Open Access
Add to BibSonomyAbstract
We presented a graphene-wrapped Cu particle hybrid system (G@Cu) to be used as a high performance surface-enhanced Raman scattering (SERS) substrate. The Cu particles wrapped by a few-layer graphene shell were directly synthesized on SiO2/Si by chemical vapor deposition assisted with thermal annealing in a mixture of methane and hydrogen. The detailed explanation on the different morphology of Cu particles induced by different thermal annealing condition was carried out with both qualitative and quantitative analysis and a series of G@Cu 3D models of Cu particles with different shape anisotropy and inter-particle gap were also built for further study of Raman enhancement mechanism. The G@Cu showed fine SERS activities, including the fluorescence quenching effect, the stability of Raman signals, chemical and optical stability, with an enhancement factor (EF) of ~1.5 × 106.
© 2016 Optical Society of America
1. Introduction
Surface-enhanced Raman scattering (SERS) has attracted considerable attention since its discovery in 1974, which is a powerful tool for ultrasensitive, single-level, and real-time detection [1–3]. Traditionally used SERS substrates are Au, Ag and Cu in the formation with rough surfaces or nanostructures, in which surface roughness or “hot spots” in nanostructures produce an enhancement of the electromagnetic field through a laser excited localized surface plasmon resonance [4–6]. This enhancement is known as electromagnetic enhancement (EM), which is proportional to fourth power of the intensity of the electromagnetic field. Besides EM, the chemical enhancement (CM) also contributes to SERS effect, which comes from charge transfer between analyst molecule and the substrate [7]. Despite considerable efforts, it is still a challenge to achieve ideal SERS substrates with good stability and reproducibility [8]. Actually, the issue of metal-molecule contact induced signal variations has become a key problem for practical applications [9, 10]. Furthermore, the high fluorescence and background noise on metal nanostructures for some molecules often limits their applications. Therefore, many efforts have been devoted to improve the chemical stability of substrate to avoid oxidation without compromising its SERS sensitivity. The atomic thickness, chemical stability, fluorescence quenching effect and CM enhancement [11–13] of graphene makes it a natural candidate material for SERS. SERS substrates with combination of metal nanoparticles and graphene have been reported [7, 14]. As for all the above-mentioned methods, the graphene-metal hybrids were obtained through transferring a graphene sheet on the surface of metal nanoparticles, which essentially belongs to physical composition, the tightly sealed structure is hard to form between metallic nanoparticles and graphene layer. The space between graphene and metal nanoparticles will inevitably cause apparent loss of electromagnetic enhancement activity because the electromagnetic enhancement efficiency decays rapidly with the distance away from the surface of the metal nanoparticles. Moreover, either dry or wet transfer can cause the graphene to be damaged [15], further reducing the homogeneity and reproducibility of the SERS substrates.
Therefore, fabrication of shell-isolated SERS substrate with graphene is a good way to solve those problems. In contrast, chemical vapor deposition (CVD) is a simpler and more efficient way to grow a thin layer of graphene on the surface of nanoparticles [16]. Considering that it is very difficult to grow high quality graphene on the Au or Ag substrate, Cu foil is a good choice for graphene growth. While for size effect of Cu particles, they cannot withstand the high temperature of graphene growth, so that a number of Cu particles are vaporized or melted, which leads to large inter-particle distance of Cu particles and low SERS activity.
In this work, we provided a direct growth approach to prepare a high performance SERS substrate with graphene-wrapped Cu particles (G@Cu) by a CVD method assisted with thermal annealing. Experimentally, characterization analysis of samples prepared with different thermal annealing condition was carried on using SEM (scanning electron microscopy), AFM (atomic force microscopy) and Raman measurement together with EF analysis. In addition, the calculated factor of fluorescence quenching (q) and oxygen content measured by EDS (Energy Dispersive Spectrometer) were used to evaluate the fluorescence quenching effect and stability of G@Cu substrate. Theoretically, detailed explanation on the different morphology of Cu particles induced by different thermal annealing condition was given from aspects of qualitative and quantitative analysis and a series of 3D G@Cu models with different shape anisotropy and inter-particle gap were built for further study of Raman enhancement mechanism.
2. Preparation
2.1. Preparation of G@Cu shell SERS substrate
The prime feature of the proposed approach includes preparation of Cu film by magnetron sputtering, formation of Cu particles in the early stage of CVD and the direct growth of graphene shell on the surface of Cu particles in the mixture of H2 and decomposed C2H4. Figure 1 shows the growth setup of G@Cu.
Fig. 1 (a) Preparation steps for the G@Cu composite as a SERS substrate, SEM images of (b) Cu film after magnetron sputtering, the height of Cu film is 192 nm, (c) G@Cu after CVD process, and (d) G@Cu after another annealing modification.
Deposition of Cu film: We deposited film consisting of 192 nm Cu onto a 500 nm SiO2 substrate using a magnetron sputtering system with a power of 100 W, pressure of 1 Pa, Ar gas of 40 sccm for 100/200/300 s.
Formation of Cu particles: The substrate was then placed in CVD system for the formation of Cu particles and graphene shell. In the early stage of CVD process, a mixed gas of H2 (40 sccm) and N2 (200 sccm) was filled into quartz tube until the inner pressure of the tube was back to atmospheric pressure. The output of the mixed gas was continuously maintained during all CVD process. Then we started to heat the tube at a heating rate of 50 °C/min until the temperature reached to 1030 °C. In this process, Cu film was evolved into particles.
Growth of graphene shell by CVD: In the following process, the temperature was maintained at 1030 °C for 10 min, and C2H4 (1 sccm) was introduced for the growth of the graphene shell. In this process, the Cu atoms were used as catalysts to decomposes C2H4, enabling a typical CVD reaction to grow a graphene layer on pre-formed Cu particles. Finally, the flow of C2H4 was switched off and the temperature was cooled to room temperature with the protection of N2 and H2. In addition, the traditional graphene grown on Cu foil via CVD was also transferred on SiO2/Si substrate as a comparison with the graphene grown by our method. The detailed description of the graphene transfer process can be found in the reference [7].
2.2. Annealing optimization of G@Cu SERS substrate
For further optimization, the formed G@Cu SERS substrate was further treated with annealing process, at temperature of 350/450/550/650/750 °C for 10/20/30/60 min with Ar and H2 as protection gas to remove metal oxides and other impurities. And the comparative samples of Cu particles were fabricated in the same process without graphene growth. For simplified, we named the samples with numbers under different annealing condition, shown in Table 1.
Table 1. Samples’ name and the corresponding preparation condition
2.3. Characterization and measurements
The morphology of samples was observed using a field emission scanning electron microscopy (FESEM, JEOL JSM-7800F). The Raman spectra were collected using a laser confocal Raman spectrometer (Horiba JY LabRAM HR Evolution), with a 100 × objective lens, numerical aperture (NA) of 0.9, work distance (WD) of 0.21 mm, and an air cooled double-frequency Nd:Yag green laser (λ = 532 nm, 50 mW with a 10% neutral density filter) with laser spot diameter of ~0.72 μm. An integration time of 2 s was used in all Raman measurement, noted that the intensity of incident light affects the plasmas generation in nano structures [17]. In order to avoid damage to the analyst molecules, a typical laser with low energy is used, there is no photoexcited plasmas in solids by athermal lattice disassembly and generation of WDM as intermediate states and nanoplasmas produced with intense laser fields of Irad ~1014–1017 W/cm2 and high-intensity (> 1018 W/cm2) photoexcitation [17]. Oxygen content on G@Cu and Cu particles substrate was obtained using an X energy spectrum detector (EDS Inca X-MaxN).
3. Results and discussion
3.1. Thermal annealing effect on morphology of samples
Scanning electron microscopy was used to investigate the surface morphology of G@Cu formed by CVD and annealed under different condition. For further analysis, image pro plus 6.0 is employed to count the size of the nanoparticles and their inter-particle distance, shown in Table 2.
Table 2. Morphological parameters of each sample
Figure 2 showed the SEM images of G@Cu substrate without (Fig. 2(a)) and with (Figs. 2(b)-2(h)) the following thermal process. It can be seen from Fig. 2(a), the G@Cu substrate directly fabricated by CVD was covered by a large number of particles with relatively uniform size and round shapes. To identify the existence of graphene shell and the size of Cu particles, a higher magnification of SEM image was adopted. As shown in the insert of Fig. 2(a), there was a layer of graphene film on the surface of Cu particles, which indicated that in the CVD process, graphene can grow on the surface of Cu particles to form the G@Cu core-shell structure. In addition, the calculated average size of these particles was ~0.62 μm and the average inter-particle distance was ~0.66 μm (in Table 2). Because of the plasmon coupling between two adjacent particles, metallic nanoparticles with narrow gaps were regarded to be essential for large Raman enhancement [18], while for the as-prepared G@Cu substrate with micron scale inter-particle distance, it is difficult to form “hot spots” [19]. So that the electromagnetic enhancement effect of G@Cu substrate directly prepared by CVD may not be good because of the large gap between Cu particles. For further optimization, the following thermal annealing process was implemented.
Fig. 2 SEM images of G@Cu composite samples (a) #1, (b)#2, (c) #3, (d)#4, (e) #5, (f)#6, (g) #7, (h) #8, and (i) #9, and the corresponding amplified images are inserted.
Figures 2(b)-2(f) showed the SEM images of G@Cu substrates annealed under different temperatures (350, 450, 550, 650, 750 °C respectively) for 20 min. The typical area of these substrates displayed island morphology, with great difference from the regular spherical structure of G@Cu without annealing. In addition, when annealed at 350 °C (Fig. 2(b)), partial Cu particles on G@Cu substrate lost their original spherical structure and turned into an ellipsoid structure. The emergence of major and minor axis of these Cu particles (0.48 μm and 0.26 μm respectively) made the inter-particle distance of Cu particles decrease to ~0.27 μm. Increased the annealing temperature to 450 °C (Fig. 2(c)), the Cu particles became more irregular and island structure covered almost all of the SEM area. Moreover, the borders of clusters were irregular and unable to distinguish which made the inter-particle distance of Cu particles decreased to ~0.20 μm. When the annealing temperature continued to increase, the island structure and inter-particle distance of Cu particles increased dramatically (1.11, 1.24 and 1.33 μm for 550, 650 and 750 °C respectively). Elongated and irregularly shaped Cu particles appeared with average major and minor axis of ∼5.01 and ∼0.56 μm for 550 °C (Fig. 2(d)), ∼6.23 and ∼0.63 μm for 650 °C (Fig. 2(e)), ∼10.02 and ∼0.88 μm for 750 °C (Fig. 2(f)). Shape anisotropy, defined as the average ratio of the minor and major axis of the nanoparticle cross section in the plane of the substrate, was 0.54 for 350 °C, 0.53 for 450 °C, 0.11 for 550 °C, 0.10 for 650 °C, and 0.09 for 750 °C, compared to 0.89 for substrate without annealing. No orientation dependences of the anisotropy were observed in the images of Fig. 2, as expected.
Figures 2(g)-2(i) depicted the SEM images of the G@Cu substrates annealed with different duration (10 min for Fig. 2(g), 30 min for Fig. 2(h), 60 min for Fig. 2(i)) at 450 °C. Together with Fig. 2(c) (annealed for 20 min at 450 °C), we can see that as the annealing duration increased from 10 to 20 min, the borders of these Cu clusters became more indistinguishable and inter-particle distance decreased from 0.27 μm to 0.20 μm, while continued increasing the annealing duration to 30 min, the distances between clusters increased to 0.22 μm, and when the duration reached 60 min, the borders of Cu particles were better distinguished and the inter-particle distance increased to 0.26 μm. Besides, we found that the annealing duration would not change the shape anisotropy, while the annealing temperature increased the shape anisotropy of Cu particles.
The difference in shape and size is due to the different levels of heat dissipation and the surface tension properties produced by annealing temperature and duration [20]. We explained the formation mechanisms of Cu particles from both qualitative analysis and quantitative analysis.
(1) Qualitative analysis: The changes in the morphologies of Cu particles at different temperatures and durations can be attributed to the varying differences of surface free energy or surface diffusion coefficient induced by temperature and duration [21].
Firstly, diffusion barrier (En) specifies the interaction among Cu particles, graphene shell and SiO2/Si. The relationship between surface diffusion coefficient (D) and the temperature (T) is given as follows [22, 23]:
Where K is the Boltzmann constant, En1 and En2 is the surface diffusion barrier of Cu on surface of graphene and SiO2/Si, respectively.Secondly, Cu particles were formed on the SiO2/Si substrate and wrapped in graphene shell, which brought the difference of surface diffusion barrier above and below Cu particles. With an increase of annealing temperature, surface diffusion coefficient of both graphene (D1) and SiO2/Si (D2) increased according to Eq. (1), and the increase speed of D1 was higher combining Eq. (1) and (2). As a result, the difference of surface diffusion coefficient (ΔD = D2 – D1) of different surface on Cu particles decreased accordingly, which led to that the spherical nanoparticles lost a clear boundary gradually, and then became elongated and irregularly shape when the annealing temperature reached a certain value. In this recrystallization process, the fusion of smaller Cu particles with adjacent large particles made the whole system achieve a stable status [24], and the distance between islands decreased firstly, and then increased accordingly.
Thirdly, keeping the annealing temperature unchanged and increasing annealing duration, we can see that all the D1, D2 and ΔD are unchanged which results in the unchanged shape anisotropy of Cu particles, despite the increase of both major axis and minor axis during the recrystallization process.
(2) Quantitative analysis: Another explanation on the formation of Cu particles can be given by the modified LSW (Lifshitz-Slyozov-Wagner) theory [25, 26]. According to Ostwald ripening, in which the Cu particles’ growth rate of major or minor axis is determined by two mechanisms in parallel: Lifshitz and Wagner diffusion-controlled and Wagner kinetic mechanisms [27]. Accordingly, the flow of atoms j consists of two parts: diffusion-controlled js and kinetic ji, that is
According to [28], the size distribution function f (d, t, T) can be represented as the product of three terms: temperature T, duration t and the relative variable u = d/dg (dg is the maximum major or minor axis of cluster):
Where with Φ being the disperse phase volume.To determine f (d, t, T) from Eq. (4) above, we need to know the temperature and duration dependence for dg. According to [26], dg is written as
Where .When x = 0, g(u) corresponds to the Wagner distribution [29]. When x = 1, g(u) transforms to the Chakraverty distribution [30]. In the 0 < x < 1 range, the size function is described by the Chakraverty-Wagner distribution g(u). When the data points are plotted along the coordinate axis but are normalized to g(u)/gmax along y axis and u = d/dg along x axis, the 11 Chakraverty-Wagner curves calculated with an interval Δx = 0.1 are suitable for comparing with experimental data (histograms of major and minor axis). To enable comparison, the experimental histogram (experimental curve) is to be represented in the same normalized form as the theoretical curve. In this case, the major or minor axis of Cu particles (d), plotted along the horizontal axis in the histogram, is transformed to variable u = d/dg, where the maximum diameter dg is taken from the experimental histogram. Along the vertical axis of the histogram, all values corresponding to the number of particles per unit volume for a given diameter within a certain range, Δd, are normalized to the histogram peak. Comparison of the experimental histogram and theoretical curve after they are preliminary plotted on the same scale shows how they match each other and the matching degree is evaluated by the area of histogram under the theoretical curves.
Figure 3 shows that the experimental histograms of Cu particles, corresponding to sample #1 to #9 with different annealing condition, can be adequately described by the Chakraverty-Wager theoretical curves at different x (from 0 to 0.8). For sample without annealing (Fig. 3(a)), the largest area of histogram under a series of theoretical curves appeared at x = 0.2 and 0.1 for major and minor axis respectively (in Table 3). This means that (1) Cu particles formed in the early stage of CVD process through Ostwald ripening by two mechanisms is in parallel and (2) the predominant growth mechanism is chemical reaction that proceeds on the surface of Cu particles during sintering. For the annealing optimization process, the experimental histograms of major/minor axis (in Figs. 3(b)-3(f)) can be described by Chakraverty-Wager theoretical curves at x = 0, 0.2, 0.6/0.7, 0.6, 0.8/0.9 for the annealing temperature from 350 to 750 °C, respectively. In addition, the experimental histograms of Cu particles at different annealing duration are also shown in Figs. 3(g)-3(i). Together with Fig. 3(c) we can see that the experimental histograms of major/minor axis can be described by Chakraverty-Wager theoretical curves at x = 0, 0, 0.7/0.8, 0.7/0.6 for the annealing duration from 10 to 60 min (450 °C), respectively. This implies that the morphology of Cu particles annealed at 350 °C for 20 min, 450 °C for 10 and 20 min is entirely controlled by chemical reaction and in this case, Eq. (3) has no diffusion component js for the distribution. With an increase of annealing temperature or duration, the effect of surface diffusion appears and then become dominance gradually.
Fig. 3 Experimental histograms of major (red line) and minor axes (blue line) for Cu particles on the as-prepared G@Cu substrates versus theoretical Chakraverty-Wagner distribution (a-i represents sample #1-#9, respectively). The dotted curves are theoretical ones, corresponding to extreme growth mechanisms controlled by either surface diffusion, Ch distribution (x = 1), or chemical reaction, W distribution (x = 0). (A color version of this figure can be viewed online.)
Table 3. The area ratios of Cu particles histogram on sample #1 to #9 under the theoretical curves
3.2. Results of the Raman enhancement experiment
In order to demonstrate the property and existence of graphene shell on G@Cu substrates after annealing, a conventional Raman measurement was performed. As a comparison, the D, G and 2D peaks of transferred graphene were clearly observed at ∼1350, ∼1582 and ∼2698 cm−1, respectively, shown in Fig. 4(a). These peaks can be regarded as the fingerprint of graphene [31, 32]. The Raman spectra showed typical features of few-layer graphene: the intensity ratio of I2D/IG = ~1.27 and the full width at half maximum (FWHM) of 2D peak was ~32 cm−1 [33, 34]. In addition, the intensity ratio of ID/IG = ~0.36, indicated the existent of defect on transferred graphene. In order to identify the exact morphology and layer number of transferred graphene, an AFM was employed. From Fig. 4(b), it can be seen that the boundaries of graphene were clear and there were wrinkles, breakages and residues caused by graphene transfer process. The height profile across graphene and bare SiO2/Si substrate was shown in Fig. 4(c). It revealed a step height of ∼1.425 nm, which agreed well with the thickness of 4~5 layer graphene [35].
Fig. 4 (a) Raman spectra of the transferred graphene, (b) AFM image of the transferred graphene on bare SiO2/Si substrate, and (c) the height profile across the section analysis from point A to B in AFM image (b) (A color version of this figure can be viewed online).
Raman spectra of nine as-prepared G@Cu samples were given in Fig. 5. In Fig. 5(a), the G peak ~1581 cm−1 and 2D peak ~2689 cm−1 were observed for G@Cu substrate without annealing. The significant variation of the position of G (down shift ~1 cm−1) and 2D peak (down shift ~10 cm−1) compared with transferred graphene substrate indicated that directly growth of graphene shell on Cu particles could bring tensile strain in graphene. Simultaneously after annealing, the position of G peak kept unchanged and 2D peak was down shift ~4 cm−1 (~1581 cm−1 for G peak and ~2694 cm−1 for 2D peak), which revealed that the diffusion of Cu atoms and the recrystallization during the following thermal anneal process can increase the compressive stress in graphene [36]. In addition, we can see from Table 4 that compared with the transferred graphene, the direct growth of graphene shell on Cu particles perform high property: the low-intensity peak around 1349 cm−1 (in the range of 15~40 a.u.) and the decreased values of ID/IG (in the range of 0.2~0.32) suggested an insignificant population of defects in the graphene shell. In addition, the G and 2D peaks of as-prepared G@Cu samples had different degrees of enhancement compared with the transferred graphene (44 and 56 a.u. for G and 2D band respectively) which was caused by the EM of Cu particles and the plasmonic coupling caused by the interaction between Cu particles and graphene shell, where hot electrons can be injected into graphene layer [7, 18, 37].
Fig. 5 Raman spectra of (a) nine as-prepared G@Cu samples and (b) R6G with concentration of 10−5 M when 9 as-prepared G@Cu samples as SERS substrates.
Table 4. Intensity of D, G, 2D peak and ID/IG values of each sample
To estimate the SERS activity of the as-prepared G@Cu substrates, 10−5 M aqueous solution of R6G was chosen as the probe molecules. As shown in Fig. 5(b), the observed Raman peak locating at ∼1187 cm−1 is attributed to the C–H and N–H bending vibration. While the band peaked at ∼1362 cm−1, ∼1505 cm−1, ∼1644 cm−1 and ∼1580 cm−1 could be assigned to the relating stretching vibrations, with the first three peaks belonging to the C–C bond and the last one to the C–O bond, respectively. The other cognizable vibration peaks, locating at ∼614 cm−1, ∼773 cm−1 and ∼1309 cm−1, respectively, are assigned to the C–C–C in-plane bending vibration, C–H out-plane bending vibration and C–C stretching vibration, correspondingly [38].
To quantitatively demonstrate the enhancement effect of SERS substrate, the representative bands at 614, 773, 1309 and 1362 cm−1 are selected to calculate the EF values with the equation [39]:
where C0 and I0 are the concentration and intensity for normal Raman measurement with 0.1 M R6G solution, respectively, CSERS and ISERS are the concentration and the average intensity of corresponding peak for the SERS measurement, respectively. The experimental results in Table 5 showed that for annealing temperature, the enhancement of R6G on different G@Cu substrates decreased orderly in the sequence of G@Cu annealed at 450 °C, 350 °C, without annealing, 550 °C, 650 °C and 750 °C which indicated that the SERS activity of G@Cu substrate increased firstly, and then decreased with the increased of annealing temperature. For annealing duration, the best enhancement effect of R6G was achieved on G@Cu substrate annealed for 20 min. The EF of each sample was consistent with the conclusion of SEM analysis in section 3.1.Table 5. EF values of the representative bands of R6G on the as-prepared SERS structures
3.3. Fluorescence quenching effect and chemical stability
Graphene, with exceptional properties of fluorescence quenching effect [40] and chemical stability [41], can be used as a substrate to suppress fluorescence background and as a protective layer of metal nano-materials to enhance the stability of SERS substrates [7, 42]. Here, a typical Raman spectra of R6G at 532 nm excitation and oxygen content measured by EDS were used to evaluate the fluorescence quenching effect and stability of Cu particles (without graphene growth) and G@Cu substrates (annealed at 450 °C, 20 min).
Raman spectra of R6G on Cu particles and G@Cu substrates were shown in Fig. 6. A strong fluorescence background was observed and the characteristic Raman peaks of R6G were almost submerged for Cu particles substrate (red line). In contrast, for R6G adsorbed on G@Cu substrate (black line), the fluorescence emission was weaker and the Raman peaks were clearly observed, revealing that graphene suppressed the fluorescence background, and created SERS spectra with a high signal-to-noise ratio. The 1580 cm−1 peak was from graphene.
Fig. 6 Raman spectra of R6G (10−5 M) when G@Cu composite as SERS substrate (black line) and Cu particles without graphene as SERS substrate (red line). (A color version of this figure can be viewed online)
For quantitative analysis of the fluorescence quenching effect of graphene shell on G@Cu substrate, the factor of fluorescence quenching (denoted as q) was defined by the following formula: q = (ACu‒AG@Cu)/ ACu, where A is the integrated area of background noise in Raman spectrum. The calculated ACu and AG@Cu for R6G from 600 to 2000 cm−1 were ∼2.5 × 106 and 2.8 × 105, respectively, which indicated that the fluorescence quenching factor q can be estimated to ∼88.8%. Moreover, the additional enhancement of the SERS signal of R6G on graphene-wrapped Cu particles was observed, which can be attributed to the molecule enrichment effect and CM derived from graphene [43].
To investigate the stability of our samples, we measured the oxygen content with EDS after samples stored in air for several days. As shown in Fig. 7(a), the oxygen content of Cu particles substrate was obviously increased (~35%) after oxidation treatment and the oxygen content tended to saturate after stored in air for 20 days. This was because of the formation of a layer of oxide film on the surface of the Cu particles, which prevented the internal oxidation reaction. For G@Cu, after the oxidation treatment of about 20 days, the oxygen content increased 5% and in the next several ten days, the oxygen content remained unchangeable, indicating excellent chemical stability of G@Cu. Since oxygen gas and moisture cannot permeate through the graphene layer, the grown graphene shell on Cu particles can effectively protect metal from oxygen damage [44]. In addition, the increase of oxygen content can be ascribed to the formation of copper oxides on the surface of Cu particles, which was confirmed by EDS measurements after oxidation stability test, shown in Fig. 7(b) (Cu particles) and Fig. 7(c) (G@Cu composite). Moreover, Raman signal of R6G molecules using our G@Cu composite (stored in air for several days) as SERS substrate was also measured.
Fig. 7 (a) Oxygen content on Cu particles without graphene (red line) and G@Cu composite (black line) after stored in air for different days, the corresponding EDS of (b) Cu particles without graphene and (c) G@Cu composite stored in air for 70 days.
3.4. Simulation
To study the Raman enhancement mechanism (EM and the plasmonic coupling effect) of Cu particles with different morphologies wrapped in graphene shell, the distribution of electromagnetic field was simulated using COMSOL Multiphysics. The simulations were carried out in vacuum (n medium = 1.0), the wavelength of incident light was 532 nm, and the incident electric field (E0 = 1 V/m) was x-polarization propagated along the z direction. In order to simplify the simulation, a double-semi-ellipsoid Cu particles system with different inter particle distances and ratios of minor axis to major axis was built. For the graphene shell, the properties of graphene were taken from the experimental results (in section 3.2), where the refractive index of five-layer-CVD-grown graphene was set to εG = 2.63 + 1.28i at 532 nm [7], calculated from a Lorentz-Drude model and the thickness was set to 1.7 nm (monolayer ~0.34 nm, five layer ~1.7 nm). Simulation results of the maximum electric field |E| max of Cu particles with different shape anisotropy and nanogap in double-Cu-nanoparticle system were shown in Figs. 8 and 9.
Fig. 8 Schematic models of graphene and double Cu particles composite structures for shape anisotropy of (a) 0.5, (b) 0.88, (c) 0.375 and (d) 0.5 with another combination (different from (a)) in COMSOL simulation. The corresponding electric field distribution on the plane x-y (z = 0,the second column) for shape anisotropy of (e) 0.5, (f) 0.88, (g) 0.375 and (h) 0.5 with another combination respectively, while (i), (j), (k) and (l) the corresponding electric field distribution on the plane x-z (y = 0, the third column) respectively. (m) The corresponding intensity distributions of electric field along the white dotted arrow direction likely in (e), and (n) the enlarged view of figure (m). (o) The corresponding intensity distributions of electric field along the red dotted arrows direction likely in (i). (A color version of this figure can be viewed online)
Fig. 9 Intensity distribution of electric field along (a) x axis in plane x-y (z = 0) and (c) z axis in plane y-z (x = 0) when the gap is 2 nm, 4 nm, 6 nm, 8 nm and 10 nm, respectively. The inset images are shown with the schematic model of graphene and double Cu particles with shape anisotropy (minor/major = 30:60) in plane x-y and y-z. (b) The enlarged view of figure (a) (red area). (A color version of this figure can be viewed online)
Detailed modeling and analysis are as follows:
Firstly, the models for analysis of different shape anisotropy induced by different annealing were built with different ratios of minor/major axis. According to our experimental results (in Table 2), the minor axis was set to 30 nm, and the changed major axis (34, 60 and 80 nm) was set to obtain different shape anisotropy (0.88, 0.5 and 0.375). It had to mention that the shape anisotropy of Cu particles was ~0.1 with annealing temperature higher than 550 °C, which increased the difficulty of our simulation. For the purpose of the study on electric field distribution effected by shape anisotropy of Cu particles, relative high shape anisotropy was set to 0.375 for comparison. In addition, the inter-particle distance was set to 2 nm for simplifying our models. The geometric dimensions of graphene shell and Cu particles with different shape anisotropy were shown in Figs. 8(a)-8(d), in which two different combination forms of Cu particles with shape anisotropy of 0.5 were given in Figs. 8(a) and 8(d), respectively.
The electric field distributions of G@Cu with different shape anisotropy and combinations were shown in Figs. 8(e)-8(l). We can see that the electric field was mainly distributed on the surface of G@Cu structure and the junction of two nanoparticles. In addition, electric field of high intensity appeared in the nanogap of two G@CuNP structures, which was usually called “hot spots”. With the increase of shape anisotropy, the maximum electric field |E| max decreased from 20 to 13.3 and Cu particles with shape anisotropy of 0.5 had the largest |E| max. This was consistent with our experimental result that G@Cu substrate annealed at 450 °C made the Cu particles to be shape anisotropy of ~0.5, with a strong EM for R6G (in section 3.2), since SERS enhancement of the substrate is mainly due to the extremely strong electric fields at the inter-particle gaps. Figure 8(h) showed the electric distribution of Cu particles with another possible combination, and the simulation result was different with the first kind of combination in Fig. 8(a). It had to mention that both of these two combinations coexisted on the as-prepared G@Cu substrate, which would bring down the uniformity of SERS signal [7]. Further analysis on the electric intensity distribution along the white and red dotted arrows in Figs. 8(e) and 8(i) was given in Figs. 8(m), 8(n) and 8(o). The high intensity of electric field appeared at the edge of G@Cu structure was due to the edge effect, and the electric intensity in the center (“hot spot”) (x = 0) was decreased in the sequence of: shape anisotropy of 0.5 (combination in Fig. 8(a)) > 0.88 > 0.375 > 0.5 (another combination in Fig. 8(d)).
Secondly, further analysis on the inter particle distance of Cu particles was studied through another model of double Cu particles with shape anisotropy of 0.5 and different inter-particle gaps. Because the simulation of micron-sized gap is time-consuming and the electromagnetic field intensity is so weak that the difference of the electromagnetic field is not obvious when the distance is changed. A series of inter-particle gaps in nanoscale (from 2 to 10 nm) were set to study the effect on the distribution of electromagnetic field in terms of theory. The intensity distribution of electric field along x axis in plane x-y (z = 0) and the enlarged view were shown in Figs. 9(a) and 9(b). The edge effect resulted in the highest intensity of electric field at the edge of G@CuNP structure, and the electric intensity in the center (“hot spot”) was decreased with an increase of the inter-particle gap. Figure 9(c) showed the intensity distribution of electric field along z axis in plane y-z (x = 0), from which we can see that the maximum electric field |E| max appeared at z = 0, and then decreased along two sides of z = 0. In addition, with an increase of inter-particle gap of Cu particles, |E| max also presented a descending trend. Both of these can be used to explain why the G@Cu substrate annealed at 450 °C with the smallest average nano-gap had the largest SERS effect on R6G compared with G@Cu without or with other anneal conditions. Because of the plasmon coupling between two adjacent particles, the increasing inter-particle distance results in damped local field enhancement in the gaps between the particles [19].
The SERS activity of G@Cu substrate was determined by shape anisotropy of Cu particles, inter-particle gap and combination of Cu particles. For G@Cu substrate annealed at 450 °C for 20 min, higher electric enhancement was brought by the formation of elongated and irregularly shaped Cu particles with shape anisotropy of ~0.53, gap of 0.2 μm.
4. Conclusion
Herein graphene-wrapped Cu nanoparticle substrates (G@Cu) were prepared as SERS substrates via a direct CVD growth method assisted with thermal annealing process. After optimization annealing process, Cu particles elongated and gradually lost the original spherical shape and the inter-particle distance can be adjusted under different annealing condition. The formation of Cu particles was analyzed through qualitative and quantitative methods and using R6G molecules as probes, the G@Cu annealed at 450 °C for 20 min showed a relatively excellent SERS enhancement activity with EF of ~1.5 × 106. The G@Cu also showed high chemical stability and the fluorescence background of Raman signal was low. Future work will be focused on control the size of Cu particle and gap between particles, in order to gain higher enhancement. Moreover, the quality of graphene (layer, coverage situation, etc.) will be improved.
Funding
National Natural Science Foundation of China (No. 61376121); National High-tech R&D Program (No. 2015AA034801); National Natural Science Foundation of Chongqing (No. CSTC2015JCYJBX 0034).
Acknowledgments
We would like to thank Analysis and Test Center of Chongqing University and Mr. X.N. Gong for the SEM and Raman spectrometer help.
References and links
1. M. Fleischmann, P. J. Hendra, and A. J. McQuillan, “Raman spectra of pyridine adsorbed at a silver electrode,” Chem. Phys. Lett. 26(2), 163–166 (1974). [CrossRef]
2. D. L. Jeanmaire and R. P. Van Duyne, “Surface Raman spectroelectrochemistry: Part I. Heterocyclic, aromatic, and aliphatic amines adsorbed on the anodized silver electrode,” J. Electroanal. Chem. Interfac. 84(1), 1–20 (1977). [CrossRef]
3. Y. Liu, Y. Hu, and J. Zhang, “Few-layer graphene-encapsulated metal nanoparticles for surface-enhanced Raman spectroscopy,” J. Phys. Chem. C 118(17), 8993–8998 (2014). [CrossRef]
4. E. Le Ru and P. Etchegoin, Principles of Surface-Enhanced Raman Spectroscopy: and related plasmonic effects (Elsevier, 2008).
5. E. C. Le Ru, E. Blackie, M. Meyer, and P. G. Etchegoin, “Surface enhanced Raman scattering enhancement factors: a comprehensive study,” J. Phys. Chem. C 111(37), 13794–13803 (2007). [CrossRef]
6. J. Zhang, X. Zhang, C. Lai, H. Zhou, and Y. Zhu, “Silver-decorated aligned CNT arrays as SERS substrates by high temperature annealing,” Opt. Express 22(18), 21157–21166 (2014). [CrossRef] [PubMed]
7. T. Gong, Y. Zhu, J. Zhang, W. Ren, J. Quan, and N. Wang, “Study on surface-enhanced Raman scattering substrates structured with hybrid Ag nanoparticles and few-layer graphene,” Carbon 87, 385–394 (2015). [CrossRef]
8. M. J. Banholzer, J. E. Millstone, L. Qin, and C. A. Mirkin, “Rationally Designed Nanostructures for Surface-Enhanced Raman Spectroscopy,” Chem. Soc. Rev. 37(5), 885–897 (2008). [CrossRef] [PubMed]
9. 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, and Z. Q. Tian, “Shell-Isolated Nanoparticle-Enhanced Raman Spectroscopy,” Nature 464(7287), 392–395 (2010). [CrossRef] [PubMed]
10. H. Yao, L. Jin, H. J. Sue, Y. Sumi, and R. Nishimura, “Facile decoration of Au nanoparticles on reduced graphene oxide surfaces via a one-step chemical functionalization approach,” J. Mater. Chem. A Mater. Energy Sustain. 1(1), 10783–10789 (2013). [CrossRef]
11. X. Ling and J. Zhang, “First-Layer Effect in Graphene-Enhanced Raman Scattering,” Small 6(18), 2020–2025 (2010). [CrossRef] [PubMed]
12. X. Ling, J. Wu, W. Xu, and J. Zhang, “Probing the effect of molecular orientation on the intensity of chemical enhancement using graphene-enhanced Raman spectroscopy,” Small 8(9), 1365–1372 (2012). [CrossRef] [PubMed]
13. W. Xu, J. Xiao, Y. Chen, Y. Chen, X. Ling, and J. Zhang, “Graphene-Veiled Gold Substrate for Surface-Enhanced Raman Spectroscopy,” Adv. Mater. 25(6), 928–933 (2013). [CrossRef] [PubMed]
14. W. Xu, N. Mao, and J. Zhang, “Graphene: a platform for surface-enhanced Raman spectroscopy,” Small 9(8), 1206–1224 (2013). [CrossRef] [PubMed]
15. G. B. Barin, Y. Song, I. de Fátima Gimenez, A. G. Souza Filho, L. S. Barreto, and J. Kong, “Optimized graphene transfer: Influence of polymethylmethacrylate (PMMA) layer concentration and baking time on graphene final performance,” Carbon 84(84), 82–90 (2015). [CrossRef]
16. S. Chen, W. Cai, R. D. Piner, J. W. Suk, Y. Wu, Y. Ren, J. Kang, and R. S. Ruoff, “Synthesis and characterization of large-area graphene and graphite films on commercial Cu-Ni alloy foils,” Nano Lett. 11(9), 3519–3525 (2011). [CrossRef] [PubMed]
17. K. Ostrikov, F. Beg, and A. Ng, “Colloquium: Nanoplasmas generated by intense radiation,” Rev. Mod. Phys. 88(1), 011001 (2016). [CrossRef]
18. A. Mcleod, K. C. Vernon, A. E. Rider, and K. Ostrikov, “Optical Coupling of Gold Nanoparticles on Vertical Graphenes to Maximize SERS Response,” Opt. Lett. 39(8), 2334–2337 (2014). [CrossRef] [PubMed]
19. K. Long, X. Luo, H. Nan, D. Du, W. Zhao, Z. Ni, and T. Qiu, “Surface-enhanced Raman scattering from graphene covered gold nanocap arrays,” J. Appl. Phys. 114(18), 183520 (2013). [CrossRef]
20. C. L. Tan, S. J. Jang, Y. M. Song, K. Alameh, and Y. T. Lee, “Bimetallic non-alloyed NPs for improving the broadband optical absorption of thin amorphous silicon substrates,” Nanoscale Res. Lett. 9(1), 181 (2014). [CrossRef] [PubMed]
21. L. Y. Ma, L. Tang, Z. L. Guan, K. He, K. An, X. C. Ma, J. F. Jia, Q. K. Xue, Y. Han, S. Huang, and F. Liu, “Quantum size effect on adatom surface diffusion,” Phys. Rev. Lett. 97(26), 266102 (2006). [CrossRef] [PubMed]
22. H. Zhou, C. Qiu, F. Yu, H. Yang, M. Chen, L. Hu, and L. Sun, “Thickness-dependent morphologies and surface-enhanced Raman scattering of Ag deposited on n-layer graphenes,” J. Phys. Chem. C 115(23), 11348–11354 (2011). [CrossRef]
23. C. W. Huang, H. Y. Lin, C. H. Huang, R. J. Shiue, W. H. Wang, C. Y. Liu, and H. C. Chui, “Layer-dependent morphologies of silver on n-layer graphene,” Nanoscale Res. Lett. 7(1), 618 (2012). [CrossRef] [PubMed]
24. J. Park, J. Joo, S. G. Kwon, Y. Jang, and T. Hyeon, “Synthesis of monodisperse spherical nanocrystals,” Angew. Chem. Int. Ed. Engl. 46(25), 4630–4660 (2007). [CrossRef] [PubMed]
25. R. Vengrenovich, B. Ivanskii, and A. Moskalyuk, “Generalized Lifshits-Slezov-Wagner distribution,” J. Exp. Theor. Phys. 104(6), 906–912 (2007). [CrossRef]
26. R. Vengrenovich, B. Ivanskyi, and A. Moskalyuk, “Generalized Chakraverty-Wagner distribution,” Ukr. J. Phys. 53(11), 1101–1109 (2008).
27. R. Vengrenovych, B. Ivanskyy, I. Panko, M. Stasyk, and I. Fesiv, “Stability of nanocrystals in 2D and 3D systems in Ostwald ripening,” Powder Metall. Met. Ceramics 54(5–6), 281–291 (2015). [CrossRef]
28. R. Vengrenovitch, “On the Ostwald ripening theory,” Acta Metall. 30(6), 1079–1086 (1982). [CrossRef]
29. C. Wagner, “Theorie der Alterung von Niederschlagen durch Umlösen (Ostwald-Reifung), Zeitschrift für Elektrochemie,” Ber. Bunsenges. Phys. Chem 65(7–8), 581–591 (1961).
30. B. Chakraverty, “Grain size distribution in thin films—1. Conservative systems,” J. Phys. Chem. Solids 28(12), 2401–2412 (1967). [CrossRef]
31. A. C. Ferrari, J. C. Meyer, V. Scardaci, C. Casiraghi, M. Lazzeri, F. Mauri, S. Piscanec, D. Jiang, K. S. Novoselov, S. Roth, and A. K. Geim, “Raman spectrum of graphene and graphene layers,” Phys. Rev. Lett. 97(18), 187401 (2006). [CrossRef] [PubMed]
32. S. Xu, B. Man, S. Jiang, W. Yue, C. Yang, M. Liu, C. Chen, and C. Zhang, “Direct growth of graphene on quartz substrates for label-free detection of adenosine triphosphate,” Nanotechnology 25(16), 165702 (2014). [CrossRef] [PubMed]
33. A. C. Ferrari, “Raman spectroscopy of graphene and graphite: disorder, electron–phonon coupling, doping and nonadiabatic effects,” Solid State Commun. 143(1–2), 47–57 (2007). [CrossRef]
34. S. C. Xu, B. Y. Man, S. Z. Jiang, C. S. Chen, C. Yang, M. Liu, X. G. Gao, Z. C. Sun, and C. Zhang, “Direct synthesis of graphene on SiO2 substrates by chemical vapor deposition,” CrystEngComm 15(10), 1840–1844 (2013). [CrossRef]
35. S. C. Xu, B. Y. Man, S. Z. Jiang, A. H. Liu, G. D. Hu, C. S. Chen, M. Liu, C. Yang, D. J. Feng, and C. Zhang, “Direct synthesis of graphene on any nonmetallic substrate based on KrF laser ablation of ordered pyrolytic graphite,” Laser Phys. Lett. 11(9), 096001 (2014). [CrossRef]
36. F. Ding, H. Ji, Y. Chen, A. Herklotz, K. Dörr, Y. Mei, A. Rastelli, and O. G. Schmidt, “Stretchable graphene: a close look at fundamental parameters through biaxial straining,” Nano Lett. 10(9), 3453–3458 (2010). [CrossRef] [PubMed]
37. C. E. Cheng, C. Y. Lin, H. Y. Chang, C. H. Huang, H. Y. Lin, C. H. Chen, C. C. Hsu, C. S. Chang, and F. S. Sen Chien, “Surface-enhanced Raman scattering of graphene with photo-assistedsynthesized gold nanoparticles,” Opt. Express 21(5), 6547e6554 (2013).
38. Y. Q. Wang, S. Ma, Q. Q. Yang, and X. J. Li, “Size-dependent SERS detection of R6G by silver nanoparticles immersion-plated on silicon nanoporous pillar array,” Appl. Surf. Sci. 258(15), 5881–5885 (2012). [CrossRef]
39. L. Q. Lu, Y. Zheng, W. G. Qu, H. Q. Yu, and A. W. Xu, “Hydrophobic teflon films as concentrators for single-molecule SERS detection,” J. Mater. Chem. 22(39), 20986–20990 (2012). [CrossRef]
40. L. Xie, X. Ling, Y. Fang, J. Zhang, and Z. Liu, “Graphene as a substrate to suppress fluorescence in resonance Raman spectroscopy,” J. Am. Chem. Soc. 131(29), 9890–9891 (2009). [CrossRef] [PubMed]
41. K. S. Novoselov, “Graphene: the magic of flat carbon,” ECS Trans. 19(5), 3–7 (2009).
42. X. Li, J. Li, X. Zhou, Y. Ma, Z. Zheng, X. Duan, and Y. Qua, “Silver nanoparticles protected by monolayer graphene as a stabilized substrate for surface enhanced Raman spectroscopy,” Carbon 66(2), 713–719 (2014). [CrossRef]
43. Q. Hao, B. Wang, J. A. Bossard, B. Kiraly, Y. Zeng, I. K. Chiang, L. Jensen, D. H. Werner, and T. J. Huang, “Surface-Enhanced Raman Scattering Study on Graphene-Coated Metallic Nanostructure Substrates,” J. Phys. Chem. C Nanomater. Interfaces 116(13), 7249–7254 (2012).
44. S. Xu, B. Man, S. Jiang, J. Wang, J. Wei, S. Xu, H. Liu, S. Gao, H. Liu, Z. Li, H. Li, and H. Qiu, “Graphene/Cu nanoparticle hybrids fabricated by chemical vapor deposition as surface-enhanced Raman scattering substrate for label-free detection of adenosine,” ACS Appl. Mater. Interfaces 7(20), 10977–10987 (2015). [CrossRef] [PubMed]
