Only the nanospheres, which have varied facets on the surfaces, have citrate as the surfactant. Detailed synthesis procedures of these nanocrystals are described in the Supporting Information. The scanning electron microscopy (SEM) and transmission electron microscopy (TEM) images and extinction curves of the synthesized particles are displayed in Figure 1. The sizes of AuNCs, AuRDs, and AuOhs are 51.0 ± 4.6, 68.7 ± 2.9, and 58.5 ± 1.7 nm, respectively.

Figure 1.

(Left to right) Schematic drawing for the crystals face-to-face bridged by Ru₂Cu(dpa)₄Cl₂, TEM, SEM images and extinction curves of without (black line) and with Ru₂Cu(dpa)₄Cl₂ at various periods (time interval ≈ 15 s) for the substrate (a) AuRD (scale bar for the TEM image is 20 nm), (b) AuOh, (c) AuNC, and (d) AuNR.

In AuRDs, after adding Ru₂Cu(dpa)₄Cl₂, the extinction curves recorded after each 15 s period display a gradual red shift; the plasmonic band originally at 539 nm moved to 683 nm then to 826 nm for forming longer chains (Figure 1a). The SEM and TEM images show that AuRDs are linked face-to-face. For AuOhs and AuNCs, a similar behavior of self-assembled and face-to-face chains in solution is observed, as shown in Figure 1b,c. For rods of size 12.1 ± 0.7 × 29.4 ± 2.7 nm² aspect ratio 2.4 ± 0.2, the longitudinal plasmonic mode lies at 627 nm and is red-shifted to 772 nm when connected by complexes. Long chains are depicted in the SEM images; the complexes anchor on both ends of the AuNRs by chlorides in solution, forming head-to-head chains.^20,22 Ying et al. reported that metallaynes with sulfide moieties can serve as connectors for gold nanoparticle assemblies.²³ Orendorff and Murphy reported that AuNR has {110}/{100} facets on the sides and {100}/{111}on the heads, and the bridging molecules mostly anchor on the {111} facets.²⁴ Similar TEM images were obtained, and the schematic drawing of the bridged surfaces is shown in Figure 1d. When the sample mixtures of AuNR/Ru₂Cu(dpa)₄Cl₂ were deposited on a silica wafer and let dry, the SEM image showed AuNR aggregates, not only head-to-head but also side-to-side structures, unlike that in the solution phase.

SERS and ECSERS Spectra of Ru₂Ni(dpa)₄Cl₂

The SERS spectra of Ru₂Ni(dpa)₄Cl₂ in the vibrational wavenumbers of 200–450 cm^–1 are displayed in Figure 2a–f. With substrate gold nanospheres (AuNSs), AuRDs, and AuNCs, the SERS spectra (Figure 2a–c) of Ru₂Ni(dpa)₄Cl₂ exhibit an intense band at 313 cm^–1, which is assigned to the Ru–Ru stretching mode, ν_{Ru–Ru str.} of the [Ru₂]⁴⁺ core, electron configuration π*⁴.¹⁵ The vibrational mode assignments of trimetal–string complexes and ligand dpa^– are described elsewhere.^15,25−27 The metal-related modes lie in the wavenumber range below 450 cm^–1. Among those, ν_{Ru–Ru str.} exhibits the most Raman intensity. For Ru₂Ni(dpa)₄Cl₂ in solid form we obtained a broad band exhibiting both [Ru₂]⁴⁺ and [Ru₂]⁵⁺ cores, indicating the crystalline samples having both valence states.¹⁵ The AuNS with varied facets has an oxidation potential of ∼1.33 V, which can reduce the diruthenium core from 5+ to 4+.¹⁵ On AuOhs, this band ν_{Ru–Ru str.} lies at 327 cm^–1 (Figure 2d); hence, the [Ru₂] core remains as 5+. The SERS spectrum (Figure 2e) of Ru₂Ni(dpa)₄Cl₂ in the AuNR solution with complexes connected on both ends of the rods using facet {111}, ν_{Ru–Ru str.} is at ∼330 cm^–1, close to that in the AuOh. If the samples were prepared dry on a silica wafer to avoid the interference of solvent signal, the SERS spectra (Figure 2f) display a broad feature for this ν_{Ru–Ru str.} band, which can be deconvoluted to two bands peaked at 313 and 327 cm^–1, separately. Accordingly, we have the mixed valence states of diruthenium cores [Ru₂]⁴⁺ and [Ru₂]⁵⁺. These aggregated rods have hot spot regions including the side-to-side, which has facets {110} and {100}; thus, a broad band feature is observed. Conclusively, with a single facet only one valence state of the [Ru₂] core appeared. These SERS data reveal that facets {100} (AuNC) and {110} (AuRD) with the bridging conformation in nanocrystals reduced the complexes to [Ru₂]⁴⁺. Facet {111} only yielded the [Ru₂]⁵⁺ core. Hence, the order of reduction reactivity is {110}, {100} > {111}.

Figure 2.

SERS spectra of Ru₂Ni(dpa)₄Cl₂ on substrates (a) AuNSs, (b) AuRDs, and (f) AuNRs prepared on a dry silicon wafer, (c) AuNCs, (d) AuOhs, and (e) AuNRs in ACN solution, recorded at an excitation wavelength of 632.8 nm. ECSERS curves (g) at −0.4, +0.3, +0.4, +0.8, and +1.1 V in 0.1 M tetrabutylammonium perchlorate (TBAP)/dichloromethane (DCM) by 785 nm excitation. The black line is the experiment data and the dash lines are the fitting bands with the peak position indicated. The pound sign denotes the ACN band and the asterisk the DCM. (h) Cyclic voltammetry in the DCM and 0.1 M TBAP is the supporting electrolyte. (i) Intensities for deconvoluted bands at 313, 327, 333, and 337 cm^–1 from the ECSERS curves and (j) absorption spectrum recorded in 0.1 M TBAP/DCM at −1.0 V.

From the cyclic voltammetry (CV) measurements, two reversible redox potentials appear at E_1/2 = +0.02 and +1.06 V for oxidation of Ru₂Ni(dpa)₄Cl₂/[Ru₂Ni(dpa)₄Cl₂]⁺ and [Ru₂Ni(dpa)₄Cl₂]⁺/[Ru₂Ni(dpa)₄Cl₂]²⁺, respectively, as shown in Figure 2h. All voltages applied in this work are reported versus the Ag/AgCl (saturated) reference electrode. These results agree with those of Huang et al.²⁸ They also reported the electroabsorption curves measured at an applied voltage of 0.16/0.32/1.2 V, exhibiting a near-infrared absorption band centered at ∼850/900/1100 nm for the major species Ru₂Ni(dpa)₄Cl₂/[Ru₂Ni(dpa)₄Cl₂]⁺/[Ru₂Ni(dpa)₄Cl₂]²⁺.²⁸ Because the near IR bands of [Ru₂] core lie close for the neutral and mono-oxidized species and the small oxidation potential Huang et al. assigned the first oxidation occurred for Ni⁺ → Ni²⁺ and the second oxidation for [Ru₂]⁵⁺ → [Ru₂]⁶⁺. We further recorded the absorption curves at −1.0 V voltage for the reduced form of [Ru₂Ni(dpa)₄Cl₂]⁻ ([Ru₂]⁴⁺ core), and found that the absorption band is blue-shifted to ∼771 nm, as shown in Figure 2j.

The ECSERS curves at an applied voltage of −0.4 to 1.1 V were recorded and are displayed in Figure 2g. We used 785 nm light as the excitation source to resonance enhance the Raman intensity of [Ru₂]^5+,6+ because they have relatively more absorption at this wavelength than the 4+ core to achieve a better signal-to-noise ratio. Before the voltage was applied, the complexes were reduced by the nanoparticles; thus, ν_{Ru–Ru str.} appeared at 313 cm^–1, which is attributed to the reduced species [Ru₂]⁴⁺[Ni]⁺. In both measurements of CV and ECSERS, the applied voltage on the reacting species is referred to a reference electrode. As the reference electrode is the same, the reported voltages are the same in both cases. After applying a voltage of 0.3–0.4 V, the samples were oxidized, and the ν_{Ru–Ru str.} band was broadened and extended to the blue region and was deconvoluted to yield the 313, 327, and 333 cm^–1 bands which are ascribed to species [Ru₂]⁴⁺[Ni]⁺, [Ru₂]⁵⁺[Ni]⁺, and [Ru₂]⁵⁺[Ni]²⁺, respectively. When the applied voltage was increased to 0.8 V, the samples were further oxidized and the band was extended to 337 cm^–1. Up to 1.1 V, only band 337 cm^–1 remained. This is assigned to the species [Ru₂]⁶⁺[Ni]²⁺. For reduction, the voltage was tuned to −0.4 V, and the ν_{Ru–Ru str.} band was broadened (313 + 327 cm^–1) to contain contributions from [Ru₂]⁴⁺[Ni]⁺ and [Ru₂]⁵⁺[Ni]⁺.

In Figure 2i, the plot of ECSERS intensities of various bands versus the applied voltage shows conversion among varied species. The intensities of the 313 and 327 cm^–1 bands decrease with the applied voltage and vanish at ∼+0.8 V. The 333 cm^–1 band appears near −0.2 V; the intensity increases first then decreases to zero at +0.8 V. As those bands decrease in intensity with voltage, the 337 cm^–1 band first appears near +0.6 V and the intensity increases with the voltage. Varied species have different absorptions in the visible range. Hence, their Raman enhancement is varied. Nevertheless, based on the variation in the band intensity, we further verify the bands at 313, 327, 333, and 337 cm^–1 to be [Ru₂]⁴⁺[Ni]⁺, [Ru₂]⁵⁺[Ni]⁺, [Ru₂]⁵⁺[Ni]²⁺, and [Ru₂]⁶⁺[Ni]²⁺, respectively. The values of ν_{Ru–Ru str.} and assignments are summarized in Table 1. Stronger bonding strength in Ru–Ru for the [Ru₂]⁵⁺[Ni]²⁺ core than for [Ru₂]⁵⁺[Ni]¹⁺ is obtained.

Table 1. List of Band Position ν_Ru–Ru-str., Electron Configuration, and Valence State of the Diruthenium Core.

metal core	Raman shift	Ru2 electron configuration	absorption λ (nm) (ε × 103 M–1 cm–1, transition)
[Ru2]4+	320a	π*2δ*2	634 (3.3, δ* → δ*L*/δ* → L*), 752 (3.6, δ → π*/π* → δ*L*)b
[Ru2]4+Ni+	313	π*4	771 (3.5)
[Ru2]5+Ni+	327	π*2δ*	850 (4.4)c
[Ru2]5+Ni2+	333	π*2δ*	900 (5.9)c
[Ru2]6+Ni2+	337	π*δ*c	1100 (9.8)c
[Ru2]4+Cu+	312	π*4	766 (1.2)
[Ru2]4+Cu+	320	π*2δ*2
[Ru2]5+Cu+	325	π*2δ*	894 (4.3)c
[Ru2]5+Cu2+	325	π*2δ*	862 (4.8)c
[Ru2]6+Cu2+	332	π*δ*c	1150 (9.0)c

^a[Ru₂(OAc)₃(bpnp)]⁺, ref (15).

^b[Ru₂(OAc)₃(bpnp)]⁺, in DCM.

^cReference (28).

We employed the time-dependent density functional theory (TD-DFT) method UBP86/def2TZVP/W06 containing a split valence and a triple zeta basis set to assist in assigning the electronic structure. All calculations were performed using the GAUSSIAN package.²⁹ The initial geometry used is the X-ray crystal structures taken from Huang et al.²⁸ then was optimized using DFT, the same method and basis set as TD-DFT. For the [Ru₂]⁵⁺ core spin S = 3/2, we obtained the electron configuration π*²δ* similar to that of Huang et al. using method B3LYP.²⁸ The vertical transition π* → δ* in the near IR lies at 916 nm, close to the observed position 850 nm. Thus, this is a d–d transition. In complex [Ru₂(OAc)₃(bpnp)]⁺[PF₆]⁻ (bpnp = 2,7-bis(2-pyridyl)-1,8-naphthyridine) which bears a [Ru₂]⁴⁺ core, spin S = 1 has ν_{Ru–Ru str.} at 320 cm^–1.¹⁵ This complex has absorption bands at 752 and 634 nm in solvent DCM (Supporting Information, Figure S2). These bands according to the TD-DFT calculations are assigned to δ → π*/π* → δ*bpnp* [molecular orbital (MO) of ligand bpnp mixed with metal δ*] and δ* → δ*bpnp*/δ* → bpnp*, respectively. The calculated band positions are 689/700 and 632/626 nm, respectively, and the electron configuration is π*²δ*². Here, the metal δ* orbital is filled; so supposedly none of the near IR π* → δ* transitions exists. This agrees with the experimental observation.

For the reduced form [Ru₂]⁴⁺, S = 1 of Ru₂Ni(dpa)₄Cl₂, we obtained a configuration π*²δ*² for the optimized structure and the bands in the red region lie at 735/771 nm, corresponding to δ* → δ*L*/π* → δ*L* (L denotes ligand dpa. MO of dpa mixed with metal δ*), transitions with the characteristics of metal-to-ligand charge transfer. These results are similar to those obtained for [Ru₂(OAc)₃(bpnp)]⁺. When we set the spin state S = 0 we obtained an optimized singlet structure derived from the same configuration π*²δ*² instead of π*⁴. In this case, the bonding strength of Ru–Ru would be expected to be similar for the same configuration. Although the TD-DFT yields the absorption positions close to the observed 771 nm, the ν_{Ru–Ru str.} observed at 312 cm^–1, a weak Ru–Ru bonding, implies an electron configuration π*⁴, instead of π*²δ*², more electron on the π* antibonding orbital. Method B3LYP was used to perform the same calculations. However, both B3LYP and BP86 methods predict the lowest energy singlet and triplet states for the [Ru₂]⁴⁺ core with the configuration π*²δ*².

Conclusively, based on the previous assignment¹⁵ and the results of calculations, ν_{Ru–Ru str.} = 313 cm^–1 is assigned to the [Ru₂]⁴⁺core with the configuration π*⁴ and 320 cm^–1 to the [Ru₂]⁴⁺core with the configuration π*²δ*². Huang et al.²⁸ suggested that the [Ru₂]⁵⁺ core has a configuration π*²δ* spin S = 3/2, and our measured Raman shift ν_{Ru–Ru str.} = 327 cm^–1. For the [Ru₂]⁶⁺ core two configurations π*δ* and π*² are considered. Ren et al.^16,17 reported Ru₂(DMBA)₄Cl₂ (DMBA = N,N′-dimethylbenzamidinate) with a Ru–Ru bond length of 2.3224 Å and assigned configuration π*². This complex has a vis–NIR band at 726 nm. However, the second oxidized complex [Ru₂Ni(dpa)₄Cl₂]²⁺ with the [Ru₂]⁶⁺ core has the NIR band at 1100 nm and Huang et al.²⁸ tentatively assigned to have π*δ*. Our measured Raman band ν_{Ru–Ru str.} = 337 cm^–1, a strong Ru–Ru bonding, indicates that the π*δ* configuration is more likely. In addition, using DFT UBP86/def2TZVP/W06, we obtained an optimized structure for the π*δ* configuration, a Ru–Ru bond length of 2.331 Å, and calculated ν_{Ru–Ru str.} = 330 cm^–1. This seems to agree with our observations.

SERS and ECSERS Spectra of Ru₂Cu(dpa)₄Cl₂

The Raman spectra of Ru₂Cu(dpa)₄Cl₂ and [Ru₂Cu(dpa)₄Cl₂]PF₆ in solid powder are displayed in Figure S3. In the neutral and mono-oxidized forms, the ν_{Ru–Ru str.} mode is assigned to the band at 325 cm^–1. They correspond to the [Ru₂]⁵⁺[Cu]⁺ and [Ru₂]⁵⁺[Cu]²⁺ cores according to Huang et al.²⁸ In this complex, the bonding strength of Ru–Ru is less affected by the third metal ion. The SERS and ECSERS spectra of Ru₂Cu(dpa)₄Cl₂ are shown in Figure 3a–g. In the SERS spectra, the ν_Ru–Ru,str. appears at 315–312 cm^–1 on the substrates AuNS and AuRD, but shifts to 320–322 cm^–1 on AuNCs, AuOhs, and AuNRs (solution phase). Besides, the SERS spectra of the dry AuNR sample display a broad band with deconvoluted bands peaked at 312 and 321 cm^–1. On the basis of the observed wavenumbers of the Ru–Ru stretching mode, our data reveal that the AuRD facet {110} reduced the diruthenium copper complex to yield the [Ru₂]⁴⁺ core with the electron configuration π*⁴ but AuNCs, AuOhs, and AuNRs to [Ru₂]⁴⁺ π*²δ*² In Ru₂Cu(dpa)₄Cl₂ all nanocrystals reduced to [Ru₂]⁴⁺ only with varied electronic configurations.

Figure 3.

SERS spectra (left) of Ru₂Cu(dpa)₄Cl₂ on the substrate (a) AuNS, (f) AuNR prepared on dry silicon wafer, (b) AuRD, (c) AuNC, (d) AuOh, and (e) AuNR in ACN solution. The pound sign denotes the ACN band and the asterisk the DCM. (g) ECSERS spectra at no volt, +0.7, +0.9, +1.1, and +1.2 V in 0.1 M TBAP/DCM. The excitation wavelength is 632.8 nm. (h) CV and (i) absorption curve (−1.3 V) recorded in 0.1 M TBAP/DCM. (Δ is attributed to the oxygen reduction.)

CV shows two reversible redox appearing at E_1/2 = +0.19, +1.07 V, and one irreversible peak at −0.6 V, Figure 3h. Similar results were obtained by Huang et al.²⁸ Similar to the nickel complex, these are assigned to oxidation [Ru₂]⁵⁺[Cu]⁺/[Ru₂]⁵⁺[Cu]²⁺ and [Ru₂]⁵⁺[Cu]²⁺/[Ru₂]⁶⁺[Cu]²⁺, and reduction [Ru₂]⁴⁺[Cu]⁺/[Ru₂]⁵⁺[Cu]⁺. The ECSERS curves (Figure 3g) recorded using 632.8 nm excitation display bands at 312–315 cm^–1 for low applied voltages, then a new band at 332 cm^–1 appears at the voltage of +1.1 V. The ν_{Ru–Ru str.} of the [Ru₂]⁴⁺ core exhibits an intense Raman signal for a wide voltage range. The 332 cm^–1 band is attributed to the [Ru₂]⁶⁺ core and the band corresponding to the [Ru₂]⁵⁺ core is unobserved here. At 785 nm excitation, the 312 cm^–1 band is displayed under the applied voltage up to 0.7 V, as shown in the Supporting Information. However, the higher wavenumber bands are weak and very broad, not as clear as those using 632.8 nm excitation. Hence, the ECSERS spectra only exhibit two species in the voltage range −1.3 to 1.2 V. The absorption curve recorded at −1.3 V has a broad band peaked at 766 nm and extended to near-infrared. From the results of ECSERS measurements, we assigned to the reduced form with [Ru₂]⁴⁺ core π*⁴.

Accordingly, using this copper complex, we have the order of reduction reactivity of the facets as {110} > {100},{111}. Summarizing the results for both complexes, the order of the facets’ reduction reactivity is {110} > {100} > {111} in gold nanocrystals. Table 2 lists the positions of ν_{Ru–Ru str.} in various Au nanocrystals and their assignments.

Table 2. List of SERS Band Positions Recorded by Different Nanostructures and Valence States of Diruthenium Moieties.

			Ru2Ni(dpa)4Cl2		Ru2Cu(dpa)4Cl2
structure		crystal facet	νRu–Ru str., cm–1a	[Ru2]n+, conf.	νRu–Ru str., cm–1a	[Ru2]n+, conf.
AuNSs		{110}, {100}, {111}	313	4+, π*4	315	4+, π*4
AuRDs		{110}	313	4+, π*4	312	4+, π*4
AuNCs		{100}	313	4+, π*4	322	4+, π*2δ*2
AuOhs		{111}	327	5+, π*2δ*	322	4+, π*2δ*2
	solution	{111}	330	5+, π*2δ*	320	4+, π*2δ*2
AuNRs	dry	{110}, {100}, {111}	313/327(broad)	4+/5+	312/321(broad)	4+

^aSERS spectra of Ru₂M(dpa)₄Cl₂ were recorded by 632.8 nm laser excitation.

Fermi Levels of Facets Versus MOs of Diruthenium

The observed reduction reactivity order is explained by the energy order of Fermi levels of gold facets. Among them, the Fermi energy of {110} should be the greatest. The Fermi level of bulk gold is around 5.53 eV.³⁰ Comparing the energies of MOs of diruthenium from the DFT calculations (reference to the vacuum level) and our experimental observation, we propose that the π* of [Ru₂]⁵⁺ in the diruthenium nickel complex lies between the Fermi levels of facets {100} and {111} to accept electrons from {100} and {110}. The electrons on δ* are also moved to π* to form [Ru₂]⁴⁺ π*⁴. The diruthenium copper complex has slightly varied energies in the high occupied MOs to differentiate the reduction reactivity of the facets. In this complex, the hot electrons from all facets can fill the half-empty δ* to form [Ru₂]⁴⁺ π*²δ*², and only {110} has Fermi energy greater than π* to yield [Ru₂]⁴⁺ π*⁴. Accordingly, the total energy of [Ru₂]⁴⁺ π*⁴ is expected to lie above [Ru₂]⁴⁺ π*²δ*². This agrees with the results of DFT calculations that for the lowest singlet state of the [Ru₂]⁴⁺ configuration π*²δ*² is obtained. Figure 4 depicts the schematic relative energy among the Fermi levels of gold facets and the π* and δ* MOs of diruthenium ions to explain the experimental observation.

Figure 4.

Illustration of the reduction reaction of metal–string complexes by AuNPs and the relative energy between the Fermi level (E_f) of gold and MOs of the diruthenium core.

Conclusions

The metal–string complexes Ru₂M(dpa)₄Cl₂ with diruthenium moiety having various valence states were used to serve as a probe to evaluate the reduction reactivity of various facets of gold nanocrystals. The high-degree uniformity of cubic, octahedral, and rhombic dodecahedral gold nanocrystals with facets {100}, {111}, and {110}, respectively, were used both as reductants in the reactions studied here and as substrates for the SERS measurements. For each facet we only observed one certain valence state of diruthenium ions, whereas in ECSERS other minor species remain in the solution. Because of varied electronic resonance enhancement in Raman intensity, the minor species might gain better SERS intensity than the major species. This might yield a wrong assignment just based on the ECSERS measurements.

Our data clearly yield the order of reduction reactivity of the facets as {110} > {100} > {111}. This order turns out to be the same as the observed order in the catalytic reactions. Varied facet indices have distinct Fermi energies; thus, redox reactions can occur differently. In these diruthenium metal–string complexes, the ν_{Ru–Ru str.} at 312, 320, 327, 333, 337 cm^–1 is assigned for [Ru₂]ⁿ⁺ cores, n = 4 electron configuration π*⁴, 4 π*²δ*², 5 π*²δ* with Ni⁺, 5 π*²δ* with Ni²⁺, and 6 π*δ*, respectively. Because of the similar ligand structure, these wavenumbers furthermore reveal the bonding strength of Ru–Ru.

Experimental Section

Materials

Complex Ru₂Ni(dpa)₄Cl₂ and Ru₂Cu(dpa)₄Cl₂ were synthesized following the methods described before.²⁸ Silver nitrate (AgNO₃), CTAB, TBAP, and gold(III) chloride trihydrate (HAuCl₄·3H₂O) were from Alfa Aesar, ACN, sodium bromide (NaBr), and sodium borohydride (NaBH₄) from Sigma-Aldrich, CTAC from Tokyo Chemical Industry, and sodium hydroxide (NaOH), potassium iodide (KI), ascorbic acid, sodium citrate, and acetone from J.T.Baker. All chemicals were used as received. Pure deionized water (Milli-Q Millipore, 18.2 MΩ/cm) was used in all the preparations.

Synthesis of AuNSs, AuNRs, AuNCs, AuRDs, and AuOhs

The AuNS was synthesized by using the citrate reduction method.³¹ The AuNS solution was used without being further purified, and the diameter determined by TEM was 17.3 ± 1.5 nm. The AuNR was prepared according to the seed-growth method.²¹ The synthesized AuNR is covered with surfactant CTAB, called AuNR(CTAB). We replaced the surrounding surfactant CTAB with CTAC through ion exchange method to form AuNR(CTAC). In brief, AuNR(CTAB) after centrifugation was redistributed by CTAC aqueous solution. This procedure was repeated twice. The gold nanocrystals of AuNCs, AuRDs, and AuOhs were synthesized following the seed growth method developed by Huang and co-workers.^32,33 All the nanoparticles’ synthesis details are in the Supporting Information.

Raman and SERS Measurements

The Raman spectra were recorded in a backscattering geometry employing an objective lens (10×). The He–Ne laser (Lasos) operated with a red light at 632.8 nm served as the excitation light source for gold nanostructure substrates. The laser power at the sample region was set at ∼10 mW (for solid samples) and ∼1 mW (for SERS samples). The scattered signal passing through an edge filter, optical fiber, and monochromator (length, 0.5 m; grating 600 grooves/mm) was recorded with a liquid-nitrogen-cooled charge-coupled device. The spectral resolution was maintained at 3 cm^–1. The integration period per scan was typically about 30 s and averaged for 5–10 scans for a spectrum.

In a typical SERS sample preparation, one or two drops of complex in alcohol solution was added into 2 mL of nanoparticle solution. Alcohol serves as a dispersing agent for anchoring molecules to the nanoparticle surface. In some cases, when nanoparticles were covered with surfactants, for removal of partial surfactants to obtain a sufficient SERS signal, the Kumar and Thomas method employing solvent exchange water/ACN (1:4) for anchoring molecules was used.¹⁹ For redox reactions, illumination by a 23 W fluorescent lamp for 10 min was applied to make sure that the sample molecules have completely reacted with the nanoparticles.

ECSERS Measurements

The ECSERS consists of a potentiostat (CH Instruments), spectroelectrochemical cell (ALS), and a Raman spectrometer. The cell was equipped with a Pt counter electrode and the gold working electrode which were coated with AuNS and sample molecules. We used the Ag/Ag⁺ reference electrode in the measurements, and the applied voltages reported is corrected to reference to Ag/AgCl (saturated) in order to compare with the results from the previous study. The 0.1 M TBAP DCM solution mixed with a very few number of metal–string complexes served as the electrolyte. For near IR light resonance Raman detection, the 785 nm commercial Raman spectrometer (BaySpec) was used and the spectral resolution was 8–10 cm^–1. In the present work, all voltages are the reported reference versus the Ag/AgCl (saturated) reference electrode.

Supporting Information Available

The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acsomega.9b00005.

• Synthesis methods, Raman and ECSERS spectra, absorption spectrum and TEM and SEM images on SERS samples (PDF)

The authors declare no competing financial interest.