Ultrasonic Dispersion of Nanoparticles for Environmental, Health and Safety Assessment

Nanotoxicology, 2010; Early Online, 1–19

Ultrasonic dispersion of nanoparticles for environmental, health and safety assessment – issues and recommendations

JULIAN S. TAUROZZI1, VINCENT A. HACKLEY1, & MARK R. WIESNER2

1Material Measurement Laboratory, National Institute of Standards and Technology, Gaithersburg, Maryland, and 2Department of Civil and Environmental Engineering, Duke University, Durham, North Carolina, USA

(Received 9 July 2010; accepted 28 September 2010)

Abstract Studies designed to investigate the environmental or biological interactions of nanoscale materials frequently rely on the use of ultrasound (sonication) to prepare test suspensions. However, the inconsistent application of ultrasonic treatment across laboratories, and the lack of process standardization can lead to signiﬁcant variability in suspension characteristics. At present, there is widespread recognition that sonication must be applied judiciously and reported in a consistent manner that is quantiﬁable and reproducible; current reporting practices generally lack these attributes. The objectives of the present work were to: (i) Survey potential sonication effects that can alter the physicochemical or biological properties of dispersed nanomaterials (within the context of toxicity testing) and discuss methods to mitigate these effects, (ii) propose a method for standardizing the measurement of sonication power, and (iii) offer a set of reporting guidelines to facilitate the reproducibility of studies involving engineered nanoparticle suspensions obtained via sonication.

Keywords: Environmental assessment, toxicology, nanoparticle, nanomaterial, ultrasonics, sonication, dispersion, suspension

Introduction in biological and environmental systems (Maynard 2006; Barnard 2006). Nanotechnology, and more specifically the emer- Moreover, the assessment of potential risks associ- For personal use only. gence of engineered nanoscale particles (ENPs), ated with ENPs requires the evaluation of their behav- shows promising potential for the development of ior in environmentally and biologically relevant advanced materials and devices in the fields of med- matrices (e.g., cell media, serum, whole blood, stan- icine, biotechnology, energy, and environmental dardized seawater). When ENPs are not already in a remediation, among others (Zhang 2003; Salata fully dispersed form, the test suspensions are typically 2004; Raimondi et al. 2005). Similarly, ENPs are obtained by dispersing dry powders or stock suspen- migrating into a wide range of consumer products, sions of the source material into the desired test many of which are now in the global market place medium prior to in vitro or in vivo evaluation (Project on Emerging Nanotechnologies 2010). How- (Murdock et al. 2008; Jiang et al. 2009; Labille ever, some of the properties that make ENPs attractive et al. 2009). In these situations, the use of ultrasonic may also pose environmental and health hazards energy, a process commonly referred to as ‘soni- (Colvin 2003; Gwinn and Vallyathan 2006; cation’, is arguably the most widely used procedure. Kreyling et al. 2006; Dobrovolskaia and McNeil Sonication refers to the application of sound energy 2007; Kahru and Dubourguier 2010). As both public at frequencies largely inaudible to the human ear Nanotoxicology Downloaded from informahealthcare.com by NIST National Institiutes of Standards & Technology on 12/03/10 and private funding for application-oriented nanoma- (higher than » 20 kHz), in order to facilitate the terial research continues to grow and nanoparticle- disruption of particle agglomerates through a process based consumer products reach the market, there is known as cavitation. Ultrasound disruption is more an increasing demand for a more comprehensive energy efficient and can achieve a higher degree of scientific understanding of the interactions of ENPs powder fragmentation, at constant specific energy,

Correspondence: Dr. Vincent A. Hackley, National Institute of Standards and Technology, 100 Bureau Dr. Mail Stop 8520, Gaithersburg, 20899, MD, USA. E-mail: [email protected]

ISSN 1743-5390 print/ISSN 1743-5404 online 2010 Informa UK, Ltd. DOI: 10.3109/17435390.2010.528846 2 J. S. Taurozzi et al.

than other conventional dispersion techniques (Park dispersion, but their consideration is beyond the et al. 1993; Hielscher 2005; Mandzy et al. 2005). scope of the present work. Sonication is a convenient, relatively inexpensive tool that is simple to operate and maintain. Consequently, Terminology sonication has found extensive application in toxicological and environmental studies, where it is used to We define at the outset several critical terms, which break down powders or re-disperse stock suspensions are used throughout the text, in order to avoid ambi- to their nanoscale constituents and allow for the guity, as these terms can have different meanings in evaluation of the properties and behavior of the dis- different scientific and technical circles. For the most persed nanoparticles in relevant systems. part, we follow definitions relevant to nanotechnology However, the inconsistent application of ultrasonic as set forth in the standard E2456 (2006) by ASTM treatment across laboratories and between individual International (formally the American Society for operators, combined with the lack of accepted process Testing and Materials). Additional guidance is standardization, has likely contributed to variability in derived from recommendations of the International observed results. Furthermore, the complex physical Union of Pure and Applied Chemistry (IUPAC and chemical phenomena that occur during sonica- 2009). tion – and which may significantly alter the dispersed For the purposes of this publication, a nanoparticle material’s properties – are frequently underappreci- is defined as a sub-classification of ultrafine particle ated, perhaps resulting in the use of suspensions that that is characterized by dimensions in the nanoscale may not meet the intended experimental design, or (i.e., between 1 and 100 nm) in at least two dimen- potentially leading to artifacts that could compromise sions, with the recognition that the more rigorous conclusions as to how the observed ENP’s ecological definition of nanoparticle also requires that these or toxicological behavior relates to the material’s particles exhibit novel size-dependent properties. inherent properties. It is the authors’ contention A primary particle is the smallest discrete identifiable that the inconsistent application and reporting of entity associated with a particle system; in this con- sonication is a potential leading contributor to vari- text, larger particle structures (e.g., aggregates and ability in test results in which sonication is utilized; agglomerates) may comprise primary particles. An this conclusion has been echoed by the results of aggregate is a discrete assemblage of primary particles recent interlaboratory studies (Roebben et al. 2010). strongly bonded together (i.e., fused, sintered, or Moreover, while not the main focus of this work, it metallically bonded), which are not easily broken is important to note that many of the sonication- apart. While the distinction is often blurred, aggre- induced physicochemical effects and recommenda- gates might also be differentiated from agglomerates, For personal use only. tions addressed herein are equally relevant to other which are assemblages of particles (including primary ENP research areas where sonication is utilized, in particles and/or smaller aggregates) held together by particular those involving the fate and transport of relatively weak forces (e.g., Van der Waals, capillary, nanomaterials in environmental systems. or electrostatic), that may break apart into smaller The objectives of the present work were to: (i) particles upon processing (e.g., using sonication or Highlight salient aspects of the ultrasonic disruption high intensity mixing). Commonly, agglomerates process and survey potential sonication-induced exhibit the same overall specific surface area (SSA) effects that may alter the physical, chemical or bio- of the constituent particles, while aggregates have a logical properties and behavior of aqueous nanopar- SSA less than that of the constituent primary particles. ticle suspensions (within the context of toxicological In a related fashion, agglomeration and aggregation testing), (ii) propose a method for standardizing the refer, respectively, to the process by which agglom- measurement of sonication power, and (iii) offer a set erates or aggregates form and grow. It should be noted of reporting guidelines to facilitate the reproducibility that, although here we define them as distinct entities, of studies involving engineered nanoparticle suspen- the terms aggregate and agglomerate are often used

Nanotoxicology Downloaded from informahealthcare.com by NIST National Institiutes of Standards & Technology on 12/03/10 sions obtained via sonication. interchangeably to denote particle assemblies. Our The authors aimed not only to indentify and discuss usage is consistent with previous recommendations key sonication issues relevant to ENP dispersion, but published in this journal (Oberdörster et al. 2007), also to address them by offering practical recommen- but is by no means universally adopted. dations. For this reason, some discussions will follow Additionally, in the present work, a suspension is a traditional literature review format, while others are deﬁned as a liquid (typically aqueous) in which par- more prescriptive in style. The authors also recognize ticles are dispersed. The term dispersion is used in the that additional sample-speciﬁc preparation steps are present context to denote the process of creating a often required to achieve the desired state of liquid in which discrete particles are homogeneously Dispersion of nanoparticles 3

distributed throughout a continuous ﬂuid phase (e.g., to obtain a suspension), and implies the intention to break down agglomerates into their principal component particles. Dispersion is also used as an adjective in combination with terms relating to this process or relating to the state of the resulting product.

Sonication basics and equipment

This brief overview highlights the basic aspects of ultrasonic generation and propagation relevant to Probe particle dispersion, and serves as background for Bath Cup the ensuing discussions; while it is not a comprehen- Figure 2. Schematic illustration of direct (left) and indirect (middle sive review of the topic, readers can refer to the cited and right) sonication conﬁgurations as described in the text. references for a more in-depth treatment of the subject. During the process of ultrasonic disruption, sound waves propagate through the liquid medium much lower energy levels than are attainable using a in alternating high and low pressure cycles at fre- probe or cup horn. In the case of bath sonicators, the quencies typically in the 20–40 kHz range. During transducer element is directly attached to the outside the low-pressure cycle (rarefaction), microscopic surface of the metal tank, and the ultrasonic waves are vapor bubbles are formed in a process known as transmitted directly to the tank surface and then into cavitation (Figure 1). The bubbles then collapse dur- the bath liquid. In a cup horn, the radiating surface of ing the high pressure cycle (compression) producing a the horn is inverted and sealed into the bottom of a localized shock wave that releases tremendous (typically) transparent plastic cup into which the mechanical and thermal energy. sample container is immersed. Ultrasonic waves can be generated in a liquid At ultrasound frequencies, the alternating forma- suspension either by immersing an ultrasound probe tion, growth and subsequent shock waves produced (transducer horn) into the suspension (direct sonica- by the collapse of bubbles result in extremely large tion), or by introducing the sample container into a localized temperatures up to 10,000 K, rapid temper- liquid that is propagating ultrasonic waves (indirect ature changes (> 105 KsÀ1), pressure bursts on the sonication). This is shown schematically in Figure 2. order of several MPa, and liquid jet streams with For personal use only. In a sonication bath or a so-called cup horn soni- speeds reaching 400 km Á h-1 (Mason 1989; Hielscher cator (indirect sonication), the sound waves must 2005). Such massive, local energy output is the basis travel through both the bath or cup liquid (typically of the disruptive effect of sonication. It must be noted water) and the wall of the sample container before that these considerable pressure and temperature reaching the suspension. In direct sonication, the differentials are a result of the cavitation process probe is in contact with the suspension, reducing and occur at the local interface of the exploding the physical barriers to wave propagation and there- bubbles; consequently, these effects are inherent to fore delivering a higher effective energy output into the sonication process and will occur whether the the suspension. Bath sonicators typically operate at sonicated container is cooled or not, or the treatment performed using a bath, cup horn, or a probe sonicator. Bubble Agglomerate fracture points In all cases, the ultrasonic device transforms elec- High trical power into vibrational energy by means of a pressure piezoelectric transducer that changes its dimensions cycle Nanotoxicology Downloaded from informahealthcare.com by NIST National Institiutes of Standards & Technology on 12/03/10 in response to an applied AC electric ﬁeld. In direct sonication, the transducer horn serves to transmit and focus the ultrasonic waves into the targeted liquid Low pressure sample. For direct sonication, the following equation cycle Implosion relates acoustic vibrational energy to probe and shock waves medium parameters (Contamine et al. 1995):

1 22 Figure 1. Schematic illustration of ultrasonic wave-induced P=ρπ cA(()21 f) a cavitation and agglomerate fracture. 2 4 J. S. Taurozzi et al.

where P is the acoustic power (W) of the ultrasound require greater power consumptions from the electric source, a is the ‘emission area’ (m2), which is the source. surface area of the emitting ultrasound source, r is the A sonicator’s stated maximum power (e.g., 600 W) liquid density (kg m-3), A is the amplitude (m) of refers to the maximum theoretical power it would be oscillation of the ultrasound probe, c is the speed of able to consume, should it require so; it does not reflect the acoustic wave in the liquid medium (m s-1) and f is the actual amount of ultrasonic energy delivered to the vibration frequency (Hz). the suspension. That is, for the same frequency and The amplitude of oscillation per unit time deter- amplitude (and thus the same sonicator setting), the mines the pressure difference between cavitational sonicator would need to consume more power to treat rarefaction and compression cycles. Larger oscillation a high viscosity suspension than it would for a low amplitudes yield larger high-low pressure gradients viscosity suspension, and if so, it could consume up to and consequently greater energy outputs. The probe’s a maximum of 600 W. It is therefore erroneous to oscillation amplitude is in turn determined by the assume that selecting the highest setting value will amount of energy transferred by the instrument’s result in the delivery of the instrument’s nominal generator to the ultrasonicator probe, which can be maximum power to the sonicated suspension. For regulated using the sonicator power settings. low viscosity media, even at the highest setting value, Conventional direct sonicators operate by self- the delivered power will be a significantly lower frac- adjusting their power consumption from the electrical tion of the instrument’s maximum power rating. source in order to maintain the vibration frequency of Similarly, the power value (measured in W) usually the transducer (e.g., quartz crystal) at a constant value shown on the instrument display reflects the power (usually 20 kHz). As it oscillates, the sonicator probe – that the instrument is consuming from the electrical connected to the vibrating transducer – will experi- source to produce the desired oscillation amplitude ence a resistance from the sonicated medium that will (from the selected setting value) in the sonicated be transmitted back to the vibrating element and medium. The consumed (i.e., displayed) power, how- detected by the internal control unit of the instru- ever, does not necessarily reflect the power that is ment. The instrument’s control unit will in turn actually delivered to the sonicated suspension, which is adjust the power consumption of the instrument to affected by the probe’s oscillation amplitude in the maintain a constant vibration frequency. Highly vis- medium (see Equation 1). Herein lies the deficiency cous media will exert a greater resistance to the with reporting either the manufacturer’s stated max- oscillating probe, and will therefore require greater imum power, the adjustable output power reading power consumption to maintain a constant oscillation provided by the instrument, or the chosen setting frequency. For example, to produce an oscillation level. The flowchart in Figure 3 illustrates the energy For personal use only. with an amplitude of 3 mm at 20 kHz in deionized transformations that occur in a conventional probe water, a sonicator may consume ‘N’ watts, while to sonicator. produce the same amplitude at 20 kHz in oil, it will The net fragmentation effect from applying ultra- consume considerably more power. sonic energy to a suspension is dependent on the total The oscillation frequency value is usually fixed for a amount of energy delivered to the sonicated medium. given instrument and cannot be changed. Changing However, not all of the produced cavitational energy the instrument power setting translates to a change in is effectively utilized in disrupting particle clusters. the vibrating amplitude – not the frequency – of the The delivered energy is consumed or dissipated by probe; that is, increasing the power setting results in several mechanisms, including thermal losses, ultra- larger probe oscillation amplitudes. To maintain a sonic degassing, and chemical reactions, such as constant oscillation frequency, larger oscillation the formation of radical species. Only a portion of amplitudes, as well as higher medium resistances, the delivered energy is actually used in breaking Nanotoxicology Downloaded from informahealthcare.com by NIST National Institiutes of Standards & Technology on 12/03/10 Electric Generator Displayed Sonication source & Suspension power probe power Transducer

Electrical, mechanical and Coupling, friction, thermal losses thermal losses

Figure 3. Flowchart illustrating the energy transformation process in a conventional probe (direct) sonicator. Dispersion of nanoparticles 5

particle-particle bonds to produce smaller particle determines whether the particles will settle out or aggregates, agglomerates, and primary particles. remain in suspension, which directly impacts assay Moreover, an excessive energy input can potentially efficacy and delivered dose. Numerous other exam- result in agglomerate formation or re-agglomeration ples of size-dependent behavior can be found in the of previously fragmented clusters, as well as induce a recent literature. variety of physicochemical alterations on the materi- Dry powders consist of particles that are bound al’s surface or to the constituents of the suspending together into macroscopic structures by both weak medium. Such sonication-specific side-effects are physical Van der Waals forces and stronger chemical addressed in the following sections, but first we bonds including particle fusion (Zachariah and examine the particle disruption process as it is the Carrier 1999; Mandzy et al. 2005; Jiang et al. primary objective of ultrasonic treatment. 2009). Commonly, these powders contain micrometer – and submicrometer – scale aggregates, which Sonication effects on particle disruption and are in turn held together by physical forces to con- aggregation stitute larger agglomerates. This is depicted schematically in Figure 4. For powders consisting of nanoscale The properties associated with ENPs, and conse- particles and aggregates, and therefore having sub- quently their distinct biological and environmental stantial interfacial contact areas per unit volume, the interactions, are strongly dependent on the particle nominally weaker Van der Waals forces can be size of the processed material. For instance, the size of extremely large, requiring the use of techniques the particles in suspension dictates – along with the such as sonication to effectively break down or disrupt particles’ surface chemistry – their biodistribution, powder agglomerates. In contrast, micrometer size elimination, cellular uptake, toxicity, and environ- primary particles can often be dispersed with moder- mental transport behavior (Rejman et al. 2004; ate mixing or stirring. Lanone and Boczkowski 2006; Pan et al. 2007; Many of the commercially available metal oxide Powers et al. 2007; Li and Huang 2008; Sonavane ENP source powders (e.g., silica, titania and alumina) et al. 2008). Additionally, the agglomeration state are synthesized by high temperature vapor phase

Sonication Aggregate induced primary particles held aggregation together by sinter necks (coagulation) Peaking trend

For personal use only. or strong bonds Primary particle

A Sinter neck Dispersed Average particle size aggregates and/or Sonication energy primary particles (power, time, energy density)

Agglomerate aggregates held together No further Smaller by Van der Waals or disruption other weak forces agglomerates & aggregates Asymptotic trend Nanotoxicology Downloaded from informahealthcare.com by NIST National Institiutes of Standards & Technology on 12/03/10

B Average particle size Sonication energy Delivered sonication energy (power, time, energy density)

Figure 4. Schematic depiction of particle structures referred to in the text, and illustration of the typical effects of sonication on particle size asa function of delivered sonication energy: (A) Asymptotic behavior (solid line) versus (B) peaking behavior (dashed line). 6 J. S. Taurozzi et al.

processes (Kraft 2005; Isfort and Rochnia 2009). In and method of manufacture, the occurrence of sur- such processes, chemical precursors react at high face defects in the powder clusters, the material’s temperatures to produce small ‘hot’ nanoscale dro- surface toughness, and the bonding nature (e.g., plets of the desired material. As these hot droplets neck diameter or interfacial contact area) of aggre- pass through the reaction zone, they collide and may gates and agglomerates, there is a system-specific coalesce fully or partially, depending on the local acoustic energy threshold that must be crossed in temperature. Full coalescence leads to a larger pri- order to achieve agglomerate and aggregate breakage. mary particle, while partial coalescence leads to the It is thus worth noting that two powders of the same formation of an aggregate of fused primary particles material, but obtained following different synthetic (bound via sinter necks), which later become ‘solid routes, may show significantly different breakage bridges’ upon particle solidification (Figure 4). Par- behavior. Material properties also affect the rate at ticle coalescence and growth continue until the reac- which fragmentation occurs, and, consequently, the tants are consumed or the process temperature is amount of time it takes at a given ultrasonic output quenched (Singhal et al. 1999; Zachariah and Carrier power and for a given amount of material to reduce all 1999; Teleki et al. 2008; Isfort and Rochnia 2009). of the powder to the smallest achievable dimensions. Metal oxide and other ceramic powders frequently Thus, in principle, the cavitation process can effec- show this configuration, wherein the ‘primary tively break down powder agglomerates. Yet, under particle’ represents the solidified globules that con- certain conditions, the applied acoustic energy can stitute powder aggregates (frequently chain-like struc- conversely induce particle agglomeration or even lead tures). Yet, it must be noted that the solid bridges that to the formation of aggregates. Coagulation in ultra- hold the primary particles together in the dry powder sonic fields can occur from enhanced particle- aggregates are often too strong to be broken via particle interactions due to the increased collision sonication or other moderate dispersion processes. frequency as well as the favorable reduction in free Therefore, in some cases, it can be unrealistic to energy that accompanies the resulting reduction in the expect sonicated powders to break down to their liquid-solid interface. As noted previously in discuss- reported, nominal primary particle size. As a result, ing the basics of ultrasonic disruption, cavitation gives powders will be effectively de-agglomerated to aggre- rise to extreme local pressure and temperature gra- gates of several primary particles (i.e., ‘primary dients, as well as shock waves and jet streams of the aggregates’), rather than isolated primary particles. order of hundreds of meters per second. Under such These aggregates should be thought of as the effective conditions, sonicated particles in the neighborhood of primary size for the ENP. Although the underlying a collapsing bubble collide with each other while formation processes in solution phase synthesis (e.g., simultaneously experiencing localized intense heating For personal use only. via sol gel or precipitation) differ from the vapor phase and subsequent cooling cycles. Depending on the process, the resulting structures often express similar effective energy delivered to the particles and the physical properties to the above described structures thermal properties of the material and medium, these (i.e., particle fusion, presence of aggregates and concomitant effects can lead to re-agglomeration or agglomerates), though typically to a lesser extent even thermally induced inter-particle fusion; that is, (Brinker and Scherer 1990). sonication-induced aggregate formation. During sonication, shock waves from cavitational Published studies have demonstrated the existence collapse are principally responsible for powder break- of this phenomenon, which is often observed as an age. The particles, in fact, serve as nuclei to initiate the increase in particle size beyond a given sonication cavitation process, which partly explains why sonica- energy threshold (Aoki et al. 1987; Vasylkiv and Sakka tion is so effective (Suslick 1988). When ultrasound 2001). In other cases, an asymptotic behavior is energy is applied to powder clusters in suspension, observed, wherein sonicated powders break down fragmentation can occur either via erosion or fracture. to particles of a given size value, beyond which no Erosion or ‘chipping’ refers to the detachment of further significant size decrease is obtained even at

Nanotoxicology Downloaded from informahealthcare.com by NIST National Institiutes of Standards & Technology on 12/03/10 particles from the surface of the parent agglomerates, higher ultrasound energy inputs (Bihari et al. 2008; while fracture or ‘splitting’ occurs when agglomerates Murdock et al. 2008). The latter behavior further partition into smaller agglomerates or aggregates due conﬁrms that, for some powders, primary particles to the propagation of cracks initiated at surface defects can be held together by bonds that cannot be broken (Figure 1). These fragmentation effects may occur by sonication. simultaneously or in isolation, depending on the pow- Interestingly, both asymptotic and coagulation der, its environment and the energy levels involved. behaviors (see Figure 4) as a function of different For a given material in powder form, depending on sonication parameters have been observed for the the physicochemical properties of the solid material same source material (e.g., TiO2, Mandzy et al. Dispersion of nanoparticles 7

2005; Jiang et al. 2009), illustrating the system- The extreme localized temperatures and pressures specific dependence (i.e., source powder, dispersing generated by the cavitation process can yield highly medium, ultrasound input) of powder sonication reactive species within a sonicated medium, in a procedures. Asymptotic and peaking (coagulation) process known as ‘sonic activation’. As an indication behaviors upon sonication have been reported for a of the importance, variety, and complexity of sonica- wide variety of engineered nanomaterials. The fol- tion induced physicochemical effects in liquid media, lowing are some examples, where the respective in recent years, sonochemistry has emerged as a distinct asymptotic or peaking energy thresholds are also discipline devoted to the study of a wide range of indicated in each case: TiO2 (rutile) – 60 s at 7 W sonication induced chemical reactions in both homo- (Bihari et al. 2008); TiO2 (anatase and rutile) – (100– geneous and heterogeneous solid and liquid systems 300) s at 750 W (Jiang et al. 2009); ZrO2 – 100 s at (Mason 1989; Mason and Peters 2003). 160 W (Vasylkiv and Sakka 2001); Au – 60 min at An example of sonic activation of particular rele- 40 W Á cm-2 (Radziuk et al. 2010); Ag – 10 s at (35–40) vance to environmental and toxicological studies is W (Murdock et al. 2008). It must be noted that in that of water sonolysis. Sonicated water partly dis- these cases the reported delivered energies do not sociates into hydrogen and hydroxide radicals, with conform to the guidelines specified herein and there- concentration and lifetime dependent on the sonica- fore may not accurately reflect the actual energy tion conditions (e.g., sonication time and energy densities to which the ENPs were subjected. density) (Makino et al. 1983; Yanagida et al. Furthermore, while the disruptive effect of sonica- 1999), that subsequently recombine to form hydrogen tion is in most cases instrumental to the achievement peroxide (Brown and Goodman 1965; Mason and of nanoscale particles, it can also result in either Peters 2003). Any materials or chemical species pres- desired or undesired physical changes that may alter ent in sonicated water will thus be exposed to low (and the inherent properties of the dispersed material. variable) concentrations of highly reactive species, A mainstream example of this phenomenon is the and as a result may undergo oxidative and other sonication induced scission of carbon nanotubes chemical transformations. (Forrest and Alexander 2007; Huang et al. 2009; Uncoated, sonicated ENPs may undergo changes Lucas et al. 2009). in their surface chemistry due to the formation of While it may not be possible to completely elimi- sonolysis radicals. As previously mentioned, the soni- nate all of the aforementioned phenomena, it is crit- cation of water can yield reactive radicals that can ical to optmimize the sonication procedure in order to initiate a variety of chemical reactions. The generated minimize the undesirable effects, while attaining the hydrogen and hydroxyl radicals, and peroxide mole- desired level of particle disruption. The recom- cules, can interact with the material’s surface, chang- For personal use only. mended optimization steps are detailed in a subse- ing its oxidation state or inducing the hydroxylation of quent section. the materials’ surface, potentially altering the material’s hydrophilicity and stability, as well as its chemical Chemical effects of sonication compatibility with other medium components. Chemisorbed hydrogen radicals generated during As with size, surface chemistry – and consequently sonication could potentially lead to the irreversible surface charge – plays a critical role in governing the disintegration of sonicated carbon nanomaterials, environmental and biological interactions of ENPs. such as fullerenes and nanotubes, via ‘hydrogen- Surface chemistry, whether it is that of the pristine induced exfoliation’ (Berber and Tomanek 2009). material or includes the effects of a surface coating or Other changes in the material’s surface chemistry, capping agent, largely determines the adsorption such as oxidation or hydroxylation, may result in behavior at the solid-solution interface, the organic/ an increased aqueous solubility and result in an inorganic phase partitioning, colloidal stability in enhanced leaching of ionic or soluble species into aqueous media, reactivity with other medium com- the medium. The potentiality of sonication-

Nanotoxicology Downloaded from informahealthcare.com by NIST National Institiutes of Standards & Technology on 12/03/10 ponents, and afﬁnity towards cell membrane walls enhanced leaching should be given special consider- (Guzman et al. 2006; Li and Huang 2008; Jiang et al. ation when testing the biological interactions of ENPs 2009; Kittler et al. 2010; Walczyk et al. 2010), to with known toxicity in their molecular or ionic form; a name just a few examples. Surface chemistry is thus a case in point is nanoscale silver (Wijnhoven et al. key parameter that affects the fate, transport, bioavail- 2009; Elzey and Grassian 2010), in which the oxi- ability, and bioaccumulation behavior of ENPs. It is dized Ag+ form is well documented to exhibit biocidal therefore critical to consider the potential for changes properties. in surface chemistry in ENP suspensions arising from Moreover, if the dispersion medium is not just the application of ultrasonic energy. water, but incorporates other chemical species, the 8 J. S. Taurozzi et al.

extent and degree of formation of radicals from both account for the effect of sonication intensity, temper- water and other medium components, and the com- ature, and sugar concentration. In all cases, a signif- plexity of their interactions with the dispersed material icant decrease in the sugars’ molecular weight at can be further magnified. Such potentiality for soni- increasing sonication times and powers was observed. cation induced changes in the surface chemistry, and Protein and enzyme sonolytical degradation has in particular its oxidation state or the hydroxylation of been broadly studied as well. Wang et al. (2009c) the material’s surface, should be given special con- examined the sonication-induced damage to bovine sideration in light of recent findings that attribute serum albumin (BSA) both in the presence and ENP’s toxic effects to oxidative stress and reactive absence of metal complexes. In all cases, even at oxygen species (ROS)-mediated events, among other sonication powers as low as 20 W, ultrasonic irradi- mechanisms (Unfried et al. 2007; Wang et al. ation was shown to damage the BSA structure, the 2009a; Kim et al. 2010). extent of which increased with increasing sonication While radical induced changes in the pristine mate- time. Coakley et al. (1973) measured the sonolytical rial’s surface might be restricted to the period of inactivation of alcohol dehydrogenase and lysozyme sonication treatment and be potentially reversible – exposed to a 20 kHz ultrasonic field as a function of although to the best of our knowledge this has sonication time (at 90 W), and measured a 70% yet to be determined – it is unquestionable that inactivation of alcohol dehydrogenase and lysozyme sonication-specific, radical-mediated changes have after 10 and 25 min of sonication, respectively. been observed as a result of sonication, as discussed A more comprehensive review and discussions on below, and can significantly alter the suspension sonication induced alterations of relevant biological medium properties or the dispersed material’s surface compounds (e.g., enzymes, DNA, biological mem- chemistry, and consequently influence its environ- branes) can be found in the works of Williams (1983) mental and biological behavior. and Riesz and Kondo (1992). If organic molecules are present when sonicating Alterations in the molecular structure of common ENP suspensions, either as components of the dis- organic compounds (e.g., sugars, proteins) as a result persing medium or as a surface coating on the ENP of sonication may elicit specific biological responses in itself, they can be irreversibly degraded via direct the animal model or assay upon inoculation with the sonolysis from the dissipated cavitational energy sonicated suspension, thereby yielding false positives (Williams 1983; Wang et al. 2009c). Furthermore, or negatives that could incorrectly be attributed to the sonication-induced active radicals, such as those pro- ENPs, rather than the degraded medium. To account duced from water sonolysis, have been shown to for this possibility, we recommend using a superna- cleave polymers creating smaller oligomers and to tant control in which the medium, in the absence of For personal use only. activate polymerization reactions (so called sonopo- ENPs, is subjected to the sonication procedure and lymerization), producing new polymeric species and tested in parallel with the ENP suspensions. irreversibly modifying the organic molecules present Furthermore, sonication-induced transformations in the suspension (Brown and Goodman 1965; in the medium could potentially result in a loss of Mason 1989). Sonication can also induce covalent the medium’s stabilizing properties, therefore reduc- bonding of organic functionalities to the ENP surface ing its ability to maintain the ENPs in suspended non- (Liu et al. 2005). The irreversible degradation agglomerated form. For example, species-specific and/or transformation of organic molecules due to albumin homologs, such as BSA, have been utilized sonication is a potentiality that should be considered as stabilizing components in dispersion media when preparing ENP suspensions for environmental intended for toxicity testing of ENPs (Bihari et al. or biological testing. 2008; Porter et al. 2008; Kittler et al. 2010; As a result of sonolysis, the sonicated medium may Tantra et al. 2010), while humic acid is commonly not retain its native properties. Degraded medium used as a medium component and organic stabilizer components, such as denatured proteins, will likely added to ENP suspensions to simulate environmental

Nanotoxicology Downloaded from informahealthcare.com by NIST National Institiutes of Standards & Technology on 12/03/10 interact differently with the ENPs or the intended matrices (Chen and Elimelech 2007; Pallem et al. assay, than they would in their pristine, non- 2009; Saleh, Pfefferle and Elimelech 2010). Typi- sonicated form. Sugars and proteins are two major cally, the albumin or humic acid is present when biological and environmental media components. the suspension is sonicated. Naddeo et al. (2007) Basedow and Ebert studied the molecular degrada- demonstrated that humic acid is degraded (either tion of dextran as a function of exposure time via oxidative mechanisms or physical aggregation) under 20 kHz, low intensity (10 W cm-2) sonication when subject to 20 kHz sonication, the extent of (Basedow and Ebert 1979), while in a later which was found to increase with sonication intensity work Lorimer et al. (1995) extended the study to and organic concentration. Preliminary studies in our Dispersion of nanoparticles 9

own laboratory on the stabilization effect of BSA on disperse the ENPs, but the extra effort will result in sonicated TiO2 powders have shown a measurable less perturbation of the chemical species present. The loss of nanoparticle stability against agglomeration formation of radicals can also be mitigated by mini- when BSA is sonicated together with the source mizing the sonication power, again prompting the powder, as opposed to when BSA is added post- need for optimizing the dispersion procedure to min- sonication. In the latter case, the particles did not imize undesired effects while obtaining the desired exhibit agglomeration over the same time frame. dispersion properties. Furthermore, nitrogen or car- These results will be published as part of a separate bon dioxide bubbling, or the addition of dry ice or research article currently in preparation. other radical scavengers, can quench free radicals Moreover, the ultrasonic degradation of organic during the sonication process and minimize their molecules has the potential to irreversibly modify potential impact (Kondo et al. 1986; Honda et al. the chemical structure of ENP functional surface 2002). coatings (e.g., polyethylene glycol [PEG], and poly- vinylpyrrolidone [PVP]), or even a complete loss of Measuring and reporting delivered power the coating itself. Kawasaki et al. (2007) studied the ultrasonic degradation of PEG as a function of son- A critical parameter for the reproducible preparation ication time at 28 kHz and 20 W, over a wide range of of sonicated ENP suspensions is the delivered soni- molecular weights. In all cases, radical mediated cation power. As previously mentioned, only a por- sonolysis caused C-O bond scission and molecule tion of the total output power is effectively consumed breakage. Taghizadeh and Bahadori (2009) measured in particle disruption. the degradation kinetics of PVP when subject to The energy loss and consequently the efficiency of 24 kHz 100 W sonication, revealing a dependence the energy transformation (or the ‘energy yield’) of the degradation rate on molecular weight and from the electrical source to the acoustic power polymer concentration. Such potential sonication- effectively received by the sonicated suspension is specific alterations of organic surface coatings may heavily instrument dependent (on the instrument’s adversely impact the ENP’s capacity to remain in particular driving circuitry, self-adjusting feedback suspension, as particle coatings often provide protec- loops, and its various electrical components). Fur- tive electrostatic or steric stabilization. Ultrasonic thermore, the displayed power omits thermal and degradation of organic surface coatings could also mechanical energy losses, which are also instrument modify the ENP’s interactions with other medium specific, that occur as the output power is transferred components, interfering with the ENP’s transport, ‘downstream’ through the sonication probe and into adsorption, or toxicological behavior. the suspension (Figure 3). For personal use only. Additionally, ENPs may act as sonolysis catalysts, Therefore, simply reporting the sonicator’spower synergistically enhancing the aforementioned effects setting value (e.g., ‘setting 5’ or ‘60%’)orthedis- when sonicated in the presence of organics. played power does not provide an accurate indica- Wang et al demonstrated the sonocatalytic effect of tion of the effective acoustic power delivered to the nano-sized silica and titanium dioxide towards the sonicated suspension, and are as such reporting degradation of BSA, showing an enhanced degrada- parameters that do not allow for transferable or tion of BSA when the ENPs were present in the reproducible results. Two different instruments sonicated solution, with respect to ENP-free controls operating at a ‘level 5’,ora‘60%’ setting can deliver (Wang et al. 2005, 2009b). significantly different effective acoustic powers to the For these reasons, the chemical complexity of the same suspension, as they may translate into different sonicated suspension and the particle coating should oscillation amplitudes for different instruments. be reduced to the greatest possible extent. Whenever Alternatively, instrument ‘A’ may consume – and possible, organic compounds should be spared from display – 50 W from the electrical source to produce sonication and the formation of radicals minimized. a probe oscillation amplitude of 3 mm, while instru-

Nanotoxicology Downloaded from informahealthcare.com by NIST National Institiutes of Standards & Technology on 12/03/10 A proposed approach for the preparation of ENP ment ‘B’ may produce a probe oscillation amplitude suspensions is, when applicable, to ﬁrst sonicate the of 2 mm by consuming 50 W, thereby delivering pristine, uncoated powder in de-ionized water (or different powers to a suspension even when showing other appropriate pH adjusted aqueous solution) the same displayed power. Published works have and then mix the sonicated suspension into the discussed and demonstrated that the power reading desired, non-sonicated, dispersion medium, or add displayed by the instrument bears poor correlation the stabilizing agent and other organic additives to the with the measured power received by the sonicated suspension after sonication. This may require signiﬁ- suspension (Contamine et al. 1995; Kimura et al. cant additional effort to properly formulate and 1996; Raso et al. 1999). This issue is broadly known 10 J. S. Taurozzi et al.

in the sonochemistry ﬁeld as ‘the reproducibility problem’ (Mason 1991; Contamine et al. 1995; Kimura et al. 1996). While there is yet no available standardized procedure for reporting the effective acoustic energy delivered to liquid suspensions, we hereby propose the calorimetric method as a simple procedure to establish a standardized, instrument-independent, transferable and reproducible method for the measurement and reporting of sonication power. It is our belief that by implementing such an approach, variabilities derived from the inconsistent application and reporting of ultrasonic disruption will be reduced.

The calorimetric method

We propose a simple and easily transferable set of experimental guidelines for the determination of calorimetric curves. The calorimetric method has been proposed to standardize sonochemistry studies and used to calculate the amount of acoustic energy delivered to a liquid medium subject to direct sonication Figure 5. Set-up for the measurement of calorimetric curves. Inset (Kimura et al. 1996; Zhu et al. 2009). Calorimetry is a shows sonicator tip and temperature probe immersion depths. relatively simple, rapid, and inexpensive procedure Sound protection enclosure was omitted for clarity of the image. that allows for the direct measurement of the acoustic energy delivered to a sonicated liquid (Mason 1991). (2) Determine the mass of the liquid using a top load- The method is based on the measurement of the ing balance. First tare the empty 600 mL beaker; temperature increase in a liquid medium over time, (3) Immerse the sonicator probe (horn) approxi- as a bulk effect of the cavitational process induced in a mately 2.5 cm (1 inch) below the liquid surface; liquid by an ultrasound probe. (4) Immerse a temperature probe connected to a For personal use only. At a given sonicator setting, the temperature temperature meter and data logger (e.g., an increase in the liquid is recorded over time and the Extech HD 2001 temperature meter and data- effective delivered power can then be determined logger coupled to a Type K immersion temper- using the following equation: ature probe). The temperature probe tip should be about 1 cm from the sonicator probe (see dT P= MCp ,()2 Figure 5); dt (5) Select a sonicator output power setting and, operating in continuous mode, record the water where P is the delivered acoustic power (W), T and t temperature increase for the initial 5 min. Dur- are temperature (K) and time (s), respectively, C is p ing sonication, ensure that the beaker does not the specific heat of the liquid (J g-1 K-1) and M is the shift position, especially when operating at high mass of liquid (g). power settings; this can be accomplished for By obtaining calorimetric curves following a stan- instance, using a clamp attached to a ring stand; dardized procedure, users can report and reproduce (6) Using the recorded values, create a temperature the power level applied to a sonicated suspension in a Nanotoxicology Downloaded from informahealthcare.com by NIST National Institiutes of Standards & Technology on 12/03/10 vs. time curve and obtain the best linear fit for manner that is easily transferrable and instrument the curve using least squares regression. independent. The following protocol is proposed for the determination of direct sonication calorimetric curves (Taurozzi et al. 2010a): 1. Certain trade names and company products are mentioned in the text or identified in illustrations in order to specify adequately the experimental (1) Fill a 600 mL cylindrical borosilicate beaker procedure and equipment used. In no case does such identification imply recommendation or endorsement by National Institute of Standards and with 500 mL of de-ionized water (resistivity Technology, nor does it imply that the products are necessarily the best >18 MW Á cm); available for the purpose. Dispersion of nanoparticles 11

With the obtained slope, calculate the delivered power these modifications to the protocol would need to be using Equation 2. The procedure should be repeated reported. for all desired power settings, relating each setting Figure 6 shows calorimetric curves and calculated level to the measured power. The power obtained via delivered power at different sonicator settings, calorimetry is the value we recommend for reporting obtained using the above method for a Branson 1 power used in sonication-based dispersion proce- 450 analog probe sonicator with a standard /2-inch dures. By doing so, laboratories using different com- probe and tip. mercial ultrasonic disruptors should be able to reproduce a similar delivered power by selecting the Additional considerations appropriate calibrated setting. It is important to note that the intended use of calorimetry curves is not to As noted previously, sonication is a highly system measure the actual fraction of power utilized for specific dispersion procedure, involving a variety of powder disruption under specific dispersion condi- complex concomitant physicochemical interactions in tions, but simply to establish a standardized proce- both the sonicated ENP and the suspension medium. dure to relate instrument setting levels to measurable Therefore, for a given system, optimal sonication powers under a fixed set of controlled conditions, and conditions can only be determined by considering thus allow for the reporting and transference of son- the effects of a variety of sonication parameters on ication power levels between laboratories and users. the dispersion state of the suspension under a broad The robustness of the method was tested under range of relevant conditions. As follows, a discussion different scenarios, including variations in tempera- of the effects of these parameters is offered along with ture probe location, water volume, and the use of recommendations for implementation of ultrasonic stirring. While variations in temperature probe loca- disruption as a preparatory tool. Two parameters, tion and stirring only resulted in minor variations of namely powder properties and sonication power, the measured power (< 5%), it should be emphasized are not discussed here as they have already been that using a different water volume, a container of a addressed in previous sections. different material or a different type of horn tip than those recommended here can result in changes in the Temperature measured temperature versus time slope for a given power setting due to variations in the rate of heat During sonication, the extreme local heating cycles transference between the sonicated water and the that take place at the micro-scale bubble interface due environment. If conditions other than those recom- to cavitation will result in bulk heating of the liquid mended here are used to measure calorimetric curves, over time. Excessive bulk heating can cause For personal use only.

1.9 Setting 1, 8.85 W 1.8 Setting 3, 32.71 W 1.7 Setting 6, 60.50 W 1.6 Setting 8, 104.1 W 1.5 Setting 10, 134.7 W 1.4

1.3

1.2 Normalized temperature (T/T0) Nanotoxicology Downloaded from informahealthcare.com by NIST National Institiutes of Standards & Technology on 12/03/10

1.1

1 0 50 100 150 200 250 300 Time (s)

Figure 6. Calorimetric curves, linear ﬁts, and corresponding calculated power values for different operational settings, obtained using a Branson 450 analog sonicator. For ease of illustration, the y-axis shows the recorded temperatures normalized to the initial temperature (T0) in each case. 12 J. S. Taurozzi et al.

evaporative loss of liquid, resulting in changes in the off) intervals. The duration of on and off intervals can sample volume after sonication. Moreover, high sus- be regulated. Operating in pulsed mode retards the pension temperatures may lead to degradation of the rate of temperature increase in the medium resulting ENP or other medium components if proper precau- from the cavitation process, minimizing unwanted tions are not employed. side effects and allowing better temperature control A common approach to minimize temperature than with continuous mode operation. driven side effects is to keep the sample from experiencing substantial high-temperature excursions by Sample volume and concentration immersing the suspension container in a cooling bath. The container should be immersed to a level roughly While sonication power and time describe the amount equal to that of the contained suspension. of energy delivered to the suspension, the work done An ice-water bath is often sufficient to keep suspen- on samples of different volumes and particle con- sions from overheating. If, however, ice water is not centrations can differ. At constant volume, higher sufficient to cool the sample under intense sonication particle concentrations result in an increased particle (e.g., the sample volume is too low, the sonication collision frequency. In principle, an increased colli- time is too long, or high sonication powers are used) sion frequency can enhance particle breakage due to an ice-salt bath could be considered as an option. an increase in particle-particle impact events. How- Working with containers made of materials with ever, if sufficient local activation and sintering ener- high thermal conductivities ensures a rapid release of gies are achieved, increased collision frequencies can the generated heat from the suspension into the also induce agglomerate or aggregate formation as surrounding cooling bath. With regard to thermal particles collide and coalesce (Figure 4). The effect of properties, the following container materials work concentration is thus dependent on both the energy best, in order of decreasing thermal conductivity: delivered into the suspension and the physiochemical aluminum, stainless steel, glass, and plastic (e.g., properties of the suspension. polyethylene). When selecting the container, consid- Particle concentration can also alter the acoustic eration must also be given to the chemical compat- properties of the suspension, affecting the medium’s ibility between the container material and the viscosity and acoustic conductivity and thus the suspension components, especially under intense effectively delivered power. However, at the concen- local heating conditions; for instance, aluminum is trations commonly used for environmental and incompatible with acidic suspensions, and glass is toxicological evaluations, typically in the 50– incompatible with alkaline solutions. 2000 mg Á mL-1 range, a significant impact on the Additionally, an aluminum foil or thin thermoplas- medium’s acoustic properties is not anticipated. For personal use only. tic film with an opening just large enough for the Where particle concentration may alter acoustic prop- probe to enter untouched, will reduce the evaporative erties is when working with concentrated stock sus- loss of liquid content, especially when sonicating in pensions prior to dilution. volatile solvents or for substantial durations. Cooling The effect of suspension volume (at constant par- the sample will also reduce evaporative loss. This ticle concentration) is often measured as energy density cover will also minimize the potential release of aero- (W s mL-1). This magnitude expresses the amount of sols generated by the sonication process. delivered energy per unit of suspension volume. In principle, at equal power and particulate concentration, higher energy densities (i.e., smaller suspension Sonication time and operation mode volumes) will result in a higher disruptive effect. It The total amount of energy (E) delivered to a sus- should be emphasized that when working with small pension not only depends on the applied power (P) volumes the temperature of the suspension will rise at but also on the total amount of time (t) that the faster rates and more intense cooling conditions may suspension is subject to the ultrasonic treatment: be required in this case. Nanotoxicology Downloaded from informahealthcare.com by NIST National Institiutes of Standards & Technology on 12/03/10 E=P× t ()3 Sonicator probe, container geometry, and tip immersion

Consequently, two suspensions treated at the same The sonicator probe is the acoustic element that power for different times can show signiﬁcantly dif- conducts the acoustic energy from the transducer ferent dispersion states. into the sonicated suspension. The amount of acous- Ultrasonic disruptors typically can operate in either tic energy transferred to the suspension will heavily continuous or pulsed mode. In pulsed mode, ultra- depend on the shape of the sonicator probe and its sonic intervals are alternated with static (sonication immersion depth in the suspension. Dispersion of nanoparticles 13

The vibrational amplitude of the horn at a given Aerosoling may be indicated by fluctuations in the power is dependent on the horn taper (Berlan and audible sound pitch during operation, fluctuating Mason 1992). Tapered probes act as vibration ampli- power readings, or by the appearance of a fine spray tude magnifiers, and consequently allow for higher in the vicinity of the probe. If aerosoling is observed, it delivered powers (see Equation 1). For example, can usually be eliminated by lowering the sonication stepped horn probes allow for up to 16-fold power probe deeper into the suspension. amplifications, while exponentially tapered probes Additionally, if surfactants are present, the suspen- can yield power densities on the order of hundreds sion could generate a foam during sonication, and the of W Á cm-2 (Mason 1989). presence of a stable foam in contact with the horn At equal sonicator power settings, microtip probes surface will interfere with the delivery of ultrasonic vibrate with larger amplitudes than conventional flat tip energy to the suspension in a self-limiting process. horns. However, microtips are less mechanically robust Pulse mode operation with long off periods will help and often limited in terms of the maximum power avoid foaming in samples subject to this effect. The setting at which they can be used. The manufacturer of appearance of foaming may require that operation is the sonicator will typically specify a maximum power ceased until the foam dissipates. Anti-foaming agents setting (based on a percentage of the maximum power can be effective, but would probably not be compat- output setting) for use with microtips. ible with toxicological studies. The way in which the sonic energy is distributed within the suspension is also heavily influenced by the Tip maintenance container geometry. For example, in one published study (Pugin 1987) acoustic energy distribution pro- As the source of ultrasonic waves, all sonicator probes files measured in round bottom flasks subject to are subject to progressive erosion due to the intense indirect sonication significantly differed from those cavitation that occurs in the immediate vicinity of the of flat bottom (e.g., Erlenmeyer) flasks subject to probe tip. Tip erosion is therefore an unavoidable side similar sonication conditions. Conical bottom, flat effect in direct sonication. There are two effects of tip bottom, and round bottom flasks will also show erosion that are relevant to the present discussion. different energy maxima and minima distribution During erosion, microscopic tip residues (typically profiles, which will in turn vary significantly for dif- titanium metal) are released from the tip into the ferent probe tapers and probe tip immersion depths. sonicated suspension, introducing an impurity and When possible, using the smallest diameter vessel potentially contaminating the suspension. Operation- that allows for the probe to be inserted without ally, tip erosion also results in a reduced energy touching the container walls is probably the best output. The extent of potential contamination from For personal use only. approach. Using smaller container diameters raises tip erosion can be evaluated by performing a control the height of the liquid and maximizes the liquid- experiment, or by measuring the amount of metal probe surface area exposed to the acoustic waves, as released into the sonicated medium (e.g., via diges- well as the container wall surface/volume ratio for tion followed by ICP-MS analysis) under relevant dissipation of heat by the cooling bath. Cylindrical sonication conditions. Erosion contamination can shaped beakers work well, especially for small be minimized by adequate and frequent tip mainte- volumes. nance procedures, as explained below, or by using Probe immersion depths between 2 and 5 cm are indirect sonication. generally recommended when operating with stan- The extent and occurrence of tip erosion depends dard horns having a flat radiating surface (tip) or with on the intensity, duration, and frequency of sonica- microtips. Probes should be placed no closer than tion cycles, as well as the chemical properties and about 1 cm from the bottom of the sample container, concentration of the sonicated suspensions. If used on and contact between the probe and the container walls a daily basis, tips should be inspected weekly for signs should always be avoided. of erosion; tip erosion can be recognized by the

Nanotoxicology Downloaded from informahealthcare.com by NIST National Institiutes of Standards & Technology on 12/03/10 appearance of a grayish matting, as opposed to the Aerosoling and foaming shinny lustrous appearance of a non-eroded tip (see Figure 7). If the sonicator probe is not sufficiently immersed in A mildly eroded (i.e., matted) tip or horn surface the suspension it can give rise to surface agitation can be reconstituted (dressed) by buffing with a very resulting in nebulization (i.e., the formation and fine grit carbide paper or emery cloth (refer to the tip release of aerosols). This could pose a risk if ENPs or sonicator manufacturer for the correct grit size; do or other potentially harmful medium components are not use conventional sand paper) (Berliner 2010). released in this manner. Substantially eroded tips or horn surfaces (i.e., with 14 J. S. Taurozzi et al.

1 Figure 7. Active surface of removable /2-inch ﬂat tips, ranging from pristine (left) to matted (center) to eroded with extensive pitting (right).

visible pitting) should be replaced, as these cannot be As shown by Equation 5, acoustic attenuation 1 reconstituted. Standard /2-inch diameter horns typ- increases with frequency for a given medium. For ically have a removable threaded tip, but larger probes this reason, to achieve equivalent energy densities, are likely to consist of a solid horn and thus greater higher power inputs are required when operating at care may be required to dress the horn’s active surface higher frequencies. For example, to achieve an acous- if it becomes matted. tic intensity of 20 W Á cm-2 at a distance of 10 cm, It is also essential to ensure that removable probe would require approximately 54% greater source tips are tightly coupled to the horn, and do not loosen intensity at 10 MHz compared with that required at as a result of prolonged operation; a loose tip will 10 kHz (Mason 1989). result in a reduced acoustic power conversion and Additionally, higher viscosities will dampen the energy yield. cavitation process, requiring higher power inputs to achieve dispersion. The medium’s density and acoustic wave speed (c) will also impact on the amount of Medium properties vibrational energy that is transformed into acoustic As previously noted, the ultrasonic energy delivered to energy (see Equation 1). a suspension is partly attenuated and dissipated by the The medium’s viscosity, density, acoustic wave suspending medium. The amount of energy lost to the speed, and chemical composition are all important medium at a given acoustic frequency is governed by parameters that impact on the amount of delivered its physicochemical properties and chemical energy that is effectively utilized to disrupt powder composition. clusters. For media relevant to environmental and For personal use only. Generally, the attenuation of acoustic energy in a biological testing, the salt content is likely to have given medium can be estimated using the following only a very small impact on the acoustic properties equation (Brown and Goodman 1965): relative to pure water, as the presence of physiological salt levels will slightly alter the density and viscosity of −2αx I=Iex 0 ()4 the solution. Overall, at ultrasound frequencies, these effects are probably insignificant compared to other where I0 is the acoustic intensity (power/area) at the factors discussed previously. source, Ix is the intensity at a distance x from the Yet, as noted previously, the medium’s chemical source and a is the absorption coefficient. composition is of particular relevance in terms of the The magnitude of the absorption coefficient, a,is occurrence of potential side reactions during sonica- dependent on the medium’s physicochemical prop- tion (e.g., radical activation, sonopolymerization) that erties and composition, as well as the acoustic fre- may alter the physicochemical properties of the son- quency. In principle, higher medium viscosities and icated medium and/or the sonicated nanomaterial. acoustic frequencies will result in larger a values. For

Nanotoxicology Downloaded from informahealthcare.com by NIST National Institiutes of Standards & Technology on 12/03/10 a a given liquid, can be estimated by using the Optimizing sonication for dispersion of ENPs following equation (Brown and Goodman 1965): As discussed in preceding sections, the treatment 2 μ f 2 α ∝ ()5 conditions required to achieve complete ENP disper- 3 ρ sion are highly system speciﬁc. It is therefore advi- sable to conduct dispersion optimization studies (e.g., -1 where m is the dynamic viscosity of the liquid (kg m measure particle size as a function of sonication -1 -1 s ), f is the acoustic frequency (s ) and r is the parameters) for the test system (Bihari et al. 2008). -3 liquid’s density (kg m ). While it would be unrealistic and impractical to offer a Dispersion of nanoparticles 15

unique set of sonication conditions for all possible can be evaluated by measuring the particle size dis- systems, the ultimate goal of the optimization proce- tribution (PSD) of the sonicated suspension. In prin- dure should be to achieve the desired degree of ciple, a reduction in the average particle size and particle dispersion with the least possible energy input, polydispersity (breadth of the distribution) are indic- in order to minimize unwanted side effects. ative of more effective dispersion. The parameter The process to determine such conditions is based optimization sequence is explained as follows. on a trial and error approach. The guidelines offered Once a starting set of values for time, concentra- below are intended as tools for the interpretation of tion, and volume has been selected, and keeping all the observed behavior at a given set of sonication other processing parameters constant, sonication conditions, and to determine the appropriate param- power is varied. Sonication powers above and below eter modifications to be made in response. the starting point should be tested. As a general As noted previously, the relationship between dis- guideline, the effect of power can be scanned in persion efficiency and the different sonication para- increments of 10 W both above and below the starting meters is not linear. The convolution of delivered sonication power. Using the power value that yields energy input and suspension properties can result the smallest measured size and lowest polydispersity, in either a net agglomerate breakage or cluster re- then sonication time should be scanned (e.g., in formation effect. increments of 30 s) above and below the starting For this reason, the optimization process should point. Operation in pulsed mode is recommended scan parameter values covering a range both above for sonication times greater than 1 min, particularly and below a starting point. The first trial set of when using small volumes (i.e., below 50 mL). sonication conditions for an untreated powder in For the optimized time, power, and selected vol- suspension should be intended to ensure that disper- ume, the effect of particle concentration both above sion can in fact be achieved. Being the first trial and and below the concentration starting point and down therefore likely not the optimal set of conditions, it to the minimum acceptable concentration for the should be designed to confirm that under reasonably desired application, may be evaluated if desired. If attainable sonication conditions, the powder in sus- the optimal concentration is above the desired value pension can be effectively dispersed. The starting for the particular application, the suspension’s PSD point of the optimization should therefore provide a should be measured following dilution to ensure there high energy input to the suspension. This is achieved has been no change. by selecting a relatively large sonication output power The above described approach for optimizing the level, a prolonged sonication time, and a high energy sonication conditions for dispersion is both simple density (small suspension volume). and practical, but the authors recognize that there For personal use only. Table I offers a general guideline of starting points may be other factors to consider or other optimization for power, time, and volume selection, which might approaches that work equally well; it should be con- be utilized to investigate new ENPs or media/ENP sidered as a recommendation not a prescription for combinations (Hielscher 2005). success. The choice for the starting powder concentration is subject to the desired application. It is also possible to Reporting guidelines start with concentrations higher than the target level, and then if needed dilute the processed suspension In light of the relevance of the above discussed para- post-sonication. meters and their significance in dictating the state and The effect of different processing parameters on the quality of suspensions obtained via sonication, we dispersion state of the suspension can be analyzed propose that the following parameters be reported independently by varying one parameter while keep- in published work involving the use of ultrasonic ing the others constant. For each parameter, the energy to disperse ENPs prior to environmental, effectiveness of the applied sonication conditions exposure, or biological testing (Table II). It should

Nanotoxicology Downloaded from informahealthcare.com by NIST National Institiutes of Standards & Technology on 12/03/10 be noted that many of these parameters or material characteristics would likely be reported anyhow, or Table I. Guidelines for sonication starting points. certainly should be according to several recent pub- Energy density Sample lications and recommendations (Murdock et al. 2008; Á (W s/mL) volume (mL) Power (W) Time (s) Warheit 2008; Minimum Information for Nanoma- <100 10 50 <20 terial Characterization [MinChar] Initiative 2009; 100–500 10 50 20–100 Boverhof and David 2010). The lengthiness of this list may preclude full dis- >500 10 50 >100 closure within a typical experimental section in a Nanotoxicology Downloaded from informahealthcare.com by NIST National Institiutes of Standards & Technology on 12/03/10 For personal use only. 16 .S arzie al. et Taurozzi S. J.

Table II. List of parameters and material characteristics recommended for reporting purposes.

Sonication ENP or source material Test medium Final suspension

Type of sonicator used Description of powder or ENP as received Composition including ionic strength or Order in which components were added (bath, probe immersion, (to the extent relevant and known) molar salt concentrations and whether or not they were present inverted cup) during sonication Sonicator make, model, For commercial ENPs, product trade name For commercial media, source, product Post-treatment storage conditions prior to rated power output, frequency, and/or product number together with and batch number(s), as applicable testing (e.g., container material, temperature, probe type/diameter (if used) batch/lot number if available exposure to light, exposure to oxygen, time) Sample container volume and type Primary particle size and method of measurement Initial pH (preferably measured, otherwise Whether endotoxin was evaluated or (e.g., TEM) nominal buffer value) procedures applied to avoid/remove endotoxins (relevant if suspension is to be used for toxicity testing) Tip/probe immersion depth (if used) Gross particle morphology (e.g., spheroidal, Dispersing agent(s) and concentration(s), Conditions of filtration and/or centrifugation tubular, fibrous, platelet, aggregate) if used used (where appropriate) Operation mode (continuous or pulsed), Specific surface area (for powders) and method Source and quality of water if test medium Post-treatment mass concentration of if pulsed indicate pulse duration of measurement was prepared or diluted on-site particle phase Total cumulative treatment time Identity of coatings or surface functionality Any relevant details relating to storage, Final pH if known dilution, or preparation of medium Power input to sample (as measured for Particle size distribution in the relevant probe sonicators: See above and also suspending medium and method Taurozzi et al. (2010a) of measurement Whether cooling bath was used; if not temperature rise during sonication Mass of powder and total volume of suspension sonicated Dispersion of nanoparticles 17

refereed journal, but could easily be accommodated environmental and toxicological studies. While we in the supplemental information supported by most do not intend to discourage the use of ultrasonic scientific journals. The recommendations may also be disruption to produce such suspensions, our goal viewed as a checklist for identifying issues that may here is to promote a rational use of sonication, in need to be addressed during the selection or devel- order to ensure that unintended side effects, such as opment of a test material or dispersion procedure. For sonolysis, are minimized and that the results obtained a more detailed and formal guidance document on using such suspensions are accurately interpreted. reporting, refer to Taurozzi et al. (2010b) or consult Additionally, by providing a set of guidelines for the references cited above. the standardized measurement and reporting of rel- Additional parameters and conditions may need to evant sonication parameters, we seek to improve the be reported in order to fully describe the test sample reproducibility of related studies and allow for a and experiment in their entirety. For commercial comprehensive peer review process, which may be sources, steps should be taken, if possible, to ensure otherwise hindered by insufficient information that the as-received material reflects the man- regarding the preparation of the tested suspensions. ufacturer’s description – for instance, comparison We do not attempt to address the nearly infinite range of powder characterization to nominal specifications of possible experimental scenarios, but instead offer a and typical values. Examples might include moisture starting point towards the reproducible preparation content by TGA, primary particle or aggregate size by and comprehensive evaluation of ENP suspensions. electron microscopy, or purity determination using analytical methods. Under certain circumstances it Acknowledgements may be prudent to test for the presence of contami- nants arising from the dispersion treatment process The authors thank Fred Klaessig of Pennsylvania Bio itself, such as additional dissolved species (e.g., metal Nano Systems, LLC, for helpful suggestions and for ions, organic carbon). It may be equally important to his critical review of the draft manuscript. The authors test for the loss of species that should be present, such would also like to thank participants of the Interna- as constituents of the original test medium. The fate tional Alliance of NanoEHS Harmonization (IANH) of dispersing agents (e.g., surfactants, proteins, for useful discussions, some of which helped motivate sugars) may be relevant for both environmental and the present work. biological testing. Declaration of interest: Other material-specific parameters that are com- This work was funded in monly mentioned as reportable include aspect ratio, part by a cooperative research agreement crystalline form and content, amorphous content, and (70NANB9H9166) between NIST and the Center For personal use only. zeta potential from electrophoretic mobility or other for the Environmental Implications of Nanotechnol- fl measure (such as the isoelectric point or zero point of ogy (CEINT). The authors report no con ict of charge) related to the particle surface charge proper- interest. The authors alone are responsible for the ties; zeta potential is both material and medium content and writing of the paper. dependent, and typically varies with pH for materials that contain acidic or basic moieties, so the pH and References ionic strength of the sample should always accompany Aoki M, Ring TA, Haggerty JS. 1987. Analysis and modeling of zeta potential results. It may also be necessary, and the ultrasonic dispersion technique. Adv Ceram Mater certainly prudent to demonstrate that the post- 2(3A):209–212. dispersion suspension is stable over the time period ASTM International (formally the American Society for Testing and Materials). 2006. Standard terminology relating to nano- of the experiment; for instance, particle size could be technology. E2456-06. West Conshohocken, PA: ASTM monitored in situ (e.g., using dynamic light scatter- International. ing) over a period of time congruent with the testing Barnard AS. 2006. Nanohazards: Knowledge is our first defence. protocol. The potential list of reportable physico- Nat Mater 5(4):245–248.

Nanotoxicology Downloaded from informahealthcare.com by NIST National Institiutes of Standards & Technology on 12/03/10 chemical endpoints is nearly unlimited, so prioritiza- Basedow AM, Ebert KH. 1979. Effects of mechanical stress on the reactivity of polymers – activation of acid-hydrolysis of dextran tion within the context of the intended use is critical. by ultrasound. Polym Bull 1(4):299–306. Berber S, Tomanek D. 2009. Hydrogen-induced disintegration Closing remarks of fullerenes and nanotubes: An ab initio study. Phys Rev B 80(7):075421–075425. Sonication has proven to be a simple and energy Berlan J, Mason TJ. 1992. Sonochemistry – from research labora- ﬁ tories to industrial plants. Ultrasonics 30(4):203–212. ef cient technique for the disruption of coarse Berliner S. 2010. Changing materials with high-intensity sound. powders and, as such, has become a mainstream Accessed from the website: http://berliner-ultrasonics.org/index. tool for the preparation of ENP suspensions for html. 18 J. S. Taurozzi et al.

Bihari P, Vippola M, Schultes S, Praetner M, Khandoga AG, dispersability and cell-biological activity of silver nanoparticles. Reichel CA, et al. 2008. Optimized dispersion of nanoparticles J Mater Chem 20(3):512–518. for biological in vitro and in vivo studies. Part Fibre Toxicol 5:14. Kondo T, Kuwabara M, Sato F, Kano E. 1986. Influence of Boverhof DR, David RM. 2010. Nanomaterial characterization: dissolved gases on chemical and biological effects of ultrasound. Considerations and needs for hazard assessment and safety Ultrasound Med Biol 12(2):151–155. evaluation. Anal Bioanal Chem 396(3):953–961. Kraft M. 2005. Modelling of particulate processes. KONA Brinker JC, Scherer GW. 1990. Sol-gel science: The physics and 23:18–35. chemistry of sol-gel processing. New York: Academic Press. Kreyling WG, Semmler-Behnke M, Moller W. 2006. Health Brown B, Goodman J. 1965. High intensity ultrasonics: Industrial implications of nanoparticles. J Nanopart Res 8(5):543–562. applications. Princeton, NJ: D. Van Nostrand Company. Labille J, Masion A, Ziarelli F, Rose J, Brant J, Villieras F, et al. Chen KL, Elimelech M. 2007. Influence of humic acid on the 2009. Hydration and dispersion of C-60 in aqueous systems: aggregation kinetics of fullerene (C-60) nanoparticles in mono- The nature of water-fullerene interactions. Langmuir valent and divalent electrolyte solutions. J Colloid Interf Sci 25(19):11232–11235. 309(1):126–134. Lanone S, Boczkowski J. 2006. Biomedical applications and poten- Coakley WT, Brown RC, James CJ, Gould RK. 1973. Inactivation tial health risks of nanomaterials: Molecular mechanisms. Curr of enzymes by ultrasonic cavitation at 20 kHz. Arch Biochem Mol Med 6(6):651–663. Biophys 159(2):722–729. Li SD, Huang L. 2008. Pharmacokinetics and biodistribution of Colvin VL. 2003. The potential environmental impact of engi- nanoparticles. Mol Pharmaceut 5(4):496–504. neered nanomaterials. Nat Biotechnol 21(10):1166–1170. Liu MH, Yang YL, Zhu T, Liu ZF. 2005. Chemical modification of Contamine RF, Wilhelm AM, Berlan J, Delmas H. 1995. Power single-walled carbon nanotubes with peroxytrifluoroacetic acid. measurement in sonochemistry. Ultrason Sonochem 2(1):S43–47. Carbon 43(7):1470–1478. Dobrovolskaia MA, McNeil SE. 2007. Immunological properties of Lorimer JP, Mason TJ, Cuthbert TC, Brookfield EA. 1995. Effect engineered nanomaterials. Nat Nanotechnol 2(8):469–478. of ultrasound on the degradation of aqueous native dextran. Elzey S, Grassian VH. 2010. Agglomeration, isolation and disso- Ultrason Sonochem 2(1):S55–57. lution of commercially manufactured silver nanoparticles in Lucas A, Zakri C, Maugey M, Pasquali M, van der Schoot P, aqueous environments. J Nanoparticle Res 12(5):1945–1958. Poulin P. 2009. Kinetics of nanotube and microfiber scission Forrest GA, Alexander AJ. 2007. A model for the dependence of under sonication. J Phys Chem C 113(48):20599–20605. carbon nanotube length on acid oxidation time. J Phys Chem C Makino K, Mossoba MM, Riesz P. 1983. Chemical effects of 111(29):10792–10798. ultrasound on aqueous-solutions – formation of hydroxyl radi- Guzman KAD, Finnegan MP, Banfield JF. 2006. Influence of cals and hydrogen atoms. J Phys Chem-Us 87(8):1369–1377. surface potential on aggregation and transport of titania nano- Mandzy N, Grulke E, Druffel T. 2005. Breakage of particles. Environ Sci Technol 40(24):7688–7693. TiO2 agglomerates in electrostatically stabilized aqueous disper- Gwinn MR, Vallyathan V. 2006. Nanoparticles: Health effects – sions. Powder Technol 160(2):121–126. pros and cons. Environ Health Persp 114(12):1818–1825. Mason TJ. 1989. Sonochemistry: Theory, applications and uses of Hielscher T. 2005. Ultrasonic production of nano-size dispersions ultrasound in chemistry. Chichester, UK: Ellis Horwood. and emulsions. ENS’05, Paris, France. Mason TJ. 1991. Practical sonochemistry: User’s guide to applica- Honda H, Zhao QL, Kondo T. 2002. Effects of dissolved gases and tions in chemistry and chemical engineering. Chichester, UK: an echo contrast agent on apoptosis induced by ultrasound and Ellis Horwood.

For personal use only. its mechanism via the mitochondria-caspase pathway. Ultra- Mason TJ, Peters D. 2003. Practical sonochemistry: Power ultrasound Med Biol 28(5):673–682. sound uses and applications. Chichester, UK: Ellis Horwood. Huang YY, Knowles TPJ, Terentjev EM. 2009. Strength of nano- Maynard AD. 2006. Nanotechnology: A research strategy for tubes, ﬁlaments, and nanowires from sonication-induced addressing risk, Woodrow Wilson International Center for Scho- scission. Adv Mater 21(38–39):3945–3948. lars – project on emerging nanotechnologies. Isfort CS, Rochnia M. 2009. Production and physico-chemical Minimum Information for Nanomaterial Characterization (Min- characterisation of nanoparticles. Toxicol Lett 186(3):148–151. Char) Initiative. 2009. Recommended minimum physical and International Union of Pure and Applied Chemistry (IUPAC). chemical parameters for characterizing nanomaterials on toxi- 2009. Compendium of chemical terminology. Version 2009- cology studies. Accessed from the website: http://characteriza- 09-07: Release 2.1.5. tionmatters.org/parameters. Jiang JK, Oberdörster G, Biswas P. 2009. Characterization of size, Murdock RC, Braydich-Stolle L, Schrand AM, Schlager JJ, surface charge, and agglomeration state of nanoparticle disper- Hussain SM. 2008. Characterization of nanomaterial dispersion sions for toxicological studies. J Nanopart Res 11(1):77–89. in solution prior to In vitro exposure using dynamic light Kahru A, Dubourguier HC. 2010. From ecotoxicology to nano- scattering technique. Toxicol Sci 101(2):239–253. ecotoxicology. Toxicology 269(2–3):105–119. Naddeo V, Belgiorno V, Napoli RMA. 2007. Behaviour of natural Kawasaki H, Takeda Y, Arakawa R. 2007. Mass spectrometric organic mater during ultrasonic irradiation. Desalination analysis for high molecular weight synthetic polymers using 210(1–3):175–182.

Nanotoxicology Downloaded from informahealthcare.com by NIST National Institiutes of Standards & Technology on 12/03/10 ultrasonic degradation and the mechanism of degradation. Oberdörster G, Stone V, Donaldson K. 2007. Toxicology of nano- Anal Chem 79(11):4182–4187. particles: A historical perspective. Nanotoxicology 1(1):2–25. Kim KT, Klaine SJ, Cho J, Kim SH, Kim SD. 2010. Oxidative Pallem VL, Stretz HA, Wells MJM. 2009. Evaluating aggregation of stress responses of Daphnia magna exposed to TiO2 nano- gold nanoparticles and humic substances using ﬂuorescence particles according to size fraction. Sci Total Environ spectroscopy. Environ Sci Technol 43(19):7531–7535. 408(10):2268–2272. Pan Y, Neuss S, Leifert A, Fischler M, Wen F, Simon U, et al. Kimura T, Sakamoto T, Leveque JM, Sohmiya H, Fujita M, 2007. Size-dependent cytotoxicity of gold nanoparticles. Small Ikeda S, et al. 1996. Standardization of ultrasonic power for 3(11):1941–1949. sonochemical reaction. Ultrason Sonochem 3(3):S157–161. Park BS, Smith DM, Thoma SG. 1993. Determination of agglo- Kittler S, Greulich C, Gebauer JS, Diendorf J, Treuel L, merate strength distributions. 4. Analysis of multimodal particle Ruiz L, et al. 2010. The inﬂuence of proteins on the size distributions. Powder Technol 76(2):125–133. Dispersion of nanoparticles 19

Porter D, Sriram K, Wolfarth M, Jefferson A, Schwegler-Berry D, Taurozzi JS, Hackley VA, Wiesner MR. 2010b. Reporting guide- Andrew M, et al. 2008. A biocompatible medium for nanopar- lines for the preparation of nanoparticle dispersions from dry ticle dispersion. Nanotoxicology 2(3):144–154. materials. Powers KW, Palazuelos M, Moudgil BM, Roberts SM. 2007. Teleki A, Wengeler R, Wengeler L, Nirschl H, Pratsinis SE. 2008. Characterization of the size, shape, and state of dispersion of Distinguishing between aggregates and agglomerates of flame- nanoparticles for toxicological studies. Nanotoxicology 1:42–51. made TiO2 by high-pressure dispersion. Powder Technol Project on Emerging Nanotechnologies. 2010. Woodrow Wilson 181(3):292–300. International Center for Scholars and the Pew Charitable Trusts. Unfried K, Albrecht C, Klotz LO, Von Mikecz A, Grether-Beck S, Accessed from the website: http://www.nanotechproject.org/ Schins RPF. 2007. Cellular responses to nanoparticles: Target inventories/consumer/. structures and mechanisms. Nanotoxicology 1(1):52–71. Pugin B. 1987. Qualitative characterization of ultrasound reactors Vasylkiv O, Sakka Y. 2001. Synthesis and colloidal processing of for heterogeneous sonochemistry. Ultrasonics 25(1):49–55. zirconia nanopowder. J Am Ceram Soc 84(11):2489–2494. Radziuk D, Grigoriev D, Zhang W, Su DS, Mohwald H, Walczyk D, Bombelli FB, Monopoli MP, Lynch I, Shchukin D. 2010. Ultrasound-assisted fusion of preformed Dawson KA. 2010. What the cell ‘sees’ in bionanoscience. gold nanoparticles. J Phys Chem C 114(4):1835–1843. J Am Chem Soc 132(16):5761–5768. Raimondi F, Scherer GG, Kotz R, Wokaun A. 2005. Nanoparticles Wang F, Gao F, Lan MB, Yuan HH, Huang YP, Liu JW. 2009a. in energy technology: Examples from electrochemistry and Oxidative stress contributes to silica nanoparticle-induced cyto- catalysis. Angew Chem Int Ed 44(15):2190–2209. toxicity in human embryonic kidney cells. Toxicol in Vitro Raso J, Manas P, Pagan R, Sala FJ. 1999. Influence of different 23(5):808–815. factors on the output power transferred into medium by ultra- Wang J, Ding N, Zhang ZH, Guo Y, Wang SX, Xu R, et al. 2009b. sound. Ultrason Sonochem 5(4):157–162. Investigation on damage of Bovine Serum Albumin (BSA) Rejman J, Oberle V, Zuhorn IS, Hoekstra D. 2004. Size-dependent catalyzed by nano-sized Silicon dioxide (SiO2) under ultrasonic internalization of particles via the pathways of clathrin-and irradiation using spectral methods. Spectrosc Spect Anal caveolae-mediated endocytosis. Biochem J 377:159–169. 29(4):1069–1073. Riesz P, Kondo T. 1992. Free-radical formation induced by ultra- Wang J, Wang YF, Gao J, Hu P, Guan HY, Zhang LQ, et al. 2009c. sound and its biological implications. Free Radical Bio Med Investigation on damage of BSA molecules under irradiation of 13(3):247–270. low frequency ultrasound in the presence of Fe-III-tartrate Roebben G, Ramirez-Garcia S, Hackley VA, Roesslein M, complexes. Ultrason Sonochem 16(1):41–49. Klaessig F, Kestens V, et al. 2010. Interlaboratory reproducibil- Wang J, Wu J, Zhang ZH, Zhang XD, Wang L, Xu L, et al. 2005. ity of size and surface charge measurements on nanoparticles Sonocatalytic damage of Bovine Serum Albumin (BSA) in the prior to biological impact assessment. J Nanopart Res, in review. presence of nanometer Titanium dioxide (TiO2) catalyst. Salata O. 2004. Applications of nanoparticles in biology and med- Chinese Chem Lett 16(8):1105–1108. icine. J Nanobiotechnol 2(1):1–6. Warheit DB. 2008. How meaningful are the results of nanotoxicity Saleh NB, Pfefferle LD, Elimelech M. 2010. Influence of biomacro- studies in the absence of adequate material characterization? molecules and humic acid on the aggregation kinetics of single- Toxicol Sci 101(2):183–185. walled carbon nanotubes. Environ Sci Technol 44(7):2412–2418. Wijnhoven SWP, Peijnenburg WJGM, Herberts CA, Hagens WI, Singhal A, Skandan G, Wang A, Glumac N, Kear BH, Oomen AG, Heugens EHW, et al. 2009. Nano-silver – a review Hunt RD. 1999. On nanoparticle aggregation during vapor of available data and knowledge gaps in human and environ-

For personal use only. phase synthesis. Nanostruct Mater 11(4):545–552. mental risk assessment. Nanotoxicology 3(2):109–178. Sonavane G, Tomoda K, Makino K. 2008. Biodistribution of Williams AR. 1983. Ultrasound: Biological effects and potential colloidal gold nanoparticles after intravenous administration: hazards. New York: Academic Press. Effect of particle size. Colloid Surf B 66(2):274–280. Yanagida H, Masubuchi Y, Minagawa K, Ogata T, Takimoto J, Suslick KS. 1988. Ultrasound: Its chemical, physical and biological Koyama K. 1999. A reaction kinetics model of water sono- effects. New York: VCH Publishers. lysis in the presence of a spin-trap. Ultrason Sonochem Taghizadeh MT, Bahadori A. 2009. Degradation kinetics of poly 5(4):133–139. (vinyl-pyrrolidone) under ultrasonic irradiation. J Polym Res Zachariah MR, Carrier MJ. 1999. Molecular dynamics computa- 16(5):545–554. tion of gas-phase nanoparticle sintering: A comparison with Tantra R, Tompkins J, Quincey P. 2010. Characterisation of the phenomenological models. J Aerosol Sci 30(9):1139–1151. de-agglomeration effects of bovine serum albumin on nanoparti- Zhang WX. 2003. Nanoscale iron particles for environ- cles in aqueous suspension. Colloids Surf B-Biointerfaces mental remediation: An overview. J Nanopart Res 5(3–4): 75(1):275–281. 323–332. Taurozzi JS, Hackley VA, Wiesner MR. 2010a. CEINT/NIST Zhu ZL, Minasny B, Field DJ. 2009. Adapting technology for Protocol for the preparation of nanoparticle dispersions from measuring soil aggregate dispersive energy using ultrasonic dis- powdered material using ultrasonic disruption. persion. Biosystems Eng 104(2):258–265. Nanotoxicology Downloaded from informahealthcare.com by NIST National Institiutes of Standards & Technology on 12/03/10 NOTICE OF CORRECTION

The Early Online version of this article published online ahead of print on 15 November 2010 contained an error on page 3. Equation 1 should have read

1 P=ρπ cA22(()21 f) a 2

This has been corrected for the current version.