At present, SERS has been developed into an advanced, non-destructive analytical technology with superior qualitative capabilities, such as the identification of organic macromolecules and biomolecules, as well as quantitative detection of trace amounts of chemicals [1-9]. The lowest detection limit even can reach single-molecule levels with some specialized treatments [10, 11]. Researchers agreed that for a material to perform well in SERS, it must has high sensitivity, excellent chemical stability, exceptional detection reproducibility, and minimal preparation costs [12-15]. So far, people have mainly used Au and Ag nanostructures as SERS substrate materials, mainly because these two types of materials have strong localized-SPR effects [16-19]. Although Au substrates have excellent SERS effect and good chemical stability, their use cost is relatively high. Ag substrates are much cheaper, however, their oxidation resistance is a flaw.

In addition to noble-metal SERS substrates, the rise of semiconductor materials has opened up new directions for the selection of materials for SERS substrates, such as WO_3-x [20, 21], WO_3-x/MoO₂ heterogeneous knot [22], MoO_3-x nanosheets [23], Cu₂O nanocrystals [24], MoTe₂ ultrathin nanosheets [25], and MoS₂ nanoparticles. [26]. Although these semiconductor SERS substrates often have better biocompatibility and lower cost than noble-metal substrates, their chemical stability is often relatively low, which limits their application. For example, W₁₈O₄₉ with abundant oxygen vacancies is easy to be oxidized in air and lose SERS activity. For another example, although the Raman enhancement factor (EF) of MoTe₂ nanosheets is as high as 1.0 × 10⁹, it is easily corroded in an acidic environment and is not suitable for many chemical tests. Therefore, the development of sensitive, stable, and low-cost SERS substrates is still an urgent problem in the field of analysis and materials research.

ReO₃ is a unique metallic transition metal oxide, which has a surprising conductivity similar to that of metal Cu [27-29]. For ReO₃, Re⁶⁺ is a d1 system (the outer electron configuration of rhenium is 5d₅6s₂), where the strong hybridization between the Re 5d and O 2p atomic orbitals significantly broadens the 5d conduction band, which gives it strong metallic properties [30, 31]. Also due to its rather high free electron density, ReO₃ exhibits a strong localized surface plasmon resonance (LSPR) at the nanoscale. This high free electron density makes ReO₃ a possible SERS substrate candidate, but few researches on their SERS properties have been reported. Herein, a simple chemical method is developed to synthesize a self-assembled microtubular ReO₃ with high chemical durability, which shows strong localized-SPR behavior and SERS effect. The plasmonic ReO₃ microtubes exhibit a Raman EF of 8.9 × 10⁵ and a lowest detection limit of 1.0 × 10⁻⁹ mol/L, which is almost comparable to some noble-metal SERS substrates. Considering that the price of ReO₃ is almost one percent of that of Au, this non-noble metal SERS substrate has great application prospects. The regular microtube structures formed by self-assembly of ReO₃ nanocubes were synthesized by a facile hydrothermal method (Fig. 1a and Fig. S1 in Supporting information). The thickness of these self-assembled microtubes is uniform, the cross-sectional diameter is 200–300 nm, the tube wall thickness is about 40 nm, and the side length of the nanocubic particles is 30–40 nm. Because the tube wall is assembled from much smaller nanocubes (Fig. 1b), the surface of the outer wall is very rough, and many "hot spots" are formed at the junction of each nanocube and nanocube (electromagnetic fields are highly concentrated at the locations). The HRTEM image of the nanocube is shown in Fig. 1c. The clear lattice fringes shown in the figure indicate that the ReO₃ nanocubes have high crystallinity, and the lattice fringes with the interplanar spacing of 0.37 nm can accurately correspond to the (100) crystal plane of the cubic phase ReO₃ nanocrystals. Through powder X-ray diffraction (XRD), we obtained the crystal phase pattern of the ReO₃ microtubes. It can be seen from Fig. 1d that the XRD pattern of the sample can be accurately referred to as cubic phase ReO₃ (JCPDS No: 33–1096). The stronger diffraction peaks also indicate that the product has high crystallinity. The energy dispersive spectrometer (EDS) data shows that the sample only contain Re and O signals (Fig. S2 in Supporting information). The Re/O ratio of the self-assembled microtubes is 1/3.14, which is almost the same as the atomic ratio of ReO₃, which further confirms that the sample is ReO₃. Furthermore, the survey spectrum of X-ray photoelectron spectroscopy (XPS) demonstrated that the sample is highly pure (Fig. S3a in Supporting information). Although the Re 4f spectrum shows a small amount of Re⁷⁺, it is still Re⁶⁺ in the main body (Fig. 1e).

We evaluated the localized SPR effect of ReO₃ microtubes in the visible light region. Fig. 1f shows the UV–vis-NIR absorption spectrum of ReO₃ microtubes in the air at room temperature. It can be seen that this material exhibits a strong localized SPR effect, and its resonant peak center is located at ~530 nm, which is very close to the commonly used Raman excitation wavelength of 514 nm or 532 nm. It is a favorable condition to generate strong SERS signals that the intrinsic resonance wavelength of the SERS substrate and the Raman excitation wavelength are as close as possible. As shown in Fig. 1f (inset), under the irradiation of a 300 W xenon lamp, the surface temperature of the ReO₃ microtube material rapidly increased from room temperature to 128 ℃ within 30 s. This excellent photothermal conversion performance further proves the strong localized SPR effect. These results suggest that these ReO₃ microtubules have a very strong SPR effect and are a potential SERS substrate candidate.

As we expected, benefited from the strong localized-SPR effect of the ReO₃ microtubes, their SERS activity is also strong. To evaluate the SERS properties of ReO₃ microtubes, the most commonly used Raman probe molecule rhodamine 6G (R6G) was chosen. Fig. 2a (red spectrum data) shows the SERS spectrum of ReO₃ microtubes of 1.0 × 10⁻⁵ mol/L R6G aqueous solution. It can be seen that the ReO₃ microtube SERS-substrate can accurately detect a series of Raman scattering peaks. The four Raman peaks R₁-R₄, which are the most easily identified characteristic peaks of R6G, are located at 612 cm⁻¹, 772 cm⁻¹, 1362 cm⁻¹, and 1651 cm⁻¹, respectively, as well as scattering peaks at other positions, are highly consistent with the Raman spectra of R6G standard (the blue spectrum data in Fig. 2a). Because these ReO₃ microtubes are in close contact with the silicon wafers, silicon wafers may also contribute to SERS. In order to eliminate this effect, we conducted a controlled experiment. The results show that when a bare silicon wafer is used as the SERS substrate (the black spectrum data in Fig. 2a), the Raman signal of R6G is not obtained, which clearly proves that these enhanced Raman signals are indeed derived from the plasmonic ReO₃ microtubes.

On one hand, the practicality of the SERS substrate is reflected in its detection sensitivity. Fig. 2b shows the Raman spectra of 5 samples of R6G with different concentrations ranging from 1.0 × 10⁻⁵ mol/L to 1.0 × 10⁻⁹ mol/L. It can be seen that in a wide range of analyte concentration, these ReO₃ microtubes as SERS substrates can show excellent enhancement effects, and the lowest detection limit for R6G is 1.0 × 10⁻⁹ mol/L. This superior SERS sensitivity can be attributed to the electromagnetic field enhancement effect caused by the vigorous localized-SPR effect of these ReO₃ microtubules. In additon, we think that the good adsorption of ReO₃ microtubes also improves the detection sensitivity of this substrate material to a certain extent. Fig. S4 (Supporting information) shows the good adsorption capacity of ReO₃ microtubes.

Another criterion for measuring the practicality of SERS substrates is the signal repeatability of substrate testing. We use R6G at a concentration of 1.0 × 10⁻⁷ mol/L as the probe molecule to test the signal reproducibility of the ReO₃ microtube-substrate. As shown in Fig. 2c, the SERS spectra obtained from 50 randomly selected measurement points show that almost the same SERS spectrum intensity can be observed at different positions, suggesting the good uniformity of the ReO₃ microtube-substrate. Figs. S5 and S6 (Supporting information) show the two-dimensional Raman mapping images of the Raman signal of the ReO₃ microtube substrate itself and the SERS signal at the 612 cm⁻¹ (R₁) characteristic peak of the 1.0 × 10⁻⁷ mol/L concentration of R6G on the ReO₃ substrate, respectively. Excellent preparation uniformity and test reproducibility are also demonstrated here. Furthermore, as shown in Fig. 2d, the signal relative standard deviation (RSD) of a total of 10,000 random points at 1362 cm⁻¹ was calculated to convincingly demonstrate the high reproducibility of the prepared ReO₃ microtube substrate. The calculated results show that the signal RSD obtained from the substrate is only about 7.7%, which is significantly better than the commercial Au colloid substrates (~20%) [32].

In addition to the above two aspects, considering that in the Raman test, the substrate will be exposed to laser radiation with different powers, so the stability of the SERS substrate is also very important. Under normal circumstances, the chemical stability of transition metal oxides in the intermediate valence state is poor, such as W₁₈O₄₉ and Cu₂O with strong SERS activity. Fortunately, ReO₃ is one of the few very stable intermediate valence transition metal oxides similar to MoO₂ [33]. After immersing ReO₃ microtubes substrate in 2 mol/L NaOH solution and HCl solution for 5 h, the substrate can still exhibit an unabated SPR effect (Fig. 3a), and its SERS effect is not affected (Fig. 3b). XRD patterns, SEM images and HRTEM images also demonstrated their high chemical stability (Figs. S7-S10 in Supporting information).

In addition to R6G, the ReO₃ microtube SERS-substrate also has excellent detection sensitivity for some risk compounds of global interest. Butylated hydroxyanisole (BHA) is widely used in fried food because of its good anti-oxidation, anti-rancidity, good thermal stability and strong durability. However, the addition of BHA in food has certain safety hazards. Once excessive, it will cause certain toxicity to the human body, and even cause cancer risks. Therefore, it has been explicitly banned in the field of food addition in many countries. It can be seen from Fig. 3c that the ReO₃ microtube SERS-substrate has a good trace detection effect on BHA. 2, 4-Dichlorophenol (hereinafter referred to as 2, 4-DCP) is a recognized carcinogen and is a toxic environmental pollutant strictly controlled by the emission standards of various countries. On the ReO₃ microtube SERS substrate, even the concentration of 2, 4-DCP was only 1.0 × 10⁻⁹ mol/L, it was sensitively detected (Fig. 3d). In addition, the controlled experiments also shown that the ReO₃ microtube-substrate also has a good detection capacity on rhodamine B (RhB) (Fig. S11 in Supporting information) and crystal violet (CV) (Fig. S12 in Supporting information), which displays a good detection universality.

Recent studies have shown that for semiconductor SERS substrates with high conductivity, the electromagnetic field enhancement caused by SPR effect and the chemical enhancement mechanism caused by interfacial charge transport play a role together, which is also applicable to the current ReO₃ substrate. Specifically, on the one hand, the strong localized-SPR effect makes ReO₃ microtubes have the physical enhancement mechanism dominated by electromagnetic hot spots similar to noble metals. In order to verify this point of view, we placed the sample in a muffle furnace and heated it in air for 0.5 h at 280 ℃ to obtain a taupe ReO_3+x (Fig. S13 in Supporting information). As the oxidation state increases, the free electron density decreases, the localized SPR peak of the sample drops sharply (Fig. 4a), and the corresponding SERS performance is also greatly reduced (Fig. 4b). However, it was found by SEM and XRD characterization that the microscopic morphology and crystal phase of the sample did not change (Figs. S14 and S15 in Supporting information), which suggests that only the change of valence state has occurred. These results above prove that the electromagnetic field enhancement caused by the SPR effect does exist. On the other hand, when ReO₃ microtubes are immersed in R6G aqueous solution for 2 h, the UV–vis-NIR absorption spectra show that compared with the R6G aqueous solution and the unmodified ReO₃ microtubes, the absorption peak of the ReO₃ microtubes modified with R6G undergoes a clear blue shift (Fig. 4c). This phenomenon demonstrates that obvious interfacial charge transfer behavior occurs between ReO₃ microtube substrate and adsorbed R6G probe molecules. Based on the above results, it can be reasonably concluded that the ReO₃ substrate is affected by the double enhancement mechanism of SPR and chemical transmission.

In addition, to understand the effect of this microtubule structure assembled from nanocubes on the SERS effect, we further compared the localized-SPR effect and SERS effect of the dispersed nanocubes and the assembled microtubes. The nanocubes for comparison have similar morphology (Fig. S16 in Supporting information) and crystalline phase (Fig. S17 in Supporting information) as the cubes contained in the microtubes. Comparative studies have found that the localized-SPR effect of the ReO₃ microtubes is obviously stronger than that of the dispersed ReO₃ nanocubes (Fig. 4d). The reason may be that when the laser is irradiated on the ReO₃ microtube, part of the light passes through the nanotube wall and is reflected multiple times in the cavity, which increases the material's absorption of laser energy. However, this situation does not appear in the dispersed ReO₃ nanocubes, which directly causes the SPR effect of the ReO₃ microtubes to be much greater than that of the ReO₃ nanocubes.

In summary, a simple chemical method is developed to grow metallic ReO₃ self-assembled microtubes on the surface of silicon wafers. The rough surface and the grooved-cavity structure make the ReO₃ microtubes an efficient non-noble metal SERS substrate. Combined with their high Raman EF (8.9 × 10⁵), sensitive detection limit (1.0 × 10⁻⁹ mol/L), high corrosion resistance, and low price, these ReO₃ microtubes have a good application prospect in the field of analytical chemistry.

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

This work received financial support from the National Natural Science Foundation of China (No. 51771175) and the Science Foundation of State Administration of market supervision (No. 2021MK164).

Supplementary material associated with this article can be found, in the online version, at doi:10.1016/j.cclet.2022.05.086.