Hierarchical MoS2-microspheres decorated with 3D AuNPs arrays for high-efficiency SERS sensing

Accepted Manuscript Title: Hierarchical MoS2 -microspheres decorated with 3D AuNPs arrays for high-efficiency SERS sensing Authors: Hengwei Qiu, Minqiang Wang, Le Li, Junjie Li, Zhi Yang, Minghui Cao PII: DOI: Reference:

S0925-4005(17)31550-2 http://dx.doi.org/10.1016/j.snb.2017.08.127 SNB 22994

To appear in:

Sensors and Actuators B

Received date: Revised date: Accepted date:

7-3-2017 27-7-2017 16-8-2017

Please cite this article as: Hengwei Qiu, Minqiang Wang, Le Li, Junjie Li, Zhi Yang, Minghui Cao, Hierarchical MoS2-microspheres decorated with 3D AuNPs arrays for high-efficiency SERS sensing, Sensors and Actuators B: Chemicalhttp://dx.doi.org/10.1016/j.snb.2017.08.127 This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

Hierarchical MoS2-microspheres decorated with 3D AuNPs arrays for highefficiency SERS sensing

Hengwei Qiu, Minqiang Wang*, Le Li, Junjie Li, Zhi Yang, Minghui Cao

Electronic Materials Research Laboratory (EMRL), Key Laboratory of Education Ministry, International Center for Dielectric Research (ICDR), The School of Electronic and Information Engineering, Xi’an Jiaotong University, Xi’an 710049, China.

* Corresponding author. E-mail address: [email protected] (M. Wang)

Highlights 

First, we present a novel route for synthesis of 3D AuNPs arrays that can generate much stronger electromagnetic enhancement than traditional 2D arrays.



Second, the MoS2-MS@CF-AuNPs can be easily integrated into various flexible substrates and, herein, we present a facile way to fabricate the wafer-scale flexible SERS substrates using the suction filtration method.

Abstract: Three-dimensional (3D) distribution Au-nanoparticles (AuNPs) arrays have enabled surface-enhanced Raman scattering (SERS) for a variety of analytical applications due to the spatial-distribution plasmon. Herein, we present a new route for synthesis of 3D SERS powder, hierarchical MoS2-microspheres (MoS2-MS) decorated with “cauliflower-like” AuNPs arrays (CF-AuNPs), for high-efficiency molecular detection. Finite element simulations have shown that the spatial-distribution nanogaps can generate intensive SERS hot spots. Using the CFAuNPs@MoS2-MS for in-situ molecular detection, not only have 10-14 M rhodamine 6G or methylene blue been identified with different laser wavelength excitation (532 and 633 nm), but also various metabolites in human early-morning urine were recognized. Moreover, the wafer-scale flexible SERS substrates consisted of CF-AuNPs@MoS2-MS inset cellulose

acetate membrane (CAM) were fabricated via suction filtration method, showing good SERS sensitivity for adenosine detection.

Keywords: 3D AuNPs arrays, hierarchical MoS2 microspheres, molecular detection, SERS

1. Introduction Noble-metal nanoparticles (NM-NPs), Au and Ag, have attracted intensive interest in recent years due to their localized surface plasmon resonance (LSPR) that can intensely amplify the ambient electromagnetic field [1]. Surface-enhanced Raman scattering (SERS) is one important application of such plasmonic effect that can produce enhanced Raman signals of surface-absorbed molecules [2-4]. The most commonly used materials for SERS substrates fabrication are Au and Ag in rough surface or nanostructures, or various semiconductors include ZnO, ZnS, TiO2, Cu2O and CuO [5-11]. Although the SERS substrates fabrication have achieved great development, the fabrication of SERS substrates with stable and high average enhancement factors (EF) over relatively large-areas remains a great challenge [12]. In the past several years, hybrids or heterostructures between NM-NPs and semiconductors have been become a hot research topic due to the integrated efficacy of electromagnetic enhancement (excited by the LSPR of NM-NPs with EF of 109 or more) and chemical enhancement (caused by the charge transfer between the probe molecules and NM-NPs/semiconductor with EF ranging from 10 to 100), which can generate stronger SERS effect than onefold efficacy [1318]. Recently, 2D atom-thick layered materials, graphene and molybdenum disulfide (MoS2) with good biocompatibility and chemical inertness, have also been demonstrated that they can give Raman signals a chemical enhancement [19-21]. Therefore, many kinds of heterostructures between 2D layered materials and NM-NPs for SERS detection have been reported, such as graphene-coated Au nano-pyramids hybrid system as the SERS substrate for single molecule detection [22], graphene-veiled Au-nanoparticles (AuNPs) as the SERS substrate for investigation of position-sensitive electromagnetic enhancement activity [23], oxygen-plasma treated MoS2 and argon-plasma treated MoS2 nanoflakes for SERS detection [24], and MoS2 nanosheets with in-situ grown AuNPs for molecular SERS detection [25]. Generally speaking, for SERS substrates, the greatest enhancement always occurs at

nanogaps between the neighboring metal nanostructures with high local-field intensity (socalled “hot spots”) [26-28]. According to this strategy, in addition to the metallic nature, the SERS enhancement is highly dependent upon the size/shape, the interparticle distance, and the distribution of the NM-NPs arrays. Thus, the structure optimization of NM-NPs to obtain an increase in the number and intensity of hot spots in a large volume to enhance SERS-activity, is a high-profile research focus. Based on this, SERS substrates with high-density 3D nanogaps (i.e., hot spots) have recently attracted intensive interest. The advantages of these 3D NM-NPs arrays include that: i) superior signal quality owing to their exponentially large surface area and substantial number of hot spots from the additional third dimension over traditional 2D SERS substrate [29]; ii) the integration of multiple photons scattered in 3D space can cause the multiple cascaded amplification and the collective excitation [30]. Up to now, many research groups have reported various 3D SERS substrates, for examples, the arrays of cone-shaped ZnO-nanorods decorated with Ag nanoparticles (AgNPs) as 3D SERS substrate with huge number of SERS “hot spots” can provide high SERS sensitivity [31], and SERS substrate based on a metal-dielectric-CNT nanowire structure shows high average enhancement with the detection sensitivity can be down to femtomolar-level [32]. More recently, ZnO-nanotaper array sacrificial templated synthesis of noble-metal building-block assembled nanotube arrays as 3D SERS substrates, not only has 10-14 M rhodamine 6G been identified, but also 10-7 M polychlorinated biphenyls (PCBs) are recognized [33]. Vertically-stacked 3D cross-point Au nanoarchitectures containing dense and regular hot spots can also be used for efficient SERS analysis [34]. Other 3D SERS substrates, such as Au coated TiO2-nanotube arrays [35], AuNPs decorated carbon-nanotube arrays [36], Au capped Si-nanowire arrays [37], Ag coated Sinanopillar arrays [38], Ag nanoislands coated glass-nanopillar arrays [39], and 3D Al hybrid nanostructure [40], also display high SERS-activity owing to their high-density and long-term stability 3D hot spots in both horizontal and vertical direction. Herein, we present a new route for synthesis of dispersible 3D SERS powder, hierarchical MoS2-microspheres (MoS2-MS) decorated with “cauliflower-like” AuNPs arrays (CF-AuNPs), for high-efficient SERS detection. The hierarchical MoS2-MS with stacked nanosheets outside the surface have inherent large specific surface area and [41], herein, they act either as the supporter for AuNPs arrays (through Au-S bond), or as the additional SERS enhancer (include

both charge transfer and dipole-dipole coupling). The detection limit for rhodamine 6G (R6G) detection is 10-14 M at both 532 and 633 nm excitation. Under the same detection condition, the detection limit for methylene blue (MB) detection is 10-14 M and 10-15M at 532 and 633 nm excitation, respectively. Moreover, a series of flexible SERS substrates consisted of CFAuNPs@MoS2-MS inset cellulose acetate membrane (CAM) (size of 47 mm, holes of 0.45 μm) were fabricated via suction filtration method, showing excellent SERS activity for adenosine detection. 2. Materials and methods 2.1 Materials Sodium molybdate dihydrate (Na2MoO4·2H2O), thiocarbamide (CH4N2S), hydrogen tetrachloroaurate (HAuCl4·3H2O, ≥47.8%), rhodamine 6G (R6G), and methylene blue (MB) were purchased from Sinopharm Chemical Reagent Co., Ltd, and used without further purification. Adenosine sample was purchased from Sangon Biotech (Shanghai). All solutions were prepared using purified water (Milli-Q, Millipore system). 2.2 Preparation of MoS2-MS The MoS2-MS were synthetized by a facile one-step solvothermal method, also can be described as the molybdenum ion sulfuration reaction proceeding in high-pressure batch autoclave at 180 °C for 24 h. Briefly, Na2MoO4·2H2O of 1.6937 g and CH4N2S of 2.2836 g were dissolved in mixed liquor of deionized water (15 mL) and ethanol (15 mL) under vigorous stirring. Then the prepared mixed liquor was added into a 30 mL high-pressure batch autoclave and put in the hot air circulating oven (BINDER Gmbh) at 180 °C for 24 h to complete the sulfuration reaction. After the sulfuration reaction finished, the reaction products were cleaned using supercentrifuge with 3 times of deionized water and 3 times of ethanol, and the bottom sediment was dried in vacuum drying oven at 60 °C for 8 h. 2.3 Preparation of different CF-AuNPs@MoS2-MS samples Briefly, under vigorous stirring, 3×10-4 mol MoS2-MS (0.048 g) was immediately mixed with different aliquots of 10 mM HAuCl4 solution samples. Herein, the volume of HAuCl4 solution samples were all 20 mL, and concentrations respectively were 1 mM, 2 mM, 3 mM and 4 mM. In this way, the molar ratio between MoS2-MS and HAuCl4 respectively were 15:1 (1-CF-AuNPs@MoS2-MS), 15:2 (2-CF-AuNPs@MoS2-MS), 15:3 (3-CF-AuNPs@MoS2-MS)

and 15:4 (4-CF-AuNPs@MoS2-MS). All the reaction mixture were heated to 60 °C for 5 min in the thermostatic reactor. Finally, the four kinds of CF-AuNPs@MoS2-MS composites were purified using supercentrifuge with 3 times of deionized water and 3 time of ethanol, and dried in vacuum drying oven. 2.4 Samples preparation and SERS measurements In addition to the human morning urine samples, all the used analytes in experiments were prepared using purified water and diluted to the required concentration. The human morning urine samples were obtained from three volunteers in fasting state, and the centrifugal supernatant fluid for detection were obtained using supercentrifuge with 10000 rpm. In the SERS experiments, same mass of powders were evenly dispersed in the molecule solution with different concentrations, and dropwise added on SiO2 substrate for detection. We at least collected 10 points with valid signals from each sample, and the displayed Raman spectra are the average spectra. Unless otherwise noted, all the SERS measurements were performed under the same conditions (integration time, laser power, focus, etc.). For all the SERS experiments, the integration time is 8 s, and the laser power is 0.35 and 0.36 mW for 633 and 532 nm excitation, respectively. 2.5 Preparation of flexible SERS substrates Briefly, CF-AuNPs@MoS2-MS with different mass were evenly dispersed in 20 ml deionized water, and filtrated into CAMs using the vacuum filtration system. The mass of the powder respectively are MoS2-MS of 0.08 g (CAM-0), CF-AuNPs@MoS2-MS of 0.005 g (CAM-1), 0.01 g (CAM-2), 0.02 g (CAM-3), 0.04 g (CAM-4) and 0.08 g (CAM-5). Finally, all the flexibles SERS substrates were dried in vacuum drying oven. 2.6 Characterizations Surface morphology of all the prepared samples were characterized with field emission scanning electron microscope (FESEM, FEI Quatan FEG 250) equipped with an energy dispersive spectrometer (EDS), information of the microstructures were obtained with fieldemission transmission electron microscope (TEM, JEOL JEM-2100, operated at 200 keV), and X-ray diffraction (XRD) patterns were characterized using a D/max-2400 diffraction spectrometer (Rigaku, Japan). The SERS experiments were performed on a HR Evolution-800 Raman microscope system (HORIBA), and the instrument is equipped with a standard 633 and

532 nm laser (laser spot ~0.8 μm). 3. Results and discussion Herein, we present a facile route to fabricate hierarchical MoS2-MS via a single-step sulfuration reaction without any surfactant, and the whole reaction process can be divided into five stages. Stage 1 is the initial stages of the sulfuration reaction (generation of MoS2 molecules and self-nucleation). In this stage, self-nucleation of MoO42- and S2- forms small MoS2 nuclei/clusters by anion exchange process, then these nuclei/clusters rapidly grow into small particles or sheets (Fig. S1a). Abundant small MoS2 particles or sheets aggregate with each other due to the continuously supply of MoO42- and S2-, and multilayer MoS2 particles or sheets will grow into disc-shapes with smooth surface and 300-500 nm in size (stage 2, Fig. S1b and Fig. S1f). In stage 3, aggregations of these MoS2 disc-shapes cause a higher surface free energy and mainly present the formation of MoS2 spherical particles with the average size of 500-1000 nm (Fig. S1c and Fig. S1g). For the newly-formed MoS2 spherical particles in the system, the surfaces of them are rather smooth, which can be attributed to the lowest chemical potential of spherical particles compare with particles in other shapes. Some of them have small rounded hump or protuberance outside the surface, which can be attributed to the “big eat small” phenomenon in crystal growth. The sulfuration reaction is keep running on the surface of MoS2 spherical particles and, simultaneously, the size-distribution become more and more uniform (stage 4, Fig. S1d and Fig. S1h). As shown in Fig. S1e, a part of the MoS2 spherical particles have transformed into hierarchical MoS2-MS, but there are still existing another part of the MoS2 spherical particles keep smooth surfaces. The sulfuration reaction continues until the chemical potential between the atoms in solution and the surface atoms of MoS2-MS reach an equilibrium state. Finally, the hierarchical MoS2-MS are in a uniform size-distribution and porous structure (stage 5, Fig. S1e and Fig. 1a). The hierarchical MoS2-MS (sizes of 1.1-1.3 μm) present unbroken and porous sphereshapes with coverage of nanosheets (large-scale FESEM image in Fig. 1a). The magnified observation in a relatively small area (Fig. S2a) and an individual hierarchical MoS2 microsphere (Fig. 1b) further display these nanosheet-subunits. TEM and HRTEM images were employed to investigate the crystal structure of these outside nanosheets (Fig. 1c-Fig. 1e), and various agglomerated presentations of layered-structure nanosheets can be observed. As

respectively marked with red and green double-lines, the inter-layer spacing of 0.66 nm and a distinct set of visible lattice fringes with an inter-planar distance of 0.27 nm can be clearly observed in the HRTEM images (Fig. 1d and Fig. 1e), which are consistent with the expanded d spacing of the (002) planes and (100) facet of hexagonal MoS2. Moreover, abundant lattice imperfections include the stacking fault, vacancy defect, and grain boundary can be observed in the HRTEM image (such as the areas marked with red circles in Fig. 1e). Actually, it has reported that the inherent lattice imperfections and grain boundary contain unsaturated sulfur atoms, where are the sites for the forming of Au-S bond [42, 43]. Three EDS mappings respectively correspond to Mo/S, Mo and S element of an individual hierarchical MoS2 microsphere in Fig. S2b, were employed to investigate the elementary composition of the hierarchical MoS2-MS (Fig. S2c-Fig. S2e), and the element distribution profile of the EDS mappings are consistent with the as-selected individual MoS2 microsphere. The EDS spectra of the hierarchical MoS2-MS is shown in Fig. S2f, and elementary composition contain Mo and S with the atomic percent of 34.15% (error of 1.77%) and 65.85% (error of 2.31%), respectively. Moreover, oxygen element is not been detected, indicating that there are no residual oxygen-contained compounds, such as MoO2 and MoO3. Fig. 2a-Fig. 2d are FESEM images of the CF-AuNPs@MoS2-MS formed with different precursor (HAuCl4) concentrations, respectively are 1 mM (Fig. 2a), 2 mM (Fig. 2b), 3 mM (Fig. 2c), and 4 mM (Fig. 2d). For the HAuCl4 concentration of 1 mM (1-CF-AuNPs@MoS2MS), three types of AuNPs have formed on the nanosheets due to the abundant lattice imperfections (Fig. 2a and Fig. S3a): i) AuNPs of about 20 nm in diameter located on the topside of the nanosheets; ii) AuNPs of 10-20 nm in diameter located on the side-surface of the nanosheets; iii) AuNPs of 10 nm in diameter located on the bottom-side between the two adjacent nanosheets. There would be no new Au3+ to complement the consumption in the growth process, resulting in a gradual decrease in Au3+ concentration in the whole reduction process and, accordingly, the size of the AuNPs decreases gradually top to bottom. For the HAuCl4 concentration of 2 mM (2-CF-AuNPs@MoS2-MS), the size of AuNPs located on the top-side increase to 50 nm (Fig. 2b), and the shortened distance between two adjacent AuNPs can efficiently improve the enhancement of electromagnetic field. The AuNPs located on the top-side of the nanosheets have the priority in growth process due to the larger contact-area

with Au3+ than the AuNPs on the side-surface, achieving via simultaneously assembling small and large AuNPs, respectively, on side-surface and the top-side followed Ostwald ripening in the later period of reduction reaction. For the HAuCl4 concentration of 3 mM (3-CFAuNPs@MoS2-MS), the size of the top-side AuNPs increases to 100 nm and the side-surface AuNPs arrays also present a 3D arrangement (Fig. 2c). The magnified FESEM further shown this 3D arrangement of the AuNPs arrays, and the side-surface AuNPs can be observed in the crevices between the top-side AuNPs (Fig. S3b). For the HAuCl4 concentration of 4 mM (4CF-AuNPs@MoS2-MS), adjacent AuNPs have aggregated into bigger AuNPs (Fig. 2d), and these too massy AuNPs may impede the SERS activity. The typical XRD pattern of MoS2-MS, 1-, 2-, 3-, and 4-CF-AuNPs@MoS2-MS are shown in Fig. 2e, respectively, marked with black, red, green, blue and pink colors. For the MoS2-MS, all peaks can be assigned to the MoS2 phase of (002), (100), (103) and (110) (JCPDS card no. 37-1492), and no residues or impurity phases can be detected, indicating the complete sulfuration of MoO42-. Through the combination of the mentioned-above HRTEM and XRD pattern, the P63/mmc (194) space group of MoS2-MS can be determined. For the XRD patterns of 1-, 2-, 3-, and 4-CF-AuNPs@MoS2-MS (Fig. 2e), the peak intensity of Au (111), (200), (220) and (311) increase with the increase of HAuCl4 concentration, indicating the increase of AuNPs content. Moreover, the peak intensity of MoS2 (002), (100), (103) and (110) all decrease with the increase of HAuCl4 concentration, which can be attributed to the increased defects with oxygen dissociate the Mo-S-Mo bond in the reaction process. Beyond that, an obvious color change was observed as the dispersion changed from yellow to purple with the concentration of HAuCl4 increasing and color change also appeared in the centrifugate (Fig. S4a). In agreement with FESEM images, and XRD patterns, the EDS spectra of the 1-, 2-, 3-, 4-CF-AuNPs@MoS2-MS also declare the change of elements involve in the reaction (Fig. S5). The peak intensity of Au element increases and, simultaneously, the peak intensity of Mo/S decreases, indicating the AuNPs have successfully grown on the hierarchical MoS2-MS. Same amount of various powders were dispersed in the same concentration and volume of MB solution (10-11 M), and dropwise added on the SiO2 substrate for SERS comparison. After a series of the SERS experiments, we finally determined that the strongest SERS ability owned to the 3-CF-AuNPs@MoS2-MS (Fig. 2f and Fig. 2g), and this is consistent with the simulations (more details are shown in Fig. S6). If there is no

special instruction, the mentioned CF-AuNPs@MoS2-MS are expressed in terms of 3-CFAuNPs@MoS2-MS. The side-view FESEM image displays the top-side AuNPs of 100nm in diameter (Fig. 3b) and, accordingly, the structure of the CF-AuNPs@MoS2-MS can be described with the first two schematics in Fig. 4a with front-view and vertical-view. It can be described that the AuNPs are bound with Au-S bond on the MoS2 nanosheets with size decreases gradually top to bottom. Importantly, the uniform-size AuNPs of 10-20 nm in diameter distribute in the interface of the adjacent CF-AuNPs@MoS2-MS can generate huge electromagnetic enhancement (Fig. 3c). In the magnified FESEM image in Fig. 3c (marked with red circle), the whole area in the interfacecanyon are covered with small AuNPs of 10nm in diameter (Fig. 3d). A part of the lower layer of AuNPs can be seen through the interspaces of the adjacent AuNPs at the top layer, and the overall structure can be described using the third schematic in Fig. 3a. TEM image can further demonstrate this 3D AgNPs arrays (Fig. 3e) and, obviously, the stacked-layers of AuNPs on MoS2 nanosheets can form 3D distribution nanogaps to support SERS activity. HRTEM image in the interface-canyon reveals close interfacial matching of the AuNPs and the disulfide nanostructures (Fig. 3f), and such a relationship could exist between the (111) planes of Au (0.235 nm) and the (103) planes (0.228 nm) of MoS2 due to a rather small mismatch of 3.1%. Moreover, it is obviously that the AuNPs direct semispherical growth onto MoS2 nanosheets rather than in the volume of the reactant solution, and further observation of the interface shows lattice fringe continuation as possibly indicative of epitaxial growth (Fig. 3g). When the probe molecules access in the surrounding surface of AuNPs, the crisscrossed 3D hot spots in nanogaps between AuNPs can provide huge enhancement effect to Raman signals under the excitation of the laser beam (the fourth schematic in Fig. 3a). In order to investigate the SERS activity of the CF-AuNPs@MoS2-MS, SERS experiments were performed under the same conditions. For the MB of 10-14M, clear Raman signals can be observed at both 633 and 532 nm excitation (Fig. S7a, the assignments of the Raman peaks are shown in Fig. S8). However, the obtained Raman signals have obvious differences at different laser wavelength excitation. For MB detection (10-14 M), the Raman intensity at 633 nm excitation are generally greater than that at 532 nm excitation, which is consistent with the fact that the absorption of MB at 633 nm is much stronger than at 532 nm

[44-46]. Moreover, there are existing some red-shift/blue-shift or merger phenomena in Raman peaks at different laser wavelength excitation. The background fluorescence at 633 nm excitation are generally stronger than that at 532 nm excitation, which can be attributed to the existence of MoS2 and the relatively stronger heat-effect belong to laser at 633 nm than 532 nm. For R6G detection (10-14 M), more merger phenomena of Raman peaks and stronger background fluorescence can be observed at 633 nm excitation than 532 nm excitation (Fig. S7b). The Raman intensity at 532 nm excitation is stronger than that at 633 nm excitation for R6G detection, which is attributed to the 633 nm is outside the absorption range of R6G [47]. For further investigation, SERS spectra were obtained from different concentrations of MB (Fig. 4a and Fig. 4b). For MB detection at 633 and 532 nm excitation, the minimum detectable concentration (MDC) can be as low as 10-15 and 10-14M, respectively. The attenuation of the Raman intensity along with the decrease of MB concentration endow the potential possibility of the CF-AuNPs@MoS2-MS in molecular quantitative detection. Although the MDC of MB at 633 nm excitation is lower than that at 532 nm excitation, the strong background noise should exert a passive influence to the signal stabilization. For the practical detection, 532 nm excitation seems more appropriate due to the weaker background noise and the nearly horizontal baseline. In order to investigate the relationship between MB concentration and Raman intensity, we selected two representative 449 and 1620 cm-1 peaks (Fig. S9a). For 449 and 1620 cm-1 peaks, the Raman intensity all present a good linear attenuation toward the decrease of the MB concentration, with the linear correction coefficient of 0.989 and 0.997, respectively. For R6G detection, the most suitable laser wavelength excitation is 532 nm with both stronger Raman intensity and weaker background noise (Fig. 4c and 4d). The MDC of R6G detection at both 633 and 532 nm excitation are 10-14 M, and two representative 615 and 1366 cm-1 peaks were selected to investigate the possible relationship between Raman intensity and R6G concentration (Fig. S9b). Similarly with MB detection, the good linearity between Raman intensity and R6G concentration, with linear correction coefficient of 0.995 and 0.997 for 615 and 1366 cm-1 peaks, respectively, indicating the potential of SERS technology in molecular quantitative detection. The good SERS activity of CF-AuNPs@MoS2-MS stems from two mechanism: i) the huge electromagnetic enhancement form AuNPs; ii) the additional chemical enhancement from MoS2 (include both charge transfer and dipole-dipole coupling),

due to the polar covalent bond (Mo-S) with the polarity in the vertical direction to the surface [21]. We also used this method for three centrifugal supernatant samples detection of the human morning urine, and variety of metabolites can be detected based on the CF-AuNPs@MoS2-MS powders (Fig. S10). The Raman peaks are in accordance with reported in the published articles, and assignments of these peaks are shown in upper-right inset in Fig. S10 [48, 49]. Obviously, the peaks are not uniform in strength, which may resulted from the differences in each volunteer's physical state. Therefore, for molecular detection, the dispersible SERS powder should be superior to the traditional rigid SERS substrate in the reliability and factuality. In the practical application of the molecular detection, multi-analyte sample is very common in many testing scenarios, and herein, single molecule detection was demonstrated with bi-analyte SERS (BIASERS) method by using two different molecules at the same time. In the BIASERS method, the CF-AuNPs@MoS2-MS were distributed evenly in an aqueous solution of MB and R6G mixture (5×10-14 M of both MB and R6G), and dropwise added on the SiO2 substrate for SERS detection at 532 nm excitation. Optical image of a randomly selected area is shown in Fig. 5a, and the distribution of SERS signals was evaluated via point by point scanning Raman spectra with the step-size of 0.5 µm in this selected-area. The Raman mapping of MB at 1620 cm-1 and R6G at 1654 cm-1 shows the same boundary with the selectedarea (marked with white circle in Fig. 5a, Fig. 5b and Fig. 5c), indicating the CFAuNPs@MoS2-MS distribution provide strong support to the hot spots distribution in the Raman mapping. The Raman intensity relatively strong in the center and relatively weak around the boundary, may be resulted from the CF-AuNPs@MoS2-MS distribution relatively thick in the center and relatively thin around the boundary. For the Raman mapping of R6G at 1654 cm-1, there are many feeble hot spots outside this selected area (the crimson areas in Fig. 5c outside the white circle), which resulted from the diffused small CF-AuNPs@MoS2-MS. The Raman spectra obtained from the green point in Fig. 5a, Raman spectra of pure MB of 5×10-14 M, and Raman spectra of pure R6G of 5×10-14 M are respectively shown top to bottom in Fig. 5d. The Raman peaks assigned to MB and R6G, respectively, are marked with the blue and red arrows, indicating the capability of the CF-AuNPs@MoS2-MS for SERS detection in the bi-analyte sample. The CF-AuNPs@MoS2-MS can be easily assembled in variety of flexible substrates, and

herein, we selected the CAM (size of 47 mm, aperture of 0.45 μm) as the supporter. A series of flexible SERS substrates consisted of CF-AuNPs@MoS2-MS inset CAM were fabricated via vacuum filtration method (Fig. 6a). For the 10-12 M MB detection on CAM-0, no legible SERS peaks but background noise can be obtained, indicating that enhancement of Raman signals only rely on MoS2-MS is far from enough (Fig. 6b). By contrast, for all the CAMs filled with CF-AuNPs@MoS2-MS (CAM-1, -2, -3, -4, -5), the clear Raman signals can be observed, indicating the significant SERS activity of the CF-AuNPs@MoS2-MS (Fig. 6b). Obviously, with increase of the thickness (i.e., powders mass), Raman intensity rises first and fall later. The reason may be that the too small powders mass can lead to the incompletely-covered on CAM surface, and the too large powders mass can prevent probe molecules from entering nanogaps. Among all the as-prepared CAMs, the CAM-4 shows the strongest SERS ability, and we thus selected CAM-4 to further investigate the SERS activity. For adenosine detection (concentration of 10-10 M), Raman peaks obtained on pure CAM are indistinguishable with only background noise existing and, by contrast, Raman peaks on CAM-4 are clear and uniform, indicating the good SERS activity of this flexible SERS substrate (Fig. S11). 4. Conclusions In conclusion, we have synthesized a novel 3D SERS powder enabled for molecular detection based on the hierarchical MoS2-MS decorated with CF-AuNPs. The average size of the CF-AuNPs@MoS2-MS can be tailored through the molar ratio between MoS2-MS and HAuCl4. The spatial-distribution nanogaps can supply intensive 3D hot spots, and not only have 10-14 M R6G and 10-15 M MB been identified with different laser wavelength excitation, but also various metabolites in human early-morning urine were recognized. Moreover, the wafer-scale flexible SERS substrates were fabricated using vacuum filtration method, showing good SERS activity for molecular detection. The high-efficiency SERS detection in various analyte samples, even multi-analyte sample, shows the practicability of CF-AuNPs@MoS2MS in biochemical detection. Acknowledgements The authors gratefully acknowledge financial support from NSFC Major Research Program on Nanomanufacturing (Grant No. 91323303), Natural Science Foundation of China (NSFC, Grant No. 51572216 and 61604122), the industrial science and technology research project in

Shanxi province (2015GY005), and 111 Program (No. B14040). We thank Ms. Yu Wang at Instrument Analysis Center of Xi’an Jiaotong University for her assistance with Raman measurements. We also thank Prof. Luyi Sun in University of Connecticut, and Prof. Shouzhen Jiang in Shandong Normal University, for their helpful discussions. The SEM and TEM works were done at International Center for Dielectric Research (ICDR), Xi'an Jiaotong University, Xi’an, China.

References [1] S. Schlucker, Surface-enhanced Raman spectroscopy: concepts and chemical applications, Angew. Chem. Int. Ed. 53 (2014) 4756-4795. [2] S. Nie, S. R. Emory, Probing single molecules and single nanoparticles by surface-enhanced Raman scattering, Science 275 (1997) 1102-1106. [3] K. Kneipp, Y. Wang, H. Kneipp, L. T. Perelman, I. Itzkan, R. R. Dasari, M. S. Feld, Single molecule detection using surface enhanced Raman scattering (SERS), Phys. Rev. Lett. 78 (1997) 1667-1670. [4] N. P. W. Pieczonka, R. F. Aroca, Single molecule analysis by surfaced-enhanced Raman scattering, Chem. Soc. Rev. 37 (2008) 946-954. [5] J. N. Anker, W. P. Hall, O. Lyandres, N. C. Shah, J. Zhao, R. P. Van Duyne, Biosensing with plasmonic nanosensors, Nat. Mater. 7 (2008) 442-453. [6] Y. Sun, K. Liu, J. Miao, Z. Wang, B. Tian, L. Zhang, Q. Li, S. Fan, K. Jiang, Highly sensitive surface-enhanced Raman scattering substrate made from superaligned carbon nanotubes, Nano Lett. 10 (2010) 1747-1753. [7] Y. Wang, W. Ruan, J. Zhang, B. Yang, W. Xu, B. Zhao, J. R. Lombardi, Direct observation of surface-enhanced Raman scattering in ZnO nanocrystals, J. Raman Spectrosc. 40 (2009) 1072-1077. [8] Y. Wang, Z. Sun, H. Hu, S. Jing, B. Zhao, W. Xu, C. Zhao, J. R. Lombardi, Raman scattering study of molecules adsorbed on ZnS nanocrystals, J. Raman Spectrosc. 38 (2007) 34-38. [9] A. Musumeci, D. Gosztola, T. Schiller, N. M. Dimitrijevic, V. Mujica, D. Martin, T. Rajh, SERS of semiconducting nanoparticles (TiO2 hybrid composites), J. Am. Chem. Soc. 131 (2009) 6040-6041. [10] A. Kudelski, W. Grochala, M. Janik-Czachor, J. Bukowska, A. Szummer, M. Dolata, Surface-enhanced Raman scattering at (SERS) copper (I) oxide, J. Raman Spectrosc. 29 (1998) 431-435. [11] Y. Wang, H. Hu, S. Jing, Y. Wang, Z. Sun, B. Zhao, C. Zhao, J. R. Lombardi, Enhanced Raman scattering as a probe for 4-mercaptopyridine surface-modified copper oxide nanocrystals, Anal. Sci. 23 (2007) 787-791.

[12] J. F. Li, Y. F. Huang, Y. Ding, Z. L. Yang, S. B. Li, X. S. Zhou, F. R. Fan, W. Zhang, Z. Y. Zhou, D. Y. Wu, B. Ren, Z. L. Wang, Z. Q. Tian, Shell-isolated nanoparticle-enhanced Raman spectroscopy, Nature 464 (2010) 392-395. [13] X. Wang, X. Kong, Y. Yu, H. Zhang, Synthesis and characterization of water-soluble and bifunctional ZnO-Au nanocomposites, J. Phys. Chem. C 111 (2007) 3836-3841. [14] J. G. Fan, Y. P. Zhao, Gold-coated nanorod arrays as highly sensitive substrates for surfaceenhanced Raman spectroscopy, Langmuir 24 (2008) 14172-14175. [15] X. Li, G. Chen, L. Yang, Z. Jin, J. Liu, Multifunctional Au-coated TiO2 nanotube arrays as recyclable SERS substrates for multifold organic pollutants detection, Adv. Funct. Mater. 20 (2010) 2815-2824. [16] S. M. Morton, L. Jensen, Understanding the molecule-surface chemical coupling in SERS, J. Am. Chem. Soc. 131 (2009) 4090-4098. [17] B. Zhang, H. Wang, L. Lu, K. Ai, G. Zhang, X. Cheng, Large-area silver-coated silicon nanowire arrays for molecular sensing using surface-enhanced Raman spectroscopy, Adv. Funct. Mater. 18 (2008) 2348-2355. [18] C. Cheng, B. Yan, S. M. Wong, X. Li, W. Zhou, T. Yu, Z. Shen, H. Yu, H. J. Fan, Largearea silver-coated silicon nanowire arrays for molecular sensing using surface-enhanced Raman spectroscopy, ACS Appl. Mater. Interfaces 2 (2010) 1824-1828. [19] X. Ling, L. Xie, Y. Fang, H. Xu, H. Zhang, J. Kong, M. S. Dresselhaus, J. Zhang, Z. Liu, Can graphene be used as a substrate for Raman enhancement?, Nano Lett. 10 (2010) 553561. [20] S. Huh, J. Park, Y. S. Kim, K. S. Kim, B. H. Hong, J. M. Nam, UV/ozone-oxidized largescale graphene platform with large chemical enhancement in surface-enhanced Raman scattering, ACS Nano 5 (2011) 9799-9806. [21] X. Ling, W. J. Fang, Y. H. Lee, P. T. Araujo, X. Zhang, J. F. Rodriguez-Nieva, Y. X. Lin, J. Zhang, J. Kong, M. S. Dresselhaus, Raman enhancement effect on two-dimensional layered materials: graphene, h‑ BN and MoS2, Nano Lett. 14 (2014) 3033-3040. [22] P. Wang, O. Liang, W. Zhang, T. Schroeder, Y. H. Xie, Ultra-sensitive graphene-plasmonic hybrid platform for label-free detection, Adv. Mater. 25 (2013) 4918-4924. [23] W. G. Xu, J. Q. Xiao, Y. F. Chen, Y. B. Chen, X. Ling, J. Zhang, Graphene-veiled gold

substrate for surface-enhanced Raman spectroscopy, Adv. Mater. 25 (2013) 928-933. [24] L. F. Sun, H. L. Hu, D. Zhan, J. X. Yan, L. Liu, J. S. Teguh, E. K. L. Yeow, P. S. Lee, Z. X. Shen, Plasma modified MoS2 nanoflakes for surface enhanced Raman scattering, Small 10 (2014) 1090-1095. [25] S. Su, C. Zhang, L. H. Yuwen, J. Chao, X. L. Zuo, X. F. Liu, C. Y. Song, C. H. Fan, L. H. Wang, Creating SERS hot spots on MoS2 nanosheets with in situ grown gold nanoparticles, ACS Appl. Mater. Interfaces 6 (2014) 18735-18741. [26] Y. Fang, N. H. Seong, D. D. Dlott, Measurement of the distribution of site enhancements in surface-enhanced Raman scattering, Science 321 (2008) 388-392. [27] S. Lal, N. K. Grady, J. Kundu, C. S. Levin, J. B. Lassiter, N. J. Halas, Tailoring plasmonic substrates for surface enhanced spectroscopies, Chem. Soc. Rev. 37 (2008) 898-911. [28] M. L. Pedano, S. Z. Li, G. C. Schatz, C. A. Mirkin, Periodic electric field enhancement along gold rods with nanogaps, Angew. Chem. Int. Ed. 49 (2010) 78-82. [29] S. Vantasin, W. Ji, Y. Tanaka, Y. Kitahama, M. Wang, K. Wongravee, H. Gatemala, S. Ekgasit, Y. Ozaki, 3D SERS imaging using chemically synthesized highly symmetric nanoporous silver microparticles, Angew. Chem. 128 (2016) 8531-8535. [30] C. Srichan, M. Ekpanyapong, M. Horprathum, P. Eiamchai, N. Nuntawong, D. Phokharatkul, P. Danvirutai, E. Bohez, A. Wisitsoraat and A. Tuantranont, Highly-sensitive surface-enhanced Raman spectroscopy (SERS)-based chemical sensor using 3D graphene foam decorated with silver nanoparticles as SERS substrate, Sci. Rep. 6 (2016) 23733. [31] H. B. Tang, G. W. Meng, Q. Huang, Z. Zhang, Z. L. Huang, C. H. Zhu, Arrays of coneshaped ZnO nanorods decorated with Ag nanoparticles as 3D surface-enhanced Raman scattering substrates for rapid detection of trace polychlorinated biphenyls, Adv. Funct. Mater. 22 (2012) 218-224. [32] A. O. Altun, S. K. Youn, N. Yazdani, T. Bond, H. G. Park, Metal-dielectric-CNT nanowires for femtomolar chemical detection by surface enhanced Raman spectroscopy, Adv. Mater. 25 (2013) 4431-4436. [33] C. H. Zhu, G. W. Meng, Q. Huang, X. J. Wang, Y. W. Qian, X. Y. Hu, H. B. Tang, N. Q. Wu, ZnO-nanotaper array sacrificial templated synthesis of noble-metal building-block assembled nanotube arrays as 3D SERS-substrates, Nano Res. 8 (2015) 957-966.

[34] J. W. Jeong, M. M. P. Arnob, K. M. Baek, S. Y. Lee, W. C. Shih, Y. S. Jung, 3D crosspoint plasmonic nanoarchitectures containing dense and regular hot spots for surfaceenhanced Raman spectroscopy analysis, Adv. Mater. 28 (2016) 8695-8704. [35] X. H. Li, G. Y. Chen, L. B. Yang, Z. Jin, J. H. Liu, Multifunctional Au-coated TiO2 nanotube arrays as recyclable SERS substrates for multifold organic pollutants detection, Adv. Funct. Mater. 20 (2010) 2815-2824. [36] S. Lee, M. G. Hahm, R. Vajtai, D. P. Hashim, T. Thurakitseree, A. C. Chipara, P. M. Ajayan, J. H. Hafner, Utilizing 3D SERS active volumes in aligned carbon nanotube scaffold substrates, Adv. Mater. 24 (2012) 5261-5266. [37] H. C. Jeon, C. J. Heo, S. Y. Lee, S. M. Yang, Hierarchically ordered arrays of noncircular silicon nanowires featured by holographic lithography toward a high-fidelity sensing platform, Adv. Funct. Mater. 22 (2012) 4268-4274. [38] M. S. Schmidt, J. Hubner, A. Boisen, Large area fabrication of leaning silicon nanopillars for surface enhanced Raman spectroscopy, Adv. Mater. 24 (2012) OP11-OP18. [39] Y. J. Oh, K. H. Jeong, Glass nanopillar arrays with nanogap-rich silver nanoislands for highly intense surface enhanced Raman scattering, Adv. Mater. 24 (2012) 2234-2237. [40] X. M. Li, M. H. Bi, L. Cui, Y. Z. Zhou, X. W. Du, S. Z. Qiao, J. Yang, 3D aluminum hybrid plasmonic nanostructures with large areas of dense hot spots and long-term stability, Adv. Funct. Mater. (2017) 1605703. [41] Y. Wang, L. Yu, X. W. Lou, Synthesis of highly uniform molybdenum-glycerate spheres and their conversion into hierarchical MoS2 hollow nanospheres for lithium-ion batteries, Angew. Chem. Int. Ed. 55 (2016) 7423-7426 [42] J. Kim, S. Byun, A. J. Smith, J. Yu, J. Huang, Enhanced electrocatalytic properties of transition-metal dichalcogenides sheets by spontaneous gold nanoparticle decoration, J. Phys. Chem. Lett. 4 (2013) 1227-1232. [43] A. Y. Polyakov, L. Yadgarov, R. Popovitz-Biro, V. A. Lebedev, I. Pinkas, R. Rosentsveig, Y. Feldman, A. E. Goldt, E. A. Goodilin, R. Tenne, Decoration of WS2 nanotubes and fullerene-like MoS2 with gold nanoparticles, J. Phys. Chem. C 118 (2014) 2161-2169. [44] K. Y. Jacobs, R. A. Schoonheydt, Spectroscopy of methylene blue-smectite suspensions, J. Colloid Interf. Sci. 220 (1999) 103-111.

[45] K. Y. Jacobs, R. A. Schoonheydt, Time dependence of the spectra of methylene blue-clay mineral suspensions, Langmuir 17 (2001) 5150-5155. [46] S. H. A. Nicolai, J. C. Rubim, Surface-enhanced resonance Raman (SERR) spectra of methylene blue adsorbed on a silver electrode, Langmuir 19 (2003) 4291-4294. [47] J. Guthmuller, B. Champagne, Resonance Raman scattering of rhodamine 6G as calculated by time-dependent density functional theory: vibronic and solvent effects, J. Phys. Chem. A 112 (2008) 3215-3223. [48] A. Keuleers, H. O. Desseyn, B. Rousseau, C. Van Alsenoy, Vibrational analysis of urea, J. Phys. Chem. A 103 (1999) 4621-4630. [49] W. R. Premasiri, R. H. Clarke, M. E. Womble, Urine analysis by laser Raman spectroscopy, Laser. Surg. Med. 28 (2001) 330-334.

Biographies Hengwei Qiu is now studying for a PhD degree in Electronic Materials Research Laboratory, Key Laboratory of the Ministry of Education & International Center for Dielectric Research, Xi’an Jiaotong University. His research interests are synthesis of graphene by using CVD technology and its applications, especially for biosensors. Minqiang Wang is a professor of Electronic Materials Research Laboratory, Key Laboratory of the Ministry of Education & International Center for Dielectric Research, Xi’an Jiaotong University. His research interests are synthesis and device applications of nanomaterials and quantum dots. Le Li is now studying for a PhD degree in Electronic Materials Research Laboratory, Key Laboratory of the Ministry of Education & International Center for Dielectric Research, Xi’an Jiaotong University. His research interests are synthesis of graphene and its applications. Junjie Li is now studying for a master degree in Electronic Materials Research Laboratory, Key Laboratory of the Ministry of Education & International Center for Dielectric Research, Xi’an Jiaotong University. His current research interest is synthesis of nanomaterials. Zhi Yang is a lecturer in Electronic Materials Research Laboratory, Key Laboratory of the Ministry of Education & International Center for Dielectric Research, Xi’an Jiaotong University. His research interests are synthesis and device applications of nanomaterials. Minghui Cao is now studying for a PhD degree in Electronic Materials Research Laboratory, Key Laboratory of the Ministry of Education & International Center for Dielectric Research, Xi’an Jiaotong University. Her research interests are 2D materials and their applications.

Fig. 1. (a) FESEM image of hierarchical MoS2-MS in a low magnification. Scale bar: 5 μm. (a) FESEM image of hierarchical MoS2-MS in a high magnification. Scale bar: 600 nm. (c) TEM image of hierarchical MoS2-MS. (d) and (e) HRTEM images of hierarchical MoS2-MS in different locations.

Fig. 2. FESEM images of the CF-AuNPs@MoS2-MS with different HAuCl4 concentrations, respectively are (a) 1 mM, (b) 2 mM, (c) 3 mM, and (d) 4 mM. Scale bar: 200 nm. (e) Typical XRD patterns of the pure MoS2-MS, 1-, 2-, 3-, and 4-CF-AuNPs@MoS2-MS. (f) SERS activity of the pure MoS2-MS, 1-, 2-, 3-, and 4-CF-AuNPs@MoS2-MS for MB detection (10-11 M). (g) Raman intensity at peaks of 447, 499, 700 and 1620 cm-1 for different composites in (f). Inset: Photograph of a cauliflower.

Fig. 3. (a) Structure schematics of CF-AuNPs@MoS2-MS in different positions. (b) Side-view FESEM images of CF-AuNPs@MoS2-MS. Scale bar: 300 nm. (c) and (d) FESEM image in the interface of adjacent CF-AuNPs@MoS2-MS. Scale bar: (c) 300 nm and (d) 100 nm. (e) TEM image of CF-AuNPs@MoS2-MS. (f) and (g) HRTEM images of CF-AuNPs@MoS2-MS.

Fig. 4. SERS spectra of MB with different concentration for laser wavelength of (a) 633nm and (b) 532 nm. SERS spectra of R6G with different concentration for laser wavelength of (c) 633 and (d) 532 nm.

Fig. 5. (a) Optical image of MB&R6G&CF-AuNPs@MoS2-MS powders (5×10-14 M of both MB and R6G) mixture on SiO2 substrate. (b) Raman mapping of MB peak at 1620 cm-1 of the point by point scanning with the step-size of 0.5 µm. (c) Raman mapping of R6G peak at 1654 cm-1 of the point by point scanning with the step-size of 0.5 µm. (d) Top to bottom: Raman spectra of the point marked with green point in (b); Raman spectra of pure MB of 5×10-14 M; Raman spectra of pure R6G of 5×10-14 M.

Fig. 6. (a) Schematic of the vacuum filtration system. (b) SERS activity of the six different CAM flexible SERS substrates for MB detection in 10-12 M.