Detection and spatial mapping of mercury contamination in water samples using a smart-phone.

Detection of environmental contamination such as trace-level toxic heavy metal ions mostly relies on bulky and costly analytical instruments. However, a considerable global need exists for portable, rapid, specific, sensitive, and cost-effective detection techniques that can be used in resource-limited and field settings. Here we introduce a smart-phone-based hand-held platform that allows the quantification of mercury(II) ions in water samples with parts per billion (ppb) level of sensitivity. For this task, we created an integrated opto-mechanical attachment to the built-in camera module of a smart-phone to digitally quantify mercury concentration using a plasmonic gold nanoparticle (Au NP) and aptamer based colorimetric transmission assay that is implemented in disposable test tubes. With this smart-phone attachment that weighs <40 g, we quantified mercury(II) ion concentration in water samples by using a two-color ratiometric method employing light-emitting diodes (LEDs) at 523 and 625 nm, where a custom-developed smart application was utilized to process each acquired transmission image on the same phone to achieve a limit of detection of ∼ 3.5 ppb. Using this smart-phone-based detection platform, we generated a mercury contamination map by measuring water samples at over 50 locations in California (USA), taken from city tap water sources, rivers, lakes, and beaches. With its cost-effective design, field-portability, and wireless data connectivity, this sensitive and specific heavy metal detection platform running on cellphones could be rather useful for distributed sensing, tracking, and sharing of water contamination information as a function of both space and time.

Since the recognition of severe neurotoxic effects of mercury in the 1960s,[1−3] the development of detection techniques for real-time and long-term monitoring of mercury contamination in environmental and biological samples has become a high priority. Various neurological effects of mercury exposure have been mainly attributed to the organic form of mercury, predominantly methylmercury (MeHg+), which is known to accumulate in the food chain[4] and cross the blood–brain barrier after human ingestion.[5,6] While such findings have added weight to the severity of organic mercury contamination, the threat of inorganic mercury, namely, mercury(II) ions (Hg2+), should not be underestimated. In fact, mercury(II) ions are the primary mercury contamination in the aquatic system and the “precursor” form of methylmercury due to bacteria-assisted biotransformation processes.[7,8] Furthermore, inorganic mercury is known to be more nephrotoxic than its organic form, as it primarily accumulates in the kidney proximal tubule cells.[9] Therefore, the detection and quantification of mercury(II) ion contamination in water systems are of paramount importance and could potentially be used to assist prevention of mercury ions from entering the food chain.

Toward this need, low nanomolar (nM) concentrations of mercury(II) ions have been traditionally detected by using spectroscopic methods, including, for example, atomic absorption spectroscopy (AAS),[10,11] inductively coupled plasma mass spectrometry (ICP-MS),[12] and atomic fluorescence spectrometry (AFS).[13] However, these approaches require complex sample preparation procedures, expensive and bulky instruments, and professionally trained personnel running the tests. Therefore, they are not well suited for rapid on-site detection of mercury and may not even be available for use in developing countries. On the other hand, recent advances in microfabrication and nanoscience have enabled the development of portable detection assays that are integrated with lab-on-a-chip platforms, showing great potential for use in resource-limited environments.[14−18] Among these technologies, gold nanoparticle (Au NP)-based colorimetric assays are emerging as alternative approaches for heavy metal detection,[19−21] providing high sensitivity, specificity, and ease of signal read-out using, for example, UV–vis spectrometers[19−21] or glass slide readers.[14] However, these existing systems that utilize NPs are still limited due to their relatively bulky instrumentation, higher costs, and lack of wireless connectivity, which are important especially for distributed sensing and spatiotemporal mapping of contamination in remote locations and field settings. As an alternative to Au NP-based plasmonic techniques, detection of subppm levels of mercury(II) ions has recently been demonstrated by using dye-embedded polymer films as colorimetric substrates that are digitized using, for example, smart-phone cameras.[22] However, this recent approach does not utilize the processing/computational power of the phone, and it has limited detection sensitivity and repeatability due to unavoidable variations in ambient light conditions and user operation and/or alignment during the image capture process.

To provide a field-portable, cost-effective, and wirelessly connected platform to sensitively quantify heavy metal ion concentration in water samples, here we report a battery-powered mobile sensing device that consists of a lightweight (∼37 g) opto-mechanical attachment to a smart-phone along with a custom-developed Android application for quantification, reporting, and sharing of detection results. This lab-on-a-phone device is based on dual-wavelength illumination using light-emitting diodes (LEDs) at 523 and 625 nm and can quantify mercury-induced subtle transmission changes of a colorimetric assay utilizing citrate-stabilized plasmonic Au NPs and aptamers (Apt) mixed within disposable test tubes. Due to the shift in the plasmonic resonance wavelength of dispersed and aggregated Au NPs in response to mercury(II) ions, we demonstrated sensitive detection of mercury contamination in water samples with a limit of detection (LOD) of ∼3.5 ppb, which has the same order of magnitude as the maximum contaminant level (MCL) of mercury(II) recommended for drinking water, i.e., 2 and 6 ppb, as established by the U.S. Environmental Protection Agency (EPA) and the World Health Organization (WHO), respectively.[23,24] With this cellphone-based colorimetric detection platform, we also demonstrated geospatial mapping of mercury(II) contamination in California by testing water samples collected at more than 50 locations, from tap water sources as well as natural sources such as rivers, lakes, and beaches. This heavy metal detection system running on smart-phones could provide a complementary addition to other mobile-phone-based imaging, sensing, and diagnostics devices[25−42] and holds significant potential for distributed sensing and spatiotemporal mapping and monitoring of mercury contamination globally. In fact, cellphone subscriptions worldwide have reached more than 7 billion by the end of 2013, and smart-phone penetration rate is globally increasing, which is estimated to reach more than 60%, 45%, and 25% by the end of 2015 in North America, Europe, and Africa, respectively.[43] Therefore, the use of mobile phones for bioanalytical measurement science as well as for reporting and sharing of results provides widely scalable, cost-effective, and yet rather powerful/competitive solutions to implement various tests and measurements even in resource-limited and field settings, which constitutes an important motivation for this work.

Results and Discussion

Optical Design of the Smart-Phone-Based Mercury Reader

We created an optical imaging interface that is mechanically attached to the existing camera module of a smart-phone to quantify mercury concentration using a colorimetric nanoparticle and aptamer assay. This attachment contains two button cells (3 V), which are used to power two LEDs, as illustrated in Figure a. These LEDs are mounted at a sufficiently large distance (∼26.5 mm) away from the location of the rectangular test tubes (containing the sample and control solutions) and are scattered by optical diffusers to ensure uniform illumination of both tubes (Figure a). The emission wavelengths of these LEDs were selected to be 523 and 625 nm to follow the shift in the extinction wavelengths of the dispersed and aggregated Au NPs, respectively. In order to obtain multispectral information from a given water sample, the green LED illuminated the bottom half while the red LED illuminated the top half of each cuvette (Figure a). To avoid a possible crosstalk between the green and red lights, the optical paths of the two LEDs were separated by an opaque clapboard before they reached the cuvettes (Figure a, inset) and passed through two rectangular apertures (6.6 × 5 mm, one for each color) that are placed in front of each cuvette. The transmitted light through the sample and control cuvettes was then collected through two other rectangular apertures of the same size to be imaged onto the digital camera of the smart-phone using a plano-convex lens (f = 28 mm) (Figure a). This external lens was chosen to yield a demagnification factor of 7× so that the two 6.6-mm-wide sample cuvettes could be simultaneously imaged within the active area of the smart-phone CMOS imager chip. All of these electrical and optical elements were consolidated in an opaque cuboid (Figure a, gray part) and coupled to a base plate (Figure a, black part) with a total weight of ∼37 g. Although the current attachment is designed for an Android phone (Samsung Galaxy S II, Figure b), the same optical reader can be implemented on other smart-phones such as an iPhone, after slight mechanical modifications in the base attachment.

Figure 1

Design of the ratiometric optical reader on a smart-phone. (a) 3D schematic illustration of the internal structure of the opto-mechanical attachment. The inset image shows the same attachment with a slightly different observation angle. (b) Photograph of the actual optical reader installed on an Android-based smart-phone. The screen of the smart-phone displays a typical image of the sample and control cuvettes when illuminated by red (625 nm) and green (523 nm) LEDs simultaneously.

Plasmonic Colorimetric Assay and Measurement of Mercury(II) Ion Concentration

Spherical Au NPs have been previously studied as novel sensing probes for mercury(II) ion detection.[19−21] The characteristic color change of Au NPs from red to purple or blue upon aggregation that is induced by mercury(II) ion binding events constitutes the basis of the Au NP-based colorimetric detection assay. However, most Au NP-based probes require a surface modification step to conjugate mercury(II)-specific ligands onto Au NPs, and the LOD varies based on the capturing ligand that is selected.[44−48] Here, we adopt an alternative approach, which utilizes the strong affinity of the thymine-rich aptamer sequence to mercury(II) ions and citrate-stabilized Au NPs as colorimetric signal transducers to generate a high detection sensitivity.[49] In this protocol, Au NPs are used without the need for surface functionalization steps, which greatly facilitates field use. In a typical mercury detection experiment, 0.64 nM Au NPs (50 nm diameter) are mixed with 3 μM aptamer (5′-TTTTTTTTTT-3′) in 20 mM Tris-HCl buffer (pH 8.0) to form the probe solution. Next, 4 μL of water sample solution is added to the probe solution and incubated for 5–10 min (see Methods section for details). Aptamer forms a protective layer on the surface of Au NPs, which prevents them from aggregation even in a high-salt environment such as 10 mM NaCl. However, this aptamer layer will be stripped off by the presence of mercury(II) ions due to the formation of more stable T-Hg2+-T complexes.[50,51] As a result, the unprotected Au NPs can undergo distinct color transition from red to blue in the presence of NaCl (Figure a), and this spectral shift is detected to quantify mercury concentration using our dual-wavelength smart-phone-based colorimetric reader.

Figure 2

Principle of dual-color dual-cuvette colorimetric detection. (a) Scheme of the mercury sensing mechanism by using plasmonic Au NPs and aptamer. (b) Representative image captured on the smart-phone under dual-wavelength illumination. The left cuvette (control) contained a mixture of Au NPs (0.64 nM) and aptamer (30 nM), while the right cuvette (sample) contained a mixture of Au NPs and aptamer plus 500 nM Hg2+ (representative of a contaminated water sample). (c) Flow of image-processing steps to compute normalized green-to-red signal ratio (i.e., normalized G/R signal).

A representative smart-phone-captured image of Au NP probe solutions with and without mercury(II) ions is depicted in Figure b. Each cuvette was illuminated by red and green LEDs at different spatial locations and separated by two rectangular apertures that are 3 mm apart from each other (Figure b). The illumination spots of the LEDs were sufficiently large to cover both the sample and control cuvettes (2 mm apart). This dual-illumination color and dual-cuvette configuration forms four readable signals in a single image frame, namely, red control (RC), red sample (RS), green control (GC), and green sample (GS) signals (Figure c). To quantify the mercury contamination in a given water sample, the acquired transmission image of these cuvettes (sample and control) is first digitally split into red (R) and green (G) channels (Figure c) to further minimize the spectral crosstalk between these two colors. The centroids of each rectangular aperture are automatically localized by a detection algorithm, and a rectangular region of interest (ROI, 400 × 300 pixels) around each of these four centroids is then used to calculate the averaged transmission signal for each ROI, yielding RC, RS, GC, and GS signals. Note that RC and RS are calculated using the red channel image, whereas GC and GS are calculated using the green channel image, both of which are digitally separated from the raw RGB image captured by the cellphone camera sensor (Figure c). The transmission intensity of the sample cuvette is further normalized to that of the control cuvette by placing two identical deionized water samples in both cuvette positions, leading to an illumination normalization factor of 1.15 for the red LED (R_Factor) and 0.98 for the green LED (G_Factor). The calibration ratio of the control cuvette (G/R_C) was obtained by taking the ratio of the GC and RC. Similarly, the calibration ratio of the sample cuvette (G/R_S) was obtained by taking the ratio of GS × G_Factor to RS × R_Factor. Finally, the ultimate normalized green-to-red signal (i.e., normalized G/R) for a given water sample was computed by taking the ratio of G/R_S to G/R_C (Figure c). These calculations are automatically implemented using a custom-designed Android application running on the same smart-phone, which will be detailed in the next section.

Android-Based Smart Application for Mercury Quantification

We created a custom-designed Android application that allows for mobile testing and sharing of mercury quantification results. After attaching the colorimetric mercury measurement device onto the smart-phone camera unit (Figure ), the user can hold the cellphone horizontally and then run mercury tests using this smart application. From the main menu of the application, the user can start a new test, create a device-specific calibration curve, view previously run tests, share the test results, and review the operating instructions (Figure a). The user can calibrate the application for attachment-specific variations by imaging, for example, mercury-contaminated control samples at known concentrations (Figure b). These calibration curves can be stored and reused by various devices/attachments. After capturing a colorimetric transmission image of the sample, the user can first preview the image on the screen before proceeding to digitally analyze/process it (Figure c). The application can also use an image file already stored on the phone memory for processing/testing. After pressing on the “Process” button, the transmission signal ratios between the sample and control regions will be automatically computed on the phone, following the image-processing steps discussed in the previous section. A previously stored calibration curve is used to convert the calculated signal ratio into the mercury concentration level of the sample (in ppb), and the results are then displayed on the screen of the phone (Figure d). The total time taken for calculating the mercury concentration on the Android phone (Samsung Galaxy S II) is <7 s. The final test results can be saved on the phone memory with a stamp of time and GPS coordinates of the test and can also be shared with a secure server for spatiotemporal mapping using, for example, a Google Maps-based interface (Figure e). With the same Android application, the results can also be reviewed as a function of time per location using a graph-based interface (Figure f).

Figure 3

Screen shots of our mercury detection application running on an Android phone. (a) Main menu; (b) calibration menu; (c) preview of a captured or selected colorimetric image before proceeding to analyze/quantify the sample; (d) display of the results; (e) spatiotemporal mapping of mercury contamination using a Google Maps-based interface; (f) tracking of mercury levels as a function of time per location.

Calibration and Specificity Tests

In our cellphone-based mercury detection platform, each normalized G/R ratio computed from a captured RGB image corresponds to a specific mercury concentration value (ppb). The Android application includes a default calibration curve, which was obtained by measuring the normalized G/R ratios of a set of known concentration mercury(II) solutions ranging from 0 to 5 μM (see Figure ). The values of these normalized G/R ratios increased as the concentration of mercury(II) ions rose above 10 nM and reached saturation at >1000 nM (Figure ). The signal increase in the 10–1000 nM range is mainly due to the aggregation of Au NPs, which is triggered by the mercury(II) ion concentration. This Au NP aggregation process relatively enhances the extinction at the red wavelength (e.g., 625 nm), while it reduces the extinction at the green wavelength (e.g., 523 nm), which is also confirmed by our UV–vis spectroscopic measurements (see the Supporting Information, Figure S1a). This plasmon-resonance-based wavelength shift occurred rapidly after around 5 min (Supporting Information, Figure S2), demonstrating a quick response time for the NP/aptamer-based colorimetric assay, making it appropriate for use in field settings. As a result of these plasmonic changes due to NP aggregation, the transmission signal of the red channel relatively decreased, whereas the transmission of the green channel increased. Therefore, the final G/R ratio of a sample increased as the mercury(II) ion concentration is increased, which is also illustrated in the calibration curve presented in Figure .

Figure 4

Dose–response curve of the Au NP and aptamer based plasmonic colorimetric assay running on a smart-phone. Each measurement at a given concentration was repeated three times. The curve was fitted by an exponential function with a coefficient of determination (R2) of 0.96. An LOD of 3.5 ppb for Hg2+ was obtained based on the G/R ratios of a control sample ([Hg2+] = 0) plus 3 times the standard deviation of the control (blue dashed line).

To determine the LOD of our smart-phone-based colorimetric assay, we measured the normalized G/R values of a control sample (i.e., [Hg2+] = 0, [Au NPs] = 0.64 nM, [Apt] = 30 nM), which resulted in a signal level of 0.940 ± 0.025 (μblank ± σblank). Our LOD was then determined by the mean of this control sample plus three times its standard deviation (μblank + 3σblank; see the blue dashed line in Figure ), which corresponds to a mercury(II) ion concentration of approximately 17.3 nM, or ∼3.5 ppb. Quite interestingly, the LOD of our smart-phone-based dual-color ratiometric platform was more than 6 times better than the LOD of the exact same assay measured by a portable UV–vis spectrometer (Ocean Optics, HR2000+), which resulted in 123 nM, or 24.6 ppb, LOD (see the Supporting Information, Figure S1b). More importantly, the LOD of mercury(II) ions using our smart-phone-based field-portable sensor has the same order of magnitude as the EPA’s mercury(II) reference concentration for drinking water (i.e., 2 ppb)[23] and also satisfies the WHO guideline value for mercury(II) concentration (i.e., 6 ppb).[24]

Next, we performed specificity tests by challenging the same colorimetric plasmonic nanoparticle and aptamer assay with different metal ions, such as Fe3+, Ca2+, Cu2+, and Pb2+, as illustrated in Figure . The concentrations of all these metal ion samples were prepared to be 500 nM, and our experiments revealed that, except mercury(II) ions, the other metal ion samples yielded a signal level that is comparable to control samples (Figure ), verifying the specificity of our assay toward detection of Hg2+. The same specificity performance was also confirmed independently by UV–vis spectroscopic measurements as summarized in the Supporting Information, Figure S1c,d.

Figure 5

Specificity tests of the Au NP and aptamer based plasmonic mercury assay for different metal ions (500 nM). Each measurement was repeated three times.

Mapping of Mercury Concentration in Water Samples in California

The performance of our smart-phone-based mercury sensor was also tested with water samples including city tap water and natural water samples collected at over 50 different locations in California. Figure summarizes our measurement results for 19 of these samples collected from various apartments (tap water), rivers, lakes, and beaches on the California coast. The results suggest that all the city tap water samples have undetectable levels of mercury(II) ions since the signal readings are at the same level as our LOD (Figure ). However, our measurements for the water samples collected from natural sources reveal higher mercury concentration levels, ranging from 3.7 to 8.6 ppb, as illustrated in Figure . The samples that are found to contain mercury(II) ion concentrations above 6 ppb, i.e., the safety level recommended by WHO,[24] are mostly from ocean samples, with the worst being from the San Francisco Bay (Figure ). Our observation that the mercury content in ocean water samples is higher compared to fresh water is probably because the ocean is at the end of mercury’s global transport pathway in the environment[52] and thus might exhibit higher pollution levels.

Figure 6

Smart-phone-based mercury detection results for 11 tap water samples and eight natural samples collected in California, USA. Each measurement was repeated three times. Note that the measurements are plotted against the G/R ratios, which makes the presented scale of the mercury concentration (ppb) nonlinear, between 0.8 and 9.1 ppb.

As one of its major advantages, our hand-held smart-phone-based mercury detection platform is also able to generate spatiotemporal contamination maps for, for example, environmental monitoring. To do so, GPS coordinates were recorded for each water sample that was tested, and all the other sample-related information such as measurement results and dates was sent to a secure server using the smart-phone application for mapping of the results. Figure a–c represent three smart-phone-generated mercury-monitoring maps, where the spatial resolution of the maps is determined by the sampling density. For instance, in Figure a, samples were collected and measured at a low density of ∼0.2 measurements/km; in Figure b and c, higher resolution was shown by increasing the sampling density to 3.3 and 20 measurements/km, respectively. Figure d–f show histogram plots corresponding to the mercury(II) concentrations that are displayed in Figure a–c with a better visualization of the variation of mercury(II) levels within a given area. Interestingly, some locations such as points B and C in Figure a and d had statistically higher mercury(II) levels than the rest with very small p values (<0.001) determined by standard Student’s t test. Further investigation of this area indicated that the red ROI in Figure a included a marina hosting yachts and boats (Figure b), which possibly form the major source of heavy metal pollution in that particular region. Mercury(II) ion concentration near the marina also formed a weak gradient (from A to T), as illustrated in Figure b and e, with the closest point to the marina having the highest mercury(II) concentration (i.e., point T in Figure b and e). Point E in Figure e was statistically lower in mercury(II) concentration (p < 0.01) compared to other locations within the same region of interest, and this observation was confirmed by higher resolution mercury(II) mapping in Figure c and f. In addition to spatial mapping of contamination, the option of monitoring the level of mercury concentration as a function of time for a specific location is also feasible using our smart-phone-based sensing platform as illustrated in Figure f.

Figure 7

Spatiotemporal mapping of mercury contamination in Los Angeles coastal area. (a–c) Geospatial mercury concentration maps with different sampling densities; (b) zoomed-in area of the red ROI in (a); (c) enlarged region of the red ROI in (b). (d–f) Corresponding mercury concentration readings in (a)–(c). All the data points were measured three times. p values were calculated via two-sample Student’s t test by setting target data set as one population and the rest of the data sets as the other. ** represents p < 0.01, and *** represents p < 0.001.

Conclusions

In summary, we introduced a sensitive and cost-effective smart-phone-based mercury(II) ion sensor platform, which utilizes a battery-powered opto-mechanical reader attached to the existing camera module of a smart-phone to digitally quantify mercury concentration using a plasmonic Au NP and aptamer based colorimetric assay. We employed a two-color ratiometric detection method using LEDs at 523 and 625 nm and a custom-developed Android application for rapid digital image processing of the captured transmission images on the same phone. The LOD of mercury(II) ions with this mobile device is found to be 3.5 ppb, which is on the same order of magnitude with the maximum allowable level of mercury(II) ions in drinkable water defined by the U.S. EPA (2 ppb)[23] and WHO (6 ppb).[24] Moreover, we generated a geospatial mercury(II) contamination map by measuring more than 50 samples collected in California from various sources including tap, river, lake, and ocean water samples. The cost-effective design, portability, and data connectivity of this sensitive heavy metal detection device integrated onto cellphones could be rather useful for distributed sensing, tracking, and sharing of water contamination information as a function of both space and time, globally.

Methods

Hardware Design

Our optical imaging system was designed for an Android phone (Samsung Galaxy S II) in Autodesk (Inventor) and printed using a 3D printer (Elite, Dimension). Two LEDs (120 degree illumination angle, SuperBrightLEDs), one green (523 nm, RL5-G16120) and one red (625 nm, RL5-G12120), illuminated the test/sample and control cuvettes simultaneously and were powered by two button cells (3 V, CR1620, Energizer). An optical diffuser (made using three sheets of A4 printer paper) was inserted between the LEDs and the cuvettes for uniform illumination of each cuvette. The transmitted light through the cuvettes was then collected by a plano-convex lens (focal length f = 28 mm, NT65–576, Edmund Optics) and imaged using the smart-phone camera (f = 4 mm). This imaging configuration provides an optical demagnification factor of 28/4 = 7-fold, which permits imaging of both the test and control cuvettes (6.6 × 6.6 mm in cross section) within the field of view of the phone’s CMOS imager chip. To avoid crosstalk of the two-color illumination, a black clapboard was used to separate the light paths of the LEDs before entering the cuvettes, and four rectangular apertures (6.6 × 5 mm) were added both in front of and behind the cuvettes to spatially filter the transmitted light at each color (i.e., red and green). The acquired images were analyzed in digitally separated red and green channels to further reject possible spectral crosstalk between red and green illumination wavelengths.

Gold Nanoparticle and Aptamer Based Colorimetric Assay

Citrate-stabilized Au NPs (50 nm) were purchased from NanoComposix. Aptamer sequence of 5′-TTTTTTTTTT-3′ was obtained from Integrated DNA Technologies. All metal salts such as mercury(II) chloride were obtained from Sigma. Stock Au NP solution in 20 mM Tris-HCl buffer (TH, pH 8.0) was prepared by centrifugation of raw Au NP-citrate solution, aspiration of the supernatant, and redispersion in TH buffer with 20× dilution to give a working concentration of 0.64 nM. Water samples collected from rivers, lakes, and beaches were filtered by a 0.2 μm polyethersulfone membrane (Whatman) to remove sand and other solid particles within the test samples. Tap water samples and calibration solutions containing mercury(II) ions prepared in deionized water were used directly without further purification. In a typical measurement procedure, 4 μL of the sample of interest was mixed with 4 μL of 3 μM aptamer (20 mM TH buffer, pH 8.0), followed by a 5 min reaction period. Next, 400 μL of Au NPs (0.64 nM) in 20 mM TH buffer solution was added and allowed to react for 5 min. Finally, 8 μL of 10 mM NaCl was added and incubated for another 10 min before being analyzed by the smart-phone device.

UV–Vis Spectroscopic Investigation of Water Samples Using a Portable Spectrometer

In our comparison measurements against the smart-phone (see the Supporting Information), a white LED (RL5-W15120, SuperBrightLEDs) was used as the light source, and the transmission signal that passed through a standard 1 cm cuvette was collected by a 600-μm-diameter optical fiber and measured by a portable spectrometer (HR2000+, Ocean Optics). The background spectrum was recorded using deionized water as a blank control sample. Each spectrum was collected with an exposure time of 1 ms and scanned 500 times for averaging in order to improve the signal-to-noise ratio of each UV–vis spectroscopic measurement.