Label-free SERS study of galvanic replacement reaction on silver nanorod surface and its application to detect trace mercury ion

It is significant to explore a rapid and highly sensitive galvanic replacement reaction (GRR) surface enhanced Raman scattering (SERS) method for detection of trace mercury ions. This article was reported a new GRR SERS analytical platform for detecting Hg(II) with label-free molecular probe Victoria blue B (VBB). In HAc-NaCl-silver nanorod (AgNR) substrate, the molecular probe VBB exhibited a strong SERS peak at 1609 cm−1. Upon addition of Hg(II), the GRR occurred between the AgNR and Hg(II), and formed a weak SERS activity of Hg2Cl2 that deposited on the AgNR surfaces to decrease the SERS intensity at 1609 cm−1. The decreased SERS intensity was linear to Hg(II) concentration in the range of 1.25–125 nmol/L, with a detection limit of 0.2 nmol/L. The GRR was studied by SERS, transmission electron microscopy and other techniques, and the GRR mechanism was discussed.

technique can detect directly the Raman signal of analyte and detect indirectly analyte by the Raman signal of dye molecular probe. As another application of SERS, label-free SERS techniques have begun to emerge as potential methods using Raman optical fingerprint. It has advantages of simplicity, rapidity, directly using the characteristics of sample's Raman signal without extra label sample processing. And best of all, it can avoid the damage to the samples. Hoan et al. 31 have demonstrated a novel label-free DNA biosensor based on molecular sentinel immobilized on nanowave/metal film over nanosphere substrate. This DNA biosensor is relatively easy to fabricate at low-cost and can specifically detect a complementary target DNA. Li et al. 32 proposed an iodide-modified Ag nanoparticles method for label-free detection of proteins. Aoune et al. 33 reported a simple, label-free detection scheme for DNA hybridization based on surface-enhanced Raman spectroscopy. Yang et al. 34 had successfully applied Ag nanoparticles for label-free detection of thiram on apple peels, and demonstrated their potential for use as a SERS-based on-site detection method for various pesticides. However, label-free SERS techniques used in detection of heavy metal ions were rarely.
Galvanic replacement reaction (GRR), which a more reactive metal is etched by a less reactive one, happened usually in several minutes 35 . It is particularly interesting due to its high tunability and the possibility to study the intricacies of alloying and dealloying in metallic nanostructures 36 . In addition, GRR provides a remarkably simple and versatile route to metal nanostructures with controllable hollow interiors and porous walls 37 . Xia et al. 38 used Ag as a template coupling with Au 3+ redox reaction to prepare Au hollow structure, and the reaction mechanisms were also explained. Li et al. 39 reported a strategy that anti-GRR occurred between ultra-small ssDNA-templated silver nanoclusters and Cu(II) ions to synthetize Ag/Cu alloy that was monitored by light scattering technique. Ye et al. 40 have obtained hetero-structured nanotubes which have high electrocatalytic activities by GRR of between Ag/AgCl core-shell nanowires and H 2 PtCl 6 . Jiang et al. 41 reported a highly sensitive and selective SERS substrate by GRR. Li et al. 42 presented a facile and general GRR route to produce silver and gold dendrites as well as other metal hierarchical micro/nanostructures (Cu, Pt, Pd, Ni and Co) on commercial aluminum foil in the presence of NaF or NH 4 F. The obtained silver and gold dendrites showed significantly SERS signals of a self-assembled monolayer of 2-naphthalenethiol and 4-mercaptobenzoic acid in aqueous solution. There are rare reports about GRR used in SERS quantitative analysis. Qin et al. 43 reported a strategy to complement the GRR between Ag nanocube and HAuCl 4 with co reduction by ascorbic acid (AA) for the formation of Ag-Au hollow nanostructures to detect 1,4-benzenedithiol SERS signal. The GRR between As nanoparticles and Au(III) ions has been reported for the first time by Pal et al. 44 . The potential of such an assembly was further exploited for SERS detection of Rhodamine 6 G, 4-mercaptopyridine and 4-aminothiophenol. An effective SERS substrate was fabricated by Fu et al. 45 via Ag dendrites on Al foil by GRR with [Ag(NH 3 ) 2 ]Cl for detecting biomolecules like folic acid, DNA and RNA which used as molecular probe directly in aqueous solution. The detection concentration for the three biomolecules have reached the level of 1 pmol/L, the symmetric silver dendrites can potentially be employed as effective SERS sensors for label-free and ultrasensitive biomolecule detection. Yi et al. 46 synthesized dendritic Ag-Pd bimetallic nanostructures on the surface of Cu foil via multistage GRR of Ag dendrites in a Na 2 PdCl 4 solution to detect fluorescent rhodamine 6G (Rh6G) molecules at a concentration of 10 −6 mol/L. A transparent Ag thin film composed of vertically aligned and single crystalline silver nanopetals with uniform distribution were fabricated by GRR 47 , the Raman intensity showed concentration-dependent behavior following the Freundlich equation, with a detection limit of 500 pmol/L crystal violet. As far as we know, there are no reports about the label-free SERS quantitative analysis of Hg(II) via the nanosurface GRR. In this article, the AgNR exhibited strong SERS activity, but the Ag/Hg 2 Cl 2 nanoparticles from the GRR of Hg(II)-AgNR had very low SERS activity that caused SERS quenching. Thus, a simple, rapid, sensitive and selective SERS quantitative method was established for detection of Hg(II) in aqueous solution.

Results
SERS spectra in nanosilver sol substrate. In 75 mmol/L HAc solution containing high concentrations of NaCl, VBB was adsorbed on the surface of nanosilver such as AgNR, AgNC, AgNT and AgNP aggregates with strong SERS activity and exhibited 12 SERS peaks (Table 1) 48 . Upon addition of trace Hg(II), the GRR took place between silver atoms and Hg(II) to form a larger size core-shell composite nanoparticles Ag core /Hg 2 Cl 2shell with low SERS activity that lead to SERS signals quenching. Therefore, with the increase of the Hg(II) concentration, the SERS intensity decreased linearly at 1609 cm −1 (Fig. 1, Fig. S1, S2, S3). SERS spectra of different molecular probe. In 75 mmol/L HAc solution, the diphenyl methane dye of VBB, triphenylmethane dye such as Rh6G, RhS and RhB, safranine dye of ST, acridine dye of AR, PTD and TPPS as molecular probes were examined that could be adsorbed on the surface of AgNR aggregates with strong SERS peaks (Table S1). Upon addition of trace Hg(II), the GRR took place and their SERS intensity decreased linearly ( Fig. 1, S4-S10). The SERS spectra of 8 molecular probes were compared when AgNR sol was used as SERS active substrate, and the assignment of SERS peaks were analyzed in details (Table S1).
Transmission electron microscopy (TEM). The TEMs of AgNP, AgNT, AgNR and AgNC showed in  Fig. 2b, silver nanoparticles are most triangle with the side length between 18-72 nm. Compared with AgNT, the size of AgNR was smaller with a diameter of 9 nm and a length of 18-45 nm, in addition there are spherical nanosilvers (Fig. 2c). The TEM of AgNR-Hg(II) system show that GRR would take place when trace Hg(II) was added. The reaction ended until the Ag template completely dissolved and obtained nanomercury with hollow structure (Fig. 2f). The energy spectra of AgNR-HgCl 2 system were recorded (Fig. 2e), in which Ag element exhibited three peaks at 2.984, 3.2 and 3.8 keV, Hg exhibited three peaks at 1.7, 2.159 and 10 keV, and Cu was ascribed to the copper network that was used for the loading sample. The results indicated that the bimetallic nanostructures were made up of Ag atoms and Hg atoms.
Influence of the reaction medium. The effect of reaction medium on the GRR was examined (Fig. 3).
When HCl was used as medium, the GRR would take place and the nanosurface Ag atom would lose electrons to form Ag + and dissolved into the solution. The Ag + reacted with Cl − to from white precipitation AgCl. With the increase of concentration of Hg(II), the solution gradually become turbid. The AgCl precipitation affected the stability and the sensitivity reduced compared with water medium, so HCl was not chosen as medium. When H 2 SO 4 was used as medium, nanosilver gathered because the acid was too strong, it was not beneficial to react fully between Hg(II) and Ag, so H 2 SO 4 was not chosen as medium too. When NaOH was used as medium, in the same way, Ag + would react with OH − to generate precipitation AgOH which is very unstable and easily broken down into Ag 2 O black precipitation. With the increase of concentration of Hg(II), it can be observed trace black precipitate at the bottom and the sensitivity reduced compared with water medium too, so NaOH was not chosen as medium. When HAc was used as medium, the effect of HAc concentration was examined and the VBB SERS intensity was improved greatly on the surface of the silver because the active sites can be improved in the weak acid environment 49 . The results showed that the Δ I value reached its maximum when the concentration was 75 mmol/L. In conclusion, a 75 mmol/L of HAc solution was chosen as medium.
Effect of the reducing agent. In the process of preparation of nanosilver, the effect of reductants such as NaBH 4 , H 2 O 2 and hydrazine hydrate was studied. Although the reductants will break down after a certain   respectively. Results showed that only Na 2 SO 3 inhibited the GRR reaction when the concentration was greater than 2 × 10 −5 mol/L because a strong reducing agent competed with the galvanic replacement between Ag and Hg(II) 50 . The SERS signals were not changed with the increase of concentration of the other reducing agents (Fig. 4). Thus the formation of Ag core /Hg 2 Cl 2shell was not reduced by this reductant.
Screen of nanosilver SERS substrates. The effect of four kinds nanosilver sol substrate was examined when VBB was used as SERS molecular probe (Fig. 5). The results showed that the Δ I value reached maximum when the concentration of AgNP was 4.5 × 10 −5 mol/L and the concentration of AgNT, AgNR and AgNC was 1.0 × 10 −4 mol/L. Thus, a 4.5 × 10 −5 mol/L AgNP, and 1.0 × 10 −4 mol/L AgNT, AgNR and AgNC were chosen. The AgNR sol substrate was most sensitive to detect Hg(II).
Analytical feature and application. Under the optimal conditions, for the Hg(II)-nanosilver-VBB system, four kinds of nanosilver sols were chosen as substrates respectively, the SERS intensity for different Hg(II) concentrations(C) were recorded and the working curves were drawn according to the relationship between C and their corresponding Δ I values. The results showed that AgNR had the highest sensitivity among the four kinds of nanosilver (Table 2). From the comparison of reported assays for SERS detection of Hg(II) [51][52][53] (Table  S3), this GRR method has the advantages of simplicity, rapidity, high sensitivity and specificity. Thus, it can be used as a new method for rapid detection of Hg(II). For the Hg(II)-AgNR-molecular probe system, 8 kinds of molecular probes were considered for detection of Hg(II), the SERS intensity for different Hg(II) concentrations(C) were recorded and their working curves were drawn according the relationship between C and their corresponding Δ I values. The Fig. 6 showed that the VBB had the highest sensitivity among the 8 molecular probes (Table 3). In the sample solution, the coexistence of metal ions are mainly sodium, potassium, calcium, magnesium, zinc, aluminum, titanium, silicon and iron which are not interfere with the determination of Hg(II). There are no gold, palladium and platinum ions in the cosmetics samples, although they interfered with the determination and had the relative errors of 18%, 20% and 15%, respectively, when they were in the same concentration level as mercury ions. So the determination results are credible. The Hg contents in six kinds of cosmetic sample have been detected by the SERS method. Then, a known amount of Hg(II) was added into the sample to obtain the recovery. The results (Table S4) showed that the relative standard deviation was in the range of 0.58-3.72%, and the recovery was in the range of 95.40-107.56% that indicated this method was accuracy. The US Food and Drug Administration (FDA) limits the amount of mercury in cosmetic products to a trace amount unavoidable under good manufacturing practice, 1 ppm (1μg/g), except in those products intended for use around the eye, which it limits to 65 ppm [54][55][56] . According to the standards, the Hg content in the detected cosmetic products up to grade.  Discussion Analytical principle. As we know, GRR takes place between a suspension of nanoscale metal templates and a salt precursor containing a relatively less active metal. The driving force of the GRR is the electrical potential difference between two metals. Generally, a GRR can be split into two half reactions, the oxidation/dissolution of a metal at the anode, and the reduction/deposition of the ions and a second metal at the cathode. It is critical that the electrochemical potential of the metal ions is higher than that of the solid metal to occur this reaction 57 . Since the standard reduction potential (SRP) of Hg 2+ /Hg 0 (0.851 V vs Standard Hydrogen Electrode, SHE) and Hg 2+ /Hg 2 2+ (0.920 V vs SHE) are higher than that of Ag + /Ag (0.7996 V vs SHE), the surface of Ag nanostructures suspended in the solution can be oxidized by Hg(II), in which produce Ag + , Hg 2 2+ and Hg 0 according to reaction (1) and (2)  In 75 mmol/L HAc solution containing high concentration of NaCl, the SERS molecular probe was adsorbed on the surface of AgNR aggregates sol with strong SERS signal. Upon addition of trace Hg(II), it would be adsorbed on the surface of AgNR. Ag atom would lose electrons to form Ag + and the Hg(II) got electrons to form Hg 2 Cl 2 on the outer layer of AgNR because of the GRR of Ag atoms and Hg(II) (Fig. 7A), meanwhile the hydrophobic Hg 2 Cl 2 and AgCl molecules formed on the surfaces. With the reaction going on, Ag atom continues losing electrons to form Ag + that dissolved into the solution and formed the larger Ag pores. The SRP of Ag + /Ag was higher than that of Hg 2 2+ /Hg 0 , and this made Ag + oxidize Hg 0 to Hg 2 2+ , and the Hg(II) form Hg 2 Cl 2 continuous precipitation. The final products were larger size core-shell composite nanoparticles with the silver nanoparticles as core and Hg 2 Cl 2 as shell (Ag core -Hg 2 Cl 2shell ). The reaction ended until the Ag template completely dissolved and obtained hollow structure Ag core /Hg 2 Cl 2shell (Fig. 7A). The AgNR was one of the strongest SERS substrates but the nanomercury was weaker than AgNR. Therefore, with the increased concentration of Hg(II), the amount of core-shell composite nanoparticles Ag core -Hg 2 Cl 2shell increased and lead to the SERS intensity decrease (Fig. 7B). Base on this ground, a simple, rapid, sensitive and selective GRR label-free SERS method was established to detect trace Hg(II) in aqueous solution.

System
Regression equation Linear range (nmol/L Hg)   Indentifying GRR product of Hg(II). In order to identifie directly the final product is Hg 0 or Hg 2 2+ with the naked eye, the concentration of reactant are increased multiply. The AgNR exhibited two SPR absorption peaks at 312 nm and 430 nm (Fig. 8a). It would take place GRR with Hg(II) to exhibit turbid white solution (Fig. 8b). But it was not observed black elemental mercury precipitation in solution. So this phenomenon showed clearly that there were no elemental mercury in the reaction products and Hg 0 has been oxidized by Ag + because the SRP of Ag + /Ag (0.7996 V vs SHE) is higher than that of Hg 2 2+ /Hg 0 (0.7973 V vs SHE). The white turbid solution was the mixture of Ag core /(Hg 2 Cl 2 ) shell and a part of Hg 2 Cl 2 in solution. The UV absorption decreased since the Hg(II) oxidized the AgNR to colorless Ag + , the SPR absorption peak of AgNR disappearing intimated that AgNR has been oxidized by Hg(II). In order to verify the final product of GRR reaction is Hg 2 2+ , a 0.24 mol/L NH 3 .H 2 O was added into the solution. Figure 8b obtained a black solution due to the reaction between ammonia and Hg 2 Cl 2 generate white precipitate Hg(NH 2 )Cl and black precipitate mercury. The Fig. 8c solution without SPR peak showed that there was no AgNR in the solution. Therefore it can be ruled out the possibility of the black solution being the excess AgNR aggregates, but the reaction of ammonia and Hg 2 Cl 2 generated black precipitate mercury. In conclusion, the final products were Ag + and Hg 2 2+ . A 500 μ L 0.3 mol/L HAc, 500 μ L AgNR or AgNC, AgNT and AgNP solution, a certain amount of HgCl 2 solution were added into a 5 mL marked-test tube respectively and mixed well. Different molecular probe such as  VBB, ST, Rh6G, RhS, RhB, AR, MH, FL, PTD and TPPS was added in the mixture respectively, and diluted to 2 mL. If VBB, VB4R, ST were used as SERS molecular probe, a 20 mmol/L NaCl was added. The mixture was transferred to a 1 cm quartz cell, the SERS intensity and the corresponding blank value I 0 without HgCl 2 were recorded. The value of Δ I = I 0 -I was obtained.
Pretreatment of samples. The vanadium pentoxide/nitric acid/sulfuric acid was selected to treat samples.
In the digestion process, the organic mercury can be decomposed completely and inorganic mercury was not loss because the pentavalent vanadium fix effectively inorganic mercury under temperature 140 °C. However, when the digestion temperature was higher than 140 °C, pentavalent vanadium will turn to tetravalent vanadium, but the fixed ability of tetravalent vanadium are far below pentavalent vanadium. Therefore 100 °C was chosen as digestion temperature. Vanadium pentoxide was reduced to blue tetravalent vanadium end of the digestion process. The pretreatment of cosmetics sample was as follows 59 : a 0.5 g sample, 0.05 g vanadium pentoxide and 7 mL nitric acid were added into a 100 mL conical flask, the mixture was heated at 100 °C to boil with stirring by magnetic heated stirrer. A 8 mL sulfuric acid was added after cool down and stop the digestion until color turns to blue-green. A small amount of water was added after solution cooling, then continue heating to get rid of the nitrogen dioxide and neutralizing excess acid in solution. The sample solution was obtained after diluting to 100 ml. A 200 μ L sample solution was used to analyze mercury content according to the procedure. Then, a known amount of HgCl 2 was added into the sample solution to obtain the recovery.