Development and application of a digestion-Raman analysis approach for studying multiwall carbon nanotube uptake in lettuce

Kamol K. Das a, Yaqi You *a, Miguel Torres b, Felipe Barrios-Masias b, Xilong Wang c, Shu Tao c, Baoshan Xing d and Yu Yang *a
aDepartment of Civil and Environmental Engineering, University of Nevada Reno, MS258, 1664 N. Virginia Street, Reno, Nevada 89557, USA. E-mail:;
bDepartment of Agriculture, Nutrition & Veterinary Science, University of Nevada Reno, MS202, 1664 N. Virginia Street, Reno, Nevada 89557, USA
cLaboratory for Earth Surface Processes, College of Urban and Environmental Sciences, Peking University, Beijing 100871, China
dStockbridge School of Agriculture, University of Massachusetts Amherst, 410 Paige Laboratory, Amherst, MA 01003, USA

Received 7th November 2017 , Accepted 11th February 2018

First published on 13th February 2018

With increasing production and use of carbon nanotubes (CNTs) and their inevitable release during the life cycle of CNT-based products, these engineered nanomaterials are likely to accumulate in environmental compartments such as wastewater and biosolids, sediments, and biosolids-amended soils. Subsequent uptake of CNTs by agricultural crops could increase the risk of human exposure through the food chain. Unambiguous detection of CNTs in crop plants is essential for food safety assessment. In this study, we developed a method for the detection of multiwall CNTs (MWCNTs) in tissues of lettuce (Lactuca sativa L.), coupling digestion and Raman analysis. Five digestion reagents, including sulfuric acid, hydrochloric acid, nitric acid, ammonium hydroxide, and hydrogen peroxide, were examined. Nitric acid showed the best performance, removing 98–99% leaf/stem/root biomass (dry weight) and minimizing matrix background signals that can interfere with MWCNT Raman signals. Application of nitric acid digestion-Raman analysis to spiked lettuce tissues suggested a detection limit of 25 mg kg−1 dry weight or lower. We then applied this method to lettuce plants grown hydroponically with 0, 5, 10, and 20 mg L−1 pristine (p-) or carboxyl-functionalized (c-) MWCNT. Both p-MWCNT and c-MWCNT were detected in the root, stem, and leaf tissues of most exposed lettuce plants, indicating uptake and translocation of both MWCNTs in this edible plant. Comparisons of the plants grown with 20 mg L−1 p-MWCNT or c-MWCNT suggested that carboxylation facilitated uptake and translocation of MWCNT in lettuce. Our results demonstrated that nitric acid digestion in conjunction with Raman analysis is an effective approach for detecting MWCNTs in food crops, contributing to the potential development of new analytical platforms for studying the environmental fate of CNTs in the soil–plant system and human exposure through the food chain.

Environmental significance

Caron nanotubes (CNTs) are currently used in a broad range of applications, but concerns have been raised about their release into the environment and subsequent uptake by agricultural plants in contaminated environments, which increases the risk of human exposure through the food chain. Currently, techniques for detecting CNTs in biological samples are limited, making it difficult to study the environmental fate of and human exposure to CNTs. The present study developed a nitric acid digestion-Raman analysis approach able to remove >98% plant biomass and reliably detect multiwall CNTs (MWCNTs) at concentrations down to 25 mg kg−1 dry weight of plant tissues. Application of this analytical approach allowed identification of uptake of pristine and carboxyl-functionalized MWCNTs by lettuce (Lactuca sativa L.) under hydroponic conditions. Analysis of different parts of the exposed plants revealed translocation of both MWCNTs from roots to leaves and suggested an effect of carboxylation on translocation. Overall, this approach was demonstrated to be effective for analyzing trace amounts of MWCNTs in large volumes of plant tissues. It could be useful to the development of new analytical platforms for studying the environmental fate of CNTs and other carbonaceous nanomaterials in the soil–plant system as well as human exposure to those nanomaterials through the food chain, with broad implications on food safety and the sustainable development of nanotechnology.


There has been a rapid increase in the production of engineered nanomaterials including carbon nanotubes (CNTs) during the past decades.1–3 Multiwall carbon nanotubes (MWCNTs), with an estimated global production capacity of 9400 tons in 2015, drive the production growth of carbonaceous nanomaterials.2 Widespread application of CNTs in both industry and consumer products will lead to their release to and accumulation in the environment. It is estimated that more than 50% of the released CNTs will enter the soil, resulting in an annual release of 20–40 tons to soils by 2030 and a concentration of 0.01–3 μg kg−1 soil.4–6

Agricultural plants can take up CNTs from contaminated soils, posing a risk to humans via the food chain exposure.7–10 A recent study using 14C-labeled MWCNTs showed that more than 1.25 mg kg−1 MWCNTs could accumulate in leaves of Arabidopsis, Oryza sativa (rice), Zea mays (maize), and Glycine max (soybean), when these plants were grown in hydroponic systems with 2.25 mg L−1 MWCNTs.7 Surface chemistry of CNTs, among many other factors, can influence plant uptake and translocation of CNTs.11 Pristine MWCNT (p-MWCNT) and carboxyl-functionalized MWCNT (c-MWCNT) showed different dispersion and accumulation patterns in maize tissues and cells.8 Interactions between CNTs and soil components also greatly influence plant uptake of CNTs.9,11,12 For example, one study showed insignificant uptake of natural organic matter-suspended MWCNT by rice plants in petri dishes.13 It is still unclear how CNTs enter the plant root system and are subsequently translocated to the aerial parts of the plant.

A complete understanding of CNT uptake by agricultural plants is largely hindered by technical difficulties in detecting and quantifying CNTs in complex matrices such as biological tissues.14 A variety of methods have been applied to biological samples, including transmission electron microscopy (TEM), ultraviolet-visible-near infrared (UV-vis-NIR) spectroscopy, near-infrared fluorescence (NIRF) spectroscopy, inductively coupled plasma-mass spectrometry (ICP-MS) for trace metal catalyst impurities, thermal gravimetric analysis-mass spectrometry (TGA-MS), microwave-radiation, programmed thermal analysis (PTA), among others.8,14–19 Each of these techniques has its advantages and shortcomings. For example, TEM allows direct observation of CNTs in plant tissues without destruction of a sample's morphology;8 however, TEM is limited in detecting or quantifying CNTs unambiguously given CNTs' morphological heterogeneity, lack of distinct electron diffraction patterns, and low concentrations in real samples.14 A microwave method developed by Irin et al. (2012) has shown promise for analyzing CNTs in biological samples based on the thermal response of CNTs to microwave radiation, but requires custom built equipement.18

Raman spectroscopy has been successfully used to analyze CNTs in animal tissues,20 and serves as a promising technique for detecting CNTs in plant tissues.21 However, structural cell wall biopolymers of plants, such as lignin and cellulose, can substantially influence the interpretation of CNT Raman spectra.22,23 For example, the Raman spectrum of lignin displays a strong peak at ∼1602 cm−1,24 close to CNTs' peak at 1570–1580 cm−1. It also shows moderate peaks and multiple shoulders around 1250–1400 cm−1, a range encompassing CNTs' peaks at 1330–1340 cm−1.25 The Raman spectrum of cellulose has moderate peaks and shoulders within 1300–1500 cm−1.24,25 Additionally, the presence of trace amounts of CNTs in large volumes of plant tissues makes it critical to sample the right spot. Therefore, further efforts are needed to mitigate matrix interferences and achieve the full potential of Raman spectroscopy for detecting and quantifying CNTs in plant tissues.

To address this challenge, we aimed to develop an analytical approach, incorporating optimized plant material digestion with Raman spectroscopic analysis of CNTs. Lettuce (Lactuca sativa L.) was chosen as a model agricultural plant. A variety of reagents were tested for their effectiveness in digesting plant tissue materials and influence on Raman signals of CNTs. The developed approach was then applied to study CNT uptake by lettuce under hydroponic conditions. Plants were exposed to p-MWCNT or c-MWCNT at different concentrations. At harvest, a number of plant physiological responses were determined, and plant tissues were analyzed for the presence of MWCNTs. By comparing results from experiments using p-MWCNT and c-MWCNT, we also explored the role of surface carboxylation in CNT uptake and translocation in lettuce. In short, our analytical approach provides a base for fully utilizing Raman spectroscopy in studying the environmental fate of CNTs in the soil–plant system.

Materials and methods

Materials of MWCNTs and suspension preparation

The research-grade p-MWCNT (NC3150) and c-MWCNT (NC3151) were obtained from Nanocyl ( (average diameter of 9.5 nm, average length of <1.0 μm for both types). Carbon purity is >95.0% for both MWCNTs, with c-MWCNT having an >10% oxygen surface concentration.26 MWCNT-containing stock suspensions (50 mg L−1) were prepared as previously reported.27 Briefly, MWCNTs were suspended in autoclaved Milli-Q water and sonicated at low power for 12 h (Branson Ultrasonic 2510, 100 W at 40 kHz). Using this protocol, c-MWCNT was well dispersed while p-MWCNT was not (Fig. S1A in the ESI). Detailed physicochemical properties of as-received and sonication-suspended MWCNTs are listed in Table S1.

Plant cultivation and exposure experiments

Lettuce (L. sativa, cv. Black Seeded Simpson) seeds were purchased from Botanical Interest, Inc. (Colorado, USA) and rinsed three times with Milli-Q water before use. Rinsed lettuce seeds were transferred into dampened rockwool to germinate in the dark at room temperature in a greenhouse for 4 days. Lettuce seedlings were grown in the greenhouse under natural conditions (30/15 °C (day/night), 15–63% daily relative humidity, natural light). Four-week-old healthy seedlings of similar size were used in the exposure experiments.

MWCNT exposure experiments were performed under hydroponic conditions for 18 days in the greenhouse. The Hoagland solution (Sigma-Aldrich Hoagland No. 2; containing 115.03 mg NH4H2PO4, 2.86 mg H3BO3, 656.4 mg Ca(NO3)2, 0.08 CuSO4·5H2O, 5.32 mg Fe2(C4H4O6)3·2H2O, 240.76 mg MgSO4, 1.81 mg MnCl2·4H2O, 0.016 mg MoO3, 606.6 mg KNO3, and 0.22 mg ZnSO4·7H2O per liter) was used as the medium. Lettuce seedlings were grown in test vials filled with 100 mL of 10% Hoagland solution (pH adjusted to 6.0–6.7) containing either p-MWCNT or c-MWCNT at 0, 5, 10, or 20 mg L−1. Aggregation of both MWCNTs was observed in the medium solution (Fig. S1B), and no special treatment was conducted. Four to five seedling replicates were used for each exposure condition. The vials were wrapped with aluminum foil and sealed with parafilm at the top. The culture solution was continuously aerated with an air pump, and 10% Hoagland solution was added into each vial every day to compensate for evapotranspiration loss that was determined by monitoring weight loss of vials. Cumulative evapotranspiration was calculated for each plant based on daily weight of each vial during the entire exposure period.

Plant physiological responses

At harvest, the plants were rinsed three times with Milli-Q water to remove attached particles, after which the plants were dissected aseptically into roots, stems, and leaves. Root tissues were further rinsed with Milli-Q water to remove residual particles, and cleaned root tissues were spread out, scanned, and analyzed with the WinRHIZO image analysis system (Regent Instruments Inc., Quebec, Canada). Root system architecture, including length distribution and number of lateral roots, was determined for each plant.28,29

Leaf cell damage was assessed by measuring electrolyte leakage using a conductivity method.29,30 Pre-weighed leaf tissues were added into a covered test tube containing 10.0 mL of Milli-Q water and shaken at room temperature and 120 rpm in the dark. After 24 h of shaking, the electrical conductivity of the solution (L0) was measured using a portable conductivity meter HACH HQ40d (Colorado, USA). Samples were then autoclaved at 121 °C for 20 min and equilibrated at room temperature, after which the final electrical conductivity (Lt) was measured. Membrane leakage was calculated using the following equation: Membrane leakage (%) = (L0/Lt) × 100%.

All plant tissues were then oven-dried at 80 °C overnight before dry mass measurement.31 Total dry biomass was determined for each plant by summing up the dry mass of roots, stems, and leaves.

Approach of plant digestion and Raman analysis for detecting MWCNTs

For method development, lettuce (L. sativa, cv. Bionda Ricciolina) plants were purchased from a local nursery and dissected into leaves, stems, and roots. The tissues were washed with Milli-Q water and oven-dried at 80 °C overnight, and then ground into fine powders. The leaf, stem and root powders were spiked with p-MWCNT or c-MWCNT at a final concentration of 2500, 250, 50, or 25 mg kg−1 dry weight. While MWCNTs spiked into plant materials may not well represent MWCNTs taken up by and translocated in living plants, given complex plant–nanotube interactions,9 this spiking procedure served well for method development. Five reagents, sulfuric acid (H2SO4, 18.4 M), hydrochloric acid (HCl, 12 M), nitric acid (HNO3, 15.8 M), hydrogen peroxide (H2O2, 9.8 M), and ammonium hydroxide (NH4OH, 14.8 M), were used for digestion of plant materials. An aliquot of each spiked plant tissue (10 mg) was mixed with one of the digestion reagents (1 mL) in a clean Corex tube, digested at 62 °C for 12 hours, and then neutralized with 2 M NaOH solution at room temperature, if necessary. These temperature and time duration were determined based on published digestion protocols, considering both the effectiveness of matrix digestion and the minimal loss of MWCNTs during digestion.32,33 After centrifugation of the resulting mixture at 14[thin space (1/6-em)]500g for 2 min, the supernatant was pipetted out and the remaining residues were re-suspended in Milli-Q water for Raman analysis.

Around 50 μL of the suspension was loaded onto a 70% ethanol-cleaned microscope slide and air-dried at room temperature. A DXR Raman microscope (Thermo Scientific) was used for analysis (excitation of 532 nm and high resolution of 2 cm−1 FWHM). Spectra over the range of 50–1870 cm−1 were collected under a 50× objective, at 10 mW laser power, with an 8 s exposure time. The spot size of the laser was set at 0.6 μm × 0.6 μm. Considering potential spatial heterogeneity of tested samples, Raman spectra were collected from ≥5 randomly selected regions within a total area of 200 μm × 200 μm for each sample. The signal-to-noise ratio was ≥5 for all spectra obtained and analyzed.

Two types of controls were included in method development. To assess the interference from plant materials, blank lettuce tissue powders were digested and analyzed for background Raman signals. This control was also used to determine biomass removal efficiency for each digestion reagent. Briefly, the suspension from each digestion treatment was filtered through a pre-weighed 0.45 μm membrane filter, and the membrane filter was oven-dried at 105 °C for 12 hours before gravimetric analysis. To test the influence of digestion processes on Raman signals of MWCNTs, pure materials of both MWCNTs were subject to the same analytical procedures. This control was also used to assess mass loss of MWCNTs after each digestion process by gravimetric analysis, which proved that 60–70% of MWCNTs remained as precipitates after digestion.

After determining the best-suited digestion reagent and the optimal digestion process, we applied the developed analytical approach to lettuce plants harvested from the exposure experiments. Leaf, stem, and root tissues from harvested plants were thoroughly rinsed, oven-dried at 80 °C overnight, and ground into fine powders. An aliquot of each plant tissue sample (around 4 mg dry weight) was mixed with 0.4 mL of the best-suited digestion reagent (HNO3) in a Corex tube, digested at 62 °C for 12 hours, neutralized with 2 M NaOH solution at room temperature. After centrifugation, the precipitates were re-suspended in Milli-Q water and analyzed by Raman as above.

TEM analysis

Plant tissues spiked with MWCNTs for method development and fresh plants harvested from the exposure experiments were further observed using TEM.34,35 The detailed processes are described in the ESI.

Data analysis and statistics

Statistical analysis was performed with R (3.2.4) and Excel 2010. For normally distributed Raman data, Levene's test was conducted to check homogeneity of variance, after which data were compared using the Student's t-test or Welch's t-test. For plant physiological data not normally distributed and with small sample sizes, comparison was performed using the Mann–Whitney U test.

Results and discussion

Developing an approach for detection of MWCNTs in plant tissues

Digestion of plant materials. Acidic, alkaline, and oxidative reagents (e.g., H2SO4, HCl, HNO3, NH4OH, and H2O2) can hydrolyze and dissolve plant cell wall biopolymers such as cellulose, hemicellulose, and lignin, thereby reducing Raman background interference.33,36,37 After 12 h digestion of blank leaf or root tissues, transparent supernatants were obtained from HNO3 treatment, while treatments with other acids (H2SO4 or HCl) resulted in dark suspensions, and treatments with H2O2 and NH4OH resulted in white and brown residues, respectively (Fig. S2). These observations were similar to a previous study where animal tissues were digested using the same chemical reagents.32 For leaf tissues, biomass removal followed the order of HNO3 (97.4 ± 0.9%), H2O2 (79.0 ± 2.7%), HCl (76.2 ± 2.6%), NH4OH (72.7 ± 9.8%), and H2SO4 (68.5 ± 9.2%) (Fig. 1A). Similar results were obtained for stem tissues, with biomass removal being 95.5 ± 4.7%, 79.3 ± 1.7%, 78.4 ± 9.3%, 76.0 ± 3.2%, and 63.8 ± 2.3% for HNO3, NH4OH, H2O2, HCl, and H2SO4 digestion, respectively (Fig. 1A). These observations were consistent with previous reports of HNO3 being able to oxidize the majority of plant materials and reduce the mass by 80.0–98.5%.33,36 Specifically, HNO3 is more oxidative than other strong acids and thus more effective in dissolving lignin, whereas H2SO4 and HCl can hydrolyze cellulose and hemicellulose.33,37,38 Another study also reported HNO3 as the most effective among the five reagents in animal tissue digestion,32 reflecting the usefulness of HNO3 as a versatile digestion reagent.
image file: c7en01047h-f1.tif
Fig. 1 (A) Residual mass of ∼10 mg (dry weight) blank lettuce leaf or stem tissues after 12 h digestion using five different reagents. Data represent means and standard deviations from two independent experiments. (B) Raman spectra of blank leaf and root tissues digested with H2SO4 (green) or HNO3 (grey). Each spectrum represents the average of five scans obtained from different sample areas. Raman spectra of other digestion treatments are in Fig. S3.

Raman analysis further indicated the effectiveness of HNO3 digestion in matrix background removal (Fig. 1B and S3). The residual materials from treatments other than HNO3 had several peaks within the range of 800–1800 cm−1, showing strong signals at 1574 cm−1, 1562 cm−1, 1590 cm−1, and 1575 cm−1 after H2SO4, HCl, H2O2, and NH4OH treatment, respectively. These peaks within 1550–1650 cm−1 were likely attributed to lignin, its transformation products, or phenolic compounds present in plant tissues,22,25,39–41 and were also seen in the spectra of undigested plant tissues (Fig. S3E). In comparison, residual materials from HNO3 digestion showed no significant Raman signals in the range of 1550–1650 cm−1, implying a relatively complete removal of lignin and similar biopolymers by HNO3. Outside the range of 1550–1650 cm−1, a sharp peak at 1070 cm−1 was seen for the residuals from HNO3 or H2SO4 digestion as well as for these reagents alone (Fig. S3D), which could be attributed to NaNO3 or Na2SO4.42 Considering all these results, among the five tested reagents, HNO3 gave the best performance in terms of plant biomass removal, particularly for lignin components that constitute ∼2% dry weight of lettuce leaves.43 Thereby HNO3 digestion minimized matrix background Raman signals in the range of 1550–1650 cm−1, which may interfere with Raman signals of MWCNTs.

Raman analysis of spiked plant tissues. Typical Raman spectra of MWCNTs showed two distinct peaks at 1330–1340 cm−1 and 1570–1580 cm−1, which are known as D-band and G-band, respectively (Fig. S4). The G-band corresponds to the sp2-hybridized graphene structure, whereas the D-band is attributed to defects in graphene sheets, structural disorder, or amorphous carbon.44,45 Reactions with digestion reagents affected the position and intensity of the D- and G-band of MWCNTs (Fig. S4). For original MWCNTs, D- and G-band peaks were positioned at 1335–1339 cm−1 and 1569–1574 cm−1 for p-MWCNT and at 1338–1344 cm−1 and 1573–1575 cm−1 for c-MWCNT. After various digestions, the peak positions of both MWCNTs were upshifted by 10–13 cm−1 (p > 0.05 for p-MWCNT and p < 0.05 for c-MWCNT, t-test), which can be ascribed to the electron transfer from π states in MWCNTs to oxygen atoms intercalated during chemical oxidation.46 Relative intensity ratio of the D-band to the G-band (ID/IG), an indicator of structural and compositional changes in CNTs,17,27,44,45,47,48 was evaluated for individual digestion treatments. For original p- and c-MWCNT, ID/IG was measured as 1.22 ± 0.03 and 1.23 ± 0.09, respectively. ID/IG changed to 1.09 ± 0.07, 1.22 ± 0.06, 1.23 ± 0.08, 1.00 ± 0.03, and 1.15 ± 0.08 for p-MWCNT treated with H2SO4, HCl, HNO3, H2O2, and NH4OH, respectively, and to 1.31 ± 0.03, 1.35 ± 0.03, 1.22 ± 0.09, 1.08 ± 0.04, and 1.08 ± 0.04 for c-MWCNT treated with H2SO4, HCl, HNO3, H2O2, and NH4OH, respectively (p < 0.05 for most treatments, t-test). It is known that strong oxidizing reagents can cause oxidation and surface functionalization of CNTs, where the resulting amorphous carbon and/or structural defects can lead to increased ID/IG ratios.44,45 It is also reported that non-acidic treatments such as NH4OH and H2O2 can facilitate the complete removal of disordered carbon from existing defect sites but do not create additional defects in MWCNTs, which could decrease ID/IG ratios.44 Despite these oxidation-induced changes in ID/IG, none of the digestion reagents significantly diminished Raman signals of MWCNTs.

For lettuce tissues that were spiked with 2500 mg kg−1 dry weight of p-MWCNT or c-MWCNT and subsequently digested by various reagents, the characteristic G-band and D-band of MWCNTs were detected in all samples (Fig. 2 and S5). The detected G- and D-band had minor blue shifts (3–5 cm−1) as compared to pure MWCNTs digested by the same reagent (p > 0.05, t-test). Moreover, residual plant materials after certain treatments seemed to influence the G-band in MWCNT spectra (Fig. 2 and S5). H2SO4 or HCl digestion of spiked tissues yielded significantly stronger G-band signals than the same digestion of pure MWCNTs (p < 0.05, t-test). Such an increase in the G-band intensity was most likely due to influences of lignin and/or its transformation products released from broken cell walls,25,40,41 which had a peak at 1562–1574 cm−1 as shown in blank plant materials digested with H2SO4 or HCl. In contrast, HNO3, H2O2, or HN4OH treatment did not have such an effect on G-band signals of MWCNTs in spiked tissues. None of the digestion reagents seemed to strongly affect the D-band intensities of MWCNTs in spiked tissues as compared to pure MWCNTs digested in the same way. After subtraction of background signals from blank plant materials, Raman spectra of MWCNTs in spiked plant tissues were close to those of pure MWCNTs (Fig. S6). These results reflected less influence of digestion treatments on the D-band spectra than the G-band spectra and were in general consistent with our previous observations on the effectiveness of plant biopolymer removal by individual reagents.

image file: c7en01047h-f2.tif
Fig. 2 Raman spectra of lettuce tissues that were spiked with 2500 mg kg−1 dry weight of p-MWCNT (A and B) or c-MWCNT (C and D) and digested with H2SO4 (A and C) or HNO3 (B and D). Triplicate spectra of leaf, stem, and root samples are shown for each condition, along with pure MWCNT digested using the same reagent. Raman spectra from other digestion treatments are in Fig. S5. Also shown are ID/IG ratios of p-MWCNT (E) or c-MWCNT (F) in spiked and digested tissues. Error bars represent the standard deviations of triplicate samples. Asterisk indicates a significant difference between pure MWCNT and spiked tissues digested using the same reagent.

A comparison of MWCNT ID/IG ratios from spiked tissues subject to different digestion treatments allowed more quantitative assessment of the influence of each treatment on Raman signals of MWCNTs in spiked tissues (Fig. 2E and F). Compared to digested pure MWCNTs, spiked plant tissues (leaf, stem, or root) that were digested by reagents other than HNO3 displayed significantly lower (H2SO4, HCl) or higher (H2O2, NH4OH) ID/IG ratios (p < 0.05, t-test). In comparison, MWCNT ID/IG ratios from spiked tissues subject to HNO3 digestion were similar to those of pure MWCNTs subject to the same digestion. TEM imaging confirmed the presence and intactness of MWCNTs in HNO3 digestion residues of spiked plant tissues, supporting our Raman analysis results (Fig. S7). Taking together, among all the reagents examined in this study, HNO3 not only removed the majority of plant biomass (98–99%), thereby minimizing matrix interference in Raman spectra, but also had negligible effects on MWCNTs' Raman signals and intactness. Under our experimental conditions, HNO3 was the optimal digestion reagent and thus used in conjunction with Raman analysis in all subsequent experiments.

Concentration-dependent detection of MWCNTs. Lettuce leaves spiked with p-MWCNT or c-MWCNT at 250, 50 or 25 mg kg−1 dry weight were analyzed using our developed approach coupling HNO3 digestion with Raman analysis. MWCNTs were successfully detected in all the spiked leaf samples (Fig. 3). Similar to tissues spiked with 2500 mg kg−1 dry weight of MWCNTs, digestion residues of leaves spiked with lower-level MWCNTs showed minor blue shifts of both D- and G-band as compared to digested pure MWCNTs (p > 0.05, t-test), due to electron transfer from carbon atoms to oxygen atoms introduced during chemical oxidation. Based on these results, our developed approach was efficient for detecting MWCNTs at concentrations down to 25 mg kg−1 dry weight in complex plant tissues. Although we did not explicitly determine the detection limit of this approach, it is expected to be within the range of Raman-based methods in general,14 lower than the detection limit of TGA-MS method,15 but higher than the detection limits of microwave radiation, PTA, NIRF and ICP-MS methods.17–19,49 This detection limit is close to estimated CNT concentrations in biosolids and sediments (mg kg−1), but higher than those estimated in soils (μg kg−1).4,50 Therefore, future research needs to lower the detection limit of this digestion-Raman analysis approach for a reliable assessment of the environmental fate of MWCNTs in the soil–plant system. Based on the G-band or D-band intensity, the concentration of p-MWCNT in plant tissues was estimated, with limits noticed (detailed in the ESI). Specifically, any mass loss of MWCNTs during digestion could affect the quantitative detection of MWCNT in real samples.
image file: c7en01047h-f3.tif
Fig. 3 Raman spectra of leaf tissues spiked with p-MWCNT (A–C) or c-MWCNT (D–F) at 250 (A and D), 50 (B and E), or 25 (C and F) mg kg−1 dry weight. Spiked leaves were digested using HNO3. Different colors represent five replicate spectra from different sample locations in each condition.

Uptake and translocation of MWCNTs in lettuce plants

Effects of MWCNTs on the growth and physiology of lettuce. Lettuce plants were grown in hydroponic systems containing up to 20 mg L−1 MWCNT for studying MWCNT uptake and translocation. These exposure levels represented the estimated MWCNT concentration range in biosolids and sediments, but could be higher than MWCNT concentrations in soils and biosolids-amended soils.4–6,50 Exposure to p-MWCNT or c-MWCNT had varying effects on lettuce growth (detailed in the ESI). During 18 days, MWCNT exposure either enhanced or inhibited cumulative evapotranspiration, depending on the type and dose of MWCNTs (Fig. 4A). Most of the exposed plants had less dry biomass than the controls, and those exposed to 10 mg L−1 c-MWCNT had the lowest dry biomass, significantly less than control plants without exposure (p < 0.03, Mann–Whitney U test) (Fig. S9). Moreover, regardless of their types and concentrations, MWCNTs greatly affected root system architecture of lettuce plants (Fig. S10 and S11), inducing the development of lateral roots (Fig. 4B1). Such an effect was significant for the 10 mg L−1 c-MWCNT treatment (p < 0.03, Mann–Whitney U test). Both MWCNTs also resulted in increased leaf membrane leakage in most cases, and this effect was significant for 10 mg L−1 c-MWCNT (1.49 times higher than the controls, p < 0.03 in Mann–Whitney U test) (Fig. 4B2).
image file: c7en01047h-f4.tif
Fig. 4 (A1 and A2) Cumulative evapotranspiration of lettuce plants grown with p-MWCNT (A1) or c-MWCNT (A2) for 18 days. Four to five plants were used in each exposure condition. Differences between exposed and control plants were statistically insignificant (detailed in the ESI). (B1 and B2) Impacts of MWCNTs on lettuce physiology: lateral root development (B1) and leaf cell damage (B2) after 18-day growth. Box-and-whisker plot shows minimum and maximum (whisker bottom and top), first and third quartile (box bottom and top), and median (line inside box) of 4–5 lettuce plants. Asterisk indicates a significant difference between exposed samples and controls.

One possible explanation of these physiological changes would be uptake of MWCNTs by lettuce plants, followed by MWCNT-induced oxidative stress in lettuce cells.9,51,52 It is known that oxidative stress could induce up-regulation of genes involved in cell division and cell wall extension, thereby accelerating local cell division and differentiation.31,53,54 This might explain enhanced lateral root formation in all exposed plants in this study. Our observation of increased leaf cell damage in exposed plants was consistent with previous reports on MWCNT-caused cell membrane disruption and leakage of cytoplasmic content,9,29 which might in turn explain reduced transpiration and biomass in some exposed plants in this study.

Detection of MWCNT uptake and translocation in lettuce. To determine MWCNT uptake, we applied the developed analytical approach to lettuce plants grown hydroponically for 18 days. The fingerprint D- and G-band of MWCNTs were detected in most root, stem, and leaf tissues of the plants exposed to either p-MWCNT or c-MWCNT, except for leaves of the plants exposed to 20 mg L−1 p-MWCNT and stems of the plants exposed to 20 mg L−1 c-MWCNT (Fig. 5 and S12 and S13). The fact that MWCNTs were identified in roots of all exposed plants suggested that uptake occurred in lettuce for both MWCNTs and at all three exposure levels. Detection of both MWCNTs in most stem and leaf samples further suggested that in most cases both MWCNTs were translocated from roots through stems to leaves. A comparison of the resulting spectra also showed that in general p-MWCNT treated plants had larger variance in Raman intensity than c-MWCNT treated plants (Fig. 5 and S12 and S13). This might suggest higher spatial heterogeneity of p-MWCNT than c-MWCNT in lettuce plants, a phenomenon likely resulted from their different dispersion states due to surface charge.
image file: c7en01047h-f5.tif
Fig. 5 Raman spectra of MWCNT identified in the leaves of lettuce plants grown with p-MWCNT (A1–A3) or c-MWCNT (B1–B3) at a concentration of 5 mg L−1 (A1 and B1), 10 mg L−1 (A2 and B2), or 20 mg L−1 (A3 and B3). Each spectrum of exposed plants represents the average of 6–10 sample areas, with standard deviations shown in grey. MWCNT was not detected in the leaves of plants exposed to 20 mg L−1 p-MWCNT (A3).

Considering different parts of the exposed lettuce plants, roots generally displayed stronger MWCNT signals than stems or leaves, regardless of the exposure level (Fig. 5 and S12 and S13). This observation was not surprising as roots were in direct contact with hydroponic solutions containing MWCNTs and any uptake of MWCNTs could result in their accumulation in root tissues, while translocation of MWCNTs from roots to stems and leaves requires traversing multiple barriers in plants.10,52 It was also consistent with a quantitative study using 14C-labeled MWCNT, which reported higher concentrations of MWCNT in plant roots than in stems or leaves after exposure.7

Raman spectra of MWCNTs identified in the exposed roots showed minor blue shifts of the D-band and G-band as compared to digested pure MWCNTs (7–9 cm−1 for p-MWCNT and 5–10 cm−1 for c-MWCNT, p > 0.05 in t-test), similar to the observations from spiked lettuce tissues (Fig.S12 and S13). Raman spectra of MWCNTs identified in the stem and leaf tissues of the exposed plants also displayed minor blue shifts of the D-band and G-band as compared to digested pure MWCNTs (<14 cm−1 for p-MWCNT and <10 cm−1 for c-MWCNT, p > 0.05 in t-test) (Fig. 5 and S12 and S13). ID/IG values of MWCNTs in roots ranged from 1.05 ± 0.05 to 1.08 ± 0.09 for the p-MWCNT treated plants and from 1.13 ± 0.08 to 1.17 ± 0.03 for the c-MWCNT treated plants, lower than those identified in the spiked plant tissues (p < 0.05 for p-MWCNT and p > 0.05 for c-MWCNT in t-test) (Fig. S14). ID/IG values of MWCNTs in stems and leaves ranged from 0.89 ± 0.07 to 1.02 ± 0.08 for the p-MWCNT treated plants and from 0.90 ± 0.04 to 1.04 ± 0.05 for the c-MWCNT treated plants, significantly lower than those measured in the spiked tissues (p < 0.05 in t-test) (Fig. S14). These observations on the ID/IG ratio might be due to differences in tissue composition between the blank plants used for spiking and the hydroponically exposed plants, as illustrated by reduced dry biomass in the exposed plants. They might also be attributed to different dispersion states of MWCNTs in tissues of dead plants and living plants.55 Alternatively, they might suggest preferential uptake and translocation of MWCNTs with less structural defects by living plants or in situ transformation of MWCNTs due to complicated plant–nanotube interactions in living plants.9

Previous studies have reported uptake of MWCNTs by plant cells, seeds, and seedlings.9,10,31,52,53 In whole plants, carbonaceous nanomaterials might enter roots along with water uptake.13 At the organ level, an increase in root ramification, as observed in all exposed plants in this study, could provide more entry sites for MWCNTs and thereby facilitate their uptake by plants.56 Other studies also observed adsorption of nanomaterials on plant lateral roots, and their movement into the root epidermis, exodermis, cortex, and endodermis.57,58 Because lateral roots penetrate the root epidermis and reach the exterior of the vascular cylinder during differentiation, they provide a potential pathway for uptake and translocation of nanomaterials.56 The gaps in the Casparian strip at the emergence of lateral roots also serve as potential entry sites for nanomaterials.59 Additionally, root damage that likely occurs in soils may provide more sites for nanomaterials to easily enter plant root system. At the cellular level, although plant cell wall is believed to act as a protective barrier against external agents entering plant cells, individual MWCNT or aggregates of MWCNTs might traverse this barrier.16,52 Nanomaterials including MWCNTs can also create holes in cell walls8,52 and be internalized during endocytosis.9,10,60 Together, these processes might have contributed to MWCNT uptake and accumulation in roots that were observed in this study, although further investigation is required to address the exact mechanisms.

Following root uptake and penetration of epidermal cells, MWCNTs could move through tissues and reach the xylem through the apoplastic or the symplastic pathway,10 and be further translocated to the aerial parts of plants through xylem.13 By detecting MWCNTs in most leaf samples, we revealed translocation of both p-MWCNT and c-MWCNT from roots to leaves in lettuce. This was confirmed by TEM analysis of leaf tissues of the exposed plants (Fig. S15), and was also supported by the observation of increased leaf membrane leakage in most exposed plants (Fig. 4B2). Recently, Zhai et al. (2015)8 reported uptake of p-MWCNT, c-MWCNT and a positively charged MWCNT by maize and soybean plants growing in hydroponic solutions containing up to 50 mg L−1 MWCNT, and subsequent translocation of all three MWCNTs from roots to leaves. Based on TEM analysis, the authors suggested that MWCNTs rapidly passed through plant stems to leaves.8 This seemed to have occurred in our lettuce plants grown with 20 mg L−1 c-MWCNT, since MWCNT Raman signals were detected in roots and leaves of those plants but not in stems (Fig. S13A3). For lettuce plants grown with 20 mg L−1 p-MWCNT, MWCNT Raman signals were detected in roots and stems but not in leaves (Fig. 5A3), which could suggest limited translocation of p-MWCNT after uptake at this exposure level. These differences between the plants grown with 20 mg L−1 c-MWCNT or p-MWCNT likely reflected a lower mobility of p-MWCNT than c-MWCNT in lettuce. In aqueous solutions such as the Hoagland medium used here for plant cultivation, p-MWCNT aggregates more easily than negatively charged c-MWCNT. Therefore, transport of p-MWCNT through roots to the xylem could be less efficient.8,10 Aggregation of p-MWCNT in roots might also clog water transport pathways and thus hinder transport of p-MWCNT to leaves.61 In line with non-detection of MWCNT signals in leaves of the plants treated with 20 mg L−1 p-MWCNT, leaf cell leakage in those plants was not significantly different from the controls (p = 1 in Mann–Whitney U test) (Fig. 4B2). In comparison, MWCNT Raman signals were detected in all parts of the plants exposed to 5 or 10 mg L−1 of either MWCNT, probably reflecting less aggregation and thus higher mobility of MWCNTs at lower concentrations. Taken together, our results suggested that carboxylation could have an important effect in the uptake and translocation of MWCNTs in lettuce. Further research is needed to examine such an effect in other agricultural plants. More importantly, soil-based investigations under environmentally relevant conditions are warranted to extend findings from hydroponic conditions in this study. It is expected that many factors of the soil matrix, such as pH, texture, organic matter, and biomolecules, will influence the aggregation and deposition and thus plant uptake and translocation of MWCNTs in real situations.12


In this study, we have developed an effective method for detecting and analyzing MWCNTs in plant tissues, coupling optimized plant digestion using HNO3 with Raman spectroscopy. By removing >98% plant biomass in the digestion step, this method greatly reduced matrix background Raman signals and facilitated the detection of trace amounts of MWCNTs in macroscopic samples. Applying this method, we demonstrated the uptake and translocation of p-MWCNT and c-MWCNT in lettuce plants under hydroponic conditions. The detection of MWCNTs in the leaves of plants grown with 20 mg L−1 c-MWCNT but not in those exposed to 20 mg L−1 p-MWCNT further suggested that carboxylation might influence the accumulation and translocation of MWCNTs in lettuce plants.

Conflicts of interest

The authors report no conflict of interest.


This work was supported by the USDA (Grant No. 2015-67018-23120), DOE (Grant No. DE-SC0014275), University of Nevada Reno (a Startup fund to Y. Yang), and the National Natural Science Foundation of China (NSFC Grant No. 41629101). The work was also partially supported by the USDA grant 2017-69007-26309. We thank Dr. Dev Chidambaram (UNR) for giving us access to the Raman microscope; Dr. Sanjai Parikh (Department of Land, Air and Water Resources, UC Davis) for helpful discussion; Dr. Mojtaba Ahmadiantehrani (UNR) and Ms. Patricia E. Kysar (Biological Electron Microscopy Core Laboratory, Molecular and Cell Biology, UC Davis) for electron microscopy technical assistance.


  1. M. F. L. De Volder, S. H. Tawfick, R. H. Baughman and A. J. Hart, Science, 2013, 339, 535–539 CrossRef CAS PubMed.
  2. A. Parish, Production and Application of Carbon Nanotubes, Carbon Nanofibers, Fullerenes, Graphene and Nanodiamonds: A Global Technology Survey and Market Analysis, Innovative Research and Products (iRAP), Inc., Stamford, CT, 2011, p. 531 Search PubMed.
  3. A. A. Keller, S. McFerran, A. Lazareva and S. Suh, J. Nanopart. Res., 2013, 15, 1692 CrossRef.
  4. F. Gottschalk, T. Y. Sun and B. Nowack, Environ. Pollut., 2013, 181, 287–300 CrossRef CAS PubMed.
  5. T. Y. Sun, N. A. Bornhoft, K. Hungerbuhler and B. Nowack, Environ. Sci. Technol., 2016, 50, 4701–4711 CrossRef CAS PubMed.
  6. T. Y. Sun, D. M. Mitrano, N. A. Bornhöft, M. Scheringer, K. Hungerbühler and B. Nowack, Environ. Sci. Technol., 2017, 51, 2854–2863 CrossRef CAS PubMed.
  7. Q. Zhao, C. Ma, J. C. White, O. P. Dhankher, X. Zhang, S. Zhang and B. Xing, Carbon, 2017, 114, 661–670 CrossRef CAS.
  8. G. Zhai, S. M. Gutowski, K. S. Walters, B. Yan and J. L. Schnoor, Environ. Sci. Technol., 2015, 49, 7380–7390 CrossRef CAS PubMed.
  9. C. M. Rico, S. Majumdar, M. Duarte-Gardea, J. R. Peralta-Videa and J. L. Gardea-Torresdey, J. Agric. Food Chem., 2011, 59, 3485–3498 CrossRef CAS PubMed.
  10. P. Miralles, T. L. Church and A. T. Harris, Environ. Sci. Technol., 2012, 46, 9224–9239 CrossRef CAS PubMed.
  11. D. Lin, X. Tian, F. Wu and B. Xing, J. Environ. Qual., 2010, 39, 1896–1908 CrossRef PubMed.
  12. A. R. Petosa, D. P. Jaisi, I. R. Quevedo, M. Elimelech and N. Tufenkji, Environ. Sci. Technol., 2010, 44, 6532–6549 CrossRef CAS PubMed.
  13. S. Lin, J. Reppert, Q. Hu, J. S. Hudson, M. L. Reid, T. A. Ratnikova, A. M. Rao, H. Luo and P. C. Ke, Small, 2009, 5, 1128–1132 CrossRef CAS PubMed.
  14. E. J. Petersen, D. X. Flores-Cervantes, T. D. Bucheli, L. C. C. Elliott, J. A. Fagan, A. Gogos, S. Hanna, R. Kagi, E. Mansfield, A. R. M. Bustos, D. L. Plata, V. Reipa, P. Westerhoff and M. R. Winchester, Environ. Sci. Technol., 2016, 50, 4587–4605 CrossRef CAS PubMed.
  15. D. L. Plata, C. M. Reddy and P. M. Gschwend, Environ. Sci. Technol., 2012, 46, 12254–12261 CrossRef CAS PubMed.
  16. E. Wild and K. C. Jones, Environ. Sci. Technol., 2009, 43, 5290–5294 CrossRef CAS PubMed.
  17. K. Doudrick, P. Herckes and P. Westerhoff, Environ. Sci. Technol., 2012, 46, 12246–12253 CrossRef CAS PubMed.
  18. F. Irin, B. Shrestha, J. E. Canãs, M. A. Saed and M. J. Green, Carbon, 2012, 50, 4441–4449 CrossRef CAS.
  19. J. H. Bisesi, J. Merten, K. Liu, A. N. Parks, A. Afrooz, J. B. Glenn, S. J. Klaine, A. S. Kane, N. B. Saleh, P. L. Ferguson and T. Sabo-Attwood, Environ. Sci. Technol., 2014, 48, 1973–1983 CrossRef CAS PubMed.
  20. Z. Liu, C. Davis, W. Cai, L. He, X. Chen and H. Dai, Proc. Natl. Acad. Sci. U. S. A., 2008, 105, 1410–1415 CrossRef CAS PubMed.
  21. M. V. Khodakovskaya, K. de Silva, D. A. Nedosekin, E. Dervishi, A. S. Biris, E. V. Shashkov, E. I. Galanzha and V. P. Zharov, Proc. Natl. Acad. Sci. U. S. A., 2011, 108, 1028–1033 CrossRef CAS PubMed.
  22. L. Q. Chu, R. Masyuko, J. V. Sweedler and P. W. Bohn, Bioresour. Technol., 2010, 101, 4919–4925 CrossRef CAS PubMed.
  23. J. Trebolazabala, M. Maguregui, H. Morillas, A. de Diego and J. M. Madariaga, Spectrochim. Acta, Part A, 2013, 105, 391–399 CrossRef CAS PubMed.
  24. U. P. Agarwal and S. A. Ralph, Appl. Spectrosc., 1997, 51, 1648–1655 CrossRef CAS.
  25. N. Gierlinger, T. Keplinger and M. Harrington, Nat. Protoc., 2012, 7, 1694–1708 CrossRef CAS PubMed.
  26. C. M. White, R. Banks, I. Hamerton and J. F. Watts, Prog. Org. Coat., 2016, 90, 44–53 CrossRef CAS.
  27. Y. You, K. K. Das, H. Guo, C.-W. Chang, M. Navas-Moreno, J. W. Chan, P. Verburg, S. R. Poulson, X. Wang, B. Xing and Y. Yang, Environ. Sci. Technol., 2017, 51, 2068–2076 CrossRef CAS PubMed.
  28. B. I. Linkohr, L. C. Williamson, A. H. Fitter and H. M. O. Leyser, Plant J., 2002, 29, 751–760 CrossRef CAS PubMed.
  29. P. Begum, R. Ikhtiari, B. Fugetsu, M. Matsuoka, T. Akasaka and F. Watari, Appl. Surf. Sci., 2012, 262, 120–124 CrossRef CAS.
  30. D. Cui, P. Zhang, Y. Ma, X. He, Y. Li, J. Zhang, Y. Zhao and Z. Zhang, Environ. Sci.: Nano, 2014, 1, 459–465 RSC.
  31. M. H. Lahiani, E. Dervishi, J. Chen, Z. Nima, A. Gaume, A. S. Biris and M. V. Khodakovskaya, ACS Appl. Mater. Interfaces, 2013, 5, 7965–7973 CAS.
  32. K. Doudrick, N. Corson, G. Oberdorster, A. C. Eder, P. Herckes, R. U. Halden and P. Westerhoff, ACS Nano, 2013, 7, 8849–8856 CrossRef CAS PubMed.
  33. I. A. Rahman, J. Ismail and H. Osman, J. Mater. Chem., 1997, 7, 1505–1509 RSC.
  34. M. J. Karnovsky, J. Cell Biol., 1965, 27, A137–A138 Search PubMed.
  35. M. A. Hayat, Principles and Techniques of Electron Microscopy: Biological Applications, Cambridge University Press, 4th edn, 2000 Search PubMed.
  36. W. Xiao and W. W. Clarkson, Biodegradation, 1997, 8, 61–66 CrossRef CAS.
  37. N. Mosier, C. Wyman, B. Dale, R. Elander, Y. Y. Lee, M. Holtzapple and M. Ladisch, Bioresour. Technol., 2005, 96, 673–686 CrossRef CAS PubMed.
  38. M. Frei, Sci. World J., 2013, 1–25 Search PubMed.
  39. M. B. Prats, B. Hauser, M. T. Heredia and N. Gierlinger, Front. Chem., 2016, 4, 10 Search PubMed.
  40. P. N. Perera, M. Schmidt, V. L. Chiang, P. J. Schuck and P. D. Adams, Anal. Bioanal. Chem., 2012, 402, 983–987 CrossRef CAS PubMed.
  41. N. Gierlinger and M. Schwanninger, Spectroscopy, 2007, 21, 69–89 CrossRef CAS.
  42. S. Gajaraj, C. Fan, M. Lin and Z. Hu, Environ. Monit. Assess., 2013, 185, 5673–5681 CrossRef CAS PubMed.
  43. J. W. Anderson and S. R. Bridges, Am. J. Clin. Nutr., 1988, 47, 440–447 CrossRef CAS PubMed.
  44. V. Datsyuk, M. Kalyva, K. Papagelis, J. Parthenios, D. Tasis, A. Siokou, I. Kallitsis and C. Galiotis, Carbon, 2008, 46, 833–840 CrossRef CAS.
  45. S. Osswald, M. Havel and Y. Gogotsi, J. Raman Spectrosc., 2007, 38, 728–736 CrossRef CAS.
  46. H. Murphy, P. Papakonstantinou and T. I. T. Okpalugo, J. Vac. Sci. Technol., B, 2006, 24, 715–720 CAS.
  47. A. C. Ferrari and J. Robertson, Phys. Rev. B: Condens. Matter, 2000, 61, 14095–14107 CrossRef CAS.
  48. M. S. Dresselhaus, A. Jorio, M. Hofmann, G. Dresselhaus and R. Saito, Nano Lett., 2010, 10, 751–758 CrossRef CAS PubMed.
  49. A. M. Cano, K. Kohl, S. Deleon, P. Payton, F. Irin, M. Saed, S. A. Shah, M. J. Green and J. E. Canãs-Carrell, Chemosphere, 2016, 152, 117–122 CrossRef CAS PubMed.
  50. A. A. Keller and A. Lazareva, Environ. Sci. Technol. Lett., 2014, 1, 65–70 CrossRef CAS.
  51. D. Lin and B. Xing, Environ. Pollut., 2007, 150, 243–250 CrossRef CAS PubMed.
  52. R. Nair, S. H. Varghese, B. G. Nair, T. Maekawa, Y. Yoshida and D. S. Kumar, Plant Sci., 2010, 179, 154–163 CrossRef CAS.
  53. M. V. Khodakovskaya, K. de Silva, A. S. Biris, E. Dervishi and H. Villagarcia, ACS Nano, 2012, 6, 2128–2135 CrossRef CAS PubMed.
  54. G. Potters, T. P. Pasternak, Y. Guisez, K. J. Palme and M. A. K. Jansen, Trends Plant Sci., 2007, 12, 98–105 CrossRef CAS PubMed.
  55. A. I. Lopez-Lorente, B. M. Simonet and M. Valcarcel, Analyst, 2014, 139, 290–298 RSC.
  56. C. Peng, D. Duan, C. Xu, Y. Chen, L. Sun, H. Zhang, X. Yuan, L. Zheng, Y. Yang, J. Yang, X. Zhen, Y. Chen and J. Shi, Environ. Pollut., 2015, 197, 99–107 CrossRef CAS PubMed.
  57. J. E. Canãs, M. Long, S. Nations, R. Vadan, L. Dai, M. Luo, R. Ambikapathi, E. H. Lee and D. Olszyk, Environ. Toxicol. Chem., 2008, 27, 1922–1931 CrossRef.
  58. G. Zhai, K. S. Walters, D. W. Peate, P. J. J. Alvarez and J. L. Schnoor, Environ. Sci. Technol. Lett., 2014, 1, 146–151 CrossRef CAS PubMed.
  59. S. Majumdar, J. R. Peralta-Videa, S. Bandyopadhyay, H. Castillo-Michel, J. A. Hernandez-Viezcas, S. Sahi and J. L. Gardea-Torresdey, J. Hazard. Mater., 2014, 278, 279–287 CrossRef CAS PubMed.
  60. Q. Liu, B. Chen, Q. Wang, X. Shi, Z. Xiao, J. Lin and X. Fang, Nano Lett., 2009, 9, 1007–1010 CrossRef CAS PubMed.
  61. J. Geisler-Lee, Q. Wang, Y. Yao, W. Zhang, M. Geisler, K. Li, Y. Huang, Y. Chen, A. Kolmakov and X. Ma, Nanotoxicology, 2013, 7, 323–337 CrossRef CAS PubMed.


Electronic supplementary information (ESI) available. See DOI: 10.1039/c7en01047h

This journal is © The Royal Society of Chemistry 2018