1. Introduction

Surface-enhanced Raman scattering (SERS) is increasingly developing as a very active research field. It can be employed to amplify weak Raman signals and detect ultra-low concentration molecules [1–3]. The advantages including high sensitivity, in situ, and non-destructiveness allow SERS to expand from basic research to biomedical testing [4], trace analysis, and materials science [2,5] fields. Among substrate-molecules systems, the primary SERS mechanisms can be grouped into electromagnetic mechanisms (EM) and chemical mechanisms (CM) [6,7]. The EM is a long-range effect related to the local electromagnetic field induced by exciting surface plasmon excitations in the noble metal (Au or Ag) nanostructures’ surface [8,9]. Several researchers have worked hard to improve the SERS activity based on the EM but the non-selective metal SERS materials have the constraints of simple molecular reactions and are challenging to reproduce findings, limiting the SERS technology’s application [10]. Fortunately, high-sensitive semiconductor substrates have been gradually developed, which greatly enriching the choice of SERS substrates. For example, graphene [11], hexagonal boron nitride (h-BN) [12], and molybdenum disulfide (MoS₂) [13] nanomaterials have been validated to be promising substrates with strong SERS activity. It has been discovered that electromagnetic interference’s effects on semiconductors can be removed due to their intrinsic separation from electromagnetic effects, providing an ideal model to investigate the CM-based SERS [14–18]. Importantly, the chemical enhancement of semiconductor substrates shows excellent selective enhancement to molecules, stemming from the vibrational coupling and photo-induced charge transfer (PICT) transitions between the molecules and semiconductor substrates [19]. This can effectively increase molecular polarizability and amplify Raman scattering cross-section of probe molecules [20]. The structures of the semiconductor materials are modulated to promote the photo-induced electron transition efficiency by increasing the electron concentration or providing additional transition pathways then to improve the enhancement factors (EFs) [21] of the semiconductor SERS substrate. Many strategies have been proposed to effectively improve the PICT process, including element doping [22], phase transformation [23,24], defect engineering [25–27], and other methods [28]. However, the complex photo-induced electron transition processes in molecule-substrate systems, including the molecules’ structure and orientation, the energy alignment between the substrate’s Fermi energy level, and the highest occupied molecular orbital (HOMO)/lowest unoccupied molecular orbital (LUMO) of the molecule, are lacking a universal principle to improve the PICT efficient for semiconductor substrates. Therefore, it is necessary to study the mechanism of CM-based SERS.

Introducing plentiful surface defects into the crystal structure can cause physicochemical modifications in the substrate, which is also an effectual strategy to boost the SERS performance [29]. In semiconductor materials, the central metal atom’s coordination is invariably undersaturated, resulting in a material composition not being stoichiometric. The non-stoichiometric structural defects (vacancies, dislocations, grain boundaries, etc.) due to the crystal structure’s incompleteness tend to improve the substrate’s chemisorption to the probe molecule and enhance the substrate-molecule charge transfer efficiency [29,30]. Wang et al. [31] demonstrated that two-dimensional (2D) semiconductors nanosheets with abundant surface defects tend to exhibit high levels of SERS performances. However, the exact mechanism of the crystal structure incompleteness promoting PICT efficiency under specific conditions, is still unexplained. 2D semiconductor nanomaterials have flat and significant specific surface areas. It was discovered that shrinking semiconductors into ultrathin 2D nanosheets promotes the material’s metal abundance and surface polarizability [32–34]. When reducing the size, the nanoflakes’ specific surface area and the number of atoms located on the surface increase quickly, demonstrating inadequate atomic coordination and high surface energy. These surface atoms easily react and lead to form surface complexes [12]. 2D semiconductor nanomaterials with excellent PICT resonances, better Raman signal uniformity, and remarkable controllability at the atomic scale are considered ideal for fundamental study and practical application extensions of the SERS effect [1,2,5].

Among the semiconductor nanomaterials, tin diselenide (SnSe₂) is a van der Waals semiconductor with a $\textrm{CdI}2$-type crystal structure, belonging to the $\textrm{P}\bar{3}\textrm{m}1$ space group, with Sn–Se–Sn sandwich layered structure [35,36]. SnSe₂ has high intrinsic electron mobility [32] and ultralow thermal conductivity [35]. Moreover, SnSe₂ displays pressure-induced periodic lattice distortion and enables novel device functionalities being a phase change memory material [37]. This means its atomic structure can reversibly switch, and it has suitable band gaps, various electronic band structures, plentiful earth reserves, biocompatibility and economy which makes it an attractive nanomaterial in SERS, gas sensors, photodetectors, and so on [32,38,39]. Interestingly, in experiments, the synthesized SnSe₂ materials are normally mixed-phase of tin selenide (SnSe) and SnSe₂ with different band structure with pure SnSe₂ [33,40]. SnSe presents an orthorhombic [41,42] or cubic [43] crystallographic phase with p-type conductivity, and SnSe₂ presents in the hexagonal crystal structure, respectively contain n-type conductivity [44]. The Sn-Se compounds (SnSe_x) have attracted much attention due to their different in the intrinsic structure. The latest research results about SnSe_x in SERS [45] show the heterostructures of SnSe_x nanoflake arrays (NFAs) combined with golds (Au) and the enhancement factor (EF) is up to 1.68 × 10⁷. However, this excellent SERS performance is achieved through the semiconductor/metal heterostructures which is the synergistic effect of CM and EM. The contribution of PICT could not be clearly separated. Inspired by above-mentioned valuable discoveries, we successfully realized SnSe_x is available for the study of CM-based SERS.

Different with the most reports focusing on highly ordered structures, in this study, we noted the presence of heterogeneous phases can significantly influence the material’s properties. The Se vacancy-induced hybrid phase SnSe_x nanosheets were employing the Chemical Vapor Deposition (CVD) approach. The physical properties of various stoichiometric SnSe_x nanosheets were characterized, and the SERS sensitivity control of ultrathin 2D non-stoichiometric defect structure was achieved. The results show that hybrid phase SnSe_x materials facilitated adsorption of molecules and photo-induced charge separation, which can attribute to the non-chemometric structural defects without element doping. The SERS activity of rhodamine (R6G) on SnSe_x nanosheets was demonstrate excellent, with a surface detection limit of 10⁻¹⁰ M, which has excellent uniformity and reproducibility. The band structure of non-stoichiometries SnSe_x was generated by valence band (VB) spectrum analysis of X-ray photoelectron spectroscopy (VB-XPS) test, which is employed to show the coupling of PICT process and promotion in the substrate-molecules system. The photoluminescence (PL) spectra indicates that the number of photo-induced electrons on the defect energy levels is greatly increased. The SnSe_x nanosheets are used as CM-based SERS substrates to detect thiram with a limit of detection (LOD) of 10⁻⁴ M. It indicates SnSe_x nanosheets have a potential application value in the trace detection of environmental pollutants. The study presented herein will contribute to optimizing and realizing the PICT efficiency of SERS in 2D layered nanomaterials, which offers a new idea for efficiently promoting the sensitivity of defect-induced hybrid phase SERS active materials.

2. Experiment section

2.1 Fabrication of SnSe_x

The Se and SnO₂ solid powders were purchased from Ketai Advanced Materials Co., Ltd (Jiangxi, China). The LiI solid powder was purchased from Macklin Biochemical Co., Ltd (Shanghai, China). The R6G and thiram solid powders were purchased from Aladdin Co. Ltd. (Shanghai, China). In the experiment, a halide-assisted CVD approach was employed to synthesize ultrathin SnSe_x nanosheet [46,47], growing on mica under atmospheric pressure. SnSe_x nanosheets were synthesized using a horizontal single-temperature zone tube muffle (Lindberg/Blue M, TF55035KC-1, Thermo Scientific, Asheville, NC). Mica is a natural 2D layered material with no hanging bonds, facilitating the van der Waals force bonding between SnSe_x nanosheets and the mica surface [48]. The high purity Se (99.5%, about 3 mg) and SnO₂ (99.999%, about 2 mg) powders were employed as the source of Se and Sn vapor. For tin, iodine is the best carrier agent [48]. LiI is a halide salt with simple chemical composition and a low melting point of only 446 °C that is volatile, so LiI (99.99%, about 4 mg) was chosen to help SnO₂ attain a specific vapor pressure at lower temperatures and vaporize to form volatile chemical intermediates. One possible route in this system is

Figure 1(a)-(b) shows the process flow diagram for the preparation of SnSe_x nanosheets, combined with the determined furnace temperature finding, the distance with a substantial temperature gradient between the Se and Sn sources (9–15 cm, represented by D) was used as a variable to regulate the Se source vapor pressure. An alumina boat containing a mixture of SnO₂ and LiI was placed in the single-temperature muffle furnace’s central temperature region, and the mass ratio of precursor solid powder was designed as Se: SnO₂: LiI = 3:2:4. The crystallization temperature of SnSe_x materials is about 220 °C [14]. The temperature rises to 550 °C with a rate of 50 °C min⁻¹ during, which Ar₂ carrier gas kept flowing at 20 sccm, and the growth time was 10 min. Then, it was cooled to room temperature naturally. At 14–16 cm position the downstream tubular furnace is illustrated by experiments to be the best precipitation temperature area. Figure S1 shows the optical image of SnSe_x synthesized on mica, demonstrating their triangle shape with lateral dimensions of about 8 µm.

 

Fig. 1. (a) Schematic illustration of SnSe_x nanosheet synthesis by the CVD approach. SnO₂ mixed with LiI is loaded in the center and Se powders are loaded in the CVD tube furnace’s upstream region. (b) Temperature profile of the furnace. The distance of “0” denotes the furnace’s heating center which was set at 550 °C.

Download Full Size | PDF

2.2 Sample characterization

The morphologies and elemental composition of prepared samples were observed by using a scanning electron microscopy (SEM, Zeiss, Sigma500) with acceleration voltage of 5 kV and energy dispersive spectrometer (EDS). The sample thickness was characterized by employing atomic force microscope (AFM, Horiba, SmartSPM), and the Fermi levels were determined using the Kelvin probe force microscopy (KPFM). The elemental and chemical state analysis was obtained by X-ray photoelectron spectroscopy (XPS, Thermo Fisher Scientific, Kalpha) operating at a vacuum of about 5 × 10⁻⁸ mba in the analysis chamber, using a monochromat Al Kα source as the X-ray source with a beam spot size of 400 µm, which energy was about 1486.6 eV, 6 mA × 12 kV, the flux energy of full-spectrum scan was 100 eV with 1 eV step, and that of the narrow-spectrum scan was 60 eV with 0.1 eV step. The lattice vibration information of SnSe_x nanosheets can be obtained from the X-ray diffraction (XRD, bihec, TD-3000). The Raman, PL, and SERS spectra were obtained from the Raman spectrometer (Raman, HORIB, HR Evolution). Selecting the 532 nm laser (0.48 mW, 10%) with a 50× (NA = 0.75) objective lens and the laser spot was about 1 µm in diameter, the acquisition time was 4 s, the integration number was 2 and the diffraction grid was 600 g/nm. The acquisition time of PL spectrum was 8 s, the number of integrations was 4. Raman-mapping images were generated using a 532 nm laser. The acquisition time for each spectrum was 1 s, and the mapping scan step was 1 µm.

2.3 Preparation of the solution and SERS measurements

The R6G (purity, 99%) molecules are selected to appraise the SERS performance due to the wavelength of incident laser (532 nm) is within the absorption band of R6G. We first dissolved a certain mass of R6G solid powder in deionized H₂O, then obtained a 1 × 10⁻³ M stock solution after 15 minutes of ultrasonic treatment. The lower concentration solution was generated by subsequent dilution of the stock solution. 2 µL R6G solution with a certain concentration was dripped on the nanosheets. The SERS spectra were collected after natural drying at room temperature. It is actually the detection of a small amounts of molecules in solid. For better proving the practical application value of SnSe_x nanosheets, we collected the SERS spectra of thiram. A certain quality of the thiram powder (purity, 99%) was dissolved in the ethanol solution (purity, 99.9%), and the 10⁻² M concentration of the thiram stock solution will be obtained after 30 min ultrasonic treatment. The lower concentration solution was generated by subsequent dilution of the 10⁻² M concentration of the thiram stock solution. The SERS test procedure of the thiram is the same as that of R6G molecule.

3. Results and discussion

3.1 Physical property characterization of SnSe_x nanosheets

The morphology of ultrathin SnSe_x nanosheets was examined by SEM. Figure 2 shows the change in the morphology of SnSe_x nanosheets as the distance between the Se and Sn sources decreases (D is in the range of 15 to 9 cm). The nanosheets’ lateral size is about 2 ∼ 6 µm, decreasing first and then increasing. The triangle’s edge has undergone a change from round to straight, and the corner has changed from truncated to missing corners and then to sharp. As the D decreases, the nanosheets’ surface defect states also have a regular variation, Fig. 2(a)-(b) undergo smoothing to screw dislocation defects. When D = 13 cm, screw dislocations appeared on the sample’s surface, and cracks appeared Fig. 2(c)-(d). When D = 12 cm, the large screw dislocations on the sample’s surface disappeared and only the crack structure appeared. With the subsequent decrease in D, the sample’s surface gradually recovered to be smooth (Fig. 2 e-g). Figure S2 shows the samples’ detailed SEM images with pronounced surface defect structures (Fig. 2 b-d). Thus, as the distance between the Se and Sn sources reduces, the defect degree on the sample’s surface increases first and then decreases. The thickness of the SnSe_x nanosheets was precisely measured by AFM. Figure 2 h indicates the AFM topographic image of one SnSe_x nanosheet at D = 13 cm, with an average thickness of about 1.85 nm (see inset height profile). The corresponding root mean square (RMS) value of the SnSe_x nanosheet is about 0.123 nm, which means the uniformity of thickness is well. The thickness of a monolayer SnSe_x film was reported to be 0.62 nm [49], and here considering the transfer process and feasible surface adsorbates, the generated nanosheet can be considered as a three-layer structure. In this study, nanosheets with three-layer thickness were selected as the study object to unify the conditions. The stoichiometric ratio of SnSe_x nanosheets was investigated using EDS. The x value in SnSe_x can directly show the Se atoms’ relative content. The marked content in Table 1 represents the ratio situation of samples with evident surface defects in the SEM images, showing that the degree of surface defects of the samples is related to the Se’s content. As the D between the Se and Sn sources reduces, the stoichiometry of all the samples does not match the stoichiometric ratio of SnSe₂, and the value of x first decreases from 1.92 to 1.35 and then increases to 1.99. This shows that the synthesized SnSe_x nanosheets have Se vacancy defects, and the change of the x value first decreases and then increases, denoting that the Se vacancy degree’s trend in the samples increases first and then decreases.

 

Fig. 2. (a-g) SEM images of the SnSe_x nanosheets grown with D in the range of 9-15 cm. (h) AFM image of three-layer SnSe_x, the height profile reveals a thickness of 1.85 nm (inset). (i) The domain shape-changing procedure’s diagram depends on the growth rates of Sn-zz terminations.

Download Full Size | PDF

Table 1. Sample number and EDS results of SnSe_x nanosheets samples.

View Table

Studies have demonstrated that the samples’ morphology is influenced by the diffusion and growth process of the CVD approach. During the diffusion process, the gas-phase atomic ratio (Sn : Se) of the precursor is a crucial factor in controlling the sample’s morphology [48]. According to the principle of crystal growth, the final shape of a 2D crystal is measured by the growth rate of the edge structures of various crystal faces. Facets that grow slowly at low energies tend to become maximal, whereas facets that grow quickly at high energies either shrink or disappear completely [50]. During the experiment, due to different temperature gradients in the tube furnace, the various positions of the Se powder directly influence the content of Se atoms in the gas-phase precursor, while the position of SnO₂ is unchanged, and the flow rate of the carrier gas remains stable. In the gas-phase of Sn, the concentration remains constant and the nucleation density at the deposition site remains unchanged. Combined with Fig. 1(b), when D = 15 cm, the gas-phase concentration of Se is lower than that of Sn due to the influence of low temperature (V_se < V_sn), and the nucleation density remains unchanged, leading to a substantial growth rate, the sample’s size becomes larger; inadequate Sn gas-phase density, the Se-zz edges of exposed unsaturated Se atoms are energetically more unstable than Sn-zz edges, and have a higher probability of encountering free Sn atoms to form bonds, growing faster and readily grows into a triangle whose edge shape is Sn-zz edge termination (Fig. 2(i)). When the concentration of Sn in the gas-phase is slightly higher than that of Se, a truncated triangle is formed [50]. With the D’s decrease and the temperature’s increase, the nucleation density of the gas-phase concentration of Se relative to the increase of the gas-phase concentration of Sn (V_se > V_sn) remains unchanged, leading to a low growth rate and a decrease in the sample’s size. Inadequate Se gas-phase, the Sn-zz edge of exposed unsaturated Sn atoms is more energy unstable than the Se-zz edge, the probability of encountering free Se atoms to form bonds is higher, and the growth is faster, the edge shape is a triangle terminated by the Se-zz edge. When D is further decreased, during the process of increasing from room temperature to the experimental temperature, the Se powder evaporates in advance due to the temperature’s increase, decreases the gas-phase concentration of Se relative to the gas-phase concentration of Sn (V_Se < V_Sn) during the growth process. The core density remains unchanged, leading to a substantial growth rate and the sample returns to a significant size. It is discovered that the triangle’s edge terminated by the Sn-zz edge is sharper and straighter. The gas-phase concentrations of Sn and Se elements are unchanged according to the temperature gradient due to the presence of experimental errors, so the formation of missing-corner triangles is likely to occur when the Sn/Se atomic content is small during the growth process.

Screw dislocation–driven (SDD) spiral growth at the low supersaturation condition is thought to be responsible for the helical dislocations’ appearance in the SnSe_x nanosheets during the growth process. [51] According to the Burton-Cabrera-Frank (BCF) crystal growth theory [52,53], the crystal growth process is governed by the local growth environment’s supersaturation. The supersaturation is defined as $\mathrm{\sigma } = \textrm{In}\left( {\frac{\textrm{c}}{{{\textrm{c}_0}}}} \right)$ [53], where σ represents the supersaturation’s degree, the c and c₀ are the local and equilibrium precursor concentration. Both c and c₀ are influenced by the furnace’s temperature profile, but c₀ is primarily influenced by the deposition temperature. Thus, the growth process is more likely to generate helical dislocation structures due to the sudden low temperature in the deposition temperature region. The Se element in the SnSe_x compound belongs to the O group, which is active and more likely to sublimate or separate to generate the elemental Se at the deposition temperature. Sn atoms have metallic properties and are more stable at the deposition temperature. Thus, the fracture defect structure on the samples’ surface may originate from the generation of Se elemental substances during the solidification process. It is concluded that the decrease in the distance between the Se and Sn sources will influence the Se precursor’s gas-phase concentration and then the relative content of Se atoms in the nanosheets, changing the sample size and surface morphology.

The information on the chemical composition, chemical valence state, and valence band energy of the SnSe_x films surface were investigated using XPS. All spectra are referenced to the C1s’ binding energy peak at 284.8 eV. Figure 3(a) shows the typical XPS survey of SnSe_x, Sn, and Se elements detected in XPS were from SnSe_x nanosheets, and the other elemental peak positions are ascribed to the presence of different elements in the mica (KMg₃(AlSi₃O₁₀) F₂) substrate. The high-resolution XPS spectra of the 3d of Sn and Se elements in SnSe_1.35 are revealed in Fig. 3(b) and 3(d). The XPS spectra deconvolution of Sn 3d which fitted by Gaussian function shows the peaks of Sn⁴⁺ 3d_3/2 and Sn⁴⁺ 3d_5/2 at 494.8 and 486.4 eV, respectively. The shoulder peaks at 494.2 eV and 485.8 eV denote Sn²⁺ 3d_3/2 and Sn²⁺ 3d_5/2, respectively. The existence of Sn⁴⁺ and Sn²⁺ in SnSe_1.35 indirectly indicated that Se vacancies were likely present on the surface of SnSe_x nanosheets, which mean two phases of SnSe₂ and SnSe in the synthesized nanosheets. The binding energy difference (ΔB.E.) of the two peaks ((Sn²⁺ 3d_3/2 - Sn²⁺ 3d_5/2) and (Sn⁴⁺ 3d_3/2 - Sn⁴⁺ 3d_5/2)) is 8.4 eV, correspond to a previous study [40]. The Se 3d spectrum of SnSe_1.35 can only be deconvoluted into two peaks at 53.85 and 52.98 eV which were attributed to Se 3d_3/2 and Se 3d_5/2, respectively (Fig. 3(d)). The high-resolution spectra of the 3d energy levels of Sn element in various non-stoichiometric-ratios SnSe_x nanoflakes were uniformly investigated. Figure 3(c) shows that the 3d_3/2 and 3d_5/2 orbitals of Sn 3d energy levels show consistent chemical shifts as the D decreases, peak position of the 3d orbital first moves toward decreasing binding energy moving and then moving to the direction with higher binding energy, the binding energy is the lowest at D = 13 cm. The chemical shift is related to the total charges on the atom, the valence charge decreases and the binding energy increases. Combined with Fig. 3(b), it is shown that at D = 13 cm, Sn atoms might lose less charges and have more valence states lower than Sn⁴⁺. Meanwhile, the lower binding energy of the Sn atom has more charges, which is beneficial to the CT. This feature also serves as an indirectly evidence for the presence of Se vacancies [54]. This indicates that the CT impact will be best when the Se vacancies are maximized (D = 13 cm). It shows that the hybrid phase SnSe_x nanosheets with Se vacancies lead to two phases of SnSe₂ and SnSe in the synthesized nanosheets.

 

Fig. 3. (a) XPS spectra of SnSe_1.35. (b, d) High-resolution XPS spectra of Sn_3d and Se_3d in SnSe_1.35. (c) High-resolution XPS spectra of Sn_3d for the different non-stoichiometric ultrathin SnSe_x nanosheets.

Download Full Size | PDF

The XRD and Raman spectra were employed to investigate the lattice vibration information of the various stoichiometric ratios SnSe_x nanosheets. In Fig. 4(a), the top of XRD patterns reveals that SnSe_x sample can be indexed with hexagonal unit cells of the $\textrm{CdI}2$-type (PDF#23-0602), while the bottom of Fig. 4(a) shows that the lattice of SnSe_x includes the SnSe phase (PDF#48-1224). The peaks at 14.4°, 40.0°, and 44.1° match the (001), (102), and (003), respectively, crystallographic planes of the $\textrm{pnma}$ space group of SnSe₂, and the peak at 29.4° matches the (011) crystallographic planes of SnSe. It proposed that SnSe_x is not a single crystalline, including the mixture phases of SnSe₂ and SnSe. The peak position’s insignificant shift may be related to the surface defect state of the sample [55]. In Fig. 4(b), the Raman spectra depicts a strong band at 185 cm⁻¹, corresponding to the A_1g (interplanar vibration) mode of the Sn atoms of the SnSe₂ nanosheets, and a weaker band at 108 cm⁻¹, matching the SnSe₂ nanosheets’ E_g (in-plane vibration) modes [36]. The E_g mode’s peak intensity is related to the thickness of the nanosheets, and the thinner the nanosheets, the weaker the E_g mode peak intensity. Furthermore, at the low frequency of 71 cm⁻¹, a Raman band corresponding to the A_g¹ vibrational mode of SnSe appears, indicating that the non-chemometric SnSe_x nanosheets we generated a mixture of SnSe₂ and SnSe. The optical images of the SnSe and SnSe_1.92 nanosheets synthesized in the experiment are shown in Figure S1 b-c. The Raman spectra of the various non-stoichiometric-ratios SnSe_x nanoflakes were compared after normalization (Fig. 4(c)), and it was discovered that the A_1g mode indicated a red shift followed by a blue shift as the D value decreased. We fitted the Raman peak shift of the A_1g mode using a linear fit, as depicted in Fig. 4(d). The Raman peak shift of the A_1g mode decreases and then increases. This implies that with the change in Se defect degree, the material’s local region band gap first expands and then narrows, and the change of E_g is closely related to the vibrational change of Se atoms in the SnSe_x lattice structure. To better show the distribution of the hybrid phase of SnSe_x nanosheets, we collected Raman-mapping imaging scans of the SnSe_x nanosheet surfaces.

 

Fig. 4. (a) XRD patterns of SnSe_1.35 nanosheet samples. (b) Raman spectra of SnSe_1.35 and SnSe at room temperature. (c) Raman spectra of various non-stoichiometric ratios SnSe_x nanosheets at room temperature. (d) The purple and green lines denote the Raman shifts of A_1g and E_g modes, and the blue line denotes the intensity of A_g¹ mode. (e) Optical image of the SnSe_1.35. (f) Raman frequency and intensity mapping of A_1g and A_g¹ vibrational modes of SnSe₂ (light blue) and SnSe (navy blue) for SnSe_1.35 nanosheets, and the green part denotes the E_g vibration mode of SnSe₂.

Download Full Size | PDF

Figure 4(f) reveals the Raman frequency and intensity mapping of A_1g (light blue) and A_g¹ (navy blue) vibrational modes of SnSe₂ and SnSe that indicate SnSe₂ grows at the edge, SnSe grows at the center. The Raman-mapping image well shows that the combination of Sn and Se first grows into SnSe₂ with a 2D model during the growth. With the increase in growth time, the Sn and Se elements at the first deposition growth site decompose to form the Se elemental substance, which is precipitated from the sample’s surface. Thus, a hybrid phase sample with SnSe₂ at the edges and SnSe at the center of the nanosheets was formed that coincides with the findings of SEM. It has been demonstrated that the Raman spectral frequency shift is an inherent property of defective systems, and defects change the energy band structure of materials and have a significant effect on the lattice structure vibration of crystals [56]. Thus, it can be considered that the presence of Se vacancy defects causes changes in the energy band structure of SnSe₂ and the turning of the A_1g Raman spectra from blue shifted to red shifted as the D value reduces.

The PL spectra are extremely sensitive to the defect structure of semiconductor materials and can offer the band gap (E_g) of defective bands to help us examine the impact of Se vacancy defects in the mixed-phase SnSe_x nanosheets’ band structure. Figure 5(a) shows the PL spectra of various stoichiometric ratios, all samples indicate defect peaks at 713 nm, but the peak shapes are asymmetric. The PL intensity demonstrated a trend of gradually increasing and then decreasing. In Fig. 5(b), the PL spectra of SnSe_1.35 can be grouped into X₁ and X₂ peaks by Gaussian function fitting, at 712.1 and 750.9 nm, where X₁ denotes the intrinsic luminescence peak of free carrier recombination, and X₂ denotes Se Emission peaks of defect states caused by vacancies (the PL fitting findings of other different stoichiometric ratios nanosheets are depicted in S3). Comparing the intensities of the X₁ peaks of the different stoichiometric ratio nanosheets in Fig. 5(c), the X₁ and X₂ peak intensities increased and then decreased. Thus, with the distance D’s decrease, the intensities of the X₁ and X₂ peaks demonstrated a trend of increasing first and then decreasing, and when D = 13 cm, the intensities of the X₁ and X₂ peaks reach the maximum value. Comparing the shifts of the X₂ peaks of the different stoichiometric ratios SnSe_x nanosheets in Fig. 5(d), the X₁ peak position shifted from 708 nm to 712 nm and then gradually to 707 nm, the X₂ peak position shifted from 743 nm to 757 nm and then decreased to 743 nm. As the D decreases, the peak position shifts of the X₁ and X₂ peaks also exhibit a trend of getting bigger first and then smaller.

 

Fig. 5. PL spectra of the (a) SnSe_x nanosheets were measured with a distance ranging from 15 cm to 9 cm. (b) Deconvolution PL spectra plot of the SnSe_1.35 fitted by the Gaussian distribution. (c) The intensity of X₁ and X₂ peaks of the various non-stoichiometric ratios SnSe_x nanosheets. (d) The position shift of X₁ and X₂ peaks of the various non-stoichiometric ratios SnSe_x nanosheets.

Download Full Size | PDF

In the PL spectra, the higher the absorption peak intensity, the more defect energy levels in the surface of sample [57]. The change in PL defect peak intensity shows that with the D’s decrease, the sample’s defect degree first increases and then decreases, which is consistent with the EDS and XPS characterization findings. The numerous surface defects case the energy levels of defect to be a springboard to assist the PICT transitions within the substrate–molecule system. The photo-induced electrons will be trapped by the defect levels and form a metastable state. It indicates that the number of photo-induced electrons on the surface defect is greatly increased to promote the PICT transitions. When bound to the probe molecules, it will facilitate the CT process from SnSe_x to molecules under laser illumination [28]. The PL peaks’ shape is not a symmetrical envelope, indicating that the sample is inconsistent in purity and has multiple defects coexisting, leading to non-uniform broadening of energy levels and thus peak broadening. This also agrees with the findings of Raman’s characterization. The shift of the peak position of the PL spectrum is due to the change in the sample’s band gap [57]. The shift of the peak positions of X₁ and X₂ depicts that the band gaps of different stoichiometric ratio nanosheets are different, and the defect states’ band gap energy level can be obtained by $\textrm{E} = 1240/\mathrm{\lambda }$ [58]. It further confirms that the band structures of the SnSe_x nanosheets are changed due to defects in the samples. It is established that with the decrease in D, the degree of Se vacancy defect is the reason that the band gap of the internal energy band structure of the sample narrows first and then broadens.

3.2. SERS performances of SnSe_x nanosheets

The R6G was used to appraise the SERS performance of the different Se vacancy-induced hybrid phase SnSe_x nanosheets. The SERS measurements are performed with an excitation source of 532 nm laser, and the corresponding SERS spectra are depicted in Fig. 6(a). Four prominent peaks at 612, 773, 1360, and 1650 cm⁻¹ of R6G molecules are assigned to C-C-C ring in-plane bend, C-H on-plane bend, aromatic C-C stretching, and C-C-C ring out-plane bend vibration modes [58], respectively. All samples demonstrated SERS activity, but the characteristic energy band variation was not uniform. The corresponding SERS intensities histograms of the characteristic energy band (613, 773, and 1360 cm^-1) for R6G in Fig. 6(b) show an overall trend of increasing first and then decreasing, the SERS intensity of 613 and 773 cm^-1 at 13 cm is obviously better than that at other positions, but the SERS intensity of 1360 and 1650 cm^-1 reaches the maximum at 11 cm. This is attributed to the selective enhancement of different vibration modes of R6G molecules by the SnSe_x nanosheets, which is an indirect embodiment of the chemical enhancement mechanism. Figure 6(c) shows the SERS spectra of R6G molecules with concentrations from 10⁻¹¹ M to 10⁻⁵ M to investigate the LOD of the SnSe_1.35 nanosheet. The R6G’s characteristic peaks at 613 cm^-1 could be distinguished even at a low concentration of 10⁻¹⁰ M. The relative intensity of the characteristic peak (613, 773, 1360 and 1650 cm^-1) of R6G was fitted well to the molecular concentration from 10⁻¹⁰ M to 10⁻⁵ M with linear curve in a log-log scale. The logistic linear fitting curve in Fig. 6(d) demonstrates to the excellent ability of quantitatively detecting R6G. Figure 6(d) exhibiting the different slopes of different Raman peaks further proves that the SERS activity of the hybrid phase SnSe_x nanosheets is attributed to a chemical enhancement mechanism.

 

Fig. 6. (a) Raman spectra of R6G and (b) The corresponding SERS intensities histograms of Raman peaks (613, 773, 1360 and 1650 cm^-1) for R6G. (c) The SERS spectra of R6G with concentrations from 10⁻¹¹ M to 10⁻⁵ M on the SnSe_1.35 nanosheet. (d) Raman intensities of R6G molecule at peak 613, 773, 1360 and 1650 cm^-1 as the function of concentration.

Download Full Size | PDF

To precisely find the SnSe_x nanosheets with the highest SERS active and show the changes of different vibration modes of R6G, the intensities of the above four prominent peaks were chosen as interior labels to compute the EF regarding R6G at a concentration of 10⁻³ M (Fig. 7(a)). Quantitatively compare SERS enhancement of the different Se vacancy-induced hybrid phase SnSe_x nanosheets (See Supporting Information for calculation process, Table S1). As the D value decreases, the EFs at 613 cm⁻¹ and 1650cm⁻¹ indicated a change in SERS activity that first increased and then decreased. The optimal SERS activities can be attained on the SnSe_1.35 nanosheet, the EF at 613 cm⁻¹ is equal to 2.6 × 10⁶ and the EF at 1650 cm⁻¹ is equal to 1.7 × 10⁶. The EFs at 1360 cm⁻¹ are similar when D = 15 - 11 cm and the EFs at 773 cm⁻¹ are less than 613 cm⁻¹ and 1360 cm⁻¹. The changes of those four characteristic bands from the R6G molecules’ modes mean SnSe_x has the selective enhancement for different Raman enhancement of R6G, indicating that the enhancement of Raman signals of R6G is CM-SERS. We selected ten random points from the SERS spectra of the R6G with 10⁻⁶ M concentration to study the SERS performance uniformity between points of nanosheets. As shown in Fig. 7(b), SERS performance has good uniformity under the same vibration mode. Meanwhile, the reproducibility of R6G with 10⁻⁶ M concentration to the SERS signals was measured on 10 batches of SnSe_x substrates prepared under the same experimental conditions (D is the same), as shown in Fig. 7(c). The relative peak intensities at 773 and 1360 cm^-1 are collected and the statistical distribution histograms are drawn according to the standard formula [9]:

 

Fig. 7. (a) The EFs of 612, 773, 1360, and 1650 cm⁻¹ Raman peak of R6G on SnSe_x nanosheets with different non-stoichiometric ratios. (b) Spot-to-spot uniformity: ten SERS spectra of the R6G molecules (10⁻⁶ M detected from ten random spots on one substrate. (c) Substrate-to-substrate reproducibility: The statistical distribution of SERS signal intensity (773 cm⁻¹ and 1360 cm^-1) of the R6G (10⁻⁶ M) peaks that was measured. (d) SERS-mapping of the SnSe_1.35 and R6G system at 613cm^-1 corresponds to Fig. 4(e-f). The corresponding SERS intensities histograms of Raman peaks (613, 773, and 1360 cm^-1) for R6G.

Download Full Size | PDF

3.3. Mechanism of the SERS enhancement of SnSe_x nanosheets

The best SERS performance nanosheet (SnSe_1.35) was employed as an example to examine the PICT of SERS enhancement mechanism. The valence band (VB) position near the Fermi level for each sample can be determined by VB-XPS analysis, as depicted in Fig. 8(a), the VB of the SnSe_1.35 is 0.7 eV below the Fermi level. The band gap (E_g) can be computed by PL spectra, and the E_g of the SnSe_1.35 is about 1.65 eV, as depicted in Fig. 7(b). Thus, the conduction band (CB) position can be indirectly derived by 3.92 eV. The KPFM analysis gives more information about the Fermi level, the original contact potential difference (CPD) image of SnSe_1.35, and the average CPD of SnSe_1.35 and the Si substrate are depicted in Fig. 8(c)-(d), respectively. One can see that the average CPD is estimated to be 230 mV for SnSe_1.35 and 315 mV for Si substrate, which is employed to measure the corresponding Fermi level of SnSe_1.35. According to the formula,

 

Fig. 8. (a) The Valence band position of SnSe_1.35 relative to the Fermi level determined by VB-XPS analysis. (b) The band gap of the different SnSe_x nanosheets (c, d) The CPD diagram and the CPD data of SnSe_1.35. (e) Mechanism diagram for SnSe_1.35 and R6G (E_CM, conduction band minimum; E_F, Fermi level; E_VM, valence band maximum; hν, the energy of excitation photon; J_C, the Electron flow (E-flow) from the substrate onto molecular levels above the LUMO in the CT₁ process; ΔJ, the E-flow from the molecular levels above the LUMO onto the energy levels in the conduction band of the substrate in the CT₂.)

Download Full Size | PDF

In this study, the complex PICT process can be inferred into two types of resonance transitions, including the defect levels’ PICT resonance and the coupled molecular levels above LUMO (CT₁) and the coupling between the molecular excited states and the CB states of the substrate (CT₂) improved by the molecular RRS. With the incident 532 nm laser in the substrate-molecule systems, in CT₁, the photon energy (hν, 2.33 eV) is both greater than the energy level difference in the E_CM with E_VM and the E_CM with LUMO (hν > E_CM - E_VM and hν > E_LUMO - E_CM). The excited electrons (J_a) partially leap to the coupled energy level above LUMO (J_c), having the SERS effect through the vibronic coupling of energy levels and magnifying the molecular polarizability tensor, and partly return to VB to be recombined (J_b). In CT₂, when hν > E_CM - E_HOMO, the excited electrons above LUMO of R6G can leap to the coupled energy levels in E_CM (ΔJ) or other partially occupied/empty defect levels in the substrate’s band gap through resonant tunneling. The RRS can take place without the molecule-substrate system only hν > E_LUMO-E_HOMO must be met, and the excited electrons inside the molecules (ΔJ_R) will magnify the R6G’s polarization tensor to contribute to the coupling between the molecular excited states and the CB states of the substrate (ΔJ) [59]. Therefore, the vibronic coupling of two resonances is responsible for the substantial SERS enhancement observed in SnSe_1.35. It can be assumed that this situation of two-phase coexistence caused by Se vacancy defects can improve molecular coupling resonance and PICT resonance, and multi-pathway and more efficient CT resonances can be expected.

3.4. Application

The excellent SERS performance of hybrid phase above studies show that SnSe_x nanosheets are suitable for trace environmental pollutant detection. The thiram (C₆H₁₂N₂S₄) is a typical pesticide with high toxicity, high risk to biology, and environment and water resources. It is widely used as a fungicide to prevent and control fungal diseases of vegetables, fruits and crops [60], so it is necessary to pesticide residue detect and behavior monitor. In previously published literature, the detection concentration of thiram solution can achieve 10⁻⁹ M with a participation of metal (Au/Ag) nanoparticles [61] or under a specific environment [62]. The enhancement of Raman signal is caused by the chemical bonds between thiram molecules and metal substrate, which leads to an increase of the cost in the detection. The SnSe_x nanosheets and thiram are combined by van der Waal force without any chemical bonds. It will not destroy the structure of the molecule, and is helpful to increase the reuse rate of the SERS substrate based on 2D materials. Figure 9(a) clearly illustrates the SERS measurement of the different concentrations of thiram solution on the SnSe_x nanosheets, the peaks at 558 and 850 cm^-1 could be distinguished at the concentration of 5 ${\times} $ 10⁻⁴ M. The peak intensity and concentration were plotted as linear curve in a log-log scale (Fig. 9(b)), the R² values were 0.91 and 0.97, respectively. This means the Se vacancy-induced hybrid-phase SnSe_x nanosheets could achieve basic quantitative detection of the thiram. It proves that the proposed CM-SERS substrate based on non-metal substrate has practical application and feasibility in the field of trace environmental pollutant detection.

 

Fig. 9. (a) The SERS peaks intensities of the thiram (C₆H₁₂N₂S₄) with various concentrations (10⁻² M – 10⁻⁴ M). (b) The calibration curve of the Raman intensity at 558 and 850 cm⁻¹ against the various concentrations of thiram at a log scale.

Download Full Size | PDF

4. Conclusions

To summarize, the vacancy defects in the hybrid phase SnSe_x nanosheets can improve the energy level match between the substrate and molecules, and enhance the PICT process with the excited light. The XPS, Raman, and PL investigations suggest the physical properties of SnSe_x nanosheets, which demonstrates hybrid phase properties as the degree of Se vacancies varied, and exists a strong resonance-enhanced Raman scattering effect of the optimized SnSe_1.35 nanosheets. It can provide a new idea for the detection of organic pollution by non-metal semiconductor SERS substrate. It is worth noting that the analysis of SERS mechanism and the improvement of SERS performance for the 2D hybrid phase nanomaterials can be extended to numerous sensing systems with defective substrates and analytes, and expand a pathway for the potential applications in different fields.