Simultaneous Determination of Size and Quantification of Gold Nanoparticles by Direct Coupling Thin layer Chromatography with Catalyzed Luminol Chemiluminescence

The increasing use of metal-based nanoparticle products has raised concerns in particular for the aquatic environment and thus the quantification of such nanomaterials released from products should be determined to assess their environmental risks. In this study, a simple, rapid and sensitive method for the determination of size and mass concentration of gold nanoparticles (AuNPs) in aqueous suspension was established by direct coupling of thin layer chromatography (TLC) with catalyzed luminol-H2O2 chemiluminescence (CL) detection. For this purpose, a moving stage was constructed to scan the chemiluminescence signal from TLC separated AuNPs. The proposed TLC-CL method allows the quantification of differently sized AuNPs (13 nm, 41 nm and 100 nm) contained in a mixture. Various experimental parameters affecting the characterization of AuNPs, such as the concentration of H2O2, the concentration and pH of the luminol solution, and the size of the spectrometer aperture were investigated. Under optimal conditions, the detection limits for AuNP size fractions of 13 nm, 41 nm and 100 nm were 38.4 μg L−1, 35.9 μg L−1 and 39.6 μg L−1, with repeatabilities (RSD, n = 7) of 7.3%, 6.9% and 8.1% respectively for 10 mg L−1 samples. The proposed method was successfully applied to the characterization of AuNP size and concentration in aqueous test samples.

nanoparticles in aqueous media. By coupling TLC with laser ablation inductively coupled plasma mass spectrometry (LA-ICP-MS), quantitative characterization of differently sized gold nanoparticles is achievable. However, the use of expensive and complex LA-ICP-MS restricts its adoption, since LA-ICP-MS is not available within most of laboratories. Therefore, it is considered important to develop a detection method which is sensitive, cheap and convenient for the study of engineered nanomaterials in environmental samples 34 .
Chemiluminescence (CL) has been established as a valuable detection technique offering low detection limits, wide linear range, high analytical throughput and simple instrumentation 36,37 . Tsogas et al. recently reported a method for the ultratrace determination of silver, gold, and iron oxide nanoparticles involving chemiluminescence detection 34 . However, the metal nanoparticles required dissociation into their precursor metal ions prior to chemiluminescence detection, which is both tedious and time consuming 37 . In recent years, it is well reported that nanoparticles can participate in CL reactions as reductants, catalysts, and luminophors 38,39 . AuNPs as catalysts in CL reactions have received much attention [40][41][42][43] , and may catalyze the decomposition of hydrogen peroxide to produce reactive oxygen species and enhance the CL by the reaction between luminol (3-aminophthalhydrazide) and hydrogen peroxide 40,44 , leading to its wide application in bioanalysis 45,46 and immunoassay 47,48 . Although the catalytic ability of AuNPs for luminol-H 2 O 2 CL reaction has been widely recognized, the use of CL for the characterization of differently sized AuNPs has not been reported.
In this study, we present a simple analytical methodology for the determination of differently sized gold nanoparticles by direct coupling of thin layer chromatography with catalyzed luminol-H 2 O 2 chemiluminescence. Gold nanoparticles, which represent one of the most widely utilized nanoparticles, were separated by thin layer chromatography. The HPTLC plate containing the separated particles was sprayed with luminol and peroxide, the chemiluminescence signals then being monitored along the TLC track. In this manner, the separated differently sized AuNPs could be characterized and quantified within 45 s. Experimental values of concentration of H 2 O 2 , the concentration of luminol and size of the spectrometer aperture were adjusted to allow sensitive and reproducible detection of different sizes of AuNPs.  40 . The coupling of CL with TLC for the analysis of nanoparticles has not been reported, so the catalytic performance of gold nanoparticles on HPTLC plates on luminol-H 2 O 2 chemiluminescence was firstly investigated. AuNPs with 13, 41 and 100 nm diameters were deposited onto HPTLC plates. After air drying, the HPTLC plate was secured on a translation stage using double-sided adhesive tape and placed into a sampling chamber firstly, and then the chemiluminescence signals were monitored immediately after the HPTLC plate was sprayed with solutions of H 2 O 2 and luminol in turn using a TLC sprayer.

Result and Discussion
As shown in Figure 1A, it was observed that in the absence of gold nanoparticles, no obvious CL emission was observed for the luminol-H 2 O 2 system. However, in the presence of gold nanoparticles, intense CL signals were detected, which confirms that gold nanoparticle enhance CL signals from the HPTLC plate. Figure 1A also suggests that the catalytic effects vary substantially with particle size with 41 nm size particles producing the best CL enhancement, which agrees with previously reported results for particles of 38 nm 40 . It was also noted that the maximum emission for all the cases occurred at a wavelength of ~425 nm. These observations are in agreement with those obtained by Zhang et al. who investigated the effects of gold nanoparticles on luminol-H 2 O 2 chemiluminescent in the aqueous media 40 . It has been observed that the presence of surfactants above a critical micelle concentration enhances the CL intensity of luminol-H 2 O 2 50 , while the presence of EDTA operates to greatly reduce the luminescence of luminol-H 2 O 2 because of formation of metal-EDTA complexes 51 . In light of these observations it was considered important to assess the influence of the TLC mobile phase, which consists of both surfactants and EDTA, on the AuNPs CL intensity. The results indicated that the presence of the mobile phase has negligible effect on the AuNPs CL signal intensity and it is therefore possible to utilize the catalytic property of gold nanoparticles on CL to identify and quantify their existence. As shown in Figure 1B, a positive correlation exists between CL signal intensity and AuNP concentration in the tested range. These results indicate that using the present experimental setup, rapid and convenient measurements of CL emission is possible and the proposed method could be used to detect and identify AuNPs by coupling TLC with chemiluminescence. Compared with ICP-MS based methods for the determination of gold nanoparticles, the proposed method is more convenient, low-cost and fast, and allows determination of different sizes of gold nanoparticles in a single run.
Effectiveness of TLC-CL for AuNP size determination. In a previous study by the authors 35 , it has been demonstrated that TLC could effectively separate different sizes of AuNPs and smaller particles migrated faster than larger ones. In order to investigate the capability of TLC-CL method in size determination of AuNPs, a mixture of 13, 41, and 100 nm AuNPs was deposited onto HPTLC plates and processed as previously described. Figure 2 depicts the characteristic CL signal profile obtained by line scanning along the TLC channel. Three characteristic peaks corresponding to 13 nm, 41 nm and 100 nm AuNPs, respectively, were observed. The HPTLC plate deposited with water only (blank) and treated with luminol and H 2 O 2 yielded a negligible CL signal. Based on these observations, it appears that by coupling TLC with CL, it is possible to separate and quantify 13 nm, 41 nm and 100 nm AuNPs in one analytical step. All results demonstrate that TLC-CL method may be applied in the size determination for AuNPs though Figure 2 does show some peak overlap indicating clean baseline separation of signals from 13 nm, 41 nm and 100 nm particle signal is not achieved. These initial results were very encouraging and further experiments were performed to improve the proposed TLC-CL method.
Scientific RepoRts | 6:24577 | DOI: 10.1038/srep24577 Optimization of Chemiluminescence Detection. To improve the CL reaction and detection conditions, the experimental parameters affecting CL detection were systematically investigated varying one parameter at a time while keeping the rest constant.

Detection time for Luminol-H 2 O 2 CL reaction catalyzed by AuNPs. CL emission is a transient phe-
nomena appearing immediately when the HPTLC plate is sprayed with luminol and peroxide, the emitted light from the reactions is also decaying with time. To guarantee both the accuracy and precision of the procedure it is therefore necessary to optimize the detection time for the CL reaction. AuNPs solutions with 13, 41 and 100 nm diameters were spotted separately onto the HPTLC plates. After air drying and treatment with luminol and H 2 O 2 the chemiluminescence signals were monitored under light of wavelength 425 nm. Figure 3 shows the CL emission intensity-time profiles of the samples so prepared. It can be observed that CL emission appears immediately the HPTLC plate is sprayed followed by a rapid signal decay. Due to the difference in catalytic efficiency of AuNPs on luminol-H 2 O 2 , the slope of the CL intensity-time curve for different sizes of AuNPs is different. Though the CL emission is higher for 41 nm AuNPs the signal decays more rapidly than for 13 nm and 100 nm AuNPs, indicating that higher catalytic efficiency leads to the faster reduction of the CL emission. After 20 s, the CL signal is observed to decay more slowly leveling off at 20-30% of the initial peak height, the  variation of CL intensity being within 6.5% from 25 s to 60 s. It is concluded that in order to achieve best precision and reproducibility, the optimal detection time for AuNPs of a range of sizes should begin at 27 s. It should be noted that slight difference in CL intensity for differently sized AuNPs was observed for this time. Under the proposed analytical scheme the scanning time for developed TLC plates, from the scanning start point to the 100 nm AuNP position is approximately 7 s. Therefore, after the process of securing and spraying, the HPTLC-plate the plate is exposed in atmosphere for approximately 20 s, to allow the signal to decay to our preferred start time.
Spectrometer aperture. While a larger optical aperture in our spectrometer will obviously increase the signal intensity, the achievable physical resolution will decrease. It is therefore necessary to optimize the size of the spectrometer aperture. Figure 4 displays sample spectra for different sizes of AuNPs separated on a single HPTLC plate obtained using apertures of size 1.38 cm, 0.85 cm, 0.61 cm and 0.26 cm respectively. Figure 4A shows that the peaks for 13 nm, 41 nm and 100 nm particles can not be distinguished when the aperture is 1.38 cm while a 0.26 cm aperture shows clear separation of the peaks, though this gain in resolution was accompanied by a loss in sensitivity. The aperture of 0.26 cm was adopted in this study since the signal loss was considered acceptable. It is also possible to improve the resolution by reducing the translation stage scan rate from 3.0 to 1.0 mm s −1 . However, significant peak distortion was observed at this slower scan rate and so to produce best resolution with acceptable sensitivity for the detection of AuNPs a scan speed of 3.0 mm s −1 was selected.
pH of luminol solution. It is reported that luminol is stabilized by protonation 52 and the luminol-H 2 O 2 CL is optimal at an alkaline pH 53 , therefore the effect of pH on CL reaction of luminol was investigated under alkaline conditions. Luminol was dissolved in NaOH solution of varying concentrations covering a pH range of 9.0-13.0. Shown in Figure 5A, AuNPs exhibited the strongest catalytic effect and maximum CL intensity at pH 12. In the range of pH 9.0-12.0, CL emission was observed to increase with increasing pH, while CL intensity decreased in the range pH 12.0-13.0. From this result, a pH of 12 was selected for the study.   Figure 5B showing the CL emission intensity increasing with luminol concentration in the range of 0-0.8 mmol L −1 but only changing a little when the concentration of luminol is above 0.8 mmol L −1 .
Similarly H 2 O 2 concentration was varied over the range 0-5.0 mol L −1 , with CL emission increasing markedly with increasing concentration of H 2 O 2 up to 1 mol L −1 but for higher concentrations the signal remained constant and even showed some decrease ( Figure 5C). The observed slight decrease at higher concentrations may be a result of emission instability. Based on these results, concentrations of 0.8 mmol L −1 luminol and 1.0 mol L −1 H 2 O 2 were selected in subsequent experiments.
Analytical performance. To assess the sensitivity of the developed method and determine its suitability for AuNP quantitation, we investigated CL of 13, 41 and 100 nm AuNPs over the concentration range 0.1-48 mg L −1 . The peak height was used as the target analytical measurement throughout. Clearly evident in Figure 6 the CL signal intensity positively correlated with AuNP concentrations in the tested range. The resultant data show a linear response with correlation coefficients better than 0.99 in all studied cases (Table 1 and Figure 7). The limits     35 . In our future research, we will surely optimize the experimental setup and experiment process to improve the separation resolution and LODs of our method. In addition, the analysis time using TLC-CL of 45 s is far superior when compared with TLC-LA-ICPMS (600 s). The relative standard deviations (RSD) for seven replicate determinations of 10 mg L −1 target species were 7.3%, 6.9% and 8.1% for 13 nm AuNPs, 41 nm AuNPs and 100 nm AuNPs, respectively. All these results demonstrate that the proposed method can provide a fast, sensitive and cost-effective method for the quantitative characterization of AuNPs.

Interference from coexisting ions and dissolved organic matters (DOM).
The potential applications of chemiluminescence (CL) in analytical chemistry take advantage of the high sensitivity and simplicity of instrumentation associated with CL-based detection. However, lack of chemical selectivity is a practical problem which may be encountered in CL analysis. Likely coexisting metal ions and DOM that might react with the CL reactant were examined for their effect on the recovery of different sizes of AuNPs. Details of the effect of coexisting ions and humic acids are summarized in Table 2 and Figures S1 (see details in Supplementary Information). The experimental results show that the recovery of AuNPs in the simulated water samples is in the range of 92.5-102.4%, indicating that coexisting ions have negligible effects on the method. It has been reported that DOM such as humic acid can associate with various NMs to form stable suspensions therefore it was considered prudent to check if the presence of humic acid, widely present in environmental water, may influence the separation and detection of AuNPs. To simulate the effect of DOM on the extraction process, commercially available humic acid (HA) was used in different concentrations with different sizes of AuNPs, the HA stock suspension (10 g L −1 ) being diluted to produce samples with HA in an environmentally relevant concentration range from 0 to 30 mg L −1 . It was observed that the recovery remained constant for HA concentrations up to 10 mg L −1 , and higher than 85% when the concentration of humic acid was below 20 mg L −1 . Only for HA levels higher than 20 mg L −1 was significantly lower recovery observed ( Figure S1, Supplementary Information). Since in most waters, especially fresh waters, the HA concentrations are typically below 20 mg L −1 , the effect of DOM on AuNP measurement may be regarded as not being significant.
The assessment of interference by likely coexisting metal ions and humic acid in environmental waters suggest analysis of citrate stabilized AuNPs will not be affected. Coupling chemiluminescence with TLC separation has successfully overcome the selectivity of chemiluminescence and has improved chemical specificity in analytical measurements of a complex nature.
Since the catalytic effect of gold nanoparticles on Luminol was first reported by Cui et al. 40 , a great variety of nanoparticles (NPs) have been found to possess similar catalytic properties, e.g., triangular AuNPs 42 , AgNPs 56 , PtNPs 57 , quantum dots 58 , and so on 37 . Therefore, the interference from AgNPs has been investigated in our experiment. The recovery of 41 nm AuNPs is approximately 85%, indicating that the AgNPs has no significant effects on the method. In our future research, we will investigate interference from different kinds of nanoparticles to make our proposed method much more practical and can be used in the field of environmental analysis.
Application to Real Water Samples. According to our study, smaller particles (13 nm) migrate faster than larger ones (100 nm) and a good linear correlation between the particle size and the migration distance is obtained (R 2 = 0.9918) 35 . Moreover, the CL intensity is linearly correlated with the concentration of AuNPs in our proposed study. Thus, we can obtain quantitative results and the size information of gold nanoparticles base on the migration distance and CL signal of gold nanoparticles perform in the experiment. The feasibility of the proposed approach was evaluated by measurement of three real water samples, tap, river and lake, spiked with AuNPs in a concentration range from 0 to 10 mg L −1 . All samples were measured after passing through a 0.45 μm filter to eliminate naturally occurring particles might interfere with the procedure. Table 3

Conclusion
The direct coupling of thin layer chromatography with chemiluminescence detection applied to the separation and quantitative characterization of differently sized AuNPs was demonstrated. Detection limits for AuNPs analyzed by the developed technique were at the pg level and the recovery from real waters confirms the feasibility of this approach. Compared with established methods, the proposed method has the advantages of simplicity, high sensitivity, convenience, fast operation and requires no complex instrument. It provides an alternative way for the quantification of AuNPs. Further work need to be performed to extend to other metal based NPs.

Methods
Instrumentation. Figure 8 shows a schematic diagram of the experimental setup of the TLC-CL appara-  Table 3. Analytical results of 13 nm, 41 nm and 100 nm AuNPs spiked in water samples. a Below the limit of detection.  Preparation and Characterization of AuNPs. The procedure described by Frens et al. 59 was adopted.
50 mL of 0.01% w/v HAuCl 4 solution was transferred to a flask and heated to boiling. With vigorous stirring, 2.0 mL of 1.0% w/v trisodium citrate solution was quickly added. The color of the solution changed from pale yellow to wine red in a few seconds. The solution was refluxed for 30 min. After cooling down, the AuNP solution was kept at 4 °C in the refrigerator. The size and monodispersity of AuNPs thus prepared were determined using a transmission electron microscope and UV-VIS spectroscopy. AuNPs of differing sizes, i.e. 40 and 100 nm, were prepared by changing the volume of trisodium citrate solution added to the HAuCl 4 solution. Figure S2 in the Supplementary Information section shows the result of TEM and UV-VIS of the self-prepared AuNPs with sizes of ~13, ~41 and ~100 nm.
Procedure for quantitative characterization of gold nanoparticles by TLC-CL. The experimental procedure for TLC development was similar to that described in the authors' previous work 35 . The sample solutions containing the differently sized AuNPs were spotted (1 μL) onto the HPTLC plates and developed in a mobile phase containing 0.2 M phosphate buffer (pH = 6.8), Triton X-114 (0.4%, w/v), EDTA (10 mM) for 20 min. After development and air drying, the HPTLC plate was secured on a stage using double-sided adhesive tape and placed into a sampling chamber, and then the HPTLC plate was sprayed first with a solution of H 2 O 2 and then luminol using a TLC sprayer. The chemiluminescence signal is transitory and the CL intensity will decaying along with time, in order to guarantee both the accuracy and precision, the above procedure must be carried out at optimal time. In our experiment, after the process of securing and spraying the HPTLC-plate, the HPTLC plate then exposed to atmosphere for approximately 20 s, and finally the chemiluminescence signals were monitored along the TLC track at an optimum scan speed of 3.0 mm s −1 . The starting position for scanning is approximately 2.0 cm away from 100 nm AuNPs, thus moving from the starting positions to 100 nm AuNPs location takes less than 10 s.