Detection, Characterization, and Abundance of Engineered

Aug 18, 2012 - Potential of Hyperspectral Imaging Microscopy for Semi-quantitative Analysis of Nanoparticle Uptake by Protozoa. Monika Mortimer , Alex...
2 downloads 13 Views 2MB Size
Download PDF
Article pubs.acs.org/est

Detection, Characterization, and Abundance of Engineered Nanoparticles in Complex Waters by Hyperspectral Imagery with Enhanced Darkfield Microscopy Appala Raju Badireddy,†,‡ Mark R. Wiesner,*,†,‡ and Jie Liu‡,§ †

Department of Civil and Environmental Engineering, Duke University, Durham, North Carolina 27708, United States Center for the Environmental Implications of NanoTechnology, Duke University, Durham, North Carolina 27708, United States § Department of Chemistry, Duke University, Durham, North Carolina 27708, United States ‡

S Supporting Information *

ABSTRACT: We introduce a novel methodology based on hyperspectral imagery with enhanced Darkfield microscopy for detection, characterization, and analysis of engineered nanoparticles in both ultrapure water and in complex waters, such as simulated-wetland ecosystem water and wastewater. Hyperspectral imagery analysis of 12 different nanoparticle sample types, scattering the obliquely incident visible and near-infrared light (VNIR: 400−1000 nm) in an enhanced Darkfield background, showed that the sample information in terms of the spatial distribution as well as spectral characteristics unique to each nanoparticle types, at a sensitivity of single nanoparticle (size ≥10 nm) can be obtained. Hyperspectral imagery and Raman spectral analyses of the silver nanoparticles (AgNPs) revealed that the apparent hydrodynamic size of the particle increased while the primary size remained unchanged in the presence of coatings, which is further confirmed by dynamic light scattering measurements. Similar in size, AgNPs with different coatings exhibited similar spectral color (or peak position) but a red-shift in the peak positions by same amount relative to Bare AgNPs was observed. In conclusion, hyperspectral imagery with enhanced Darkfield microscopy can be a promising tool for detection and characterization of engineered nanoparticles in environmental systems, facilitating studies on fate and transformation of these particles in various types of water samples.

■

INTRODUCTION Nanoparticles may be categorized as originating from engineered, natural, or incidental sources.1 Nanoparticles from one or more of these categories may be present in complex environmental media, such as biota, water, soil, or air, where their detection, differentiation, and quantification is complicated by potential interferences from the environmental or physiological matrix.2 A recent study predicted that annually tons of engineered nanomaterials would be increasingly produced; for example, the upper and lower bounds for AgNPs were 2.2 and 20 ton/year, respectively, which will eventually contribute to unintentional accumulation of engineered nanoparticles in environmental waters.3 Indeed, detection, measurement, and differentiation of engineered nanoparticles in complex environmental media is a key challenge in understating the potential impact that engineered nanoparticles may have on human health and the environment.1 Commonly applied methods for nanoparticle characterization and analysis involving light scattering, fractionation and separation, microscopic, and spectroscopic techniques, typically involve sample preparation and manipulation that may introduce artifacts.1,4,5 Nanoparticles are not only sensitive to © 2012 American Chemical Society

the physical and chemical properties of their environment (e.g., pH, ionic strength, organic and inorganic content of water) but also to transformations resulting from sample preparation (such as, clumping or aggregating) and analysis using electron beams, ions, and X-rays.4 For example: electron beams have been shown to severely impact the morphology of gold and tin (in SiO2) nanoparticles and oxidation of nano zerovalent iron (nZVI) and iron oxide shells;6−8 differences in ion sputtering of suspended and supported carbon nanotubes (CNT) have been shown to affect the measurements;9 The chemical state of ceria nanoparticles have been affected by aging, coating, and solution chemistry;10−12 X-ray beams have been implicated in inducing charge on nanoparticle surfaces;10 and, the quantum coupling and engagement in quantum dots and plasmonic nanoparticles likely to be induced as result of local concentration effects.13−15 There is a need for characterization and analytical tools that can deal with heterogeneous samples with minimum sample preparation and nanoparticle alteration.5 Enhanced Darkfield Received: Revised: Accepted: Published: 10081

November 21, 2011 August 8, 2012 August 17, 2012 August 18, 2012 dx.doi.org/10.1021/es204140s | Environ. Sci. Technol. 2012, 46, 10081−10088

Environmental Science & Technology

Article

Similar analysis was performed on ultrapure water samples that were dosed with three types of nanoparticles, Ag-PVP55, TiO2 NPs, and CeO2 NPs, at 0.1 mM. From these samples, a 50 μL drop was deposited on a clean glass slide and partially dried for 30 min meanwhile intermittently the drop was spread on the slide using the pipet tip to minimize nanoparticle aggregation due to drying. These samples were then covered with a coverslip and analyzed using hyperspectral imaging microscope. Hyperspectral Imaging. Samples were examined using enhanced Darkfield transmission optical microscope (Olympus BX41) equipped with hyperspectral imaging spectrophotometer (CytoViva Hyperspectral Imaging System (HSI), Auburn, AL). CytoViva HSI consisted of three principal components: (1) a concentric imaging spectrophotometer that was capable of recording high quality spectrum (high signal-to-noise ratio) in visible and near-infrared (VNIR: 400−1000 nm) wavelengths at a high spectral resolution of 1.5 nm with 10 nm scan size and pixel size 25 nm; this allowed the visualization of nanoparticles that were smaller than 50 nm, (2) a customized hyperspectral imager (mounted on a microscope and controlled by Environment for Visualization software (ENVI 4.4 version) from Exelis Visual Information Solutions, Inc.) that can extract complete spectral information from single or multiple pixels, (3) a motorized stage that was guided by HSI system for synchronizing sample movement with hyperspectral image scanner, and (4) the microscope was equipped with a novel illumination system with complete Koehler (fixed) and Critical (adjustable) illumination, which made the nanoparticles appear brighter, thus obviated the need for staining agent or a contrast agent to visualize the sample.29 Image Processing and Analysis. Image processing and analysis involved the following steps:30 First, each image was processed to contain only unsaturated pixels (those pixels with digital numbers between 0 and 255) by removing interferences from background pixels due to glass slide and saturated pixels. Second, a training set (spectral endmembers or spectral library) for the nanoparticles was obtained from the sample image either by choosing the pixels that best represent the nanoparticles or by using n-dimensional visualization algorithm; the hyperspectral image of nanoparticles in ultrapure water sample served as the guide for choosing the endmembers associated with the nanoparticles of interest in the sample image. All images were acquired at identical gain (gain = 5) and exposure time (250 ms). Finally, image classification algorithms such as spectral angle mapper (SAM) and mixture tuned matched filtering (MTMF) were used for mapping and estimating the relative abundance of nanoparticles of interest in a given sample. The details of the steps involved in hyperspectral image analysis are given in the SI (see Image Processing and Analysis). Raman Spectroscopy. A commercial Raman spectrometer (Horiba LabRam ARAMIS) equipped with an 8mW HeNe laser and 100× objective lens was used for collecting Raman spectra at wavelength 632.8 nm. A high-resolution grating (1800 grooves) was used for spectra acquisition over a range of wavenumbers spanning 1800−500 cm−1. Dynamic Light Scattering. Hydrodynamic size of nanoparticles was determined using dynamic light scattering (DLS; ALV-GmbH, Germany; scattering angle set at 90 degrees).

microscopy with hyperspectral image analysis of nanoparticles under wet conditions has potential to minimize alterations to the nanoparticles during the characterization. Enhanced Darkfield microscope equipped with hyperspectral imaging (HSI) spectrometer (a.k.a., CytoViva Hyperspectral imaging microscope) is a recent advancement over currently existing classical optical microscopes and can be a promising tool for detection and characterization of engineered nanoparticles in environmental systems. Either Koehler or Critical illumination is achieved with conventional Darkfield. Under enhanced Darkfield conditions, particles appear 150-fold brighter due to Koehler and main features of Critical illumination by collimated light source at oblique angles.16 The spectrometer allows analysis of scattered light at pixel-bypixel level, with each pixel defined by a gray value or intensity with an associated coordinate in the image. Unique feature of this hyperspectral imagery technique is that samples under wet conditions can be imaged by acquiring hundreds of contiguous wavelengths or bands (bandwidth 1.5 nm) producing extensive spatial and spectral (chemical) data for each pixel.17 Hyperspectral imaging with airborne remote sensing systems is widely employed for surveillance, target detection and identification by the military. Also, this imaging technique is commonly used for mineral exploration and environmental monitoring, product quality assurance and inspection, food safety, forensics, and health care and medical research.17−19 Recent studies have only utilized enhanced Darkfield microscopy for tracking the nanoparticles (e.g., silver, gold, and single walled carbon nanotubes) in various eukaryotic cells.20−28 The interferences and limits of the technique are not well understood in the application of identifying and quantifying nanoparticles in mixtures or in complex media. Using enhanced Darkfield based hyperspectral imagery technique twelve different nanoparticles were characterized under wet conditions in visible and near-infrared wavelengths and the ability of this imagery technique was tested for detecting, differentiating, and estimating the relative abundance of mixture of different types of nanoparticles in ultrapure water. The impact of coatings (sorbed organic molecules) on the spectral characteristics of silver nanoparticles (AgNPs) in ultrapure water was investigated. Finally, the ability of the technique to detect and quantify the AgNPs in two, more complex, aquatic media (water obtained from a simulated wetland ecosystem water (referred to as mesocosm water) and secondary wastewater effluent) was investigated.

■

EXPERIMENTAL SECTION Nanoparticles and Water Sample. Nanoparticles with sizes ranging from 8 to 150 nm were used for this study (see Table S1 of Supporting Information (SI)). Ag-PVP10, TiO2 NPs, CeO2 NPs, nC60, pMWCNT, SWCNT-COOH, and SWCNT-GA stock suspensions were prepared by sonication (power: 89−95 W for 20 min in a ice bath) using Sonicator 4000 (Misonix, Qsonica LLS series) equipped with titanium horn and a regular half-inch titanium tip. Simulated wetland ecosystem (mesocosm) water was obtained from Center for the Environmental Implications of NanoTechnology (CEINT) research facility at Duke University and the wastewater was collected from the secondary clarifier at the Brown Water Treatment Plant in Durham County, North Carolina. For the purpose of case studies mesocosm water and wastewater were dosed with 2.5 mg/L Ag-PVP55 nanoparticles and the samples were imaged using hyperspectral imagery.

■

RESULTS AND DISCUSSION A series of initial experiments were performed to evaluate the ability of this technique to estimate the relative abundance(s) 10082

dx.doi.org/10.1021/es204140s | Environ. Sci. Technol. 2012, 46, 10081−10088

Environmental Science & Technology

Article

Figure 1. Hyperspectral images and the spectral signatures of nanoparticles in visible and near-infrared wavelengths (VNIR: 400−1000 nm) are shown here. All images were acquired using 100× objective/1.3 oil iris with a zoom in of 6×. The spectra were collected using 512 scans at a spectral resolution of 1.5 nm. Bilinear curve fitting was applied to all the above images to obtain smooth spherical objects. The primary size of the nanoparticles is listed in SI Table S1. Scale: 3 cm = 50 μm.

Differentiation of Mixtures of Nanoparticles in Ultrapure Water. Hyperspectral image and the corresponding MNF image (RGB from first three bands) of the mixture of nanoparticles in ultrapure water are shown in Figure 2A and B. After segregating the noise in the image, the nanoparticles of various colors can be seen in Figure 2B, which certainly represented the most spectrally active nanoparticles in the sample. Using hyperspectral image analysis the spectra of the nanoparticles in the sample were matched with the existing nanoparticle endmember library, mostly user built due to lack of public databases particularly relevant to nanoparticles. The image analysis of Figure 2A revealed the presence of seven distinct endmembers, which are characteristic of nanoparticle type and size. As a first guess all the colored dots were due to Ag-PVP55 (due to SPR) whereas as large yellowish-to-white colored objects were due to TiO2 and CeO2 NPs aggregates (see Figure 2A). The endmember spectra that had a >90% match with the spectral libraries of Ag-PVP55, TiO2, and CeO2 were used in mapping analysis. MTMF maps revealed that chosen endmembers accounted almost all the spectrally active materials that have >75% match with the image endmembers. The percentage abundance of Ag-PVP55, TiO2 NPs, and CeO2 NPs are 1.234%, 0.077%, and 0.113%, respectively. Using Ag-PVP55 in mesocosm water as an example a detailed procedure for nanoparticle detection and estimation of relative abundance was described in the SI (see Ag-PVP55 in simulated wetland ecosystem water, Figure S2). Effect of Coatings on VNIR Spectra of Silver Nanoparticles in Ultrapure Water. VNIR and Raman spectra of AgNPs with citrate, GA, PVP10, PVP55 coatings and without coating (Ag Bare) are shown in Figure 3A and B. From Figure 3a it is clear that Ag Bare and Ag-Cit exhibited same spectral features except that dominant peak of Ag-Cit red-shifted from 510 to 550 nm. This suggests that citrate coating (70% (w/w)) increased the apparent size of AgNPs, which was also confirmed by hydrodynamic size (26 ± 1 nm) of AgNPs measured by dynamic light scattering. Therefore, the primary size of these nanoparticles is similar to the size of Ag Bare

and differentiate nanoparticles under relatively ideal conditions where only a single type of nanoparticle or a mixture of three types of nanoparticles was present, and the suspending medium was that of ultrapure water. These experiments were accompanied by experiments in which a single type of nanoparticle, Ag NPs were suspended in a two relatively complex aquatic media, a mesocosm water and a clarified wastewater. VNIR Spectral Characterization of Suspensions of Nanoparticles in Ultrapure Water Differing by Composition and Size. A series of suspensions of a single type of nanoparticle in ultrapure water were characterized. The sixtimes zoomed-in (600× magnification) hyperspectral images of QD655 nm, AuNP, Ag-PVP10, Ag-PVP55, Ag-GA, Ag-Cit, TiO2 NPs, CeO2 NPs, nC60, SWCNT-GA, SWCNT-COOH, and pMWCNT and the corresponding VNIR spectra with unique features (shown below each image) associated with each types of the nanoparticles are shown in Figure 1 (A−D) and SI Figure S1 (E−L). The spherical “halos” appeared colored due to polydisperse nature of spherical nanoparticles (Figure 1A-D) and the quantum confinement (QD655 nm) or surface plasmon resonance (AuNP, Ag-PVP10, Ag-PVP55, Ag-GA, and Ag-Cit) effects. TiO2 NPs, CeO2 NPs, nC60, SWCNT (−GA and −COOH), and pMWCNT were visible in the enhanced Darkfield due to the optimized Koehler and Critical illumination, which is not possible with conventional Darkfield microscopy. The nanoparticles with violet-blue, green, and red have sizes approximately, 10 nm, 20 nm, and 40 nm, respectively. These observations are consistent with the results from a recent study which reported that AgNPs with blue, green, and red colors had sizes 5−15 nm, 16−30 nm, and 31− 46 nm, respectively,.31,32 Pure colors (especially red, green, and blue: RGB) are associated with primary size nanoparticles whereas secondary colors are associated with nanoparticle aggregates. This technique is convenient to visualize metal nanoparticles such as AgNPs as small as 10 nm because they offer the highest quantum yield of Rayleigh scattering, whose scattering intensity is proportional to the volume of the nanoparticles.32−34 10083

dx.doi.org/10.1021/es204140s | Environ. Sci. Technol. 2012, 46, 10081−10088

Environmental Science & Technology

Article

Figure 2. Mixture tuned matched filtering (MTMF) analysis of hyperspectral image of mixture of Ag-PVP55, TiO2, and CeO2 nanoparticles in ultrapure water is shown here. (A) A hyperspectral image, (B) Minimum noise fraction (MNF) image, (C) Endmembers and their relative percentage abundances, and (D) Hyperspectral image showing the location of each of the endmembers in the image scene. Images were acquired using 100× objective/1.3 oil iris. Scale: 2 cm = 200 μm.

Figure 3. (A) Visible and near-infrared (VNIR) spectra of AgNPs (green colored) with and without coatings were obtained using hyperspectral imager, (B) Raman spectra of AgNPs (λ = 632.8 nm) in the presence and absence of coatings are shown here, and (C) the mean hydrodynamic size of AgNPs is shown next to green colored AgNPs here.

GA, secondary peaks were observed to the left (blue region of the spectra) of peak maxima (dominant peak). Peak maxima of AgNP with PVP10, PVP55, and GA coating were identical (550 nm) and red-shifted compared to Ag Bare peak (510 nm). As the size of the coatings increased Cit < PVP10 < PVP55 < GA the scattering ability of AgNP also increased significantly, which

(Figure 3C: identical green color suggests similar sizes), but due to differential swelling properties of the coatings in water larger hydrodynamic sizes were observed for AgNPs. This was further confirmed by Raman spectra, which also showed significant similarities in spectral features (1800−500 cm−1) for both Ag Bare and Ag-Cit. In the presence of PVP10, PVP55, and 10084

dx.doi.org/10.1021/es204140s | Environ. Sci. Technol. 2012, 46, 10081−10088

Environmental Science & Technology

Article

Figure 4. Hyperspectral image of Ag-PVP55 nanoparticles in mesocosm (A), 600× zoom in of a portion of the sample image (A), (C) n-dimensional visualization analysis of endmembers, and (D) Endmembers of Ag-PVP55 nanoparticles of different sizes the sample. Images were acquired using 100× objective/1.3 oil iris. Scale: 2 cm = 200 μm.

was comprised of aggregates of natural colloids seen as flocs or clusters in the image. The purest pixels (spectrally unique pixels) were utilized by n-D visualizer to derive the image endmembers, which are represented as a colored crosses in the n-D scatter plot shown in Figure 4C. The endmembers that showed a match of greater than 75% with spectral library of AgPVP55 are shown in Figure 4D and only these endmembers were considered to represent the Ag-PVP55 in the mesocosm water. SAM and MTMF analyses revealed that Ag-PVP55 nanoparticles were present in the water at an abundance of 2.597% and 2.278%, respectively. The percentage abundance of individual endmembers is summarized in Table 1. The similarity in percentage abundances from two independent mapping methods revealed that the relative abundances of AgNPs in waters samples can be estimated reliably using hyperspectral imagery analysis. Ag-PVP55 Nanoparticles in Clarified Wastewater. Hyperspectral image of 2.5 ppm Ag-PVP55 in clarified wastewater sample, shown in Figure 5A, was analyzed exactly the same way as described for Ag-PVP55 in mesocosm water. The RGB image of first three MNF bands is shown in Figure 5B. This image shows most coherent images having spectrally distinct signatures, which were identified by the presence of

was observed in the form of secondary peaks in blue region of the spectra. Raman spectral analysis of Ag Bare and Ag-Cit showed significant similarities confirming that citrate molecules were present in close proximity but likely not bonded to the surface. Small differences in spectral peaks of Ag-(GA, PVP10, and PVP55) were attributed to chemical composition of the coatings but major features remained similar to Ag Bare spectrum. VNIR and Raman spectra highlighted the differences in the chemical composition of each coating by giving rise to new spectral features that were unique to each type of coating. Due to the high spectral resolution (1.5 nm) of VNIR wavelengths of the hyperspectral imaging system molecular level interactions between the nanoparticle surface and chemical moieties can be obtained under environmentally relevant conditions. Ag-PVP55 Nanoparticles in Mesocosm Water. Hyperspectral image analysis of 2.5 ppm Ag-PVP55 suspended in water obtained from a simulated wetland ecosystem (mesocosm water) revealed that detection and estimation of relative abundance is possible (Figure 4). Detailed steps involved in image processing are shown in SI Figure S2 (A−J) and only typical images are shown in Figure 4. Figure 4A and B shows the spatial distribution of Ag-PVP55 nanoparticles (colored dots) in mesocosm water sample, which 10085

dx.doi.org/10.1021/es204140s | Environ. Sci. Technol. 2012, 46, 10081−10088

Environmental Science & Technology

Article

Ag-PVP55 endmembers and their percentage abundances in the wastewater sample are shown in the Figure 5C. Figure 5D shows the map of the endmembers in the hyperspectral image of the sample. Comparison of MTMF map with MNF or hyperspectral image showed that the Ag-PVP55 endmembers were identified with good accuracy where almost all spectrally active colored dots were accounted with good match across all the images (Figure 5A, B, and D). The total percentage abundance of Ag-PVP55 in the wastewater sample was 1.825%. Although the same concentration of Ag-PVP55 was added to both wastewater and mesocosm water the difference in percentage abundance was attributed to the differences in composition of the water and Ag-PVP55 sorbing to various types of colloids, which ultimately determines the nature of association of Ag-PVP55 nanoparticles with natural colloids. Note that initial concentration of Ag-PVP55 in mesocosm water and wastewater was at 2.5 mg/L (i.e., ∼0.25%(w/w) on mass basis) but the hyperspectral image analysis revealed that 2.2−2.5% and 1.8% (pixel/pixel) as the Ag abundances (an order of magnitude higher), respectively. The differences between the initial and estimated concentrations are attributed to the immediate association of AgNPs to colloids and the water chemistry, which plays a significant role determining the spatial distribution and concentration of AgNPs in mesocosm

Table 1. Number of Pixels Classified As Each Endmember and Percentage of Total for the Spectral Angle Mapper (SAM) and Mixture Tuned Matched Filtering (MTMF) Methods; All the Endmembers Correspond to Ag-PVP55 Nanoparticles endmember unclassified endmember endmember endmember endmember endmember endmember endmember endmember endmember endmember endmember endmember

1 2 3 4 5 6 7 8 9 10 11 12

SAM results

MTMF results

no. of points

% total

no. of points

% total

155 844 19 11 14 232 33 4 44 15 13 28 3730 13

97.403 0.012 0.007 0.009 0.145 0.021 0.003 0.028 0.009 0.008 0.018 2.331 0.008

156 355 12 6 1 187 35 4 44 10 15 22 3304 5

97.722 0.007 0.004 0.001 0.117 0.022 0.003 0.028 0.006 0.009 0.014 2.065 0.003

various colored dots, the brighter or lighter ones represent spectrally active or inactive materials, respectively.

Figure 5. Mixture tuned matched filtering (MTMF) analysis of hyperspectral image of Ag-PVP55 nanoparticles in wastewater is shown here. (A) AgPVP55 dosed in wastewater. (B) MNF image of Ag-PVP55 dosed in wastewater. (C) Image endmember and their relative percentage abundances in the image scene. (D) Map of endmembers in the hyperspectral image. Images were acquired using 100× objective/1.3 oil iris. Scale: 2 cm = 200 μm. 10086

dx.doi.org/10.1021/es204140s | Environ. Sci. Technol. 2012, 46, 10081−10088

Environmental Science & Technology

■

water and wastewater. Despite the differences, the image acquisition and analysis revealed that it is possible to map spatially and estimate the relative abundances of nanoparticles (e.g., Ag-PVP55) or mixtures of nanoparticles in ultrapure and complex water samples without any requirement for sample preparation. Unlike metal-based nanoparticles, it is difficult to analyze carbonaceous nanoparticles in the complex waters due to presence of large quantities of carbon-based colloids in the water. For metal based nanoparticles the sizes close to 10 nm or higher were detected (Figure 1) at environmentally relevant concentrations (2.5 mg/L) but most of the nanoparticle dispersions used in this study were polydisperse, which made it difficult to measure the number concentrations accurately. Furthermore, although the color of the green silver nanoparticles (Figure 3c) was similar but the boundary of the green “halos” was not exactly spherical, which is attributed to the differences in types of coating materials (both chemically and structurally distinct) used. The methodology and the findings of this study suggests that environmental water samples can be easily analyzed under sample’s native conditions wherein the nanoparticles and natural colloids were present in partially wet condition without destroying the sample, which is often not achieved with analysis using electron, ion, or X-ray beams. Furthermore, since the pixel size is 25 nm (at 100× magnification) and single or different types of individual nanoparticle(s) occupying each pixel in field-of-view 350−400 μm can be analyzed to obtain the complete spectral information at a spectral resolution of 1.5 nm and the spatial distribution. In case of mixtures of nanoparticle (s) and natural colloids present in a pixel, their composite spectra can be unmixed by classification algorithms, such as mixture tuned matched filtering techniques, allowing the identification and relative abundance of individual materials occupying any pixel in the image. Since the environmental water samples are complex the majority of nanoparticles usually exist as heteroaggregates and homoaggregates to a minor extent, therefore, the hyperspectral imagery would only analyze samples interacting with light in the field-of-view. As with any optical microscopy based approach it is not possible to estimate the accurate concentration of polydisperse nanoparticles in complex water due to interferences from natural background colloids but it is possible to obtain local relative abundance of nanoparticles in the sample, which maybe sometimes be higher than the actual concentration (Figures 4 and 5). This was demonstrated from the analysis of Ag-PVP55 in mesocosm water and wastewater, which showed that the local relative abundance of AgNPs to be 10-fold higher than the actual sample concentration. Analysis of AgNPs with and without coatings showed that the differences in nanoparticle spectral features and changes in hydrodynamic size caused by the coating materials can be discerned using VNIR spectra. In ultrapure water the method detection limit for AgNPs is ∼25 μg/L. Further studies are underway to determine nanoparticle size and number concentration using a microscopy slide with a marked grid and establish the detection limits in environmentally relevant matrices. The findings suggest that enhanced Darkfield microscopy based hyperspectral image analysis could be a very promising tool to study the fate and transformation of engineered nanoparticles in complex water samples. Hyperspectral imagery can serve as preliminary detection, characterization, and semiquantification tool for analyzing engineered nanoparticles in unaltered water samples.

Article

ASSOCIATED CONTENT

S Supporting Information *

Size information and VNIR characteristics of nanoparticles, more details on image processing and analysis, details of AgNP analysis in simulated surface water. This material is available free of charge via the Internet at http://pubs.acs.org.

■

AUTHOR INFORMATION

Corresponding Author

*Phone 919-660-5292; e-mail: [email protected]. Notes

The authors declare no competing financial interest.

■

ACKNOWLEDGMENTS This work was funded through the Center for the Environmental Implications of NanoTechnology (CEINT) by the NSF and the EPA under NSF Cooperative Agreement Number EF0830093. Any opinions, findings, conclusions or recommendations expressed in this material are those of the author(s) and do not necessarily reflect the views of the NSF or the EPA. This work has not been subjected to EPA review and no official endorsement should be inferred. We thank Yingwen Cheng for providing us with Ag-PVP55 and Ag-GA, and Dr. Stella Marinakos for providing us with Ag-Cit.

■

REFERENCES

(1) Hassellöv, M.; Kaegi, R. In Environmental and Human Health Impacts of Nanotechnology; Lead, J. R., Smith, E., Eds.; Blackwell Publishing Ltd.: Chichester, West Sussex, 2009; pp 211−266. (2) Wiesner, M. R.; Bottero, J. Y. Environmental Nanotechnology: Applications and Impacts of Nanomaterials; McGraw Hill: New York, 2007. (3) Hendren, C. O.; Mesnard, X.; Droge, J.; Wiesner, M. R. Estimating production data for five engineered nanomaterials as a basis for exposure assessment. Environ. Sci. Technol. 2011, 45 (7), 2562− 2569. (4) Baer, D. R.; Amonette, J. E.; Engelhard, M. H.; Gaspar, D. J.; Karakoti, A. S.; Kuchibhatla, S.; Nachimuthu, P.; Nurmi, J. T.; Qiang, Y.; Sarathy, V.; Seal, S.; Sharma, A.; Tratnyek, P. G.; Wang, C.-M. Charaterization challenges for nanomaterials. Surf. Interface Anal. 2008, 40, 529−537. (5) Tiede, K.; Boxall, A. B. A.; Tear, S. P.; Lewis, J.; David, H.; Hassellöv, M. Detection and characterization of engineered nanoparticles in food and the environment. Food Addit. Contam. 2008, 25, 795−821. (6) Smith, D. J.; Petford-Long, A. K.; Wallenberg, L. R.; Bovin, J. O. Dynamic atomic-level rearrangement in small gold particles. Science 1986, 233, 872−875. (7) Yacamán, M. J.; Ascencio, J. A.; Liu, H. B.; Gardea-Torresdey, J. Structure shape and stability of nanometric sized particles. J. Vac. Sci. Technol., B: Microelectron. Nanometer Struct.–Process., Meas., Phenom. 2001, 19, 1091. (8) Zhao, J. P.; Chen, Z. Y.; Cai, X. J.; Rabalais, J. W. Annealing effect on the surface plasmon resonance absorption of a Ti-SiO2 nanoparticle composite. J. Vac. Sci. Technol., B: Microelectron. Nanometer Struct.– Process., Meas., Phenom. 2006, 24, 1104−1108. (9) Jung, Y. J.; Homma, Y.; Vajtai, R.; Kobayashi, Y.; Ogino, T.; Ajayan, P. M., Straightening suspended single walled carbon nanotubes by ion irradiation. Nano Lett. 2004, 4, (1109). (10) Baer, D. R.; Engelhard, M. H.; Gaspar, D. J.; Matson, D. W.; Pecher, K. H.; Williams, J. R.; Wang, C. M., Challenges in applying surface analysis methods to nanoparticles and nanostructured materials. J. Surf. Anal. 2005, 12, (2). (11) Karakoti, A. S.; Kuchibhatla, S. V. N. T.; Suresh Babu, K.; Seal, S. Direct synthesis of nanoceria in aqueous polyhydroxyl solutions. J. Phys. Chem. C 2007, 111, 17232−17240. 10087

dx.doi.org/10.1021/es204140s | Environ. Sci. Technol. 2012, 46, 10081−10088

Environmental Science & Technology

Article

(12) Kuchibhatla, S. V. N. T.; Karakoti, A. S.; Seal, S. Hierarchical assembly of inorganic nanostructure building blocks to octahedral superstructures  a true template-free self-assembly. Nanotechnology 2007, 18, 075303−307. (13) Bayer, M.; Hawrylak, P.; Hinzer, K.; Fafard, S.; Korkusinski, M.; Wasilewski, Z. R.; Stern, O.; Forchel, A. Coupling and entangling of quantum states in quantum dot molecules. Science 2001, 291 (5503), 451−453. (14) Norman, T. J., Jr.; Grant, C. D.; Magana, D.; Zhang, J. Z. Near infrared optical absorption of gold nanoparticle aggregates. J. Phys. Chem. B 2002, 106 (28), 7005−7012. (15) Reinhard, B. M.; Siu, M.; Agarwal, H.; Alivisatos, A. P.; Liphardt, J. Calibration of dynamic molecular rulers based on plasmon coupling between gold nanoparticles. Nano Lett. 2005, 5 (11), 2246−2252. (16) Beach, J. Hyperspectral Imaging. BioOptics World 2009, 3. (17) Grahn, H.; Geladi, P. Techniques and Applications of Hyperspectral Image Analysis; John Wiley and Sons Ltd: Chichester, West Sussex, 2009; p 361. (18) Clark, R. N.; Swayze, G. A. Mapping minerals, amorphous materials, environmental materials, vegetation, water, ice, and snow, and other materials: The USGS Tricorder Algorithm. In Summaries of the Fifth Annual JPL Airborne Earth Science Workshop; JPL Publication 95−1, 1995; Vol. 1, pp 39−40. (19) Clark, R. N.; Swayze, G. A.; Gallagher, A. Mapping the mineralogy and lithology of Canyonlands, Utah with imaging spectrometer data and the multiple spectral feature mapping algorithm. In Summaries of the Third Annual JPL Airborne Geoscience Workshop; JPL Publication 92−14, 1992; Vol. 1, pp 11−13. (20) Meyer, J. N.; Lord, C. A.; Yang, X. Y.; Turner, E. A.; Badireddy, A. R.; Marinakos, S. M.; Chilkoti, A.; Wiesner, M. R.; Auffan, M. Intracellular uptake and associated toxicity of silver nanoparticles in Caenorhabditis elegans. Aquat. Toxicol. 2010, 100, 140−150. (21) Wang, L.; Castranova, V.; Mishra, A.; Chen, B.; Mercer, R. R.; Schwegler-Berry, D.; Rojanasakul, Y. Dispersion of single-walled carbon nanotubes by a natural lung surfactant for pulmonary in vitro and in vivo toxicity studies. Part. Fibre Toxicol. 2010, 19 (7), 31. (22) Weinkauf, H.; Brehm-Stecher, B. F. Enhanced Darkfield microscopy for rapid artifact-free detection of nanoparticle binding to Candida albicans cells and hyphae. Biotechnol. J. 2009, 4, 871−879. (23) Zhang, Q.; Hitchins, V. M.; Schrand, A. M.; Hussain, S. M.; Goering, P. L. Uptake of gold nanoparticles in murine macrophage cells without cytotoxicity or production of pro-inflammatory mediators. Nanotoxicology 2011, 5, 284−295. (24) Laborda, F.; Jimenez-Lamana, J.; Bolea, E.; Castillo, J. R. Selective identification, characterization and determination of dissolved silver(i) and silver nanoparticles based on single particle detection by inductively coupled plasma mass spectrometry. J. Anal. At. Spectrom. 2011, 26 (7), 1362−1371. (25) Lee, K. J.; Nallathamby, P. D.; Browning, L. M.; Osgood, C. J.; Xu, X. H. N. In vivo imaging of transport and biocompatibility of single silver nanoparticles in early development of zebrafish embryos. Acs Nano 2007, 1 (2), 133−143. (26) Li, Y.; Jing, C.; Zhang, L.; Long, Y. T. Resonance scattering particles as biological nanosensors in vitro and in vivo. Chem. Soc. Rev. 2012, 41 (2), 632−642. (27) Prikulis, J.; Svedberg, F.; Kall, M.; Enger, J.; Ramser, K.; Goksor, M.; Hanstorp, D. Optical spectroscopy of single trapped metal nanoparticles in solution. Nano Lett. 2004, 4 (1), 115−118. (28) Yu, X.; Araki, Y.; Iwami, K.; Umeda, N. Measurement of nanoparticle sizes by conventional optical microscopy with standing evanescent field illumination. Opt. Lett. 2008, 33 (23), 2794−2796. (29) Vainrub, A.; Pustovyy, O.; Vodyanoy, V. Resoltuion of 90 nm (x/5) in an optical transmission microscope with an annular condenser. Opt. Lett. 2006, 31 (19), 2855−2857. (30) ENVI Advanced Spectral Analysis. http://www.exelisvis.com. (31) Hua, M.; Novob, C.; Funstonb, A.; Wangc, H.; Stalevad, H.; Zouc, S.; Mulvaneyb, P.; Xiae, Y.; Hartlandd, G. V. Dark-field microscopy studies of single metal nanoparticles: understanding the

factors that influence the linewidth of the localized surface plasmon resonance. J. Mater. Chem. 2008, 18 (17), 1949−1960. (32) Lee, K. J.; Nallathamby, P. D.; Browning, L. M.; Osgood, C. J.; Xu, X.-H. N. In vivo imaging of transport and biocompatibility of single silver nanoparticles in early development of zebrafish embryos. ACS Nano 2007, 1 (2), 133−143. (33) Bohren, C.; Huffman, D. Absorption and Scattering of Light by Small Particles; Wiley: New York, NY, 1983. (34) Mulvaney, P. Surface plasmon spectroscopy of nanosized metal particles. Langmuir 1996, 12, 788−800.

10088

dx.doi.org/10.1021/es204140s | Environ. Sci. Technol. 2012, 46, 10081−10088