Recyclable Multifunctional Magnetic Fe 3 O 4 @SiO 2 @Au Core/Shell Nanoparticles for SERS Detection of Hg (II)

: Mercury ions can be enriched along the food chain and even low concentrations of mercury ions can seriously affect human health and the environment. Therefore, rapid, sensitive, and highly selective detection of mercury ions is of great significance. In this work, we synthesized Fe 3 O 4 @SiO 2 @Au three-layer core/shell nanoparticles, and then modified 4-MPy (4-mercaptopyridine) to form a SERS sensor. Mercury ions in water can be easily captured by 4-MPy which were used as the reporter molecules, and the concentration of mercury ions can be evaluated based on the spectral changes (intensiﬁcation and reduction of peaks) from 4-MPy. After the mercury ion was combined with the pyridine ring, the peak intensity at 1093 cm − 1 increased with the concentration of mercury ion in the range of 10 ppm–1 ppb, while the Raman intensity ratio I (416 cm − 1 )/I (436 cm − 1 ) decreased with the increase of mercury ion concentration. This magnetically separatable and recyclable SERS sensor demonstrates good stability, accuracy, and anti-interference ability and shows the potential to detect actual samples. Furthermore, we demonstrate that the probe is applicable for Hg 2+ imaging in macrophage cells.


Introduction
Heavy metal ions are among the main pollutants of environmental waters, in particular, the Hg (II) ions have been a concern for a long time due to its serious damage to human health and the environment [1,2]. According to the legislative controls directed by U. S. Environmental Protection Agency (USEPA), the maximum concentration of Hg (II) in drinking water is limited to 2 µg/L (2 ppb, 10 nM) [3]. In China, according to the National Drinking Water Quality Standards (GB5749-2006), the concentration limit of Hg (II) in drinking water is 1 µg/L (1 ppb, 5 nM) [4]. The pollution of the Hg 2+ ions can prevent nutrient absorption and photosynthesis progresses in plants [5] and cause chronic poisoning of animals [6]. The Hg 2+ exposure, even at relatively low concentrations, impairs human health and causes serious diseases including broken immune systems [7,8], kidney damage [9,10], nervous injury [11,12], memory disorders [13], and cognitive impairment [14]. Routine detection methods for trace Hg (II) include atomic fluorescence spectrometry (AFS) [15] inductively coupled plasma mass spectrometry (ICP-MS) [16], and atomic absorption/emission spectroscopy (AAS/AES) [17,18]. However, most of these techniques require higher-cost instruments or tedious sample processing, which made them high cost, time -consuming, and unsuitable for on-site analysis. Therefore, it is quite electronic distribution in 4-MPy, which caused a number of characteristic spectral changes including peak blue shift and Raman signal enhancement. With the addition of the Hg ions, the SERS signal of 4-MPy changes dramatically, including the appearance of a new peak located at 416 cm −1 and the Raman signal intensity enhancement at 1093 cm −1 . In order to figure out the reasons for these Raman signal variations, the hybrid density functional (h-DFT) calculation is introduced to validate the electron distribution in pyridine and the reorientation of 4-MPy on the Au surface. Making use of the peak intensity enhancement at 1093 cm −1 , we show that indirect determination of mercury ion concentration can be achieved with a good linear relationship between 10 ppm-1 ppb. In addition, the intensity ratio at I (416 cm −1 )/I(436 cm −1 ) can also be used to judge the concentration of mercury ions. This allows the presented sensor to have multiple judgment rules for identifying Hg (II) ions. In addition, the sensor has excellent stability, selectivity, reproducibility, and recyclability. We have also successfully applied this sensor to Hg (II) ions monitoring in macrophages through Raman imaging.

Instrumentations
Scanning electron microscope (SEM) Hitachi SU8010 and transmission electron microscope (TEM, JEM-2010) were utilized for surface morphology, core/shell structure, and element composition. Measurement was implemented with a LabRAM HR Evolution spectrometer (HORIBA, Kyoto, Japan), equipped with a 785 nm laser (0.7 mW on sample), and a 50 × long working distance objective. The exposure time for each point was 30 s. The Raman mapping of SERS sensor test was carried out at 1 µm step size, with an integration time of 10 s per spectrum. For statistical analysis, Raman scans of 3600 points were recorded to obtain a large number of nanoparticles for the data set.

Synthesis of Fe 3 O 4 @SiO 2 @Au NPs
The synthesis process includes three stages: (1)  The magnetic particles were synthesized by using a non-autoclave solvent-thermal method according to the reported method with minor modifications [38]. Typically, 40 mL ethylene glycol and 0.65 g FeCl 3 were sequentially added into a 50 mL Erlenmeyer flask. After dissolution of the precursor, 1.0 g PSSMA and 0.1 mL H 2 O were added. Magnetic stirring was used for 1 h to obtain a homogeneous mixture, and then, 0.6 g NaOH was added to mixture. Finally, the magnetic bar was removed, and the solution was heated to 190 • C in Muffle furnace for 9 h followed by cooling to room temperature. The obtained black powder was washed with ethanol and water alternately. Finally, Fe 3 O 4 MNPs were re-dispersed in 30 mL H 2 O.

Synthesis of Fe 3 O 4 @SiO 2 MNPs
The method of synthesizing Fe 3 O 4 @SiO 2 is based on the improved Stöber method of previous research [39]. The thickness of the SiO 2 shell can be controlled by adjusting the molar ratio of TEOS, water, and ammonia. Typically, in a three-necked flask, 12 mL of aqueous dispersion of Fe 3 O 4 NPs, 80 mL of ethanol, and 0.8 mL of ammonium hydroxide were mixed together and sonicated for 20 min. With mechanical stirring, 0.5 mL TEOS was dropwise injected into the solution.  [40]. The surface of silica nanoparticles can be easily modified by PEI. Typically, 0.5 g PEI was dissolved in 50 mL of deionized water by ultrasonication for 15 min. Then, 15 mL Fe 3 O 4 @SiO 2 was dispersed in the PEI solution under mechanical stirring at room temperature for 10 h, during which PEI gradually self-assembled on the silica cores. The obtained Fe 3 O 4 @SiO 2 -PEI MNPs were centrifuged and washed 3 times with deionized water to remove the excess PEI. Finally, the PEI-functionalized Fe 3 O 4 @SiO 2 MNPs were dispersed in 30 mL water.

Preparation of Gold Nanoparticles
The preparation method of 1-3 nm gold seeds follow the method developed by Duff et al. [41]. Specifically, 5 mL of 1 M NaOH was added to 450 mL of ultrapure water and 10 mL of 50 mM THPC was added under magnetic stirring. After 10 min, 0.36 mL of 1.0 M HAuCl 4 ·3H 2 O was added and after stirring for 30 min, the gold seed solution was stored and aged at 4 • C for a week.

Preparation of Gold-Attached Fe 3 O 4 @SiO 2 MNPs
According to the fact that negatively charged gold nanoparticles can be attached to the PEI functionalized silica core, the gold-attached Fe 3 O 4 @SiO 2 nanocomposites were prepared through self-assembly [42]. Firstly, 10 mL Fe 3 O 4 @SiO 2 -PEI and 75 mL gold seed were added into 75 mL water, and the mixture reacted overnight to form SiO 2 @PEI-Au seed. Then, washed three times with water to remove excess unattached small gold nanoclusters, in order to avoid the growth of pure gold particles during the formation of the gold shell. Finally, the Fe 3 O 4 @SiO 2 -Au MNPs were dispersed in 6 mL water.

Preparation of Three-Layer Core/Shell Fe 3 O 4 @SiO 2 @Au Magnetic Nanocomposites
The three-layer core/shell Fe 3 O 4 @SiO 2 @Au MNPs were prepared by self-assembly, seed-mediated growth process and chemical reduction. A kind of plating solution containing reducible gold salt must be prepared in advance. In a conical flask, 0.1 g of potassium carbonate was dissolved in 400 mL of water. After stirring for 15 min, 6 mL of 1% HAuCl 4 in water were added to obtain HAuCl 4 plating solution. HAuCl 4 plating solution was stored for a minimum of 48 h in the dark [43]. Finally, 1 mL Fe 3 O 4 @SiO 2 -Au was injected into 100 mL HAuCl 4 plating solution with mechanical stirring. After ultrasonic for 5 min, 30 µL formaldehyde were added to reduce the HAuCl 4 . When the reaction solution turns purple-red, the reaction was stopped and the product was collected by magnetic separation and wash with water three times to obtain the three-layer core/shell Fe 3 O 4 @SiO 2 @Au magnetic nanocomposites [44].

Enhancement Factor (EF) Calculation
The magnetic nanoparticles were immersed in 4-MPy solution at different concentrations to prepare functionalized SERS sensor. As shown in Figure S1, a concentration of 10 −5 M 4-MPy solution modified magnetic nanoparticles obtained the best SERS signal.
On another aspect, the SERS signals of the 4-MPy modified on the MNPs surface and the normal Raman signals of 4-MPy were compared. The average EF value was calculated according to the previous literature [45]: The baseline correction was conducted by the Labspec 6 software. The intensity of the Raman peak located 1093 cm −1 was adopted to calculate the EF value, and the estimated average EF value was about 5.24 × 10 4 .

SERS Measurements of Hg (II) Ions
The magnetic SERS sensor was incubated with different concentrations of Hg (II) ions for 30 min. The reaction was stopped by removing the core/shell nanoparticles from the mixture through magnetic separation. After washing with pure water twice, 5 µL of the nanoparticle's suspension was dropped on aluminum foil paper for Raman recording. The solution of the metal ions metal ions (Zn 2+ , Ca 2+ , Na + , Mg 2+ , Mn 2+ , Pb 2+ , Ba 2+ , Cu 2+ , Ag + , Ni 2+ , Cr 3+ and Co 2+ ), anions (PO 4 3− , NO 3 − , Cl − , F − , CO 3 2− , CH 3 COO − , H 2 PO 4 − , HPO 4 2− ), protein (BSA) and humic acid were tested as the same procedure [46]. For the reusing performance test, an EDTA solution (10 −3 M) was incubated with the nanoparticles solution for 1 h to refresh the nanoparticles. Different concentrations of Hg (II) ions (10 ppm-1 ppb) were mixed with the actual water samples collected from Chaohu lake to simulate polluted water. The particle suspension in the lake water was removed by double-layer filter paper, and then filtered with a 0.22 µm microporous filter [47,48]. The Raman test of the simulated samples was under the same condition.
For detection of mercury ions in cells, 4 × 10 5 cells were seeded in the CaF 2 coverslips in fresh medium for 24 h. The cells were then exposed to Fe 3 O 4 @SiO 2 @Au/4MPy in serum-free medium. After 24 h incubation, the culture medium was removed, and the cells were quickly maintained with PBS and Hg (II) ions for 1 h. Cells were fixed with 10% neutral buffered formalin (with approximately 4% of formaldehyde) for 15min. As prepared cells were used for the cell Raman mapping. Due to getting high SERS signal from the nanoparticles, the 785 nm laser (0.75 mW, integration time 25 s, 50 × LWD objective lens) was adopted to do the mapping. Additionally, then, for the same mapping area, 532 nm laser (5 mW, integration time 20 s, 50 × LWD objective lens) was used for cell mapping. All the Raman data was treated by Labspec 6 including baseline correction, smoothing, and normalization.

Fabrication and Characterization of Fe 3 O 4 @SiO 2 @Au
The ferroferric oxide nanoparticles were synthesized by an improved solvent-thermal method [51]. This method adjusts the amount of NaOH to change the crystallinity of ferroferric oxide without high temperature. Therefore, ordinary glass containers such as conical flasks can be used instead of using a hydrothermal kettle for the reaction. Compared with the polytetrafluoroethylene liner, the glass container is easier to clean. Since the reaction is not sensitive to the presence of limited oxygen, thus avoiding the tedious work of deoxidation. In addition, the size of the magnetic particles can be controlled by controlling the amount of H 2 O added.
Through the Stöber reaction, silicon dioxide can be readily coated on Fe 3 O 4 to form Fe 3 O 4 @SiO 2 core/shell microspheres. The thickness of the shell can be easily adjusted by changing the amount of EtOH added. The SiO 2 shell layer can avoid the aggregation and oxidation of Fe 3 O 4 particles in the harsh liquid environment, and provide a site for Au seed deposition. The core/shell microspheres maintain the morphological characteristics of pure Fe 3 O 4 in addition to maintaining a larger particle size of about 220 nm. The positive charges and primary amine groups on the PEI can achieve rapid assembly of negatively charged particles. In this way, the positively charged PEI layer adsorbed negatively charged 1-3 nm Au NPs on the surface of Fe 3 O 4 @SiO 2 -PEI NPs. These small AuNPs were firmly attached to Fe 3 O 4 @SiO 2 -PEI NPs through the covalent bonds between -NH. Then, based on the seed-mediated method, using 1-3 nm AuNPs as the binding sites, the HAuCl4 plating solution was reduced by formaldehyde to form a continuous gold shell. Finally, Fe 3 O 4 @SiO 2 @Au was soaked in a 4-MPy solution for 0.5 h to obtain a layer of 4-MPy modification. The synthetic route of composite microspheres is shown in Figure 1

Fabrication and Characterization of Fe3O4@SiO2@Au
The ferroferric oxide nanoparticles were synthesized by an improved solvent-the mal method [51]. This method adjusts the amount of NaOH to change the crystallinity ferroferric oxide without high temperature. Therefore, ordinary glass containers such conical flasks can be used instead of using a hydrothermal kettle for the reaction. Com pared with the polytetrafluoroethylene liner, the glass container is easier to clean. Sin the reaction is not sensitive to the presence of limited oxygen, thus avoiding the tedio work of deoxidation. In addition, the size of the magnetic particles can be controlled b controlling the amount of H2O added.
Through the Stöber reaction, silicon dioxide can be readily coated on Fe3O4 to for Fe3O4@SiO2 core/shell microspheres. The thickness of the shell can be easily adjusted b changing the amount of EtOH added. The SiO2 shell layer can avoid the aggregation an oxidation of Fe3O4 particles in the harsh liquid environment, and provide a site for A seed deposition. The core/shell microspheres maintain the morphological characteristi of pure Fe3O4 in addition to maintaining a larger particle size of about 220 nm. The positi charges and primary amine groups on the PEI can achieve rapid assembly of negative charged particles. In this way, the positively charged PEI layer adsorbed negative charged 1-3 nm Au NPs on the surface of Fe3O4@SiO2-PEI NPs. These small AuNPs we firmly attached to Fe3O4@SiO2-PEI NPs through the covalent bonds between -NH. The based on the seed-mediated method, using 1-3 nm AuNPs as the binding sites, t HAuCl4 plating solution was reduced by formaldehyde to form a continuous gold she Finally, Fe3O4@SiO2@Au was soaked in a 4−MPy solution for 0.5 h to obtain a layer 4−MPy modification. The synthetic route of composite microspheres is shown in Figure   Figure 1. Synthesis route of Fe3O4@SiO2@Au/4-MPy composite microspheres.
TEM and SEM ( Figure 2) were used to analyze and inspect the morphology of sam ples obtained at different stages. It can be seen from SEM that the synthesized nanopar cles have relatively uniform particle sizes and good dispersibility. From the results of t particle size calculation, it can be concluded that the average diameter of the Fe3O4 core 185 nm, the thickness of the silica shell is 13 nm, and the gold shell is 8 nm. TEM an element maps demonstrate the presence of Si and Au, which further proved the successf synthesis of a three-layer core/shell structure, being in agreement with the SEM observ tions. TEM and SEM ( Figure 2) were used to analyze and inspect the morphology of samples obtained at different stages. It can be seen from SEM that the synthesized nanoparticles have relatively uniform particle sizes and good dispersibility. From the results of the particle size calculation, it can be concluded that the average diameter of the Fe 3 O 4 core is 185 nm, the thickness of the silica shell is 13 nm, and the gold shell is 8 nm. TEM and element maps demonstrate the presence of Si and Au, which further proved the successful synthesis of a three-layer core/shell structure, being in agreement with the SEM observations.

SERS Measurement of Hg (II) Ions
Scheme 1 shows the actual detection flow chart. The magnetic SERS nanoparticles were incubated with the water sample for 0.5 h, and then magnetically separated for Raman measurements. Figure 3 compares the Raman spectra of magnetic SERS nanoparticles before and after the addition of mercury ions. It can be seen that after the combination of mercury ions, the overall intensity of the 4−MPy was enhanced, as proved by the peaks at 1012 cm −1 , 1575 cm −1 , and 1093 cm −1 . At 416 cm −1 and 436 cm −1 , there was a blue shift of the peaks. Additionally, the attribution of Raman peaks of 4−MPy modified on magnetic microspheres can be found in Table 1. The corresponding intensity ratio of 416 cm −1 /436 cm −1 can be also used as a reference for the quantification of Hg (II) ions. Therefore, unlike traditional methods that only rely on changes in spectral intensity of the characteristic fingerprints, we can accurately identify Hg (II) ions through multiple features including the intensity and the ratio value of the peaks.

SERS Measurement of Hg (II) Ions
Scheme 1 shows the actual detection flow chart. The magnetic SERS nanoparticles were incubated with the water sample for 0.5 h, and then magnetically separated for Raman measurements. Figure 3 compares the Raman spectra of magnetic SERS nanoparticles before and after the addition of mercury ions. It can be seen that after the combination of mercury ions, the overall intensity of the 4-MPy was enhanced, as proved by the peaks at 1012 cm −1 , 1575 cm −1 , and 1093 cm −1 . At 416 cm −1 and 436 cm −1 , there was a blue shift of the peaks. Additionally, the attribution of Raman peaks of 4-MPy modified on magnetic microspheres can be found in Table 1. The corresponding intensity ratio of 416 cm −1 /436 cm −1 can be also used as a reference for the quantification of Hg (II) ions. Therefore, unlike traditional methods that only rely on changes in spectral intensity of the characteristic fingerprints, we can accurately identify Hg (II) ions through multiple features including the intensity and the ratio value of the peaks.

SERS Measurement of Hg (II) Ions
Scheme 1 shows the actual detection flow chart. The magnetic SERS nanopart were incubated with the water sample for 0.5 h, and then magnetically separated fo man measurements. Figure 3 compares the Raman spectra of magnetic SERS nanop cles before and after the addition of mercury ions. It can be seen that after the combin of mercury ions, the overall intensity of the 4−MPy was enhanced, as proved by the p at 1012 cm −1 , 1575 cm −1 , and 1093 cm −1 . At 416 cm −1 and 436 cm −1 , there was a blue sh the peaks. Additionally, the attribution of Raman peaks of 4−MPy modified on mag microspheres can be found in Table 1. The corresponding intensity ratio of 416 cm − cm −1 can be also used as a reference for the quantification of Hg (II) ions. Therefore, u traditional methods that only rely on changes in spectral intensity of the characte fingerprints, we can accurately identify Hg (II) ions through multiple features inclu the intensity and the ratio value of the peaks.    The Raman signal uniformity of the Fe3O4@SiO2@Au/4−MPy microspheres is larly important for quantitative analysis and the Raman mapping experiment rected conducted so as to confirm the signal uniformity. As shown in Figure 4, the overlapped SERS spectra without addition of mercury ions show good reprodu After treatment with mercury ions, the fluctuation of the SERS spectrum increases s which may be induced by the structure changes produced by the incorporation cury ions. For Raman mapping measurement, an area of 60 μm × 60 μm was su with a step size of 1 μm. The total 3600 spectrum were recorded and the relative st deviations (RSD) was calculated to be 14% and 19%, respectively.  The Raman signal uniformity of the Fe 3 O 4 @SiO 2 @Au/4-MPy microspheres is particularly important for quantitative analysis and the Raman mapping experiment was directed conducted so as to confirm the signal uniformity. As shown in Figure 4, the highly overlapped SERS spectra without addition of mercury ions show good reproducibility. After treatment with mercury ions, the fluctuation of the SERS spectrum increases slightly, which may be induced by the structure changes produced by the incorporation of mercury ions. For Raman mapping measurement, an area of 60 µm × 60 µm was surveyed with a step size of 1 µm. The total 3600 spectrum were recorded and the relative standard deviations (RSD) was calculated to be 14% and 19%, respectively. In order to quantitatively analyze Hg 2+ ions in water, aqueous solutions with different mercury ion concentration were prepared for SERS measurement. Figure 5 depicts the typical SERS spectra of Fe3O4@SiO2@Au/4−MPy microspheres upon exposure to different concentrations of mercury ions. It can be seen that the peak intensity at 1093 cm −1 increases with the increase of the mercury ion concentration lineally. For quantitative evaluation there is a good linear relationship between 10 −5 and 10 −9 . In addition, since EDTA has a chelating ability for mercury ions, the MNPs sensor can be treated with EDTA to achieve reusability. [56] Figure 6 clearly shows that after incubation with EDTA, the Raman intensity of 1093 cm −1 can be recovered as raw level Additionally, then, with the addition of Hg 2+ , the 1093 cm −1 shift demonstrates similar intensity as before. It can be seen from the figure that the sensor can be recycled more than In order to quantitatively analyze Hg 2+ ions in water, aqueous solutions with different mercury ion concentration were prepared for SERS measurement. Figure 5 depicts the typical SERS spectra of Fe 3 O 4 @SiO 2 @Au/4-MPy microspheres upon exposure to different concentrations of mercury ions. It can be seen that the peak intensity at 1093 cm −1 increases with the increase of the mercury ion concentration lineally. For quantitative evaluation, there is a good linear relationship between 10 −5 and 10 −9 . In order to quantitatively analyze Hg 2+ ions in water, aqueous solutions with different mercury ion concentration were prepared for SERS measurement. Figure 5 depicts the typical SERS spectra of Fe3O4@SiO2@Au/4−MPy microspheres upon exposure to different concentrations of mercury ions. It can be seen that the peak intensity at 1093 cm −1 increases with the increase of the mercury ion concentration lineally. For quantitative evaluation, there is a good linear relationship between 10 −5 and 10 −9 . In addition, since EDTA has a chelating ability for mercury ions, the MNPs sensor can be treated with EDTA to achieve reusability. [56] Figure 6 clearly shows that after incubation with EDTA, the Raman intensity of 1093 cm −1 can be recovered as raw level. Additionally, then, with the addition of Hg 2+ , the 1093 cm −1 shift demonstrates similar intensity as before. It can be seen from the figure that the sensor can be recycled more than four times. We also checked the stability of the Raman chips. As shown in Figure S4, the Raman signals remain stable within 55 days of storage in water. In addition, since EDTA has a chelating ability for mercury ions, the MNPs sensor can be treated with EDTA to achieve reusability. [56] Figure 6 clearly shows that after incubation with EDTA, the Raman intensity of 1093 cm −1 can be recovered as raw level. Additionally, then, with the addition of Hg 2+ , the 1093 cm −1 shift demonstrates similar intensity as before. It can be seen from the figure that the sensor can be recycled more than four times. We also checked the stability of the Raman chips. As shown in Figure S4, the Raman signals remain stable within 55 days of storage in water.

The Explanation of 416 cm −1 Raman Band by Hybrid Density Functional (h-DFT)
In order to study the effect of Hg ions on the Raman vibrational bands of 4−MPy, h-DFT was used. The geometries of different absorbing configurations were constructed and optimized at the M062X hybrid density function in combination with the lanl2dz basis set on metal atoms (Au and Hg) and 6−311G (d, p) on other atoms (C, N, S and H) using Gaussian 16 package. The simulated Raman spectra were shown in Figure 7. With the interaction of Hg 2+ , the vertical adsorption of 4−MPy with the presence of Hg 2+ can enhance the vibration intensity of the peak located at 1093 cm −1 attributed to pyridine breathing mode. Due to the attractive π−π interaction between 4−MPy molecules, it has a high surface coverage. The π−π interaction also leads to the multi-dentate chelation of Hg 2+ bonding to the N atoms from several pyridine molecules. After multi-dentate chelation, the peak of C−S stretching mode has a blue shift, which is consistent with the experimental trend.

The Explanation of 416 cm −1 Raman Band by Hybrid Density Functional (h-DFT)
In order to study the effect of Hg ions on the Raman vibrational bands of 4−MPy, h-DFT was used. The geometries of different absorbing configurations were constructed and optimized at the M062X hybrid density function in combination with the lanl2dz basis set on metal atoms (Au and Hg) and 6−311G (d, p) on other atoms (C, N, S and H) using Gaussian 16 package. The simulated Raman spectra were shown in Figure 7. With the interaction of Hg 2+ , the vertical adsorption of 4−MPy with the presence of Hg 2+ can enhance the vibration intensity of the peak located at 1093 cm −1 attributed to pyridine breathing mode. Due to the attractive π−π interaction between 4−MPy molecules, it has a high surface coverage. The π−π interaction also leads to the multi-dentate chelation of Hg 2+ bonding to the N atoms from several pyridine molecules. After multi-dentate chelation, the peak of C−S stretching mode has a blue shift, which is consistent with the experimental trend. With the interaction of Hg 2+ , the vertical adsorption of 4-MPy with the presence of Hg 2+ can enhance the vibration intensity of the peak located at 1093 cm −1 attributed to pyridine breathing mode. Due to the attractive π − π interaction between 4-MPy molecules, it has a high surface coverage. The π − π interaction also leads to the multi-dentate chelation of Hg 2+ bonding to the N atoms from several pyridine molecules. After multidentate chelation, the peak of C-S stretching mode has a blue shift, which is consistent with the experimental trend.
To find out the SERS performance of Fe 3 O 4 @SiO 2 @Au MNPs, the electric field distribution map of the single MNP and two MNPs nearby were simulated by FDTD Solutions (Lumerical Solution Inc., Vancouver, Canada) and the corresponding results were shown in Figure S2. Obviously, Fe 3 O 4 @SiO 2 /Au MNP dimer gap has obtained strong SERS intensity, which is beneficial to enhance the Raman signals of the analytes.

The Selectivity Analysis of as Prepared Sensors
Taking into account the application of this magnetic sensor in actual detection, we also conducted a series of selective experiments of the SERS sensor to Hg 2+ . A stable MPy-Hg 2+ -MPy complex can be formed between 4-MPy and Hg 2+ which provides excellent selectivity over other interfering metal ions for the detection of Hg 2+ . The sensor's selectivity was tested including the following ions (Ag + , Ba 2+ , Co 2+ , Cr 3+ , Mg 2+ , Mn 2+ , Ni 2+ , Ca 2+ , Na + , and Pb 2+ with a concentration of 100 ppm) and the possible coexistence of anions (CH 3 COO − , CO 3 2− , H 2 PO 4 2− , HPO 4 2− , NO 3 2− ), BSA and HA in the water. As shown in Figure 8, the Raman intensity ratio of I (416 cm −1 )/I (436 cm −1 ) acted as a reference, the results show that the SERS intensity ratio used to detect mercury ions is significantly stronger than the interference factors in other environments. To find out the SERS performance of Fe3O4@SiO2@Au MNPs, the electric field distribution map of the single MNP and two MNPs nearby were simulated by FDTD Solutions (Lumerical Solution Inc., Vancouver, Canada) and the corresponding results were shown in Figure S2. Obviously, Fe3O4@SiO2/Au MNP dimer gap has obtained strong SERS intensity, which is beneficial to enhance the Raman signals of the analytes.

The Selectivity Analysis of as Prepared Sensors
Taking into account the application of this magnetic sensor in actual detection, we also conducted a series of selective experiments of the SERS sensor to Hg 2+ . A stable MPy-Hg 2+ -MPy complex can be formed between 4−MPy and Hg 2+ which provides excellent selectivity over other interfering metal ions for the detection of Hg 2+ . The sensor's selectivity was tested including the following ions (Ag + , Ba 2+ , Co 2+ , Cr 3+ , Mg 2+ , Mn 2+ , Ni 2+ , Ca 2+ , Na + , and Pb 2+ with a concentration of 100 ppm) and the possible coexistence of anions (CH3COO − , CO3 2− , H2PO4 2− , HPO4 2− , NO3 2− ), BSA and HA in the water. As shown in Figure  8, the Raman intensity ratio of I (416 cm −1 )/I (436 cm −1 ) acted as a reference, the results show that the SERS intensity ratio used to detect mercury ions is significantly stronger than the interference factors in other environments.

The Application of Hg 2+ Detection in Real Water Samples
We also conducted actual sample detection tests by incorporating mercury ions into actual water solutions. Filter paper and then 0.22 μm filter membrane were introduced to remove the particle precipitation in the actual water samples. Then, the SERS sensor was immersed in the actual Hg ions contained water sample and incubated for 30 min, and the sensor was separated by magnetism. Figure 9 shows the results of one sample with addition of mercury ions at the concentration of 10 −5 M. Obviously, the raman peak intensity located at 1093 cm −1 of the sample with addition of Hg 2+ is higher, which is consistent with the foregoing experiments. This experiment clearly confirmed that the SERS sensor has ability to detect mercury ions in actual environmental water.

The Application of Hg 2+ Detection in Real Water Samples
We also conducted actual sample detection tests by incorporating mercury ions into actual water solutions. Filter paper and then 0.22 µm filter membrane were introduced to remove the particle precipitation in the actual water samples. Then, the SERS sensor was immersed in the actual Hg ions contained water sample and incubated for 30 min, and the sensor was separated by magnetism. Figure 9 shows the results of one sample with addition of mercury ions at the concentration of 10 −5 M. Obviously, the raman peak intensity located at 1093 cm −1 of the sample with addition of Hg 2+ is higher, which is consistent with the foregoing experiments. This experiment clearly confirmed that the SERS sensor has ability to detect mercury ions in actual environmental water.

Test of Detection of Hg 2+ for Cell Analysis
Furthermore, the magnetic SERS sensor was employed for cell analysis. The macrophages were incubated with a magnetic nanoprobe. After the probe was endocytosed by the macrophages, the macrophages were treated with mercury ions containing PBS for 1h, and then the Raman mapping was performed. The corresponding images were shown in Figure 10. The results show that the intensity ratio of 416 cm −1 to 436 cm −1 was significantly decreased and the intensity of the 1093 cm −1 was enhanced with the addition of Hg ions. Due to the complex composition of the cell, the signal of the SERS material in the cell will inevitably be affected. The dual criteria of the intensity change of 1093 cm −1 and the ratio change of 416 cm −1 and 436 cm −1 can better identify the existence of Hg ions.

Test of Detection of Hg 2+ for Cell Analysis
Furthermore, the magnetic SERS sensor was employed for cell analysis. The macrophages were incubated with a magnetic nanoprobe. After the probe was endocytosed by the macrophages, the macrophages were treated with mercury ions containing PBS for 1h, and then the Raman mapping was performed. The corresponding images were shown in Figure 10. The results show that the intensity ratio of 416 cm −1 to 436 cm −1 was significantly decreased and the intensity of the 1093 cm −1 was enhanced with the addition of Hg ions. Due to the complex composition of the cell, the signal of the SERS material in the cell will inevitably be affected. The dual criteria of the intensity change of 1093 cm −1 and the ratio change of 416 cm −1 and 436 cm −1 can better identify the existence of Hg ions.

Test of Detection of Hg 2+ for Cell Analysis
Furthermore, the magnetic SERS sensor was employed for cell analysis. The macrophages were incubated with a magnetic nanoprobe. After the probe was endocytosed by the macrophages, the macrophages were treated with mercury ions containing PBS for 1h, and then the Raman mapping was performed. The corresponding images were shown in Figure 10. The results show that the intensity ratio of 416 cm −1 to 436 cm −1 was significantly decreased and the intensity of the 1093 cm −1 was enhanced with the addition of Hg ions. Due to the complex composition of the cell, the signal of the SERS material in the cell will inevitably be affected. The dual criteria of the intensity change of 1093 cm −1 and the ratio change of 416 cm −1 and 436 cm −1 can better identify the existence of Hg ions.

Conclusions
In summary, in this work, a new field-detectable SERS sensor was developed to detect Hg ions in water. The new nano sensor has excellent anti-interference and environmental adaptability and shows high sensitivity and selectivity to Hg ions. The Fe 3 O 4 core provides good magnetism, SiO 2 is used to protect Fe 3 O 4 from external interference, the dense gold shell provides a good SERS enhancement effect, and 4-MPy is used for surface modification. 4-MPy can combine with Hg 2+ to cause spectral changes. In addition, after EDTA treatment, the sensor can be recycled and reused. Importantly, the Raman image of the Hg 2+ iontreated macrophages confirms the MNPs sensors work effectively at the cell level. Based on these advantages, the nano sensor provides a quick and easy method for on-site detection of Hg ions.
Supplementary Materials: The following supporting information can be downloaded at: https: //www.mdpi.com/article/10.3390/chemosensors11060347/s1, Table S1: Comparison between this work and other reported results for detection of mercury ion; Figure S1 The Raman spectra of different concentrations of 4-MPy modified on magnetic microspheres. Figure