Modulation and Study of Photoblinking Behavior in Dye Doped Silver-Silica Core-Shell Nanoparticles For Localization Super-Resolution Microscopy

Abstract

Blinking of fluorescent nanoparticles is a compelling phenomenon with widely debated mechanisms. The ability to inhibit or control blinking is important for applications in the field of optical, semiconductor and fluorescent imaging. Self-blinking nanomaterials are also attractive labels for localization-based super-resolution microscopy. In this work, we have synthesized silver core silica nanoparticles (Ag@SiO₂) doped with Rhodamine 110 and studied the parameters that affect blinking. We found that under nitrogen rich conditions the nanoparticles shifted towards higher duty cycles. Also, it was found that hydrated nanoparticles showed a less drastic response to nitrogen rich conditions as compared to dried nanoparticles, indicating that surrounding matrix played a role in the response of nanoparticles to molecular oxygen. Further, the blinking is not a multi-body phenomena, super-resolution localization combined with intensity histogram analysis confirmed that single particles are emitting.

Introduction

In fluorescent imaging, the blinking of nanoparticles is the reversible cycling of fluorescent emission between an ‘switch on’ and ‘switch off’ state. During the past decade, numerous studies on blinking of nanostructures have been reported^1–4. A detailed study of blinking can make nanostructures useful as fluorescent probes for fluorescent imaging and for super-resolution imaging such as blink microscopy.^5–6

Blinking is typically viewed as a hindrance to fluorescent imaging and optical biosensors where continuous brightness is desired. In such cases, receding of the blinking is needed. While in super-resolution imaging based on localization, where photo-switchability is required, controlled blinking can be beneficial^7–10. The time scale of blinking can range from several microseconds to seconds, exceeding the excited state lifetime which typically lies between 1–10 nanoseconds.

The cause of blinking phenomena is widely debated and still an active area of study. For semiconductor quantum dots and metal nanoparticles, blinking is attributed to plasmonic interactions^11–12; while in fluorescent probes, one of the widely studied mechanisms considers the triplet state to be the prevalent reason behind blinking^13–15. This mechanism considers that the dark state is caused either due to tunneling of charges from an excited singlet state to ‘charge traps’ in the nearby areas or by intercrossing of charges from triplet states to singlet states¹⁶. Yet another mechanism, this one based on charge transfer^17–19, hypothesizes that the ‘off’ or dark state depends on the polarity of the surrounding matrix¹⁷. Other mechanisms based on conformational changes of single molecules^20–22 have also been reported. It has been shown that the proximity of a fluorophore to metal can help suppress blinking by changing the emission and recombination rates^23–25.We report here a blinking phenomenon observed in fluorescent dye doped silver core silica shell (Ag@SiO₂) nanoparticles. The metal core of the silica nanoparticles improves the fluorescence emission efficiency due to Metal Enhanced Fluorescence (MEF)^26–28 and are promising systems to increase dye sensitivity. The dye labeled core-shell nanoparticles are inert to cellular environment, non-toxic towards target biological cells, and subsequent conjugation of affinity ligands does not hamper the nanoparticle’s optical properties^29–30. We demonstrate that the blinking observed in these nanoparticles can be controlled by changing the molecular oxygen content as well as the surrounding medium of the nanoparticles. Blink microscopy^{5–6, 31} localization of the nanoparticles showed individual emitting particles, which was confirmed by intensity histograms. Such a comprehensive study of these core-shell nanoparticles can open new doors for them to be used as promising fluorescent probes.

Results and Discussion

Single Particle Observation and Localization

Clumping of nanoparticles has been an issue for long-term storage of the nanoparticles, and there was some concern that the nanoparticles would aggregate during analysis. This aggregation could cause errors when analyzing blinking behavior. In order to confirm that we were in fact imaging single particles we used the ThunderSTORM³² ImageJ plugin, for super-resolution localization of the blinking nanoparticles. Figure 1A–B shows two blinking nanoparticles before and after ThunderSTORM analysis showing that both are single nanoparticles rather than clusters. The diameter of a representative nanoparticle was measured before and after ThunderSTORM analysis, shown in Figure 1(C) and 1(D). The size of the nanoparticle measured after the ThunderSTORM processing approaches the size determined by TEM and DLS.

Figure 1:

An image of representative nanoparticles before STORM localization and a super-resolution image of the same nanoparticles after STORM analysis, confirming that they are single nanoparticles rather than clusters, shown as bright white spots on ab black background. Below are graphs showing the sizes of three nanoparticles before (right) and after (left) ThunderSTORM analysis. The FWHM of the peaks on the before graph are not possible to determine. The FWHM of the first peak is 0.35 μm.

Intensity histogram analysis was also employed to ensure the observed objects were individual particles. The distribution of intensity values around the mean intensity is small enough that multiple occupancies could be resolved. In this work, multiple occupancies were not observed.

Spectroscopic Characterization

The dye doped, core-shell nanoparticles were synthesized by a two-step process as shown in Figure 3. Silver nanoparticles were synthesized by a citrate-reduction method followed by the growth of silica shell doped with Rhodamine 110 using a modified Stöber synthesis method^33–34. Rhodamine 110 was chosen due to our interest in synthesizing bright nanoparticles that fluoresced under green light excitation. Silica provides a medium for maintaining optimal distance between the fluorophores and the metal core for MEF³⁵. In this synthesis technique, the dye molecules are encapsulated non-covalently in the silica matrix. It should be noted that the stability of the nanoparticles is compromised for long term use compared to covalently bonded dye molecules, however this method ensures simplicity of synthesis and non-toxicity of the nanoparticle^36–37. Core-shell nanoparticles lacking any dye were also synthesized to verify that the dye was not the cause of blinking. Nanoparticles without dye were observed to blink, although the addition of dye allows for tuning of the emission intensity and wavelength. A video of dye-free, blinking nanoparticles is available in Supporting Information.

Figure 3:

Two step synthesis method for Rhodamine 110 dope Ag@SiO₂ nanoparticles. Silver colloids were formed by reduction of silver ions followed by a sol-gel silica surface incorporated with Rhodamine 110 dye.

The fluorescent Ag@SiO₂ nanoparticles were found to be uniformly distributed and well dispersed when imaged with an inverted microscope (Figure 4(a)). Transmission Electron Microscopy (TEM) showed that the nanoparticles had a symmetrical core shell structure with a core size of ~85 nm (Figure 4(b)). The nanoparticle size distribution was uniform with a diameter of ~100±20 nm confirmed by both Dynamic Light Scattering (DLS) (Figure 4(c)) and TEM. The Fluorescence measurements (Figure 4(d)) when excited at 475 nm source, show an emission maximum at ~522 nm which validates the presence of Rhodamine 110. An absorbance peak at ~570 nm was observed in the UV Vis measurements, which were a possible result of convolutions from silica shell and the silver core.³⁸

Figure 4:

(a) A microscope white-light image of the Ag@SiO₂ nanoparticles (b) TEM image of representative Ag@SiO₂ nanoparticles. The silver nanoparticles core is clearly observed, as well the sol-gel shell. The image shows two nanoparticles clumped together; this is because of the time elapsed between completion of the synthesis and the imaging (approximately 12–14 hours), the more time passes the more they clump. (c) DLS measurement of the size distribution of the nanoparticles (d) FL measurements of the nanoparticles (e) UV Vis absorption spectra of the nanoparticles

A blinking behavior was observed (supported information) in the synthesized core-shell nanoparticles which has not been reported previously^29–30. Figure 5(a–d) are frames captured from a time-lapse video of nanoparticles over a span of 20s. The video of blinking nanoparticles can be found in the supplemental materials. As seen from the figures, some of the nanoparticles appear in frames and then disappear or diminish in brightness over the subsequent frames. A z-axis profile analysis is shown in Figure 5 (e, f) for further support of the blinking behavior. This blinking rate and duty cycle were not uniform over the nanoparticle distribution. A video of blinking nanoparticles is available as supporting information. To better understand the detailed behavior of this phenomenon, various experimental parameters such as oxygenation and hydration were altered and the effects have been discussed in the following sections.

Figure 5:

(a-d) A time lapse of a representative sample of blinking nanoparticles. The rate and duty cycle are not uniform across the sample. Two such nanoparticle examples have been highlighted within white square and circular boarders. (e) z-axis profile of the nanoparticle highlighted by the white square. (f) z-axis profile of the nanoparticle highlighted by the white circle. The video was taken at a rate of10 fps (supporting information).

Excitation and Nitrogen Effects on Blinking

The nanoparticle blinking behavior was studied under variation of excitation source power, dye concentration, N₂ content, and O₂ content. The power of the excitation source was altered by changing the current (ranging 500 mA-6000 mA). Duty cycle was calculated for three nanoparticles per image frame and averaging the data. Figure 7(a) shows the plot for duty cycle versus power plots for both dried and hydrated nanoparticles with a dye concentration of 1 mM. Figure 7(b) is a plot demonstrating the duty cycle calculations for dried and hydrated nanoparticles under different dye concentrations. Fig 7(c) shows such a plot under increasing O₂ content conditions. The data with no O₂ content was obtained by exposing the sample chamber to N₂ for 20 min, the 20% O₂ content was the data collected under ambient laboratory conditions. As suggested by the plots, no clear underlying patterns were observed by changing the three parameters. Also, data from dry and hydrated nanoparticles were not significantly different from each other. This was statistically verified by T-tests performed on the available data which have been included in the Supplemental Information.

Figure 7:

Distribution of duty cycle with varying (a) power of excitation source (b) concentration of dye solution, and (c) O₂ content in the nanoparticles for both dried and hydrated nanoparticles. T-tests were used to determine the statistical significance of the data. The dependence studies showed that blinking was independent of these three parameters for both dried and hydrated nanoparticles.

Figure 7(b) suggests that quenching of dye molecules due to increase in dye concentration is not a significant contributor in the blinking mechanism. The results from Figure 6(c), (d) had shown a clear trend of increase in the duty cycle with increasing N₂ exposure time which is contrary to the plot in Figure 7 (c) where no variation observed with increase in O₂ content for both dried and hydrated NPs. This could be due to a possible linkage of blinking to nitrogen defects caused at the surface of the nanoparticles. This is a preliminary conclusion drawn based on the performed experimental results. Nitrogen defect induced blinking behavior has been reported previously in nanodiamonds³⁹ but to our knowledge no such literature findings are available for metal core nanoparticles. Further investigations are required to fully reinforce these findings and to obtain a better understanding of the nitrogen depletion based blinking mechanism.

Figure 6:

(a) Behavior of Ag@SiO₂ /wR110 under N2 atmosphere, (b) Histograms for duty cycle of dried nanoparticles and hydrated nanoparticles on glass slide respectively (c), (d) Pie chart showing changes in duty cycle during the first 5 minutes under N₂ atmosphere for dried and hydrated nanoparticles, respectively (e) Plot showing dependence of 0–20%, 60–80%, and 80–100% duty cycle on N₂ exposure time with their corresponding fits for dried nanoparticles (f) Plot showing dependence of 80–100% duty cycle on N₂ exposure time with their corresponding fits for hydrated nanoparticles.

As it has been reported that oxygen affects fluorescence blinking,^{17, 40–42} we studied the blinking behavior by observing the fluorescent Ag@SiO₂ nanoparticles under N₂ flow. 100 μL colloidal solution of nanoparticles was deposited on a microscope glass slide and dried in an oven at 50°C. Two samples were studied under N₂ atmosphere, one was hydrated with a drop of DI water on top of the dried particles, the other was left unhydrated. The sample was placed in an atmospheric chamber and the nanoparticles were exposed to nitrogen. Images were captured using a 60x microscope objective at an interval of 5 minutes for the next 20 minutes. Before exposure time N₂ images were captured as well, referred to as t=t₀ (t₀ obtained under 20% O₂). Previous studies in our lab show the system equilibrates within 5 minutes.⁴³ ImageJ software was used for analyzing the data and analysis of each time interval was performed using 100 nanoparticles.

As seen in Figure 6(a) and 6(b), histograms of the duty cycle for the dried and hydrated nanoparticles show contrasting behavior under N₂ atmosphere. Dried nanoparticles show an increase in duty cycle within the first 5 minutes under N₂ conditions as seen in Figure 6(c). This is not the case for the hydrated nanoparticles, where there is no significant change (Figure 6(d)). To compare and visualize any underlying trend in duty cycles for both the cases, the bin counts from the histogram were counted between 0–100% and plotted against N₂ exposure time. For the dried nanoparticles (Figure 6(e)), it was found that the duty cycles between 0–20% showed an exponential decay with increasing N₂ exposure time. While duty cycles between 60–80% and 80–100% showed a positive linear and power law dependence increase respectively. The population of blinking particles also increases. For the hydrated nanoparticles, similar calculations did not show any underlying trend in duty cycles except between 80–100% (Figure 6(f)). For the hydrated nanoparticles the duty cycle in the 80–100% region show a logistic behavior demonstrating that between 5–15 minutes there is a linear increase in the number of nanoparticles that stay ‘on’ followed by a slow increase in this number with further N₂ exposure time. Therefore, for the dried nanoparticles, under nitrogen rich conditions there was a 50% reduction in the 0–20% ranged duty cycle and more particles were distributed towards higher duty cycles as compared to hydrated ones. This striking difference in the response of dried and hydrated nanoparticles under N₂ environment indicates the possible linkage of blinking to surrounding matrix¹⁷.

Experiments were also carried out using the bare cores, the whole nanoparticle without the dye, and a SiO₂ nanobubble containing dye; blinking behavior of each was observed under both blue and green light. The related videos can be found in the Supplemental Materials. Blinking was observed in both the bare cores and the nanoparticle without the dye; the SiO₂ nanobubbles occasionally turned on or off, however this was extremely infrequent. We will explore this phenomenon in future work. The duty cycles under green light excitation when dry of the bare cores, the nanoparticles without dye, and the SiO₂ nanobubbles was 17±19%, 10±8%, and 55±27%, respectively. This set of control experiments was conducted to show the fluorescence and photoblinking behavior of the nanoparticle components. The related videos are available in the Supplementary Materials.

Conclusion

In summary, we reported a blinking phenomenon observed in synthesized Rhodamine110 doped Ag@SiO₂ nanoparticles and a detailed study of the experimental parameters affecting it. Blinking was found to be independent of excitation source power and dye concentration. Higher duty cycles were favored under nitrogen exposure conditions while oxygen exposure didn’t show any considerable impact on the duty cycles; this effect was more pronounced for dried nanoparticles as compared to hydrated ones. This indicated a possible correlation of blinking to nitrogen defects on the nanoparticle’s surface, though complete control of the core-shell nanoparticles ‘on’ and ‘off’ could not be demonstrated but longer duration of the ‘on’ state could be achieved. These findings can be applied to imaging techniques where frequent blinking and toxicity are an issue. Further understanding of this nitrogen defect induced blinking will lead to the development of MEF Ag@SiO₂ nanoparticles with controlled photo-switching. The control tests all show some blinking behavior, though not to the degree of the complete nanoparticle, which indicates that both the Rhodamine 110 and the silver nanoparticle core contribute to the blinking behavior of the completed nanoparticle, suggesting that modification of the nanoparticle structure can be used to control the blinking. The measurement of single nanoparticles was verified using the STORM localization, indicating these particles can potentially be used for super-resolution imaging as well as eliminating the potential of nanoparticle cross-talk originating the blinking phenomenon.

Methods

Materials:

Silver Nitrate (>99.9%) was purchased from Alfa Aesar, Tetraethoxysilane (TEOS, 98%) was purchased from Acros Organics. Ammonium Hydroxide (28–30%) was purchased from Thermofisher Scientific. Rhodamine 110 was purchased from Sigma-Aldrich and Ethyl Alcohol (190–200% proof) was supplied by Greenfield Global and purchased from the TTU Chemistry Stockroom. The sodium chloride was purchased from Sigma-Aldrich.

Synthesis of Fluorescent Ag@SiO₂ Nanoparticles:

For preparation of Ag nanoparticles, 9 mg of Silver nitrate was added to 49 mL of DI water in a 250 mL Erlenmeyer flask with a glass stopper. The solution was heated under magnetic stirring, until it started to boil. At that point, a solution of 10 mg Trisodium citrate dissolved in 1 mL DI water was added drop by drop into the boiling solution. The mixture was kept boiling and stirring for 30 minutes, making sure the solution did not evaporate, until the color changed to greenish brown. The solution was then cooled to room temperature and centrifuged at 500 rpm for 1 hour to remove any large particles.

The top solution was then carefully aspirated, dispersed in 200 mL ethanol, and sonicated for 30 minutes to disperse the Ag nanoparticles well. To adjust the pH to ~ 9, 4 mL of ammonium hydroxide (28–30%) was added to the Ag colloid solution. Next, fluorescent molecules embedded in the sol-gel coating as SiO₂ was formed on top of the Ag nanoparticles. For the sol-gel layer formation, the Ag colloid solution was set to stirring, and 22.5 μL of TEOS (Tetraethyl orthosilicate) in 10 mL ethanol was added drop by drop to the solution at a rate of 2.5 mL/hr. Simultaneously, 1mM Rhodamine 110 solution in ethanol was also added to Ag colloidal solution. The solution was then incubated overnight at room temperature for the formation of the fluorescent sol-gel coating. Finally, the next day the nanoparticle solution was centrifuged four times at 3500 rpm (2800 g-force) for 30 min, and the final solution was suspended in water for future experiments.

Synthesis of Controls:

The bare nanoparticle cores were synthesized using the same silver nitrate and trisodium citrate solutions discussed above. The nanoparticles without dye were synthesized using the same methods discussed above without the addition of the Rhodamine 110. The nanobubbles were synthesized by taking 500 μL of the completed nanoparticle solution and adding 48 mg of sodium chloride and stirring with a magnetic stir bar overnight. A TEM image of these nanobubbles is shown in Figure 8.

Figure 8:

TEM image of the nanobubble control. Due to aggregation of the nanobubbles, a single nanobubble has been highlighted (white circle). TEM shows complete removal of the nanoparticle core and distortion of the spherical shape of the complete nanoparticle due to loss of the core.

Characterization Methods:

The nanoparticles were imaged using a Nikon TI Inverted Microscope with a 100x oil objective of NA 1.49 or an Olympus IX71 Inverted Micrsocope with a 60x, 0.67 dry objective. An epifluorescent light source with a wavelength of 508 nm was used to excite the nanoparticles, for which a reddish orange fluorescence was observed. The camera used for imaging was a Zyla 5.5 with a pixel size of 6.5 μm × 6.5 μm or a Thorlabs Quantulux sCMOS with a pixel size of 5×5 μm. For fluorescent imaging experiments the colloidal suspension of nanoparticles in water were dried on a glass coverslip at 50°C.

The nanoparticle size distribution measurements were done by Dynamic Light Scattering (DLS) with Zetratrac Particle Size Analyzer (Malvern Inc.) at the TTU Material Characterization center. For UV-Vis absorption measurements an Agilent Cary 8453 Spectrophotometer was used and the wavelengths from 250–700 nm were measured. A Shimadzu RF-5301PC Spectrofluorimeter was used for the Fluorescence spectra measurements. For the experiments an excitation wavelength of 475 nm was used, and the emission was measured in the range of 500 – 600 nm.

TEM and Duty Cycle

Transmission electron microscopy (TEM) using a Hitachi H8100 TEM microscope was used to verify the DLS measurements of the size of the nanoparticles as well as to determine the thickness of the silica shell. TEM was also used to verify removal of the nanoparticle core after synthesis of the nanobubbles. The sample was dried onto a small disk used to load the sample into the electron microscope and images of the nanoparticles were analyzed by the TEM specialist to determine the size of the whole nanoparticle and the thickness of the silica shell.

The noise threshold of the z-axis graphs was determined using the following equation.

$$\mathit{\text{threshold}}=\stackrel{-}{x}+3\sigma$$

Each data point consisted of an image sequence (frames acquired versus time). The average value and standard deviation were found for an area of the z-axis plot of the image stack where the nanoparticles were in the “off” state. Three standard deviations were added to the average and this was considered the threshold for the background noise; any value above this threshold was considered to be the nanoparticle in the “on” state. The duty cycle was calculated as follows.

$$DC\mathrm{\%}=\frac{on}{\mathit{\text{total}}}\times 100$$

Localization Microscopy of Single Nanoparticles:

The image stack was analyzed using the ThunderSTORM ImageJ plugin,³² which took the image stack and created a single composite image showing the sub-diffraction localization of the nanoparticles, a portion of which is shown in Figure 1. Since the particles self-blink between on and off states, only a single excitation source is required.^44–45 The use of a single excitation source simplifies experimentation, as multiple light sources, modulators, and shutters are not needed.

Figure 2:

Intensity histogram of all the nanoparticles in a representative image sequence. The distribution is not bimodal, which would be expected if clusters and single particles were both being imaged, indicating that only single particles were imaged. The intensity value distribution around the mean of 160 counts is small enough that multiple occupancies—if present—would be resolved.