SEARCH

SEARCH BY CITATION

Keywords:

  • TiO2;
  • Nanoparticle;
  • Metal contamination;
  • Dynamic light scattering

Abstract

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS AND DISCUSSION
  6. CONCLUSIONS
  7. Acknowledgements
  8. REFERENCES

While conducting toxicity tests with nano titanium dioxide, the authors found that test suspensions were being contaminated with aluminum and titanium from tip erosion during direct sonication. The contaminating alloy particles had a measurable size distribution and zeta potential using dynamic light scattering, which changed the measured characteristics of the suspensions. Caution should be used when employing direct sonication for preparing test suspensions due to potential interferences of these particles in toxicological assessments. Environ. Toxicol. Chem. 2013;32:889–893. © 2013 SETAC


INTRODUCTION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS AND DISCUSSION
  6. CONCLUSIONS
  7. Acknowledgements
  8. REFERENCES

Engineered nanoparticles (ENPs) have been recognized as valuable components of novel technologies and consumer products due to their unique physical, chemical, and electrical properties 1, 2. The properties that make these particles functionally unique also may make them uniquely toxic to biological systems 3, 4. Several reports have linked toxicity of ENPs to characteristics such as surface charge, surface area, and catalytic capabilities, rather than to the classic disruption of metabolic processes caused by soluble chemicals or bulk materials of the same composition 4–6. Unfortunately, much of the research published on the toxic effects of ENPs supplies little characterization information beyond that provided by the manufacturer 7, making it difficult to establish ENP attributes related to toxicity 8, 9. Commonly, dynamic light scattering/phase analysis light scattering (DLS/PALS) measurements of mean hydrodynamic diameter (dH) and zeta potential (ζ) of ENP suspensions are employed due to their ease of operation and relatively low cost 10, 11.

The purpose of our overall research program is to examine the toxicity of ENPs, such as titanium dioxide (TiO2), to terrestrial organisms and ecosystems under ecologically relevant concentrations, which are likely to be very low. Titanium dioxide was examined, in part, because of its prevalence in toxicity research and its increasing presence in consumer products. Specifically, TiO2 nanoparticles are used in numerous products such as sunscreens, paints, coatings, and other materials due to their photocatalytic and light diffusing abilities (http://www.nanotechproject.org/inventories/consumer/).

To examine the inherent toxicity of TiO2 particles, we chose to generate ENP test suspensions with minimal confounding factors (e.g., surfactants or organic material otherwise present in growth media). Our standard mixing protocol was developed following procedures reported in the literature 12, 13, including the use of direct sonication. It is known that erosion of the sonicator probe occurs with use, due to the high intensity of energy generated through cavitation 12–14. However, the potential confounding issues related to ENP characterization and subsequent toxicity assessments caused by such contamination have not been recognized.

In the present study, we report results showing that direct sonication can lead to contamination of test suspensions from the degradation and erosion of the sonicator probe. We show that probe degradation can produce particles similar in size to the ENP agglomerates and with measureable ζ, which may interfere with the physicochemical characteristics of ENP suspensions as determined through DLS/PALS. Furthermore, we suggest caution in interpreting results from toxicity studies using direct sonication based on the potential for unknown underlying toxic effects of probe particulates indiscernible from those of the ENPs themselves.

MATERIALS AND METHODS

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS AND DISCUSSION
  6. CONCLUSIONS
  7. Acknowledgements
  8. REFERENCES

Nanoparticle test suspension preparation

Titanium dioxide P25 (Evonik Degussa) primary nanoparticles (mean particle size 21 nm, 99.5% purity) were used. Dry P25 powder was weighed and mixed with a 0.2 µm-filtered (Nalgene disposable 1-L unit) 1 mM KCl solution (prepared using nanopure water at >18.2 MΩ). Test suspensions were prepared at low concentrations (5–500 mg/L) to mimic anticipated exposure levels found in terrestrial ecosystems. Suspensions were prepared in a 1-mM KCl solution to promote stability and improve measurability of zeta potential. For the purposes of the present study, most of the results presented describe our findings for 50-mg/L suspensions. Suspensions were sonicated using a Misonix S4000 digital sonicator (QSonica) with a Ti alloy probe. Batches of TiO2-nanoparticle were sonicated in 100- to 250-mL aliquots for 30 min (80/20% pulse on/off) using a 1.27-cm probe at 50% amplitude that supplied approximately 75 watts of applied power. As the probe aged through use, it was cleaned and polished routinely, and it was replaced on recognition of surface pitting that cannot be reduced with polishing, as suggested by the manufacturer. An ice bath was used to reduce suspension heating during sonication. All suspensions were vigorously shaken just prior to use for toxicity testing or for characterization measurements to promote uniform distribution of particles in the dispersions. Measurements of suspension pH were taken using an Accumet XL15 meter in conjunction with zeta-potential measurements to ensure suspension consistency between batches.

Characterization of test suspensions using DLS/PALS

A ZetaPALS Zeta Potential Analyzer (Brookhaven Instrument) was used to measure both particle size and ζ for all ENP test suspensions using DLS and PALS. Hydrodynamic diameter (dH) was determined by generating a diffusion coefficient (cumulants method), applied to the Stokes-Einstein equation

  • equation image

where D is the diffusion coefficient, kB is Boltzmann's constant, T is temperature in Kelvin, η is the viscosity of the suspension medium, and r is the radius of a spherical particle. From this relationship, particle size distribution can be determined. Samples were analyzed using a 661-nm laser, a scattering angle of 90°, and a TiO2 refractive index (real) of 2.7. Zeta potential was determined using PALS, through a measurement of electrophoretic mobility using the Smoluchowski limit.

Confirmation of exposure through scanning electron microscopic imaging and X-ray analysis

In addition to characterizing ENP suspensions, plant tissue was harvested to confirm the presence of TiO2-nanoparticle after suspensions were applied to the test plants. Samples were mounted on carbon adhesive tape and analyzed on an environmental scanning electron microscope (FEI Quanta ESEM; 15 kV, 50 Pa, spot size 3) at varying magnification. This allowed us to visualize the presence of the TiO2-nanoparticle aggregates, their size, and their distribution on the tissue surface. Particle elemental composition was determined using silicon drift detector energy dispersive X-ray spectrometry, which is coupled with the SEM, capturing X-ray spectra generated from the SEM electron beam interacting with the sample.

Quantitative elemental analysis of ENP suspensions

We employed electron probe microanalysis (Cameca S50 Microprobe, 15 keV) featuring wavelength dispersive spectrometry Bragg spectrometers to verify the composition of suspensions. A dropcast of suspension was made by dispersing a small volume onto a pure silicon wafer (mounted on either a nickel-plated aluminum or a stainless steel holder) using a Pasteur pipette and allowing it to evaporate. A dropcast of the neat primary TiO2-nanoparticle was made by depositing a slurry of our TiO2-nanoparticle powder and ethanol onto a silicon wafer. Aggregates were isolated within a visible field (+6,000× magnification), and a minimum of three random, discrete points were subjected to an electron beam to generate an X-ray spectrum for each sample. Direct comparison to reference material–generated spectra was used to calculate relative atomic mass percentages for each element present.

RESULTS AND DISCUSSION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS AND DISCUSSION
  6. CONCLUSIONS
  7. Acknowledgements
  8. REFERENCES

Discovery of contamination

While preparing and characterizing our ENP suspensions at concentrations ranging between 5 and 500 mg/L, we noted that despite using identical procedures for preparing and sonicating our test suspensions, the measured characteristics were unexpectedly highly variable between weekly batches. For example, the mean dH of 5 mg TiO2/L suspensions was about 1,000 nm, with a considerably broad size distribution including macroscale particles. Following replacement with a new sonicator probe, the first suspension prepared had a dH of 366 ± 28 nm.

Many ENPs may agglomerate (i.e., produce assemblages of particles having larger dimensions than of the primary particles) 15 when placed in a liquid. Typically, the preparation of stable ENP test suspensions involves manipulating a combination of physical disruption of agglomerates, steric stabilization using surfactants, and changes in electrostatic repulsion using counter-ions 16, 17. While each of these manipulations offers the possibility to improve ENP suspension stability, and therefore characterization, each also may increase the possibility of introducing confounding variables into the system. Although issues with reproducibility between batches can be explained in part by an inherent lack of stability in TiO2-nanoparticle suspensions, reproducibility was only one factor that alerted us to the potential issue with characterization. Concomitantly, dark gray particulates were observed at the base of stock suspensions over time.

During SEM analysis of plant tissue exposed to our TiO2-nanoparticle suspensions, we identified several large agglomerates that we subsequently examined using silicon drift detector energy dispersive X-ray spectrometry to confirm that these agglomerates contained Ti. Elemental spectral analysis confirmed the presence of not only Ti but also Al and other metallic elements (Fig. 1). A rough quantification using Noran System Six software indicated that a 2-µm agglomerate contained approximately 6% Al by mass, which is considerably higher than can be explained by trace amounts in the environment. We then concluded that the contamination was either present in our primary ENP source material or introduced at some point during suspension preparation.

thumbnail image

Figure 1. Silicon drift detector energy dispersive X-ray spectrometry spectra of plant tissue exposed to sonicated TiO2-nanoparticle suspensions (15 kV beam, 50 Pa, spot size 3, magnification 18,000 ×). The NSS software detected the presence of Al and other metallic elements in a 2-µm-sized Ti-agglomerate (elements other than Al attributable to KCl solution in which Ti is suspended and biologically relevant compounds from tissue).

Download figure to PowerPoint

Concerns with ENP suspension preparation: Sonication

With direct sonication, a metal sonicator probe is submerged directly in the ENP-containing test suspension. Several factors have been reported to alter the effectiveness of sonication, including changes in specific energy, probe diameter/shape, duration of mixing, temperature, and surface tension of the liquid 10–13. In general, it is suggested that the length of sonication time be limited to reduce unanticipated ENP responses to the consequent heat being generated and to reduce particle–particle interactions, which have been proposed to enhance the formation of agglomerates 15. Though these shortcomings have been noted in previous studies, we are not aware of any group having discussed the potential issues associated with the presence of probe particles on DLS/PALS measurements.

The suspensions we used with the lowest concentration of TiO2 (5 mg/L) were below the measurement limits of our DLS instrument, which likely contributed to some reproducibility issues; therefore, our subsequent analyses focused on ENP suspensions with concentrations of 50 mg TiO2/L and greater. While DLS/PALS are sensitive methods, they also have shortcomings and are highly dependent on consistent, accurate, and reproducible sample preparation. In addition, the DLS/PALS technique relies exclusively on light-scattering properties of the material being examined; therefore, the techniques cannot distinguish between ENPs and foreign material. Despite these shortcomings, it appeared that replacing the eroded sonicator probe with a new one somehow altered the measured characteristics of our ENP test suspension, thereby changing our mean dH and ζ.

Tests to identify the source of contamination

We determined four possible ways that Al contamination could be introduced into our test system. Contamination could have come from (1) the water source, which in this case was nanopure water; (2) the KCl used to regulate the ionic strength of our test suspensions; (3) the ENP source material; and (4) the sonication procedure. We utilized electron probe microanalysis to identify the step in which contamination was introduced. Results from the electron probe microanalysis indicated the absence of Al in all samples of nanopure water, KCl solutions, and native particles, suggesting that possibilities 1 through 3 above were not the source of the Al. Additionally, no background Al was detected from the nickel-plated aluminum sample holder in any samples. When we examined suspensions that underwent sonication, however, we found that the particles contained Ti and approximately 8% Al by mass (8.21 ± 1.55%, mean ± standard deviation, n = 6). The electron probe microanalysis result agreed with the composition of the Ti alloy (grade 5, Al6V4) used to make the sonicator probe, as confirmed by a Qsonica technical representative.

To further confirm the absence of contamination in our source material, we used X-ray diffraction (Rigaku Ultima IV) to examine the dry primary Degussa P25 TiO2-nanoparticle powder. A microgram sample of powder was loaded into a beryllium membrane–topped, zero-background sample holder and then analyzed. Anatase and rutile TiO2 phase patterns were confirmed with PDXL database software (Rigaku) (Fig. 2). We concluded, therefore, that some particle agglomerates observed with SEM were particles of an Al-containing Ti alloy that had been added to the test suspensions during sonication.

thumbnail image

Figure 2. X-ray diffraction phase pattern of dry TiO2-nanoparticle powder in Be sample holder; Cu Kα, 20 mA, 2 deg/min scan speed, 5 to 80 deg range, 2Θ measurement axis. The PDXL software estimated 89.9% anatase, 10.1% rutile TiO2.

Download figure to PowerPoint

Contaminant characterization using light scattering

Although erosion of the tip of the probe is known to occur as a result of high-frequency vibrations over time 13, we are unaware of literature discussing the potential for it to interfere with light-scattering measurements. To optimize mixing protocols for our experiments, we examined solutions of KCl under varying conditions: different durations of sonication, applied power of sonication, probe type (replaceable tip and solid), and probe age. For both a new and an aged probe (under the same duration, power, and probe type [replaceable tip]) suspensions contained a significantly greater number of particles than KCl without sonication (new probe 287 ± 41.6 vs 8.17 ± 3.16 kilocounts per second, aged probe 362 ± 28.5 vs 4.93 ± 1.03 kilocounts per second, mean ± standard error [SE], n = 9 and 3, respectively; analysis of variance p < 0.05). Particles represented a broad size distribution for all solutions, from 50 to 3,500 nm. Regardless of probe age, Ti alloy particles were present in solutions, showing the potential for contamination even with a new probe. The changing state of the sonicator probe through use, even after a single sonication, can be visualized through matting and pitting of the alloy surface (Fig. 3).

thumbnail image

Figure 3. Images of Ti alloy sonicator probe (1.27 cm, grade 5 Ti alloy) under varying conditions: (left) new, unused probe; (middle) probe after single use, matted finish; (right) aged probe with obvious pitting on surface.

Download figure to PowerPoint

In a sample of filtered KCl solution with minimal interference due to dust, there were too few particles to generate a meaningful zeta-potential value (i.e., –2.66 ± 1.18 mV, mean ± SE, n = 30). This same issue of measurability was observed in samples below 10 mg TiO2/L (5 mg TiO2/L ζ –2.37 ± 1.62 mV, mean ± SE, n = 20). However, a KCl solution of Ti-alloy particles eroded from the sonicator probe appeared to have a negative mean ζ that was consistent regardless of sonication conditions (i.e., aged probe ζ –28.6 ± 0.77 mV vs new probe ζ –28.7 ± 3.34 mV, mean ± SE, n = 9). The presence of a measureable zeta potential suggests that the Al-containing alloy material may act as an additional population of particles, potentially interacting with nanoparticles in suspension and altering the overall characteristics of a test suspension.

Influence of contamination on ENP suspensions and implications for toxicity response studies

As an additional test to examine the potential influence of sonicator probe contamination on our ENP test suspensions, we used an indirect sonication method (14-cm-diameter probe [cup horn] contained in a chilled reservoir in which a suspension-containing vessel is submerged) for comparison. Although direct and indirect sonication methods are not directly comparable, we optimized the cup horn method to replicate the duration, delivered power, and temperature conditions that we used with direct sonication. The amount of delivered power (watts) was determined in a separate test using the calorimetric method in which the change in temperature of a given mass of water is measured over time 12. For this test we used a suspension of 50 mg TiO2/L as it falls within the measurement limits of the ZetaPALS Analyzer and is a concentration commonly used in our lab. Suspensions were prepared in a 1-mM KCl solution at a measured pH of 6.5 to 7. They were characterized for particle size distribution and ζ and compared to a 50-mg TiO2/L and 1-mM KCl solution (no ENPs) directly sonicated using an aged probe.

Distinct populations of particles were observed for each suspension, which were related to the presence of contamination (Fig. 4). In the absence of any contaminant from direct sonication, approximately 88% of TiO2 particles were ≤300 nm in diameter, with roughly 50% of the population represented by a particle size of ≤50 nm. The smallest particles detected were approximately 40 nm. Direct sonication and the introduction of contamination resulted in an increase in mean particle size whereby roughly 50% of particles were ≤300 nm. No particles less than 50 nm were measured in the direct sonication suspension, suggesting that the proportion of larger particles introduced through sonication was enough to mask the presence of ENPs or to alter particle–particle interactions in suspension. It is possible that with aging of the sonicator probe, the proportion of contaminants to ENPs increases such that additional changes in size distribution may be observed. We did not see the same extent of variability in physicochemical characteristics between batches of test suspensions of higher ENP concentration, suggesting that the effect of contamination on characterization may decrease as ENP concentration increases. However, we note that at higher suspension concentrations, the accuracy of DLS also decreases due to multiple scattering events 18.

thumbnail image

Figure 4. Cumulative distribution function, C(d), of suspensions with respect to particle size distribution: 50 mg TiO2/L with indirect sonication (n = 3), 50 mg TiO2/L sonicated with a Ti-alloy probe (n = 3), and three consecutive sonications of KCl with a Ti-alloy probe (no ENP, n = 9). ENP = engineered nanoparticle.

Download figure to PowerPoint

Along with measurable alterations to particle size distribution and apparent mean dH, ζ measures were significantly affected by the presence of contamination. In the absence of contamination, the ζ of a 50-mg TiO2/L suspension was approximately +27.5 ± 0.32 mV (mean ± SE, n = 250). This closely resembled the anticipated value of +30 mV for a well-dispersed TiO2-nanoparticle test suspension 19, 20. Direct sonication of a 50-mg TiO2/L suspension with a Ti-alloy probe resulted in a significantly lower ζ of +21.3 ± 0.29 mV (mean ± SE, n = 250; analysis of variance p < 0.05). The introduction of contamination also resulted in a less stable suspension as measured by PALS. In a heterogeneous suspension with contaminants, ENPs are more likely to exist in a different agglomerated state from neat nanoparticles, which may lead to unexpected interactions with toxicity test subjects. Because DLS/PALS makes no elemental distinction between particles and there is apparent overlap in size distributions between ENPs and Ti-alloy particles, it would be nearly impossible to distinguish between the two populations of particles or their relative proportions within a test suspension. We believe the most problematic issue in characterizing ENPs in many toxicity testing laboratories may be that the presence of Al contamination is not easily recognized, requiring supplemental analytical techniques such as silicon drift detector energy dispersive X-ray spectrometry.

The presence of contamination in test suspensions prepared using direct sonication methods also may help to explain confounding results between studies that employ different mixing protocols with the same ENPs. For instance, several studies suggest that increased sonication time results in the formation of ENP agglomerates through increased particle interactions 14. This increase in observed aggregates may be due to an increased presence of Ti-alloy particles with extended sonication time, without necessarily affecting the ENP state.

Lastly, Ti-alloy contamination in nanoparticle suspensions may also result in direct toxicity to test organisms through the presence of ionic aluminum species (e.g., Al3+). When aluminum-bearing materials are present in suspension, the formation of Al3+ increases significantly at pH below 5.5 and is the predominant aluminum ion in solution below pH 4.7 21, 22. Even though Al3+ is not the only ionic aluminum species at these pH levels, it is the single best indicator of aluminum toxicity to plants 23. The pH of nanoparticle suspensions is often lowered to improve stability, but this may result in release of Al3+ to test media 20. Consequently, a toxicity response due to the presence of Al may be incorrectly thought to be an effect due to nanoparticle exposure.

CONCLUSIONS

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS AND DISCUSSION
  6. CONCLUSIONS
  7. Acknowledgements
  8. REFERENCES

Our results suggest that the potential for contamination resulting from sonicator tip erosion during direct sonication may make it difficult to accurately characterize suspensions using DLS/PALS and can increase the likelihood of introducing soluble metals into nanoparticle test suspensions. The amount of contamination appears to vary depending on the duration of sonication, the age of the probe, and the mixing protocol employed. The contamination is readily identifiable with more sophisticated analytical approaches, but the majority of toxicology studies employ DLS/PALS as the primary, or only, method of suspension characterization. As a result, we caution toxicology researchers to be aware of the potential influence of this contamination on their ability to accurately characterize their test suspensions.

While mixing higher-concentration nanoparticle suspensions may be one way to avoid many of the measurement interferences when using DLS, the presence of contamination may alter ENP states within a test suspension, which would go undetected based on the technique's limitations. However, using suspensions of higher ENP concentrations will not eliminate the possible introduction of soluble Al ions at low pH, which can be toxic in biological matrices. We suggest that in toxicology studies in which more sophisticated techniques for characterization are not available, it may be advisable to employ alternative methods to direct sonication to prepare nanoparticle suspensions.

Acknowledgements

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS AND DISCUSSION
  6. CONCLUSIONS
  7. Acknowledgements
  8. REFERENCES

We acknowledge J. Donovan at the Center for Advanced Materials Characterization in Oregon (CAMCOR) for his advice and support. We also thank L. Tumburu and M. Plocher for assistance with the Arabidopsis plants. We thank V. Hackley, A. Poda, and J. Taurozzi for their valuable reviews of a previous version of the manuscript. The information in this document was funded by the U.S. Environmental Protection Agency. It has been subjected to the agency's peer and administrative review, and it has been approved for publication as a U.S. Environmental Protection Agency document. Mention of trade names or commercial products does not constitute endorsement or recommendation for use.

REFERENCES

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS AND DISCUSSION
  6. CONCLUSIONS
  7. Acknowledgements
  8. REFERENCES
  • 1
    Theodore L, Kunz RG. 2005. Nanotechnology: Environmental Implications and Solutions. John Wiley and Sons, Hoboken, NJ, USA.
  • 2
    U. S. Environmental Protection Agency. 2007. Nanotechnology white paper. EPA 100/B-07/001. Final report. Washington, DC.
  • 3
    Oberdörster G, Oberdörster E, Oberdörster J. 2005. Nanotoxicology: An emerging discipline evolving from studies of ultrafine particles. Environ Health Perspect 113:823839.
  • 4
    Nel A, Xia T, Madler L, Li N. 2006. Toxic potential of materials at the nanolevel. Science 311:622627.
  • 5
    Navarro E, Baun A, Behra R, Hartmann NB, Filser J, Miao AJ, Quigg A, Santschi PH, Sigg L. 2008. Environmental behavior and ecotoxicity of engineered nanoparticles to algae, plants, and fungi. Ecotoxicology 17:372386.
  • 6
    Buzea C, Pacheco II, Robbie K. 2007. Nanomaterials and nanoparticles: Sources and toxicity. Biointerphases 2:MR17MR71.
  • 7
    Stone V, Nowack B, Baun A, Brink N, Kammer F, Dusinska M, Handy R, Hankin S, Hassellöv M, Joner E, Fernandes T. 2010. Nanomaterials for environmental studies: Classification, reference material issues, and strategies for physico-chemical characterisation. Sci Total Environ 408:17451754.
  • 8
    Warheit DB, Hoke RA, Finlay C, Donner EM, Reed KL, Sayes CM. 2007. Development of a base set of toxicity tests using ultrafine TiO2 particles as a component of nanoparticle risk management. Toxicol Lett 171:99110.
  • 9
    Menard AD, Drobne D, Jemec A. 2011. Ecotoxicity of nanosized TiO2. Review of in vivo data. Environ Pollut 159:677684.
  • 10
    Bihari PM, Vippola M, Schultes S, Praetner M, Khandoga A, Reichel C, Coester C, Tuomi T, Rehberg M, Krombach F. 2008. Optimized dispersion of nanoparticles for biological in vitro and in vivo studies. Particle and Fibre Toxicology 5:14.
  • 11
    Ma Y, Kuang L, He X, Bai W, Ding Y, Zhang Z, Zhao Y, Chai Z. 2010. Effects of rare earth oxide nanoparticles on root elongation of plants. Chemosphere 78:273279.
  • 12
    National Institute of Standards and Technology. 2010. CEINT/NIST Protocol for preparation of nanoparticle dispersions from powdered material using ultrasonic disruption—Version 1. U.S. Department of Commerce, Washington, DC.
  • 13
    Taurozzi JS, Hackley VA, Wiesner MR. 2011. Ultrasonic dispersion of nanoparticles for environmental, health and safety assessment—Issues and recommendations. Nanotoxicology 5:711729.
  • 14
    Qsonica. Sonicator Ultrasonic Processor: Part No. S-4000 Operation Manual. Newtown, CT, USA.
  • 15
    Mandzy N, Grulke E, Druffel T. 2005. Breakage of TiO2 agglomerates in electrostatically stabilized aqueous dispersions. Powder Technol 160:121126.
  • 16
    Chowdhury I, Hong Y, Walker SL. 2010. Container to characterization: Impacts of metal oxide handling, preparation, and solution chemistry on particle stability. Colloids Surf, A 368:9195.
  • 17
    Jiang J, Oberdörster G, Biswas P. 2009. Characterization of size, surface charge, and agglomeration state of nanoparticle dispersions for toxicological studies. J Nanopart Res 11:7789.
  • 18
    Kaszuba M, Corbett J, Watson FM, Jones A. 2010. High-concentration zeta potential measurements using light-scattering techniques. Philos Trans R Soc London A 368:44394451.
  • 19
    Hunter RJ. 2001. Foundations of Colloid Science. Oxford University Press, New York, NY, USA.
  • 20
    Berg JM, Romoser A, Nivedita B, Zebda R, Sayes CM. 2009. The relationship between pH and zeta potential of 30 nm metal oxide nanoparticle suspensions relevant to in vitro toxicological evaluations. Nanotoxicology 3:276283.
  • 21
    Bohn HL, McNeal BL, O'Connor GA. 1979. Soil Chemistry. John Wiley and Sons, New York, NY, USA.
  • 22
    McBride MB. 1994. Environmental Chemistry of Soils. Oxford University Press, New York, NY, USA.
  • 23
    Parker DR, Kinraide TB, Zelazny LW. 1988. Aluminum speciation and phytotoxicity in dilute hydroxy-aluminum solutions. Soil Sci Soc Am J 52:438444.