One-step synthesis of highly-branched gold nanostructures and its application in fabrication of SERS-active substrates Open Access

SERS has been attracting attention as a promising technique in identifying single molecule characterized by molecular fingerprint specificity and high sensitivity. Generally speaking, the enhancement can be attributed to electromagnetic (EM) enhancement and chemical (CE) enhancement.^1,2 Surface plasmon resonance (SPR) is attributed to collective oscillations of conduction electrons around the nanostructures when a particular wavelength of light radiates onto the surface of roughened metal nanostructures. The enhancement via SPR is referred as EM enhancement.³ Unlike EM effect, CE enhancement can be ascribed to the bonding or interaction between the adsorbed molecules and the metal surface. In the process light-induced charge transfer is vital.⁴ To further understand the contribution of EM and CE mechanism to SERS, Zhao et al. used time-dependent density functional theory (TDDFT) to analyze the Raman scattering of pyridine-Ag₂₀ model. Ag₂₀ tetrahedron had two different binding sites, which were denoted as S-complex and V-complex. The former was at its faces which stood for a {111} surface of fcc silver. And the latter was at its vertexes which stood for ad-atom site. Thus, different chemical environments could be simulated through the two adsorption ways, leaving alone the requirement for changes in Ag cluster. By comparing the simulated normal Raman spectra of pyridine alone with the Raman spectra of S-complex and V-complex with respect to charge transfer resonance and resonance within Ag cluster, they found that enhancement of CE and EM were about 10⁴ and 10⁵-10⁶ respectively.⁵ As seen, EM enhancement was much larger than CE enhancement. With regard to EM enhancement, Xu et al. found that when the nanostructures in a dimer were closely placed (<1 nm) under an appropriate wavelength of incoming light which was polarized along the dimer axis, the EM of the interstitial region could be enhanced greatly. The interstitial region was known as “hot spot”. Similar regions were also found to be close to the sharp tip of the droplet-shaped particles, which could be ascribed to the increasing surface charge density at the tip.⁶ Thanks to hot spots at the tips, the sphere with a tip displayed a much larger SERS enhancement as compared with the sphere with the same size.⁷ To fabricate SERS-active substrates with plenty of hot spots, different nanostructures have been investigated, including nanostars, porous nanosheet and polyvinylpyrrolidone (PVP)-Ag nanocubes films.^8–10 However, the controllable fabrication of a reproducible SERS substrate still limits the application of SERS in single molecules’ quantitative analysis. By introducing anisotropy into particle shape, the substrates can be more reproducible than that of the aggregated nanospheres, for they have sharp tips which serve as “hot spots” compared with these at the interstitial region.¹¹ 

Gold nanostructures have been studied intensively because of their remarkable performance in molecule detection, catalysis and biomedical imaging.^2,12–14 Up to now, gold nanostructures with various shapes from zero-dimensional (rods¹⁵) to three-dimensional (nanostars, multipods, nanoflowers^16–20) structures have been produced. However, to produce gold nanostructures of better surface properties, efforts have been devoted into the synthesis of particles with specific morphology. Highly-branched gold nanostructures with shape edges are the ones that attract much attention for their unique optical properties. Zhu et al. reported on the synthesis of gold nanostars with fractal structure in the presence of ascorbic acid (AA). In the study, a secondary growth of new branches was observed which was accompanied by enhancement of SERS effects, indicating SERS enhancement was associated with the branch number.¹⁹ Han et al. also utilized AA to reduce HAuCl₄ in aqueous solution of poly diallyl dimethyl ammonium chloride (PPDA) for preparation of dendritic gold nanostructures, which exhibited much better SERS sensitivity than the flowerlike and urchinlike nanoparticles.²⁰ Su et al. also reported a protocol using N-2-hydroxyethylpiperazine-N′-2-ethanesulfonic acid (HEPES) to synthesize gold nanostars with different branch lengths. By comparison, it was found that the nanostars with the longest branch presented the most intense Raman signal.¹⁰ 

To synthesize highly-branched gold nanostructures, proper reductants and shape-directing agents are necessary for induction of growth along specific facets.¹⁸ In this paper, we described a simple one-step method for production of highly-branched gold nanostructures (HGNs) whose size lies in the submicron range, and the branches in the nanometer range in the presence of dopamine hydrochloride. Compared with seed-mediated growth method, the one-step synthesis method has several superiorities like briefness, high efficiency, but it is also characteristic by “uncontrollable”. However, in this method by altering the amount of the reagents and reaction temperature we managed to tune the morphology and size of the HGNs, whose optical properties were confirmed by visible-near infrared (Vis-NIR) absorbance spectroscopy. Afterwards SERS efficiency of the 4-MBA labeled HGNs prepared already was compared and the one with the largest SERS enhancement was adopted to assemble the SERS-active substrates via electrostatically assisted APTES-functionalized surface-assembly method. To find whether the substrates could be widely used in molecule detection in the future, the reproducibility and sensitivity of the substrates were all investigated.

Ultra-pure water used was from Milli-Q (Millipore, America, resistivity >18M) source. Chloroauric acid tetrahydrate (HAuCl₄·4H₂O), dopamine hydrochloride, and APTES were all purchased from Yangzhou Younuo Chemicals Co., Ltd. (China). 4-MBA was purchased from Yangzhou Noah Chemical Co., Ltd. (China). All the glassware used were cleaned with aqua regia and deionized water.

0.4 mL HAuCl₄ solution (50 mM) was put into a beaker with 10 mL of ultra-pure water under vigorous stirring, followed by addition of 2 mL dopamine hydrochloride solution (53 mM). Then the solution was heated slowly until the temperature reached 60 °C. During the addition, the color of the solution turned from light yellow to dark green and finally to orange-red within minutes. At last, the temperature was fixed at 60 °C and let it continue to react about 30 minutes to yield HGNs.

The effect of dopamine hydrochloride on the synthesis of nanostructures was studied by varying the amount of dopamine hydrochloride from 0.4 to 3.0 mL, whereas the amount of HAuCl₄ was fixed at 0.4 mL and the final reaction temperature was fixed at 60 °C. In the study of the effect of HAuCl₄, the amount of HAuCl₄ was set from 0.1 to 0.6 mL, whereas the amount of dopamine hydrochloride was fixed at 2.0 mL and the final reaction temperature was fixed at 60 °C. The effect of reaction temperature was studied by varying the final reaction temperature from 40 to 80 °C, whereas the amount of HAuCl₄ and dopamine hydrochloride was fixed at 0.4 mL and 2.0 mL respectively.

The fabrication of APTES-functionalized silicon wafers was shown in Scheme 1. Briefly, before APTES-coating the substrates were washed with acetone and ultra-pure water successively, followed by cleaned with ethanol for three times. After drying at 80 °C, the silicon wafers were immersed in 1% (v/v) ethanol solution of APTES for 12 h. To remove the residual silane molecules and alkoxy groups, the wafers were rinsed in ethanol and ultrasonically cleaned for three times and let it dry at 80 °C.

For deposition of HGNs layers, the aminated silicon wafers placed horizontally into the HGNs solution prepared above. After successful adsorption of HGNs onto the silicon wafers, all the substrates were washed with ultra-pure water for one time and left to dry at room temperature. Thus, the clean APTES-functionalized SERS substrates were prepared.

The analyte 4-MBA was used for SERS measurements with its thiol groups absorbed onto the surface of the gold nanostructures. To find out the best candidate for the following experiments, 4-MBA ethanol solution was mixed with the HGNs solutions mentioned in Control experiments. For adsorption of 4-MBA onto the HGNs, the mixture was left to stand for 4 h. Then the mixture was put into a glass made of quartz followed by detection of its Raman spectra.

To investigate the detection capacity of the HGNs substrates, 4-MBA solutions at different concentrations were dripped onto the substrates with a pipette. After the analyte molecule adsorbed onto the surface of the substrates, rinsed the substrates with ultra-pure water for three times before SERS measurement. The Raman spectra were all measured by a Renishaw Invia microscope Raman spectrometer which was made up of a Leica confocal microscope and a He-Ne laser (785 nm) with laser power of 5 mW focusing on the samples. The exposure time was 10 s. To evaluate the uniformity of the substrates, SERS mapping was obtained by a Renishaw Invia microscope Raman spectrometer with laser power of 5 mW. The scan area was 50×50 μm.

The Vis-NIR absorbance spectra were obtained by a Cary UV-5000 spectrometer. Scanning electron microscope (SEM) images were captured by a S-4800 II field-emission scanning electron microscope at 3.0 KV. The shape of the nanostructures was characterized by a Tecnai 12 transmission electron microscope at 60 KV (Philips). To have an observation of the particles’ crystalline structure high-resolution TEM images (HRTEM) and selected area electron diffraction (SAED) images were captured with a Tecnai G2 F30 S-Twin TEM at 200 KV (FEI). The EDX spectrum of the HGNs was characterized by a S-4800 II field-emission scanning electron microscope at 20.0 KV.

The structure of the as-prepared HGNs was characterized by SEM and TEM. The Fig. 1(A) showed the HGNs with the average size about 600 nm. Thus, the HGNs had much larger size than those previously reported.^16–18 To have a better observation of the HGNs, a high-magnification SEM image was observed. As shown in Fig. 1(B), the HGNs had more than 10 branches with three-dimensional morphology which were similar to many towers extending radially from a center. In each branch, there were lots of secondary, star-like leaves stretching in several specific directions. The representative TEM image of the HGNs was shown in Fig. 1(C). The morphology of the HGNs was almost the same and no other shapes were found. And the crystalline structure of the HGNs was characteristic by HRTEM (Fig. 1(D)), which showed that the interplanar spacing of the tips was measured to be 0.240 nm, indicating the HGNs grew preferentially on {111} planes. To further observe the morphology of the HGNs, SAED image was also taken (Fig. 1(E)). It showed that the HGNs grew in random orientations, such as {111}, {200}, {220}, {311} crystalline facet. The typical Vis-NIR absorbance spectrum of the HGNs was displayed in Fig. 1(F). As shown, the absorbance spectrum exhibited a relatively wide but strong SPR band around 470 nm in the visible range. Similar Vis-NIR absorbance spectra of highly branched nanostructures had been reported by You’s and Qi’s group with the presence of a wide band at about 500 nm.^20,21 To determine the chemical composition of the HGNs, the associated energy-dispersive X-ray spectroscopy (EDX) spectrum of the HGNs in Fig. 1(G) was captured (Fig. 1(H)). Only the peaks associated with Au were found apart from Si peak originating from silicon wafers, indicating the HGNs were highly pure. The HGNs had rough surface determined by sharp branches and the valleys, which acted as potential “hot spots” for enhancing the electromagnetic field around the particle.

In the reaction, we hypothesized that the amount of the reactants (HAuCl₄ and dopamine hydrochloride) and reaction conditions might affect the size and morphology of the nanostructures. Dopamine hydrochloride is used as a weak reducing agent in the preparation of nanomaterials. It was applied to reduce HAuCl₄ for preparation of HGNs in this method. As shown in Fig. 2(A), the structures of the HGNs prepared by using different amounts of dopamine hydrochloride (53 mM) were compared. Obviously the nanostructures prepared by using 0.4 mL dopamine hydrochloride were of immature, quasi-spherical shape with the diameter ranging from 50 to 200 nm. The poor homogeneity indicated the amount of dopamine hydrochloride was lacking. As the amount of dopamine hydrochloride was doubled (0.8 mL), flower-like particles with short tips but without secondary leaves emerged as a kind of transitive morphology. And compared with the nanostructures produced with 0.4 mL, the particle size increased to about 400 nm. When the amount of dopamine hydrochloride reached 2.0 mL, the nanostructures grew in kinetics-favored facet, leading to the formation of HGNs about 500 nm in size. With further increase of the amount of dopamine hydrochloride, HGNs with larger size formed. The possible growth process of the HGNs was proposed as follows. Briefly, gold nuclei were produced when HAuCl₄ was reduced by dopamine hydrochloride. Then oriented aggregation of Au⁰ onto the surface of the nuclei led to the formation of the tips. With the tips continuously stretching, the HGNs were finally obtained.

To investigate the influence of the morphology and size of the nanostructures on their optical properties, the Vis-NIR absorbance spectra of the solutions prepared above were combined in Fig. 2(B) for comparison. A red-shift of the plasmon band from 452 to 485 nm was found as the size of the particles got larger and the branches got longer. It was consistent with the calculations conducted by Hao et al.²² They demonstrated that the plasmon bands of branched particles were very sensitive to the branches’ length and sharpness. To clarify it specifically, the nanostructures solutions prepared at different amounts of dopamine hydrochloride were mixed with ethanol solution of 4-MBA (1×10^-3 M) at the ratio of 9/1, followed by detection of its Raman spectrum. As shown in Fig. 2(C), the SERS spectra were characteristic of 4-MBA with bands at 1080 and 1594 cm^-1, which could be ascribed to the in-plane ring breathing coupled with ν (C-S) and ν (C-C) mode respectively. And the band at 1080 cm^-1 was sensitive to the changes in the amount of dopamine hydrochloride, so the intensity at 1080 cm^-1 was chosen as the comparative standard. Besides, there were other bands of 4-MBA such as bands at 417 cm^-1, assigned to the ν (C-S), at 691 and 1181 cm^-1, which could be ascribed to C-H out-of-plane deformation and ν (C-H).^23,24 One could find that SERS was sensitive to both the wavelength and intensity of the plasmon band. From left to right, the first three samples had the plasmon band of 452, 460 and 473 respectively, and their intensities increased gradually. The corresponding Raman intensities got intense rapidly with increase of the branch length (i.e. the wavelength of the plasmon band) and the concentration of the nanostructures (i.e. the intensity of the plasmon band). Similar results have been reported by Zhu et al.²⁵ It’s well-known that sharp tips can serve as “hot spots” which causes a larger SERS enhancement. Therefore, the SERS enhancement in the circumstance might mainly be ascribed to long and densely arranged branches on the SGNs surface, which were absent in the quasi-spherical particles.²⁶ Moreover, a much larger surface area of the SGNs made it possible for more analyte molecules to adsorb.²⁷ However, when the HGNs were prepared by using 3.0 mL dopamine hydrochloride, the Raman intensity declined accompanied with a red shift of the plasmon band by 12 nm and the decreased plasmon band intensity. The Raman intensity could be weaken by reducing the particle concentration for the decrease of the particle concentration inevitably led to the decrease of the gaps between the particles which could also serve as “hot spots”.²⁸ Therefore it could be speculated the decrease of gaps between the nanostructures was the key factor leading to the decrease of Raman intensity. To conclude, the HGNs prepared by using 2.0 mL dopamine hydrochloride exhibited the largest SERS enhancement.

In this section, effect of different amounts of HAuCl₄ (50 mM) on the synthesis of the HGNs was discussed. The morphology and structure of the HGNs prepared by using different amounts of HAuCl₄ was characterized by SEM images and TEM images (Fig. 3(A)). With increase of the amount of HAuCl₄, a notable increase of HGNs’ diameter from 200 to 600 nm accompanied by the lengthening of the branches could be observed. However, the shape of these nanostructures was very similar. The increase of HAuCl₄ promotes the reduction of Au³ to Au⁰ leading to more Au⁰ adsorbed to the surface of the nanostructures. Therefore, the size of the nanostructures got larger and the branch length got longer.²⁹ Fig. 3(B) showed the Vis-NIR absorbance spectra of the HGNs produced above. As the size of the HGNs got larger and the branches got longer, the plasmon band red-shifted from 440 to 488 nm. The Raman spectra of 4-MBA labeled HGNs prepared by using different amounts of HAuCl₄ were shown in Fig. 3(C). When the amount of HAuCl₄ was below 0.4 mL, a gradual increase of the Raman intensity was observed accompanied by the increase of both the plasmon band wavelength (i.e. the branch length) and intensity (i.e. the particle concentration of HGNs). The high density of long branches could account for the enhancement. Whereas, when the amount of HAuCl₄ reached 0.6 mL the band intensity decreased which was mainly caused by the decrease of the HGNs concentration. Thus, to optimize the substrates’ SERS effect, the amount of HAuCl₄ was set at 0.4 mL.

In many synthesis reactions, temperature is an important parameter since it influences the HGNs’ growth rates and nucleation.³⁰ To explore reaction temperature’s influence on the morphology of the HGNs, the reaction temperature was set from 40 to 80 °C with the amount of HAuCl₄ and dopamine hydrochloride fixed at 0.4 mL and 2.0 mL respectively. Fig. 4(A) showed the Vis-NIR absorbance spectra of HGNs prepared at different reaction temperatures. When the temperature was below 60 °C, the intensity of the plasmon band increased (i.e. the concentration of the HGNs) with rise of the reaction temperature. However, the intensity of plasmon band turned to decreasing above 60 °C, which indicated the multifunction of reaction temperature in the reaction. Considering that the four samples were all synthesized with the same reagents for 30 minutes and other experiment conditions were controlled, we could conclude that at temperatures below 60 °C lifting reaction temperature promoted the nucleation of the HGNs but played a depressing role when it was above 60 °C. While at 80 °C a blue-shift of the absorption band was observed, which was speculated to be shortening of the branch arising from surface melting.^30,31 Moreover, the width of the band also got larger at 80 °C inferring a decline of the homogeneity of the HGNs. The SERS activities of 4-MBA labeled (1×10^-4 M) HGNs synthesized at different temperatures were also compared in Fig. 4(B). When the temperature was below 60 °C, the Raman intensity at 1080 cm^-1 rose as the intensity of the plasmon band increased which was ascribed to the rise of the concentration of HGNs. While at 80 °C, Raman intensity decreased as the intensity of the plasmon band declined, indicating the concentration of HGNs was decreasing. Besides, the blue-shift of the plasmon band might also contribute to decrease of Raman intensity. Therefore, to obtain the largest SERS enhancement, the reaction temperature was set at 60 °C.

Based on the above results, the HGNs prepared by using 2.0 mL dopamine hydrochloride (53 mM), 0.4 mL HAuCl₄ (50 mM) were used for the fabrication of SERS-active substrates. APTES with -NH³⁺ is the most commonly used aminosilane groups which can boost the adsorption of nanostructures with inherent negative charge,³² so it was employed to functionalize the silicon wafers assisted by electrostatic interaction for adsorption of HGNs’ layers. In terms of the adsorption procedure, it was found that when the silicon wafers were immersed vertically into the HGNs solution, the nanostructures could hardly adsorb onto the substrates, different from the methods mentioned by Su et al.¹⁰ Instead we succeeded in depositing the particles by directly placing the wafers horizontally into the solutions, which might be associated with the large size of the HGNs. After the substrate was taken out of the container and dried at 80 °C, the HGNs’ structure and distribution were observed by using SEM. Fig. 5(A) showed the SEM image of the substrate which was made up of well-shaped, densely-packed HGNs. To have the uniformity of the surface SERS signal, the SERS mapping of 4-MBA (1×10^-4 M) labeled substrate was recorded (Fig. 5(B)). The scan area was 50×50 μm, the step size was 2 mm, the laser power was 5 mW, and the acquisition time at each point was 10 s. The Raman intensity of the peak at 1080 cm^-1 was mapped at each grid point on the HGNs substrates. As we could see, the intensities of the characteristic peak at 1080 cm^-1 showed good reproducibility. The Raman spectra of five randomly chosen points on the substrate were also detected (Fig. 5(C)). As shown in Figs. 5(B) and 5(C), the characteristic peak of 4-MBA (1080 cm^-1 and 1594 cm^-1 band) did not show obvious difference. In order to intuitively illustrate the difference between the points, we made a statistical analysis of the Raman intensities of the characteristic peak at 1080 cm^-1. The RSD of the peak intensities at 1080 cm^-1 was 3.57% (<20%), demonstrating the APTES-functionalized substrates with the HGNs were characteristic of good reproducibility.

Besides reproducibility, sensitivity of the substrates was another important factor for SERS detection. To figure out the detection capability of the HGNs substrates, 4-MBA at different concentrations were introduced to the substrates. Afterwards the corresponding Raman spectra of 4-MBA labeled HGNs substrates were recorded (Fig. 6(A)). As we can see, different concentrations of 4-MBA labeled HGNs substrates all showed characteristic peaks except the analyte at the concentration of 1×10^-10 M. So 5×10^-10 M was taken as the detection limit of the HGNs substrates. As shown in Fig. 6(B), the intensities at low concentrations (from 5×10^-10 to 1×10^-9 M) of 4-MBA changed slightly, which might be ascribed to the systematic error during the measurement. And the inset demonstrated the linear relationships between the Raman intensity of the characteristic peak at 1080 cm^-1 and the concentration of 4-MBA from 1×10^-8 to 1×10^-4 M. Its regression equation could be expressed as y=15658lgC+133289. In order to demonstrate the advantages of the HGNs substrates over the non-SERS ones, 1×10^-9 M of 4-MBA solution was directly dripped onto the surface of the substrates, followed by detection of its Raman spectrum, as shown in Fig. 6(C). AEF was calculated according to the reported method:¹⁰ 

where I_SERS corresponded to the Raman intensity for the SERS-active substrates with the HGNs at a certain concentration of analyte C_SERS and I_NS corresponded to the Raman intensity for the non-SERS substrates at a certain concentration of analyte C_NS. When C_SERS was 1×10^-9 M and C_NS was 1×10^-2 M, AEF was calculated to be 7×10⁷, indicating the SERS-active substrate has a good application prospect for high-sensitive detection of biomolecules.

To conclude, we have developed a simple one-step synthesis of anisotropic HGNs in high yields. The HGNs had more than ten branches with many secondary leaves on it. The HRTEM demonstrated the tips of these particles grew preferentially on {111} lattice plane. By varying the amounts of the reagents, controllable production of HGNs with diameter ranging from 200 to 600 nm has been achieved. In this method, the presence of dopamine hydrochloride was vital for its role not only as a reductant but also as a shape-directing agent. Besides the reagents, reaction temperature was another important parameter. Lifting reaction temperature could accelerate the nucleation when it was below 60°C. Afterwards the Raman spectra of the samples prepared above were all investigated to find the best candidate for assembly of SERS-active substrates, and the results indicated that the HGNs produced with 2.0 mL dopamine hydrochloride (53 mM) and 0.4 mL HAuCl₄ (50 mM) at 60 °C exhibited the most intense Raman signal. After assembling of the HGNs, the substrates have shown outstanding SERS sensitivity with a detection limit of 5×10^-10 M of 4-MBA. Moreover, the uniformity of the APTES-functionalized substrates has also been proved remarkable with RSD of 3.57 %. From a practical point of view, the APTES-functionalized HGNs substrates held great potential to be used in different fields like disease diagnosis, chemical analysis owing to its high reproducibility and sensitivity.

This research was funded by the National Natural Science Foundation of China (No. 81701825, No. 81770018 and No. 21605005), Social Development Foundation of Jiangsu (No. BE2018684), Natural Science Foundation of the Jiangsu Higher Education Institutions of China (No. 17KJB416012) and the Postdoctoral Science Foundation in Jiangsu Province (No. 1701141C).