Observation of single-molecule Raman spectroscopy enabled by synergic electromagnetic and chemical enhancement

,


Introduction
Raman spectroscopy is a traditional and powerful means to observe vibrational, rotational, and other low-frequency modes in materials via inelastic scattering against light, and enable molecule and material analysis and identification [1][2][3][4][5][6][7].After 90 years of development, it still attracts tremendous interest to probe single-molecule (SM) Raman where I R and I E are the Raman intensity for N vol molecules and enhanced Raman intensity for N surf molecules, respectively.Since Nie, Kneipp and their co-workers claimed SM Raman detection for the first time in 1997 [11,12], and Xu's team reported a 10 10 EF for single hemoglobin molecule in 1999 [13], all through the scheme of surface-enhanced Raman spectroscopy (SERS) mediated by surface plasmon polariton (SPP) excitation and action, multiple researches paving towards the road of SM-SERS have been accomplished [14][15][16][17][18][19][20][21][22][23][24], among these, the highest EF is 10 12 [21,25].
It is generally acknowledged that Raman signal needs a huge EF of ~ 14-15 orders of magnitude to maintain a signal intensity comparable with molecule fluorescence, which allows one to truly enter the regime of single-molecule detection sensitivity [11,12].This criterion has set a huge obstacle to frustrate and obstruct most contemporary macroscopic, mesoscopic, and microscopic strategies of Raman scattering enhancement [7], including two most prestigious schemes, the electromagnetic enhancement (EME) and chemical enhancement (CME) [26][27][28].EME is mainly induced by mesoscopic local surface plasmon resonance (LSPR) effect in metallic nanoparticles and nanostructures with their free electron gas violently interacting with incident laser light, leading to occurrence of great electric field intensity and Raman enhancement in specific nanoscale regions called hot spot [29,30].CME is mainly a microscopic effect where the molecule Raman cross section (or Raman activity) is enlarged by the formation of loose chemical bond between molecule and surrounding background and consequent action of additional pathways of charge-transfer resonance [30,31].
The above argument can be placed into a more clarified physical model.The signal intensity of enhanced Raman scattering is expressed as.
where A is a coefficient related in practice with the collection efficiency of the optical system used to collect the Raman signal, I 0 (r 0 , ω) is the intensity of incident light, α(ω R , ω) is the Raman polarizability (proportional to Raman activity or cross section), of the detected molecule, and |E(r 0 ,ω)| 4 |E 0 (r 0 ,ω)| 4 corresponds to the famous local field enhancement factor [9]. Equation ( 2) has implied that one can harness the macroscopic strategy of Raman instrument improvement (first term), the microscopic strategy of molecule Raman polarizability enhancement (second term, CME), and the mesoscopic strategy of local field enhancement (third term, EME) to push up the overall enhancement of Raman detection down to single-molecule (SM) level.Since the space for macroscopic instrument performance improvement is very limited, one should focus to dig into mesoscopic EME and microscopic CME for SM Raman detection.Previous studies have shown that EME can become large in SERS substrates made from noble metal Au and Ag single nanoparticles with sharp tips, corners, and edges [32,33], or nanoparticle dimers or aggregates with nanoscale gap between them [34][35][36], or in TERS configuration with a sharp Ag/Au tip sitting above planar Ag/Au substrates [37][38][39].So far, the biggest EME occurs in Ag/Au nanoparticle dimers or aggregates, which can reach [10][11] (1) orders of magnitude, still 3-5 orders of magnitude shorter than the SM detection limit.Compared with EME, CME achieves a relatively low enhancement factor.In usual Ag/ Au nanoparticle SERS substrates, CME is present together with EFE, but the magnitude is only 10-100.In some carbon nano-porous substrates [40], CME can reach a very high value of 10 6 , but unfortunately, the CME is negligibly small here.It is clear that either EME or CME alone is unable to make SM Raman detection, but synergic action of simultaneous strong EME and CME might be feasible to bring reality the dream of SM Raman detection.
In this work we harness both EME and CME to reach a very high level of Raman enhancement simultaneously.We successfully design and implement a 2D materialplasmon nanogap composite nanoscale system and make unambiguous experimental observation of single-molecule Raman spectroscopy of RhB molecule.The RhB molecules are absorbed on two-dimensional monolayer WS 2 sheets, within the plasmonic nanogap formed by gold nanoparticle sitting upon 2 nm SiO 2 coated gold thin film on a 5mm × 5mm square silicon wafer as shown in Fig. 1.Raman spectroscopy down to an extremely dilute value of 10 -18 mol/L can be clearly visible, and the statistical enhancement factor could reach 16 orders of magnitude compared with the reference detection sample of silicon plate with a detection limit of 10 -2 mol/L.The electromagnetic enhancement comes from local plasmon resonance induced at the nanogap, which could reach ~ 10-11 orders of magnitude, while the chemical enhancement comes from monolayer WS2 2D material, which could reach 4-5 orders of magnitudes.

Results
The first step to confirm the power of our SM-SERS strategy is to find a good 2D material exhibiting strong CME.We proceeded a series of tests on Mxenes, black phosphorus and transition-metal dichalcogenides (TMDCs), which are promising to have Raman enhancement.A kind of TMDCs WS 2 is proved to be capable of 10 4 ~ 10 5 Raman magnification for RhB, while the other tested materials did not show enhancement factor higher than 10 2 , so we estimate WS 2 as the most promising CME substance for RhB SM Raman probing.RhB molecules are attached on monolayer WS 2 , and afterwards laid on a 2 nm SiO 2 coated gold thin film.Finally, gold nanoparticles are added to form nanogap with the Au plating (Fig. 2a).Optical microscope image shows the tight attachment of WS 2 slices on the base (Fig. 2b), and scanning electron microscopy (SEM) picture in Fig. 2c indicates that the Au nano-particles are distributed barely evenly on the monolayer WS 2 (each with area of one square micrometer).The low wave number Raman spectra as illustrated in Fig. S4b indicates that the WS 2 sheet in the working hot spot is monolayer as the intensity of 356 cm −1 ( E 1 2g (Ŵ )) is higher than 417.5 cm −1 ( A 1g ( Ŵ )) [41].Energy dispersive spectroscopy (EDS) elemental mapping clearly demonstrates that the tungsten element is distributed on the gold film uniformly (Fig. 2d), in another word, the monolayer WS 2 is uniformly attached to the base [42].
The spectra in Fig. 3 illustrates the CME effect of monolayer WS 2 for RhB with various concentrations, which indicate the lowest detectable threshold of pure RhB is 10 -2 M in a silicon substrate with neither EME nor CME for Raman scattering.With the aid of the WS 2 CME, the detection limit of RhB molecule can reach 10 -7 M, meanwhile the Raman intensity of 10 -6 M RhB on WS 2 is of same order of magnitude with 10 -2 M sole RhB on silicon wafer.Thus, the CME factor of monolayer WS 2 can be defined as 10 4 ~ 10 5 .
CME for WS 2 based SERS is dependent on the electronic structure of the interface between the RhB molecules and the WS 2 , therefore the molecules layer closest by the substrate is the key crucial to CME, which is generally defined as "first layer effect" [43].The RhB molecule contains a chloride anion and a cation framework, thus the molecule frameworks can attach tightly to the WS 2 , which has abundant negative charge on the surface.Moreover, the charge transfer (CT) mechanism tends to appear between the adsorbed RhB molecules and WS 2 mono-layer material.On the basis of frontier molecular orbital theory, the electrons with different energies in molecules will be assigned into different molecular orbital energy levels [44], among which the highest occupied molecular orbital (HOMO) and the lowest unoccupied molecular orbital (LUMO) are dominating in the interaction with other matters [45,46].As discussed in Supplementary Information S1, the calculated molecular orbitals of RhB and energy band of monolayer WS 2 indicates that the valence band of WS 2 is totally included between the HOMO and LUMO of RhB.As a result, charge carriers in the valence band can provide charges to the HOMO through charge transfer (CT), and enhance LUMO via CT resonance in the meantime (Supplementary Information, Figs.S1 and S2).The extremely high carrier mobility of WS 2 (Supplementary Information, Table S1) further amplifies the effect of CT and CT resonance, so that finally a CME of 10 4 ~ 10 5 can be achieved [47][48][49][50].
Because of the charge transfer between the RhB molecules and WS 2 , several peaks in the WS 2 enhanced Raman spectra of RhB fluctuate or/and shift compared with the normal spectrum, which are declared by grey dashed lines in Fig. 3. Let "X" represents the xanthene ring, "M" represents the methyl, and "D" represents the diethylamino group, respectively.WS 2 enhanced Raman spectra of RhB differ from the ordinary spectrum in the following details.The bands of C X -C X -C X bending vibrations are not shifted to different wavenumbers, while the broad band at 613 cm -1 turns to an extremely narrow and strong peak and the band at 632 cm -1 decreases dramatically.In the case of the C X -H in-plane bending vibrations, the band at 717 cm -1 almost disappears, the band at 1136 cm -1 gets weaker and shifts to the lower wavenumbers.In the mean time, the band at 761 cm -1 standing for C X -H out-of-plane bend is shifted to higher wavenumbers at 773 cm -1 , while the C X -C X stretching vibrations (1193, 1367, 1510, 1560 cm −1 ) are shifted to lower wavenumbers (1182, 1359, 1499, 1542 cm −1 ).Meanwhile the C X -N stretching band appears around 1389 cm -1 .The above results suggest the spatial orientation reform of the RhB molecule to the WS 2 substrates, and CT enhancement takes place on the interface [51].It can be explained that the "X" is approximately parallel to the surface when RhB is adsorbed on WS 2 , hence, the C X -C X stretching bands are highly enhanced other than the C X -N stretching band and the C X -H bending vibrations [52].
Figure 3 also displays a typical experimental result of the EME effect for the Au plasmonic nanogap against Raman scattering for RhB molecule.The yellow curve in the fifth panel of Fig. 3 indicates that the EME factor of RhB by the LSPR excited within the Au nanogap nanocavity can reach a very high level of about 4.5 x 10 9 .For a more quantitative comparison and justification, we have carried out three-dimensional finite-difference time-domain (3D-FDTD) simulations to evaluate the electromagnetic enhancement factors with and without the WS 2 layer and the RhB molecule.The results are discussed in details in Supplementary Information S2.The simulation results show that the existence of WS 2 layer can enhance the effect of local electromagnetic field, so can the Rhodamine B molecule.Summarily, the maximum electromagnetic enhancement factor can approach to the order of 10 9 according to the FDTD simulations.However, in practical situations, the maximum EF values have the capacity to reach an even larger orders of magnitude when taking into account more complex sample surface and optical field pattern under experiment conditions.For instance, the roughness and shape anomaly of the gold nanoparticles and plating could lead to a larger EME.Besides, molecular Raman polarizability could become higher due to the quantum confinement from WS 2 film and nano-gap upon RhB molecule.Consequently, the overall EME factor can be levitated up to the level of 10 10 -10 11 .In fact, this physical picture has been supported to some extent by the comparison between experiment and theory for pure Au nanogap EME without WS 2 incorporated.Notice that the experimental EME factor can reach 4.5 x 10 9 , while the theoretical EME factor is at least one order of magnitude lower.If one considers possible larger Raman polarizability of RhB molecule under tight quantum confinement within the nanogap, the calculated EME factor may increase to some extent and better match the experimental EME factor.
The above results reveal that with matched 2D semiconductor materials, the Raman signal of molecule can be amplified 10 5 times due to the chemical interaction between substances, known as the CME of SERS.Whereas the EME mechanism does not rely on the microscopic-level light-matter interactions, it is more like an environment construction: noble metallic nanoparticles and nanostructures form nanogaps with hot spots, where enormous electric field intensity and Raman enhancement take place in this specific nanoscale regions due to the LSPR [53,54].Thus, through appropriate combination, the CME and EME can make a synergy to gain single-molecule level Raman detection power.A series of control-experiments is carried out to verify the feasibility.
The experiments illustrating the SM-SERS power for the Au nanaogap-WS 2 2D material substrate are taken on 5 mm × 5 mm chips for each sample.Every chip is added by 30 μL RhB solution with different concentrations.When the concentration reaches down to 10 -18 M, there are only 18 molecules on one chip on average.The distribution of Au nanoparticles in samples with gold nanogap is about one every square micron.The scanning scope for the samples is 40 μm × 60 μm, the sampling step is 1 μm.The diam- eter of facula area is easily calculated as 1.298 μm.The weakest detected Raman signal of 10 -6 ~ 10 -20 M RhB samples are shown in Fig. 4a, which obviously indicates that with the lowest concentration detection limit as 10 -18 M, the 2D WS 2 material incorporated gold nanogap SERS substrates have successfully entered the realm of SM Raman detection.
It is noteworthy that under ultra-low concentrations, even when the amount of total molecules is increasing, the Raman signal remains in the SM level.This is because only the molecules which have obtained both EME and CME simultaneously can be detected.Besides, as the above results have shown, the detection limit of CME is just 10 -7 M, and within the incident laser region, there is only one EME hot spot which can contain just an individual RhB molecule [12].Therefore, when the RhB concentration reaches 10 -6 M, the signal intensity begins to grow and the value is approximately equal to the sum of former-mentioned SM signal and 10 -6 M RhB on WS 2 in Fig. 3, as the other exposed molecules initiate to contribute observable Raman signal together with the single molecules sitting inside the hot-spot.A 40 μm × 60 μm range in the center of each sample is scanned, and the achieved Raman mapping pictures of the intensity at 1650 cm −1 peak are exhibited in Fig. 4b.Furthermore, the whole chip of 10 -20 M sample is completely scanned, and no signal could be detected.When the quantity of molecules rises, the amount of active hot spot also increases to induce more SM signals.Taken together, by the synergy of CME and EME, unambiguous RhB SM Raman detection has been achieved.Under extremely low concentration, RhB molecules are dispersed in the form of single molecule.Because the LSPR in nanogap only occurs in a small region for only a single molecule, Raman signal detected in low concentration samples can only be attributed to SM signals.Meanwhile, the increase of molecule amount will result in more active hot spots to take action and thus more intense Raman signal to be detected.As discussed in Supplementary Information S3, we have compared the Raman mapping picture and the corresponding photomicrograph (Fig. S4a as an example), and confirmed that a particle sitting right upon the molecule absorbed WS 2 is the essential requirement to achieve SM-SERS signal, in another word, the synergy effect of EME and CME is vital to SM-SERS.To further verify the repeatability of our method, we randomly chose two points on several samples with signal detected.The detected curves are shown in Fig. S4b.The signal to noise ratio (Fig. S4c) of these curves are calculated as the ratio of the mean value to the standard deviation of them, which is fluctuating between 6 ~ 8.5.

Discussion
In conclusion, we have dug deeply into the two major enhancement mechanism for SERS, the EME and CME, explored to maximize the strength of each mechanism, and more importantly, and found out the optimal synergetic action of them, in a hope to fulfill the condition of SM Raman detection.We have found experimentally that 2D material monolayer WS 2 offers a very promising CME with enhancement factor up to 10 5 for absorbed RhB molecules, due to strong CT and CT resonance between the highmobility WS 2 substrate and RhB molecules.Besides, it is well-established that the plasmonic nanogap made from gold nanoparticles sitting on the gold thin film can offer a very promising EME with enhancement factor up to 10 10 for molecules absorbed within the nanogap hot spot under an appropriate gap separation size.We have designed and implemented a hybrid configuration of plasmonic nanogap formed by a gold nanoparticle sitting on the 2D monolayer WS 2 and 2 nm SiO 2 coated gold thin film, which in principle should enable synergic action of EME and CME with their individual enhancement factors multiplied.Indeed, our experiments have unambiguously shown that Raman spectroscopy for RhB molecules absorbed on the hybrid plasmonic nanogap can still be clearly observed with an extremely low concentration down to 10 -18 mol/L, and has successfully entered the realm of SM Raman spectroscopy.The statistical enhancement factor could reach an unprecedented high value as ~ 10 16 for this hybrid nano-device when compared with the reference detection sample of silicon plate.Moreover, the designed and implemented SM Raman base with so outstanding power of SM detection is very easy to fabricate.Thus, further researches on different molecules and applications are promising.Such a synergic route of Raman enhancement devices could open up a new frontier of single molecule science, allowing detection, identification, and monitor of single molecules and their spatial-temporal evolution under various internal and external stimuli.

Materials
Rhodamine B is obtained from Shanghai Adamas Reagent Co., Ltd.Mono-layer WS 2 dispersion is purchased from Nanjing MKNANO Tech.Co., Ltd.Gold nanoparticle colloid is achieved from Nanjing XFNANO Materials Tech Co., Ltd.The gold, chromium and SiO 2 sputtering target are acquired from Beijing Dream Material Technology Co., Ltd.All reagents are of analytical grade and used directly without further purification.Deionized water is produced through a Millipore water purification system (Milli-Q, Millipore) and used throughout the study.

Instruments and measurements
The SEM image and EDS elemental mapping are taken by a GemniSEM 500 electron microscope.The HR-TEM images are recorded using a JEOL-2010 electron microscope.The gold, chromium and SiO 2 coating on the SM Raman base are sputtered by an MSP-300B magnetron sputtering instrument.The thickness of coatings is measured by a Bruker DektakXT step profiler.All the Raman spectra are gained from a Renishaw inVia Raman spectrometer, laser: 532 nm edge (mode: Confocal), grating: 2400 l/mm (vis), exposure time for each point: 0.5 s, laser power: 2.5 mW, slit opening 20 μm, objective × 50 L.

Fabrication of the RhB SM Raman base
20 nm chromium, 300 nm gold and 2 nm SiO 2 are sputtered on a 5 mm × 5 mm square silicon slice, successively.0.2 mL WS 2 (0.1 mg/mL) dispersion is mixed with 2.8 mL RhB solution with specific concentration, the mixture is ultrasonic stirred for 30 min to make the RhB molecules absorbed adequately by the 2D material.Afterwards, 20 μL of the mixture is dropped uniformly on the coated slice.after 2 h vacuum drying under 40 °C, 20 μL gold nanoparticle (6 μg/mL) colloid is added on the aforementioned slice.Eventually, via 2 h vacuum drying beyond 40 °C, the RhB SM Raman base can be created.

Fabrication of the reference samples
RhB solutions with specific concentration is ultrasonic stirred for 30 min.Afterwards, 20 μL of the solutions are dropped uniformly on bare silicon plate.Finally, the samples are 2 h vacuum dried under 40 °C.

Fig. 1
Fig. 1 Schematic diagram of RhB SM-SERS Raman detection.The golden sphere on the top refers to the attached Au nanoparticle.The green region below is the hot spot of excitation laser.The black (carbon), mazarine (nitrogen) and red (oxygen) ball-and-stick model presents the RhB molecule.The yellow (sulfur) and violet (tungsten) ball-and-stick model shows the mono-layer WS 2 .The blue flash pattern illustrates the charge transfer between the RhB molecule and mono-layer WS 2 .The subjacent cuboid shows the silicon slice with several coatings

Fig. 2
Fig. 2 Fabrication and micromorphology of RhB SM Raman base.a Preparation progress of the sample.b Optical microscope and (c) SEM image of the base surface.d EDS elemental mapping of the gold nanoparticle on the base

Fig. 3
Fig.3Experimental characterization of WS 2 chemical enhancement and Au plasmonic nanogap electromagnetic enhancement for RhB Raman scattering.From top to bottom panels show the WS 2 enhanced (brown, red, orange and green lines) Raman spectra, Au nanogap enhanced Raman spectra (yellow line) and normal Raman spectra of RhB (blue line)

Fig. 4
Fig. 4 Experimental characterization of SM-SERS substrate.a Raman spectra of RhB with variable concentration in SM Raman base.b Raman mapping of the intensity at 1650 cm −1 peak of RhB at concentrations 10 -6 ~ 10 -20 M