In Situ Direct Monitoring of the Morphological Transformation of Single Au Nanostars Induced by Iodide through Dual-Laser Dark-Field Microscopy: Unexpected Mechanism and Sensing Applications

Single nanoparticle imaging is a significant technique to help reveal the reaction mechanism and provides insight into the nanoparticle transformation. Here, we monitor the in situ morphological transformation of Au nanostars (GNSs) induced by iodide (I−) in real time using dark-field microscopy (DFM) with 638 nm red (R) and 534 nm green (G) laser coillumination. The two lasers are selected because the longitudinal localized surface plasmon resonance of GNSs is located at 638 nm and that for GNSs after transformation is at 534 nm. Interestingly, I− can interact with GNSs directly without the engagement of other reagents, and upon increasing I− concentrations, GNSs undergo color changes from red to orange, yellow, and green under DFM. Accordingly, green/red channel intensities (G/R ratios) are extracted by obtaining red and green channel intensities of single nanoparticles to weigh the morphological changes and quantify I−. A single nanoparticle sensor is constructed for I− detection with a detection limit of 6.9 nM. Finally, a novel mechanism is proposed to elucidate this shape transformation. I− absorbed onto the surface of GNSs binds with Au atoms to form AuI−, lowering the energy of its bond with other Au atoms, which facilitates the diffusion of this atom across the nanoparticle surface to low-energy sites at the concaves, thus deforming to spherical Au nanoparticles.


Introduction
Real-time imaging of single nanoparticles during their reactions and motions is vital to understanding their chemical [1][2][3][4], physical [5,6], and biological properties [7,8]. Although techniques such as atomic force microscopy (AFM) [9][10][11], liquid cell transmission electron microscopy (TEM) [12][13][14], and scanning electron microscopy (SEM) [15,16] are useful for monitoring single nanoparticles, they are expensive, exhibit low temporal resolution, and typically cannot be conducted under biophysical conditions. Thus, inadequate detailed kinetic information, inaccurate physical and chemical properties, and deactivation of biological function are problematic. Many techniques enabling the monitoring of realtime and in situ dynamic behaviors of single particles based on their optical properties have become important tools for understanding the physical and chemical properties and bifunctionality of nanoparticles [17][18][19]. Dark-field microscopy (DFM) is innately suitable for monitoring single plasmonic nanoparticles with a high signal-to-noise ratio and high spatial and temporal resolution. It is extensively applied in imaging [20,21], sensing [22][23][24], and drug and gene carriers [25]. It demonstrates promising prospects for understanding the details and mechanisms of certain chemical phenomena [26][27][28]. For example, DFM

Apparatus
The UV-VIS absorption spectra of the GNSs were obtained using a UV-2700 spectrometer (Shimadzu, Kyoto, Japan). The morphologies of the GNSs were observed using a Titan G2 60-300 microscope (FEI, Thermo Fisher Scientific, Hillsboro, OR, USA). XPS was conducted on a K-alpha multichine surface analyzer from Thermo Scientific (Waltham, MA, USA) with Al Kα radiation as the X-ray source and a pass energy of 100 eV. Zeta potentials were measured using a zeta potential analyzer (Nano-ZS90, Malvern, UK). For DFM imaging, a Nikon Ni-U upright microscope equipped with a 100 W tungsten halogen lamp, an oil immersion dark-field condenser (numerical aperture (NA) = 1.20-1.43), and a 40× Plan Fluor objective lens was used. A DP73 single-chip true-color charge-coupled device (CCD) camera (Olympus, Japan) was mounted on the top of the microscope to capture images. Red (638 nm, 100 mW) and green lasers (534 nm, 100 mW) from Laserland (Wuhan, China) were used to replace the 100 W tungsten halogen lamp for illumination to improve the image quality of the individual GNSs.

Preparation of GNSs
The GNSs were synthesized using a seedless approach according to a previous study [53]. Briefly, 40 mL of the HEPES buffer (75 mmol/L, pH 7.4) was mixed with 59.18 mL of deionized water in a volumetric flask, and then 823 µL of HAuCl 4 solution (24.28 mmol/L) was added. After shaking for 20-30 s, the solution was left undisturbed in a water bath at 25 • C for 30 min. The solution color slightly changed from light yellow to purple and finally to greenish blue, indicating the successful formation of GNSs. To improve the stability of the GNSs and prolong their storage time, the pH of the solution was adjusted to 9.6 ± 0.1 with a 1 mol/L NaOH solution.

Shape Transition of GNSs Monitored under UV-VIS Spectroscopy
The I − -induced morphological variation of GNSs was investigated by UV-VIS spectroscopy. Briefly, aliquots (1 mL) of the GNS solution were added into a 2 mL centrifuge tube, and 1 mL each of the I − solutions at varying concentrations (0, 0.3, 0.5, 1, 1.3, 1.5, 1.6, 1.7, 1.8, 1.9, and 2 µmol/L) was added. After incubation at room temperature for 90 min, the UV-VIS spectra were recorded. Moreover, the time-dependent UV-VIS spectra were also recorded at an interval of 5 min. To confirm whether other anions can have a similar effect, 2 µmol/L of Cl − , S 2 O 3 2− , OH − , SO 4 2− , S 2− , CH 3 COO − , CO 3 2− , NO 3− , and AA solutions was added to each GNS solution separately, followed by the recording of their UV-VIS spectra after a 90 min incubation.

Real-Time Monitoring of the Morphological Transformation of Single GNSs Induced by I − under DFM
To anchor GNSs onto the surface, coverslips were cleaned and treated by salinization via soaking in a piranha solution (H 2 SO 4 :H 2 O 2 = 3:1 in volume) for 1 h. Then, they were extracted and washed with ultrapure water. After drying in a stream of N 2 , the coverslips were placed on a steel shelf and maintained at 105 • C for 2 h to thoroughly remove the water molecules on their surfaces. Subsequently, the coverslips were immersed in ethanol containing 2.5% APTES and left undisturbed for 1 h. Next, they were rinsed twice with ethanol and immersed in ultrapure water for further use.
To monitor the transformation process of single GNSs by DFM, GNSs (10 µL) were cast onto an APTES-modified coverslip surface. After 3 min, the coverslips were rinsed with ultrapure water and reversely placed on a glass slide for further DFM imaging. The images of these GNSs were captured under exposure times of 200, 400, 600, and 800 ms through traditional DFM with a 100 W tungsten lamp. GNSs with 0.5 µmol/L of I − were similarly recorded after incubation for 90 min. To improve image quality, green and red lasers at 534 and 638 nm, respectively, were integrated. After passing through a beam expansion, they were applied as light sources for illumination. The GNSs without and with 0.5 µmol/L of I − were imaged at an exposure time of 50 ms by DFM. The obtained color images of the GNSs were split into red, green, and blue channels using Image J (1.8.0, National Institutes of Health). The G and R intensities of the single GNSs were extracted, and G/R ratios were calculated to assess the morphological changes of the GNSs during their interactions with I − .

Sensitive Detection of I − by DFM Imaging of Single GNS Nanoparticles
To investigate the sensitivity, the coverslips loaded with GNSs were soaked separately in a series of KI solutions with various concentrations (e.g., final concentrations of 0, 20, 50, 70, 100, 200, 300, 400, and 500 nM). The GNSs immobilized on the coverslip act as an individual signal-response output sensor, which is not affected by the average value in the homogeneous system and has higher sensitivity. Subsequently, the coverslips were reversely placed on a glass side, and the color images of single nanoparticles were captured through dual-laser DFM. The color DFM images were split into RGB channels using Image J software. The G and R intensities of single GNSs were extracted, and the G/R ratios were calculated to assess I − content.

Selectivity
To investigate the selectivity of our assay toward I − , the above procedure was repeated by replacing with other anions, including CO 3 2− , S 2− , OH − , S 2 O 3 2− , Cl − , SO 4 2− , NO 3 − , CH 3 COO − , and AA, followed by capturing their DFM images. All the experiments were repeated three times.

I − Detection in Real Sample
To evaluate the application of our assays, we tested iodine contents in table salts and seaweeds that were bought from a supermarket. In these samples, iodine usually exists in the form of organic iodine, IO 3 − , and so on, rather than I − . IO 3 − is the main species in table salts; we needed to transform iodine to iodide. Briefly, 1 g of iodized salt was dissolved in 99 mL distilled water, and 1.0 mL of 20 mM AA was added. The mixture solution was heated at 50 • C for 20 min, ensuring that the IO 3 − was reduced to I − [54]. The treatment of seaweeds was according to GB 5009.267-2016 with some modifications. Briefly, 2-3 g of samples was put into a crucible, followed by adding 5 mL of 0.47 M Na 2 CO 3 solution. Then, the mixture was heated until there was smoke and was placed in a muffle furnace at 600 • C for 4 h and then taken out after the temperature dropped to 200 • C. Subsequently, the ashed sample was transferred to a 100 mL volumetric flask by adding 5 mL water and washing three times. Finally, 1.0 mL of 20 mM AA was added, and the solution was kept at 60 • C for 15 min. If the detected iodide content exceeded the detection range, the sample should be diluted by adding distilled water and redetected. For detection, 750 µL of sample solution was mixed with 750 µL GNS solution (pH 6.8) for 90 min. An amount of 10 µL of solution was added dropwise to the surface of the APTES-modified coverslip and observed under DFM. All analysis results were repeated three times.

Results and Discussion
3.1. Transformation of GNSs after the Introduction of I − I − can induce the morphological deformation of GNSs, as shown in Figure 1 and as suggested in previous studies [43,50]. Two plasmonic absorption peaks of the GNSs at 520 and 638 nm were assigned to their transverse and longitudinal LSPR bands, respectively ( Figure 1a) [53,55,56]. The longitudinal LSPR is closely related to the number, length, and size of the branch [57]. After incubation with I − (2 µM) for 90 min, the GNS solution exhibited one absorption peak at 534 nm (Figure 1b), revealing that the GNSs were transformed into spherical Au nanoparticles. Correspondingly, the solution changed from greenish blue to red. The TEM image (Figure 1c) demonstrated that the as-prepared GNSs had an average diameter of 39.4 ± 1.5 nm with 2-5 branches. After incubation with I − , spherical Au nanoparticles with an average diameter of 23.4 ± 0.8 nm were observed ( Figure 1d). The time-dependent UV-VIS spectra (Supporting Information Figure S1) of the GNS solution showed that during the transformation, the longitudinal LSPR experienced gradual blue shifts. These results confirmed the shape transition of the GNSs induced by I − ; however, they did not provide sufficient information to support whether or not the shape changed through the aggregation/fusion/fragmentation processes, as suggested in previous studies [43,50].
the ashed sample was transferred to a 100 mL volumetric flask by adding 5 mL water and washing three times. Finally, 1.0 mL of 20 mM AA was added, and the solution was kept at 60 °C for 15 min. If the detected iodide content exceeded the detection range, the sample should be diluted by adding distilled water and redetected. For detection, 750 µL of sample solution was mixed with 750 µL GNS solution (pH 6.8) for 90 min. An amount of 10 µL of solution was added dropwise to the surface of the APTES-modified coverslip and observed under DFM. All analysis results were repeated three times.

Transformation of GNSs after the Introduction of I −
I − can induce the morphological deformation of GNSs, as shown in Figure 1 and as suggested in previous studies [43,50]. Two plasmonic absorption peaks of the GNSs at 520 and 638 nm were assigned to their transverse and longitudinal LSPR bands, respectively ( Figure 1a) [53,55,56]. The longitudinal LSPR is closely related to the number, length, and size of the branch [57]. After incubation with I − (2 µM) for 90 min, the GNS solution exhibited one absorption peak at 534 nm (Figure 1b), revealing that the GNSs were transformed into spherical Au nanoparticles. Correspondingly, the solution changed from greenish blue to red. The TEM image (Figure 1c) demonstrated that the as-prepared GNSs had an average diameter of 39.4 ± 1.5 nm with 2-5 branches. After incubation with I − , spherical Au nanoparticles with an average diameter of 23.4 ± 0.8 nm were observed ( Figure 1d). The time-dependent UV-VIS spectra (Supporting Information Figure S1) of the GNS solution showed that during the transformation, the longitudinal LSPR experienced gradual blue shifts. These results confirmed the shape transition of the GNSs induced by I − ; however, they did not provide sufficient information to support whether or not the shape changed through the aggregation/fusion/fragmentation processes, as suggested in previous studies [43,50].

Enhanced Image Quality of Single GNSs through DFM with G and R Lasers
Using the conventional DFM with a tungsten lamp (100 W), single GNSs in the absence and presence of 2 µM I − were imaged at various exposure times (200-800 ms) (Figure 2a). It was practically impossible to observe the GNSs under an exposure time of 200 ms. At 400 ms, GNSs exhibited dim red spots. With an increase in exposure time, brightness was greater. However, when the exposure time was raised to 800 ms, multiple spots changed to orange, which should be red according to their longitudinal LSPR. However, after the shape transformation to spheres with the presence of I − , dim green spots were visualized with an exposure time at 400 ms. Nanoparticles exhibited green spots at 600 ms, and more yellow-green spots were observed at 800 ms. Based on spectral change in the LSPR from 638 to 534 nm, the color of the nanoparticles should have changed from red to green. The unusual color variations were attributable to two reasons. One is that the wavelength of the tungsten light source covered a range of 360-760 nm, and the light collection efficiency of the CCD was wavelength dependent [58]. The other one is that GNSs had a small size distribution (Figure 1c). Long exposure time changed the saturation, inevitably causing a color change. Furthermore, observing small GNSs at a short exposure time was challenging. Furthermore, the color of nanoparticles recorded under DFM with conventional illumination can hardly reflect the morphological changes. In addition, the long exposure time (>400 ms) required for conventional DFM limited the temporal resolution of GNSs.
changed from red to green. The unusual color variations were attributable to two reasons. One is that the wavelength of the tungsten light source covered a range of 360-760 nm, and the light collection efficiency of the CCD was wavelength dependent [58]. The other one is that GNSs had a small size distribution (Figure 1c). Long exposure time changed the saturation, inevitably causing a color change. Furthermore, observing small GNSs at a short exposure time was challenging. Furthermore, the color of nanoparticles recorded under DFM with conventional illumination can hardly reflect the morphological changes. In addition, the long exposure time (>400 ms) required for conventional DFM limited the temporal resolution of GNSs.
To enhance image quality, R and G lasers with wavelengths of 638 and 534 nm were integrated to replace the 100 W tungsten lamp to monitor the deformation of the GNSs induced by I − (Figure 2b). The lasers were selected because of the LSPR variation during the transformation. At a short exposure time of 50 ms, the color of all GNSs was bright red, and after the transformation, it was green (Figure 2c). The improved imaging quality was attributed to the merits of lasers, including their strong light intensity and monochromatic properties [58]. Additionally, a short exposure time was beneficial for better temporal resolution. Moreover, because only G and R lasers were employed, the scattering spots were in the clear-cut color gamut comprising red and green. Because of the wavelength of light scattering, the color variations are highly correlated with morphological change. Therefore, the R and G laser illuminations are effective for observing small nanoparticles in the transformation process. To enhance image quality, R and G lasers with wavelengths of 638 and 534 nm were integrated to replace the 100 W tungsten lamp to monitor the deformation of the GNSs induced by I − (Figure 2b). The lasers were selected because of the LSPR variation during the transformation. At a short exposure time of 50 ms, the color of all GNSs was bright red, and after the transformation, it was green (Figure 2c). The improved imaging quality was attributed to the merits of lasers, including their strong light intensity and monochromatic properties [58]. Additionally, a short exposure time was beneficial for better temporal resolution. Moreover, because only G and R lasers were employed, the scattering spots were in the clear-cut color gamut comprising red and green. Because of the wavelength of light scattering, the color variations are highly correlated with morphological change. Therefore, the R and G laser illuminations are effective for observing small nanoparticles in the transformation process. Figure 3a shows that after adding I − (0.5 µM), the red GNS spots gradually changed to orange, yellow, and green, revealing that I − induced the dynamic morphological changes. After 30 min, 33.5% of the spots turned orange. When reaction time was prolonged to 90 min, practically all nanoparticles became green. The results showed inhomogeneous reaction rates among different GNSs, suggesting that reactions, like that between small molecules, follow kinetic models. In addition, a few particles appeared after incubation since the particles suspended in the channel were then adsorbed onto the slide. To quantitatively weigh the changes, we split the color images into RGB channels, and the G and R channel intensities of single GNSs were extracted [23,58]. Figure 3b shows images of the GNSs in the R and G channel, displaying evident changes in their intensities after transformation. Therefore, it is feasible to use G/R ratios to record the reaction kinetic. To check whether G/R ratios are identical to spectral change, we placed a transmission grating beam splitter (TGBS, 70 lines/mm) in the light collection path to acquire the scattering spectrum of the single GNSs [59]. Results showed that the changes in the G/R ratio obtained by splitting images with/without the TGBS were identical (Supporting Information Figure S2). More importantly, during the process, diverse behaviors were observed. Figure 3c,d shows four representative examples. For particle 1, the G/R ratio started to increase at 20 min, became fast within 30-60 min, turned slow during 60-75 min, and became fast again then. For particle 2, the G/R ratio generally increased within 0-100 min. Interestingly, the G/R ratio of particle 3 increased during 30-60 min, slightly declined within 60-75 min, and rose to the maximum, exhibiting variable reaction rates. Particle 4 exhibited a continuous ascending G/R ratio over 15-90 min. Fascinatingly, during transformation, the scattering intensities of particles 2-4 decreased, and that for particle 1 increased (Figure 3e). The diverse kinetics among various GNSs occurring among nanoparticles was probably due to the difference in the sizes and shapes as well as active sites and surface energies. Above all, the GNSs underwent shape transformations individually, without aggregation and fragmentation.

Sensitivity and Selectivity
The scattering images of single GNSs at increasing concentrations (0-500 nM) of I − were recorded (Figure 4a) and demonstrated concentration-dependent kinetics. With an increase in the I − concentration, single GNSs turned from red to orange, yellow, and green. The color change in the single GNSs was effortlessly distinguished under microscopy when the I − concentration exceeded 50 nM. Note that such a color change was due to the direct interaction between I − and GNRs without the engagement of other reagents, indicating an assay with high convenience and practicality. To quantify I − , G/R ratios were statistically calculated from the intensities of 150 single GNSs. A linear relationship was gained between the G/R ratio and I − concentration (Figure 4b). With I − concentration over the range of 0-500 nM, a linear equation was constructed: G/R = 0.054 + 0.0079 × C iodide with a correlation coefficient (R 2 ) of 0.993. Owing to the high anisotropy of the GNSs, the Au atoms at the edges exhibited high chemical activity. Thus, they interacted relatively readily with I − . Only with additional I − , the Au atoms with low chemical activities can interact with I − . It is reasonable that Au atoms on the nanoparticle surface are not in the complete lattice; thus, they have few neighbors and exhibit high chemical activity. The LOD was 6.9 nM (LOD = 3σ/slope, where σ is the standard deviation of five blank samples), which is more sensitive than the reported fluorescence and colorimetric methods (Table 1).

Sensitivity and Selectivity
The scattering images of single GNSs at increasing concentrations (0-500 nM) of I − were recorded (Figure 4a)  solubility product constant values for Au2S and Au2S2O3 were 1.6 × 10 −73 and 3.2 × 1 whereas that for AuI − was 1.6 × 10 −23 [62,63]. Although the adsorption of S 2− and S2O3 2 the Au surface was more substantial than I − , they could not trigger the sh transformation of the GNSs. We can infer that the morphological change was becaus not only the strong affinity to Au but also specific properties of the AuI − complexes a being adsorbed onto the surface.  Other anions, such as Cl − and S 2 O 3 2− , failed to induce the GNS shape transformation (Figure 4c,d), thus suggesting minimum interference. Dissimilar to I − , other anions, including Cl − , S 2 O 3 2− , OH − , SO 4 2− , S 2− , CH 3 COO − , CO 3 2− , NO 3 − , and AA, cannot induce GNS shape deformation, as also confirmed in the UV-VIS absorption spectra (Supporting Information Figure S3). S 2− and S 2 O 3 2− exhibited high affinity to Au [60,61], and the solubility product constant values for Au 2 S and Au 2 S 2 O 3 were 1.6 × 10 −73 and 3.2 × 10 −27 , whereas that for AuI − was 1.6 × 10 −23 [62,63]. Although the adsorption of S 2− and S 2 O 3 2− on the Au surface was more substantial than I − , they could not trigger the shape transformation of the GNSs. We can infer that the morphological change was because of not only the strong affinity to Au but also specific properties of the AuI − complexes after being adsorbed onto the surface.

Mechanism for I − -Inducing Transformation of GNSs
According to previous studies [41,43], the morphological transformation of Au nanoparticles induced by I − involved aggregation, fusion, and fragmentation through the Ostwald ripening process based on TEM images and UV-VIS spectra. However, in this study, we observed that the transformation from GNSs to spherical nanoparticles was a self-fusion process through adsorption, binding, and migration (Scheme 1). We propose that I − can be spontaneously absorbed on the surface, and the formation of AuI − enabled the migration of Au atoms from high-energy sites to low-energy sites. To support this hypothesis, we recorded the XPS spectra of GNSs in the presence and absence of 2.0 µM I − , as shown in Figure 5. A new peak at 618 eV was responsible for the surface I − (Figure 5a). The binding energy (BE) for Au 4f 7/2 of the GNSs located at 82.27 eV shifted to 82.77 eV after its interaction with I − , thus supporting their interaction without alternating the oxidation state of Au (Figure 5b) [69]. After being absorbed, the BE for I 3d 5/2 located at 618.13 eV indicated the existence of negative valence of I − (Figure 5c) [70]. The zeta potential of the GNSs decreased from −49.

Single GNS Sensor for the Quantification of I − Real Sample Detection
To promote the application of this probe and method in real sample d salt and seaweeds were used to monitor the I − content. For table salts, we iodine in salts to iodide. AA was also effective in reducing IO3 − to I − . The seaweeds and a complex vitamin tablet was according to GB 5009.267-20 modifications. Considering the high sensitivity of our assay, the seaweeds 1000 times using distilled water. For detection, 750 µL of sample solution w Scheme 1. The mechanism for the shape transformation of GNSs induced by iodide.
the replacement of surface molecule HEPES by I − [71]. High-resolution TEM images of the GNSs without and with 2 µM of I − were taken (Figure 5e,f). Lattice spacings of 0.235 and 0.205 nm were observed on the GNSs, representing typical (111) and (200) crystal facets, respectively. However, after being transformed to spherical Au nanoparticles, only (111) was observed. I − did not enter the interior of the Au nanoparticles. Moreover, the deformed spherical nanoparticles were particularly round without edges or corners.  To examine whether the transformation was not attributed to the electron injection process, we added extra NaBH 4 and starch to the reacting solution to examine whether I 2 was formed (Supporting Information Figure S4). Owing to the shape transformation, the GNS solution changed from greenish blue to red after adding I − . In the presence of 0.1 mM of NaBH 4 with a strong reducing ability to prevent the formation of iodine, the GNSs still underwent a shape change, revealing that iodine was not responsible for the shape transformation. I 2 can interact with starch to form blue complexes, but no change was observed on the UV-VIS curve of GNSs with the presence of extra starch, thus again confirming that iodine was not formed. Additionally, in the absence of light, 2 µM of I − still induced GNS transformation, ruling out the essential role of photo irradiation in the process.
Single nanoparticle imaging directly confirmed that aggregation, fusion, and fragmentation did not occur during the transformation of GNSs to spherical Au nanoparticles. To investigate whether other shaped nanoparticles experienced similar changes, Au nanorods, Au nanoprisms, and large GNSs (~100 nm, capped by citrate) were prepared as per previous studies with slight modifications (Supporting Information Figure S5) [72,73]. The results showed that 2 µM of I only induced the transformation of the large GNSs. Thus, the specific shape of GNSs with various corners and branches was critical for the shape deformation, for the existence of highly active Au atoms on the edges of the GNSs. The GNSs capped with citrate and HEPES underwent transformation, suggesting that the surface capping agents did not account for the shape transformation. The growth of large spherical Au nanoparticles was not observed in this study. Therefore, we ruled out the occurrence of Ostwald ripening [74,75]. Our results are consistent with a previous study in that the chemical adsorption of I − onto the GNSs played an important role in nanoparticle transformation [76]. However, it was not sufficient, and the AuI − formation took effect for shape transition because other ions such as S 2− with strong chemical adsorption could not induce this phenomenon. Thermodynamically, I − induced the shape transformation in a spontaneous process, which was primarily because spherical Au nanoparticles were more energetically favored than star-shaped particles. Moreover, the Au atoms on the nanoparticle surface were energetically less stable than those in the interior because all atoms inside bonded to 12 neighbors, while the atoms on the surface were bound to fewer. Therefore, the mechanism for the transformation was that I − adsorbed onto the surface of the GNSs was bound to an Au atom to form AuI − , lowering the energy of its bond with other Au atoms, which facilitated the diffusion of this atom across the nanoparticle surface to low-energy sites.

Single GNS Sensor for the Quantification of I − Real Sample Detection
To promote the application of this probe and method in real sample detection, table salt and seaweeds were used to monitor the I − content. For table salts, we transformed iodine in salts to iodide. AA was also effective in reducing IO 3 − to I − . The treatment of seaweeds and a complex vitamin tablet was according to GB 5009.267-2016 with some modifications. Considering the high sensitivity of our assay, the seaweeds were diluted 1000 times using distilled water. For detection, 750 µL of sample solution was mixed with 750 µL GNS solution (pH 6.8) for 90 min. An amount of 10 µL of the solution was dropped onto the surface of the APTES-modified coverslip and observed under DFM. The iodide contents in table salt, seaweeds, and complex vitamin tablets were precisely quantified, as shown in Table 2. The iodized salt contained 22.8 mg/kg of iodide, meeting the requirement of the standard. The iodide content in seaweeds was as high as 1620.4 mg/kg. In addition, the sample recoveries varied from 98.5% to 106.5%, and the relative standard deviation (RSD) was in the range of 3.2%-6.7%, which verified the reliability of the GNSs for real complex biological samples. Note: ND denotes not detected.

Conclusions
Here, DFM with R and G laser illumination afforded high-quality scattering images of single GNSs, thus providing detailed information during the shape transformation of GNSs induced by I − . With increasing I − concentration, the color of the single nanoparticles changed from red to orange, yellow, and green. The G/R ratios reflected the degree of morphological transformation, which was dependent on the I − concentration. A single nanoparticle sensor was developed for I − quantification with a LOD of 6.9 nM. Importantly, the I − contents in table salt, seaweeds, and a complex vitamin tablet were quantified with good accuracy. The processes of aggregation, fusion, and fragmentation did not occur, dissimilar to the results of previous reports. The I − -induced transformation of Au nanoparticles was investigated by XPS, high-resolution TEM, and zeta potential measurements. It confirmed that the valence of I − did not change during the transformation process, and Ostwald ripening was not responsible for shape transition. The strong absorption and binding of I − onto the Au surface were critical for shape transformation. However, the specific properties of AuI − played more critical roles, because the interaction of I − with the Au atom lowered the energy of its bond with other Au atoms, which facilitated the diffusion of this atom across the nanoparticle surface to low-energy sites.