Continuous dry dispersion of multi-walled carbon nanotubes to aerosols with high concentrations of individual fibers

Continuous dry dispersion of multi-walled carbon nanotubes to aerosols with high concentrations... JNanopartRes (2018) 20:154 https://doi.org/10.1007/s11051-018-4262-y RESEARCH PAPER Continuous dry dispersion of multi-walled carbon nanotubes to aerosols with high concentrations of individual fibers Barbara Katrin Simonow & Daniela Wenzlaff & Asmus Meyer-Plath & Nico Dziurowitz & Carmen Thim & Jana Thiel & Mikolaj Jandy & Sabine Plitzko Received: 23 November 2016 /Accepted: 25 May 2018 The Author(s) 2018 Abstract The assessment of the toxicity of airborne morphological composition of the aerosol, its fiber con- nanofibers is an important task. It relies on toxicological tent as well as fiber length and diameter distributions. inhalation studies and validated exposure measurement High fractions of individual fibers of up to 34% were techniques. Both require nanofiber-containing aerosols obtained, which shows the setup to be capable of dis- of known morphological composition and controlled persing also highly tangled MWCNT agglomerates fraction of individual fibers. Here, a dry powder disper- effectively. sion method is presented that operates with mixtures of nanofibers and microscale beads. Aerosolization exper- Keywords Multi-walled carbon nanotubes Aerosol iments of mixtures of multi-walled carbon nanotubes . . generation Dry powder dispersion Aerosol (MWCNTs) and glass beads that were continuously fed morphology into a Venturi nozzle enabled high generation rates of aerosols composed of individual and agglomerate nano- fiber structures. The aerosol process achieved good Introduction stability over more than 2 h with respect to concentra- tion and aerodynamic size distribution. Its operation The unique properties of carbon nanotubes (CNTs) make duration is limited only by the reservoir volume of the them a widely studied material with promising commer- cyclone used to separate the beads from the aerosol. The cial applications. The production capacity and use of aerosol concentration can be controlled by changing the CNTs have been reported to steadily increase (De mass ratio of MWCNTs and glass beads or by adapting Volder et al. 2013). In accord to the fiber toxicology the mass feed rate to the nozzle. For two agglomerated paradigm (Pott and Friedrichs 1972;Stanton and MWCNT materials, aerosol concentrations ranged from Wrench 1972), their fibrous morphology and 1700 to 64,000 nano-objects per cm .Comprehensive biodurability have raised concerns about potential lung scanning electron microscope analysis of filter samples carcinogenicity especially for multi-walled CNTs was performed to categorize and determine the (MWCNTs). While highly tangled MWCNT agglomer- ates with granular morphology have been reported to Electronic supplementary material The online version of this show no fiber-toxic effect (Pauluhn 2009), inhalation article (https://doi.org/10.1007/s11051-018-4262-y) contains supplementary material, which is available to authorized users. exposure studies of well-dispersed MWCNTs have ob- served asbestos-like carcinogenicity (Oberdörster et al. : : : B. K. Simonow D. Wenzlaff A. Meyer-Plath (*) 2015). Mercer et al. (2013) have shown that individual : : : : N. Dziurowitz C. Thim J. Thiel M. Jandy S. Plitzko and short MWCNT structures may translocate to sec- Federal Institute of Occupational Safety and Health (BAuA), ondary organs like the liver. Also mesothelial injury, Nöldnerstraße 40 - 42, 10317 Berlin, Germany e-mail: meyer-plath.asmus@baua.bund.de characteristic for asbestos, was observed for MWCNTs 154 Page 2 of 17 J Nanopart Res (2018) 20:154 after intraperitoneal injection of well-dispersed CNT A challenging and time-consuming step for liquid suspensions (Nagai et al. 2011; Rittinghausen et al. CNT dispersions however is the preparation of homog- 2014). The inhalative exposure to individual and fibrous enous and temporally stable CNT suspensions needed CNT structures is therefore considered a health risk, in for these methods. To obtain sufficient wettability of particular for individual MWCNTs and CNT bundles CNTs in water and other polar liquids and to promote with a geometry according to the World Health Organi- their dispersibility and stability in the solvent, signifi- zation’s (WHO) fiber definition of a length exceeding cant amounts of surfactants have to be added (Ryman- 5 μm, a diameter below 3 μm, and an aspect ratio larger Rasmussen et al. 2009; Pauluhn and Rosenbruch 2015). than 3:1 (WHO 1985). In addition, high energy liquid dispersion techniques However, the assessment of such nanoscale fibrous generally must be applied, including ultrasonication objects in air is challenging. None of the direct-reading (Shvedova et al. 2005; Ahn et al. 2011), high shear force measurement instruments currently used to assess air- mixing, or mechanical milling (Ryman-Rasmussen et al. borne nano-object exposures is capable of differentiat- 2009;Setoet al. 2010). High energy dispersion tech- ing granular and fibrous morphologies. In addition, their niques may give rise to significant shortening of fibers. response to nanofibers and low-density CNT agglomer- Chemical functionalization of CNTs (Han et al. 2010; ates is mostly unknown. It has been shown that in-depth Bahk et al. 2013;Wanget al. 2015) may increase the interpretation of scanning mobility particle sizer degree and stability of CNT suspensions. Such pre- (SMPS) and aerodynamic particle sizer (APS) data re- treatments for liquid dispersion may have considerable quires good knowledge of the aerosol morphology when effects on the physicochemical material properties. As dealing with fiber-containing aerosol (Brockmann and additional drawback, they generally result in aerosols Rader 1990;Chen et al. 2016). Systematic challenging not of pristine but surfactant-coated objects. Therefore, of such instruments with well-characterized fiber-con- aerosol generation techniques that can start from pris- taining aerosols is therefore required to study principle tine, chemically and morphologically unmodified CNT limitations and response characteristics. material appear desirable. For such inhalation toxicological and instrument re- Dry powder dispersion techniques can avoid the sponse studies, purely fiber-containing aerosols with wetting and stabilization problem of liquid-powder pro- stable concentrations of morphologically uniform and cessing. They have been successfully applied for the fully de-agglomerated nanofibers are desirable but cur- generation aerosols from pristine CNTs. Batch tech- rently not available. Considerable work is required for niques use a single batch of material that is agitated to the characterization of fiber-containing aerosols. They transfer energy to CNT agglomerates in order to release need to be sampled and analyzed microscopically at fibers, fiber fragments, and agglomerates into an air high resolution (Chen et al. 2012). flow. Powder agitation can be applied by slow or rapid Various methods for CNT aerosol generation have oscillating motion using loudspeakers (McKinney et al. been developed and reported in the past. They are either 2009; Porter et al. 2010) and so-called vortex (Lee et al. based on dry dispersion of CNT powders (Fujitani et al. 2010; Dazon et al. 2017) or linear shakers (Spurny et al. 2009;McKinney et al. 2009; Myojo et al. 2009;Lee 1975; Fujitani et al. 2009; Plitzko et al. 2010). They et al. 2010; Plitzko et al. 2010;Chen et al. 2012;Voand have been reported to be capable of releasing micro- and Zhuang 2013;O’Shaughnessy et al. 2014; Vo et al. nanoscale MWCNT structures and individual fibers. 2014) or on the spraying of liquid CNT suspensions Although stable aerosol concentration was reported to (Jennerjohn et al. 2010; Seto et al. 2010; Ahn et al. have been generated with some of these batch-style 2011; Bahk et al. 2013; Su and Cheng 2014; Wang systems (Fujitani et al. 2009;McKinney et al. 2009), et al. 2015). Pneumatic spray systems were successfully perpetual agitation of a powder batch gradually trans- used to produce individual and well-dispersed MWCNT forms the morphology of the powder. The applied ener- aerosols in high and stable concentrations (Lee et al. gy may effect both loosening and compactifying CNTs 2010; Ahn et al. 2011; Su and Cheng 2014). and their agglomerates by agglomerate abrasion and breakup or agglomerate interlocking and fiber entangle- ment processes, respectively. Batch-style dry techniques 1 may therefore require careful process optimization in The term Bobject^ will be used in the following when both granular and fibrous particles are addressed. order to maintain a perpetually agitated CNT material JNanopartRes (2018) 20:154 Page 3 of 17 154 in stable dusting condition. For some CNT material Meanwhile, the production of Baytubes has been batches, such conditions may not be achieved. Especial- discontinued. The material NM04003a from the JRC ly loosely agglomerated MWCNTs may initially release Nanomaterial Repository might serve as a substitute high aerosol concentrations. They however may exhibit (Totaro et al. 2016). SEM images of the starting mate- a tendency to interlock and compactify with progressing rials are shown in Fig. 1.TEM images of thetwo agitation duration, which would result in changes in MWCNT materials are included in the supplementary aerosol concentration and morphology. Both parameters information. are however important to assess the temporal stability of Swarcoflex glass beads (GB) (diameter 400–600 μm, aerosol generators used for instrument challenging and roundness ≥ 80%, SWARCO M. Swarovski GmbH) exposure studies. were used as received. They were mixed with the This work presents a dry dispersion technique and MWCNTs in order to transform the MWCNT powder comprehensive aerosol characterization approach to into a free-flowing powder of reduced MWCNT mass provide test aerosols that contain morphologically concentration. Prior to CNT aerosolization experiments, well-characterized high concentrations of individual the aerosol particle concentration generated by aerosol- and agglomerated fibrous structures. The method ization of different microscale glass bead types was operates with a continuous flow of fresh CNT material. studied. It does not suffer from aging effects occurring in batch For the preparation of mixtures with glass beads, agitating aerosol generators. The flow of fresh material MWCNT powder of a specific mass was added to is provided from mixtures of pristine CNTs and micro- 1 kg of glass beads in a cylindrical glass bottle of 1 l scale beads. The beads not only dilute the CNT material volume. In this work, the MWCNT concentration of the to the desired mass concentration but also transform it to mixture ranged from 10 mg/kg to 4 g/kg. The MWCNT powder was then mixed with the glass beads by slow a free-flowing powder mixture. This facilitates con- trolled continuous feeding to the aerosolization unit. revolution of the closed bottle around its axis at about Here, a Venturi nozzle generates high rates of aerosols 10–15 rpm. After typically 5 to 15 min of mixing, a from the mixture. In the following, the performance of visually homogenous, gray colored mixture was the technique is evaluated for two MWCNT materials of obtained. different agglomerate and fiber morphology. The gener- ated aerosols were injected into a home-built exposure Aerosol generation chamber that was designed with focus on high spatial aerosol uniformity and a volume flow large enough to For MWCNT aerosol generation, a dry dispersion sys- supply multiple measurement or exposure devices si- tem was developed that disperses mixtures of MWCNTs multaneously with CNT-containing aerosols. and glass beads by means of a Venturi nozzle. The mixtures were fed volumetrically into a notch in a horizontally rotating turntable. The turntable transported Materials and methods the mixture to the Venturi side inlet, where it was sucked in and was blown through the Venturi nozzle into the Materials feed line. In contrast to shaker-based vibration systems that operate on perpetually agitated batches, the system In this study, two different types of MWCNT materials presented here avoids morphological powder aging ef- were tested: ARIGM001 (industrial grade; outer diame- fects from perpetual agitation by continuously supply- ter(OD)10–30 nm; length < 15 μm; purity > 80%; Arry ing pristine CNT material to the nozzle. A downstream International Group Ltd.) and Baytubes C150P (95% two-stage cyclone system separated airborne glass carbon, OD 13 nm; median length > 1 μm; provided by beads and larger MWCNT agglomerates from the aero- Bayer Material Science AG), denoted as Baytubes in the sol fraction that was then introduced to the exposure following. Both were used without prior purification or chamber (see Fig. 2). additional treatment. Baytubes were chosen, as they are A modified SAG 410/U control unit (Topas GmbH, known to form powders of highly tangled, multi micron- Dresden, Germany) was used to control the rotational sized granular structure (Pauluhn 2009) and therefore speed of the turntable in the range from 8 to 12 rpm as tend to exhibit a low dust release propensity (cf. Fig. 1). well as the pressure of dry and particle free air that was fed 154 Page 4 of 17 J Nanopart Res (2018) 20:154 Fig. 1 Scanning electron microscope images of the starting material ARIGM001 (left) and Baytubes C150P (right). The lower right image shows the surface of a large agglomerate into the primary Venturi nozzle inlet at 4–5bar. The During operation, the feeding rate of the MWCNT/ MWCNT/GB mixture was poured from the reservoir bot- GB mixture into the nozzle was controlled via the speed tle into a stainless steel funnel. Its opening was adjusted of the turntable. Continuous weighting of the first cy- directly above the notch of 10-mm width and 5-mm depth, clone’s metal container enabled mass-controlled feed- cut into the upper surface of the turntable. The MWCNT/ ing. For the present work, feeding rates of 10–13 g/min GB mixture flowing into this notch was continuously of MWCNT/GB mixtures were used. transported towards the side inlet of a Venturi nozzle The use of several bottles and larger storage containers (ISO 5011) with an inlet nozzle diameter of 0.7 mm for the CNT/GB mixture enables long-term operation and an outlet flow of 15–20 L/min. The aerosolized mix- of the generator. For the experiments presented ture was fed into a home-built two-stage cyclone system here, for materials safety reasons, the metal container consisting of a metal container of 5 L volume, 8.5 cm of the first cyclone had no revolving door outlet as is diameter, and 22.8 cm height as first cyclone stage. A commonly used for industrial cyclones and allows for modified FSP 2 sampling head (DEHA Haan & Wittmer emptying the cyclone during operation. Therefore, the GmbH, Heimsheim, Germany) with integrated SIMPEDS operation time of our experiments was limited to about cyclone (Harris and Maguire 1968) served as second 3 h to prevent overload of the first cyclone at a mass of cyclone stage. A 300 μm stainless steel filter mesh in the about 2.5 kg glass beads. outlet of the first cyclone prevented coarser objects to enter the second stage filter in case of cyclone overload. The Instrumentation tangential inlet of 8 mm diameter and the central riser pipe outlet of 20 mm diameter and 60 mm length of the first Experimental details of the chamber used for the aero- cyclone very effectively removed the majority of glass solization and aerosol homogeneity studies are given in beads and larger CNT agglomerates from the aerosolized the Supplementary Information. powder mixture. The secondary cyclone served to further The particle number concentration in the exposure narrow down the object size distribution. chamber was monitored with a Grimm CPC model JNanopartRes (2018) 20:154 Page 5 of 17 154 Fig. 2 Schematic diagram illustrating the components and work- cyclone system separates the airborne fraction from the glass beads ing principle of the dry dispersion aerosol generator and its con- and larger MWCNT agglomerates. The aerosol composition in the nection to the exposure chamber. A mixture of glass beads and exposure chamber was monitored and sampled with a set of gas MWCNT material is continuously transported to a Venturi noz- lances, see Supplementary Information zle that aerosoloizes the mixture. Downstream, a two-stage 5.403 (Grimm Aerosol GmbH, Ainring, Germany), op- Depending on aerosol concentration, sampling times erating with a sampling flow rate of 0.3 L/min. Mea- of 5–60 min and flow rates of 2–3 L/min were applied. surement of the electrical mobility size distribution in All samplers and measurement instruments were the range of 10 to 1,100 nm was performed with a connected to stainless steel sampling gas lances of the Grimm Scanning Mobility Particle Sizer (SMPS) chamber using antistatic silicone tubes supplied by consisting of a CPC model 5.403 and an electrostatic Grimm Aerosol GmbH (Asbach et al. 2016). differential mobility analyzer (DMA) of BVienna^ type model L-DMA. The SMPS was likewise operating with Experimental procedure a sampling flow rate of 0.3 L/min at a scanning cycle time of 7 min. SMPS data analysis was done with the Before the start of an experiment, the exposure chamber software provided by the manufacturer (Grimm Soft- was flushed with dried and HEPA-14-filtered com- ware 5.477). Large objects with aerodynamic diameters pressed air until a number concentration of 0–60 parti- from 0.5–20 μm were detected with an Aerodynamic cles/cm and a relative humidity of 20–26% was Particle Sizer (APS) model TSI 3321 (TSI GmbH, Aa- reached inside of the chamber. Next, the aerosol gener- chen, Germany) equipped with a 100:1 diluting stage ator was connected to the aerosol inlet of the chamber model TSI 3302A. The APS size distribution samples and the air flow to the Venturi nozzle inlet was adjusted. were taken with scanning times of 59 s at a sampling After the start of the turntable, the MWCNT/GB mixture flow rate of 5 L/min. was poured into the funnel of the aerosol generator to For studies of morphology distributions, aerosols initiate aerosol generation (t = 0). For the present setup, were sampled on track etch membrane filters (gold- a maximum of 2.5 kg MWCNT/GB could be aerosol- coated polycarbonate with 37-mm diameter and a pore ized continuously. At the standard air flow rate of 20 L/ size of nominal 200 nm supplied by APC GmbH, min, completely flushing the 400 L exposure chamber Eschborn, Germany, using PGP sampler units (DEHA with aerosol required about 20 min. This is reflected in Haan & Wittmer GmbH, Heimsheim, Germany). the temporal development visible in Fig. 6. Plateau 154 Page 6 of 17 J Nanopart Res (2018) 20:154 concentration values were not reached before about central area of the filter was imaged at a magnification 30 min of operation. All mean concentration values of ×3000, an accelerating voltage of 3 kVand a working given in the following were obtained at minimum distance of 6.1 nm resulting in a minimum feature 30 min after start of operation. detection size of 8.3 nm (edge size of a pixel). For each All silicone tubes used for chamber and instrument sample, images with a standard resolution of 5120 × connections were cleaned or replaced after approximate- 3840 pixels and 1344 μm area were acquired at 10 to ly 3–5 experiments. Prior to injecting a new type of 15 randomly chosen filter positions and were subjected MWCNT material, the inner surfaces of the exposure to subsequent morphological characterization. chamber and all measurement lances were cleaned and all All objects imaged by SEM were categorized visual- silicone tubes were replaced to avoid cross contamination. ly according to their shape, structure, and degree of agglomeration into one of the seven object categories Determination of the system's background shown in Fig. 4. These categories do not cover all possible aerosol morphologies. They were chosen ac- concentration cording to the subject of the present study that addresses the dispersion state of fibrous and agglomerated fibrous The background concentration arising from the use of glass beads for the aerosolization process was deter- materials as generated by our setup. We therefore dif- mined for each delivered glass bead lot of about 50 kg ferentiated between objects with aspect ratios greater mass. For this purpose, 3 kg of pure glass beads was (high aspect ratio, HAR) or smaller (low aspect ratio, aerosolized with the aerosol generator at a feeding rate of LAR) than 3:1 as well as between objects with or 12 to 13 g/min. The resulting particle number concen- without visual fibrous structures. For objects with fi- tration in the exposure chamber was monitored over brous structures, we further distinguished between indi- time. After typically 30 min, a concentration plateau vidual fibers, weakly bounded agglomerates with a was reached and maintained as long as the glass beads countable number of elements, named clusters, and highly agglomerated structures with an uncountable were aerosolized. For the present study, the glass bead lot No. 058 was used. It reached a median plateau number of elements, named agglomerates in the follow- ing. Particulate, non-fibrous objects were not distin- concentration of (1143 ± 185) particles/cm ,according to CPC monitoring, averaged from minute 30 to 120. guished with respect to their agglomeration state. Fibers As shown in Fig. 3, the background particles were non- and fiber-containing agglomerates attached to granular fibrous, compact particles of relatively low SEM image objects were categorized as fibrous objects since contrast. They were easily distinguishable from fibrous attaching catalyst and catalyst support particles are a MWCNT particles during morphological analysis. common phenomenon for industrial grade CNT materials. The image manipulation software GIMP (version Electron microscopic analysis of sampled aerosols 2.8.6, GNU Image Manipulation Program) was used to visually detect and count all objects found on SEM Analyses of aerosols sampled on track etch membrane filters were performed with a scanning electron micro- images in accordance with their category. GIMP’spath tool together with an in-house developed Bscript-fu^ scope (SEM, Hitachi SU8030, Hitachi High- Technologies Europe GmbH, Krefeld, Germany). A plugin was used to measure the geometrical length of Fig. 3 Images of background particles originating from glass bead lot No. 058 during dispersion of mixtures with ARIGM001. The particles were collected on a silicon wafer by electrostatic precipitation. The white scale bar has a length of 500 nm JNanopartRes (2018) 20:154 Page 7 of 17 154 Objects free of Un-countable Individual Countable Fibres Fibers Fibres Objects HAR Agglomerates HAR Fibre HAR Fibre Cluster Individual Fibres Agglomerates LAR Particles are not distinguished LAR Fibre LAR Particles and LAR Fibre from LAR Agglomerates Cluster LAR Agglomerates Agglomerates Fig. 4 Illustration of the seven object categories used to categorize the SEM imaged objects by their visual morphology, their structure, and their degree of agglomeration individual fibers by drawing a path along the fiber and MWCNT agglomerates. They survived even if rotation- converting its length in pixels to nanometers using the al mixing durations exceeded 1 h. This suggests that ball image resolution. milling effects of the glass beads did not effectively micronize all MWCNT agglomerates. This was espe- cially the case for Baytubes MWCNTs, which were Results and discussion synthesized in highly tangled form and large agglomer- ates (see Fig. 1). Such local inhomogeneities could Mixing of MWCNTs with glass beads introduce short-term fluctuations of the aerosol concen- tration that might not be significant in large exposure To test the performance of the dry dispersion aerosoli- chambers. zation system, mixtures were prepared of glass beads The miscibility and transport behavior of MWCNTs and glass beads may be promoted by at- with either pristine ARIGM001 or pristine Baytubes MWCNTs in different mass ratios (see Table 1). Hori- tractive interaction between glass and MWCNTs. zontal rotations of the cylindrical mixing bottle around Such attraction may result from van-der-Waals inter- its rotational axis for a few minutes achieved an appar- action (dispersion forces) and from electrostatic ently homogenous distribution of the MWCNT material forces. Dispersion forces are governed by the contact in the glass beads. However, as the variations of the area between two object surfaces and are material inde- temporal stability of the aerosols generated from a con- pendent (Hamaker 1937), whereas triboelectrical charge tinuous material in Fig. 6 suggest, the homogeneity separation can lead to strong electrostatic interaction achieved by such a mixing approach cannot be assessed depending on material pairing (Matsusaka et al. with high accuracy by visual inspection alone. In some 2010). If at least one partner is an electrical insulator, batches, MWCNT-enriched or depleted zones appear to separated surface charges of opposite sign cannot re- have survived in the mixture. In future experiments, the combine easily if created by friction (tribocharging) use of alternative, e.g., wobble mixing techniques, and between to two materials of differing electronegativ- of lifting blades inside of the mixing bottle should be ity. Glass is known to exhibit a strong triboelectrically studied to further improve the mixing homogeneity. electropositive character, whereas the CNTs may Close-up visual inspection also revealed small-scale serve as electron donor partner. SEM analysis of a local inhomogeneities in the mixture caused by larger glass bead mixture with ARIGM001 showed that High aspect ratio Low aspect ratio > 3:1 < 3:1 154 Page 8 of 17 J Nanopart Res (2018) 20:154 Table 1 Overview of the prepared mixtures of MWCNTs and glass beads together with their experimental parameters for aerosolization Experiment MWCNT material MWCNT mass fraction Experimental parameters for aerosolization [g/kg] Nozzle feeding rate Nozzle outlet flow Mean object number b 3 [g/min] [L/min] concentration [#/cm ] A10 ARIGM001 0.01 11.0 16.9 4000 ± 320 A11 ARIGM001 0.01 12.0 19.8 3600 ± 130 A12 ARIGM001 0.01 13.9 19.9 2900 ± 175 A20 ARIGM001 0.06 14.7 19.9 6600 ± 580 A21 ARIGM001 0.06 10.3 19.8 8200 ± 90 A30 ARIGM001 0.20 12.4 19.8 52,000 ± 2300 A31 ARIGM001 0.20 9.7 19.8 46,000 ± 2200 A40 ARIGM001 0.33 11.0 16.9 64,000 ± 8200 B10 Baytubes 0.40 10.5 19.6 1700 ± 130 B20 Baytubes 4.00 10.5 19.6 7500 ± 1300 MWCNT mass per glass bead mass in the mixture Determined 30 min after start. Standard derivation given single MWCNT fibers and even agglomerates adhere of interaction forces and the optimum material pairing to glass bead surfaces (see Fig. 5). Other bead mate- are presently unclear. Excessive adherence of individ- rials with different triboelectrical potential, surface ual fibers should be avoided since it may lower the roughness, or polar and dispersive surface energy concentration of fiber morphologies in the aerosol. components may show different nanotube adherence A benefit of dry dispersion of MWCNTs in glass performance. Other interesting bead materials could beads is the long-term storability of dry powder mix- be, e.g., stainless steel, zirconia, or polymers like tures, since, different to liquid dispersions, no sedimen- polystyrene or polytetrafluoroethylene. Higher sur- tation of MWCNTs among glass beads occurs. Using a face roughness beads may promote not only adher- protective inert gas atmosphere may reduce possible ence to the bead but also CNT abrasion from agglom- oxidative material aging during mixing and storage in erate surfaces during mixing. The optimum strength a gas-tight bottle. Fig. 5 SEM image of MWCNTs attaching to a glass bead surface MWCNT aggregate found in a mixture of ARIGM001 and glass beads. The sample was coated with an iridium layer of 4-nm thickness prior to SEM analysis MWCNT fiber Glass bead surface A JNanopartRes (2018) 20:154 Page 9 of 17 154 Temporal development of the aerosol concentration 1.5× the approximate filling time, sufficiently stable concentration levels were observed for most tested The aerosol concentration generated with the studied MWCNT/GB mixtures. In dependence of the studied aerosolization system was monitored with a CPC over MWCNT material and mass ratio in the mixture, mean time. Figure 6 shows the results for the experiments object number concentrations from (1,723 ± 134) to A11, A12, A13, A40, B10, and B20, which were con- (64,033 ± 8,209) #/cm were measured for the plateau taining mixtures of glass beads and MWCNTs either of region. These concentration levels were maintained as type pristine ARIGM001 (BA^ experiments) or of type long as the MWCNT/GB mixture was continuously Baytubes (BB^ experiments) in concentrations as speci- transported to the Venturi nozzle. In our experiments, fied in Table 1. Once the feeding of mixtures was started we were able to produce MWCNT aerosols for up to (t ≔ 0), the number concentration in the chamber was 140 min, before aerosolization had to be stopped to increasing on a time scale of about 20 min, correspond- empty the collecting reservoir of the first cyclone stage. ing to the filling time of a volume of 400 L at an aerosol Longer aerosolization times are possible if larger flow of 20 L/min. After about 30 min, which is about reservoir volumes are used. Fig. 6 Temporal development of the concentration of MWCNT confidence level of the data for integration over the total experi- aerosols, as measured by CPC, during dry dispersion aerosoliza- ment time starting from minute 30. The relative errors for this tion experiments, cf. Table 1. Gray areas indicate the 1-sigma timespan are given as δ values in the diagram 154 Page 10 of 17 J Nanopart Res (2018) 20:154 At comparable mixture mass feed rate of about 11 g/ Interestingly, the APS size distributions in the range of min, the concentration of MWCNT material in the glass 500 to 1100 nm were significant lower than those mea- bead mixture had a significant influence on the object sured with SMPS. A comparable disparity in measure- number concentration in the resulting aerosols, which ments was reported from Baron et al. (2008)for can be seen by comparing the results for the mixtures SWCNT aerosols. The authors assumed that the partly A40 and A10, containing ARIGM001 with 0.33 and uncertain charging and light absorption behavior of 0.01 g/kg, respectively. The higher concentrated mixture these low-density, open structured, and conductive ob- A40 produced about 15 times higher aerosol concentra- jects may be the reason for such an observation. tion than mixture A10. A similar trend can be seen for Within the studied time ranges, no significant chang- the high and the low concentrated Baytubes mixtures es in the size distributions of the MWCNTaerosols were B10 and B20, respectively, even though the difference monitored, as can be seen in Fig. 7(A2–D2) showing the in object number concentration is less pronounced. SMPS size distributions as a function of time. Although It became evident that the output concentration was small fluctuations in the total number concentration had strongly dependent on the type of MWCNT material: an influence on the signal intensities, geometrical mean Mixtures with Baytubes formed significantly lower diameters as well as the distribution width were found to aerosol concentrations than those with ARIGM001 with lie in similar ranges for the compared aerosols. This is respect to the specific MWCNT mass feed rate. particularly apparent for the unimodal size distribution Baytubes were synthesized in large agglomerates of of the higher concentrated ARIGM001 aerosol in highly tangled structure (see Fig. 1). They most likely Fig. 7(B2). Here, the SMPS size distribution was mon- require higher energies for break up and detangling into itored for 70 min without showing significant changes in aerosolizable fragments than the smaller and more distribution width and mean aerodynamic diameter. loosely agglomerated material structure of ARIGM001. The dry dispersion aerosolization system studied The results show that our aerosolization method al- here has the potential to produce nanofiber aerosols of lows controlling the object number output concentra- highly stable output with respect to both concentration tions by changing the MWCNT mass fraction in the and size distribution for arbitrary times of operation. We glass bead mixture. This is comparable to atomizing believe that the powder mixture feeding system is the differently concentrated liquid CNT suspensions. key to long-term aerosol stability. It continuously sup- plies fresh nanofiber material to the Venturi nozzle, Temporal development of the aerodynamic size which prevents aging effects occurring in perpetually distribution agitated powder beds, but requires the mixture of nano- fibers and beads to be sufficiently homogenous. A feed- Parallel to number concentration monitoring with CPC, back loop that monitors the concentration in the aerosol supply line and controls the feeding rate via the turntable size distributions of generated aerosols were monitored with SMPS and APS over time. Figure 7(A1–D1) shows speed could correct for nanofiber concentration inho- the mean log-normal SMPS and APS size distribution of mogeneities in the bead mixture. each aerosol, averaged over several scans and measured in time ranges of sufficiently stable output concentra- Morphological characterization of the MWCNT tion. For ARIGM001 aerosols, unimodal size distribu- aerosols by SEM analysis tions with comparable geometrical mean diameters of 126 nm and respectively 122 nm were measured The generated aerosols were sampled on tracketch (Fig. 7(A1–B1)). A broader multimodal size distribution membrane filters. SEM images of the filters, as shown was found for both Baytubes aerosols (Fig. 7(C1–D1)). representatively in Fig. 8 for an ARIGM001 and a The distributions of Baytubes aerosols exhibit a tail Baytubes aerosol, were used to categorize all observed towards larger aerodynamic diameters. It indicates that, aerosol objects into one of the seven object categories of especially at this high mass concentration of 4 g/kg, Fig. 4. For each sample, between 1000 and 3000 objects Baytubes were far less effectively dispersed than were analyzed to characterize the aerosol morphology ARIGM001 mixtures. SEM images of the sampled aero- and to determine the concentration of individual fibers. sol in Fig. 8 confirm the presence of more microscale The results of this categorization are summarized in agglomerates for Baytubes than for ARIGM001. Table 2. For both ARIGM001 samples (A10 and A40) JNanopartRes (2018) 20:154 Page 11 of 17 154 Fig. 7 Mean log-normal SMPS and APS size distributions of the concentrations (left,A1–D1). The x-y-z plots on the right show MWCNT aerosols from mixture A10, A40 and B10, B20 as the SMPS size distributions for the same time ranges of all aerosols measured during time ranges of sufficiently stable number as function of time (right,A2–D2) 154 Page 12 of 17 J Nanopart Res (2018) 20:154 B1 A1 ARIGM001 Baytubes® B1 S8 A2 S2 S1 S3 S5 S4 S7 S6 A2 B2 A3 B3 A3 B3 350 nm 350 nm S8 S1 S2 S3 S4 S5 S6 S7 Fig. 8 SEM images of collected ARIGM001 (left)and Baytubes exhibit a mean tube diameter of about 14 nm. Therefore individual (right) aerosol with magnified areas A1–A3 and B1–B3. Both nanofibers are not easily identified in the upper right group of aerosols were well dispersed and contained of a high fraction out images B1–B3 of individual fibers. The insets S1–S7 were added since Baytubes as well as for the Baytubes sample B20, more than 30% designed to count and characterize low aspect ratio of the collected aerosol objects were individual particles, it appeared therefore important to determine MWCNT fibers. Moreover, almost 50% of all objects the amount of individual nanofibers and HAR objects in on these samples had a high aspect ratio equal or greater a test aerosol for the interpretation of instrument re- than 3:1. Only for experiment B10 lower ratios of HAR sponses. Possible limitations of the data obtained from objects and individual fibers were found. This was most the categorization approach applied here may arise from probably caused by the ratio of CNT-related and glass missing information on the diameter and size of the bead-related contributions to the aerosol. At the low categorized objects. This prevents estimating mass con- aerosol concentration of experiment B10, the finite glass centrations from SEM images. In addition, the thin and bead background became significant: LAR objects in- rather short Baytubes fibers found on samples B10 and creased to about 60%. These findings reveal the neces- B20 were categorized the same way as much thicker and sity to further reduce the background contribution from longer ARIGM001 fibers on sample A10 and A40. the glass beads. However, the toxicological relevance of the two fiber The morphological distinction between HAR and morphologies may be quite different to micron-sized pow- LAR objects was introduced here since the aerosol der structures (Pauluhn 2009). Occupational hygiene con- generation method is intended to be used for systematic trol may therefore require detecting and distinguishing in- laboratory studies on nanoparticle instrument response dividual and agglomerate fiber morphologies. It will be a to CNT-containing aerosols of known morphological challenge of future research and development to substitute distribution. As most nanoparticle instruments are the very laborious task of detecting and categorizing JNanopartRes (2018) 20:154 Page 13 of 17 154 Table 2 Results of the morphological categorization for the samples A10, A40 and B10, B20. All SEM imaged particles were categorized by their visual shape, structure and estimated degree of agglomeration into 7 particle categories Percentage of particle category for the sample A10 [%] N = 3355 A40 [%] N = 3327 B10 [%] N = 1110 B20 [%] N =2147 Total number of the categorized particles HAR particle agglomerates 3.7 2.8 1.3 1.8 HAR fiber agglomerates 4.4 5.2 1.8 6.6 HAR fiber clusters 6.6 9.0 4.1 4.8 Individual fibers 32.1 30.9 21.2 34.5 LAR particle agglomerates and individual particle objects 49.5 44.6 62.1 40.1 LAR fibers agglomerates 3.1 5.1 7.0 9.0 LAR fiber clusters 0.7 2.5 2.5 3.2 Total HAR objects 46.7 47.8 28.4 47.7 Total LAR objects 53.3 52.2 71.6 52.3 nanofibers on aerosol filter samples, which was performed aerosol to be depending on the MWCNT material. Fur- here, by an online fiber detection technique. ther experiments with other types of CNT materials are The approach followed in the present work gave recommended to study weather this aerosolization tech- valuable insights into the HAR object release propensity nique is suited for the generation of narrower length of two different MWCNTs types under studied disper- or size distributions. Also variations of the air flow rate sion conditions. The very similar individual fiber release through the Venturi nozzle that governs dispersive ratio came quite unexpected and shows that the applied forces should be studied in future. dispersion energies were high enough to break individ- The aerosolization efficiency (aerosol mass released ual fibers off the surface of agglomerates and to split up vs. CNT mass used) of our dry dispersion technique larger agglomerates. Such breaking of tangled fibers could not be determined here. The mass concentration during dispersion shifts the fiber length distribution to of the generated nanoscale aerosols was too low to be smaller values. However, due to the highly tangled state determined by weighing. A second approach to estimat- of the starting materials, the length distributions before ing the aerosol mass based on particle size distribution and after dispersion cannot be compared, since the data obtained from SMPS and APS is not reliable for lengths of tangled nanotubes cannot be measured reli- fibrous aerosols. Firstly, since the main peak(s) of the ably in the agglomerated state. SMPS size distributions in our experiments resulted For further aerosol characterization of the sampled from individual fibers but do not provide fiber length MWCNT aerosols, the geometrical length and diameter information and therefore do not provide mass informa- of all imaged and categorized fibers on sample A10 and tion. Secondly, for A10, A40, and B10, the contribution B20 were determined and plotted as scatter plot and of microscale agglomerates in the right hand tail of the histograms in Fig. 9. Log-normal peak fitting of the SMPS and APS size distributions is insignificant in our geometric fiber length distributions gave an average data especially for ARIGM001, see insets in Fig. 7(A1– length of around 350 nm for ARIGM001 (parameter w D1). It is nonetheless expected to dominate the total in Fig. 9), whereas Baytubes had a shorter length of aerosol mass since microscale agglomerates contain about 200 nm. The majority of analyzed MWCNTs (> many tangled CNTs. For experiment B20, the strong 90%) exhibited a length below 5 μm. Especially discrepancy between the SMPS and APS reading gives Baytubes formed predominantly short fibers below rise to uncertainties on the shape of the size distribution of agglomerates. A third possible approach to estimating 1.5 μm. This small length is believed to be related to the large agglomerate size and high degree of tangle- the aerosol mass could rely on measuring the volume of ment of Baytubes (cf. Fig. 1). Tangled and interlocked agglomerates and the length and diameter of individual fibers must to be broken to be released from the fiber fibers imaged with SEM. This is a very laborious task network and to become observable as individual fibers. and requires assuming agglomerate densities. Such den- The results show the fiber length distribution in the sity assumption is necessary for any mass estimation 154 Page 14 of 17 J Nanopart Res (2018) 20:154 Fig. 9 Scatter plots and histograms of the geometrical fiber diam- for Baytubes (right, N = 118) with log-normal (length) and Gauss- eter and length data pairs of individual fibers found in the aerosol ian (diameter) peak fitting sample A10 for ARIGM001 (left, N = 176) and aerosol sample B20 approach based on agglomerate geometric or mobility Analogies between dry and liquid aerosolization diameter data. For partially dispersed CNT materials, these densities are unknown. Future experiments will It is instructive to review principle analogies between aim at characterizing aerosols using a so-called nano- the dry dispersion concept presented here and atomizing particle mass classifier (nano-PMC) that was developed concepts using liquid dispersions. It shows that similar recently (Broßell et al. 2015). In combination with a components and process steps are required both for dry differential mass analyzer, it allows measuring the mass and liquid dispersion. All of them require individual and density of aerosol particles. optimization for best aerosol performance. Themicroscalebeadsusedherecorrespond to a cheap and non-toxic solvent in liquid suspensions. Like Reproducibility of the aerosolization technique a solvent, the beads allow nanofiber material dilution in replicate experiments and concentration control. For many material variants, The aerosolization technique developed here was also studied with respect to reproducibility of the generated concentration. For this comparison, all ARIGM001 mix- tures were aerosolized under similar experimental con- ditions. The resulting aerosol number concentrations measured from minute 30 to 60 after start of powder feeding are compared statistically using box-whisker plotting in Fig. 10. For the evaluated time span, the aerosols of each group of MWCNT/GB mixture with 0.01, 0.06, and 0.20 g/kg in Fig. 10 showed a fair to good degree of reproducibility. For mixtures with equal MWCNT amount, aerosols with mean object number concentra- tions in the same size range resulted. A maximum standard derivation of 25% in mean object number was found. A feedback loop that monitors the concen- Fig. 10 Box-whisker plot for the temporal development of the concentrations generated from ARIGM001 and glass bead mix- tration in the aerosol supply line and controls the feeding tures during minute 30 to 60 after start of feeding (box 25/75%; rate via the turntable speed could correct for CNT inho- whisker 1.5; mean: diamond, median: notch). The vertical separa- mogeneities in the glass bead mixture and further im- tors group the three compared mixtures of 0.01, 0.06, or 0.20 g/kg prove temporal stability and reproducibility. mass ratio MWCNT/GB JNanopartRes (2018) 20:154 Page 15 of 17 154 spherical beads can achieve transforming sticky and a flow rate of 20 L/min. The obtained MWCNTaerosols clogging low-density nanofiber materials into free- contained a high fraction of individual fibers of up to flowing powder mixtures. Their liquid-like character fa- 34%. The toxicological relevance of the obtained fiber cilitates continuous feeding to the atomizing unit, e.g., a aerosols was assessed by quantifying to the content of Venturi nozzle, and controlling the mass feed rate. fibers of WHO geometry. The thicker and more loosely Obtaining homogenous mixtures of nanofiber mate- agglomerated material ARIGM001 resulted in a higher rials in beads may require optimization of mixing move- content of WHO fibers in the aerosol than Baytubes, ments which are generally accompanied by ball milling both in relative and absolute concentration. effects on nanofiber agglomerates. The resulting energy Compared to liquid dispersion and aerosolization of transfer to the nanofiber materials is in analogy with, CNTs, dry mixing with glass bead exhibits a number of e.g., ultrasonication of nanofibers in liquid suspension. benefits. It requires neither chemical nor mechanical Successful and temporally stable mixing results of nano- pre-treatment of the CNTs. Since dry mixtures do not fibers and beads however do not require pre-treating of sediment, no possibly toxic organic solvents and surfac- the nanofibers with chemical functionalization or sur- tants are required to achieve stable dispersions in glass factant coating as in liquid dispersion. beads. Glass beads are an inexpensive dilution material Even the polarity of a solvent, which is highly critical and may be re-used. Similar to dryer columns for liquid for preparing liquid suspensions, bears analogies to dry atomization, also our dry dispersion technique requires mixing: Optimum triboelectrical pairing of powder an additional cyclone stage for downstream separation components might result in an adhering nanofiber corona of glass beads and CNTs. The dry aerosolization process on the bead surface and further stabilize nanofibers in was found to allow high aerosol stability. Observed the bead mixture via attractive electrostatic forces concentration fluctuations were attributed to partially (Matsusaka et al. 2010). inhomogeneous mixing of CNTs and glass beads, a Finally, the bead-filtering cyclone stage used here is process step that requires further improvement. an analogue of a drying unit necessary for obtaining dry Future work should be devoted to aerosols from atomized liquid dispersion. & Wind-sifting glass beads prior to use in order to reduce the glass fragment concentration. Conclusion and perspective & Studying triboelectric charge separation during mixing and its role in attaching nanoscale fragments Providing fiber aerosols of controlled individual fiber to beads. concentration for fiber toxicological inhalation studies & Optimizing dry mixing to further improve the intended as well as for assessing aerosol measurement perfor- mixing with glass bead and to control ball milling effects of MWCNTs. mance and exposure limit control strategies is an impor- tant task. & Improving process stability via a feedback loop of Here, a dry dispersion aerosolization technique was the CPC concentration to the powder feeding rate. developed that uses mixtures of nanofiber materials and & Estimating the energy transfer necessary to breakup microscale beads that are fed continuously through a or disentangle CNT agglomerates. Venturi nozzle to generate nanofiber-containing aero- & Varying the air flow rate through the Venturi nozzle sols. The technique was studied for two morphological- that governs dispersive forces. ly very different MWCNT materials in glass bead mix- & Further narrowing of the morphological distribution tures of varied mass ratio. Aerosols were characterized of the aerosol by the use of less tangled materials. by CPC, SMPS, APS, and morphological analysis of filter samples with SEM to determine and categorize the morphological composition of the generated aerosols, Acknowledgments The authors thank Guillermo Orts-Gil for relative and absolute concentrations of individual fibers TEM analysis and Nils Kujath for the technical assistance. as well as fiber length and diameter distributions. Funding This work was funded by the German Federal Ministry The generator allows controlling the aerosol concen- of Education and Research (BMBF), FKZ 03X0127B, and was tration in an extremely wide range. Aerosols of up to conducted in the frame of the SIINN ERA-Net project NanoIndEx 10 3 2×10 individual nanofibers per m were generated at in cooperation with other project partners who funded by the 154 Page 16 of 17 J Nanopart Res (2018) 20:154 French National Funding Agency for Research (ANR), the British De Volder MF, Tawfick SH, Baughman RH, Hart AJ (2013) Technology Strategy Board (TBS), and the Swiss TEMAS AG. Carbon nanotubes: present and future commercial applica- tions. Science 339(6119):535–539 Compliance with ethical standards All relevant ethical stan- Fujitani Y, Furuyama A, Hirano S (2009) Generation of airborne dards were satisfied. multi-walled carbon nanotubes for inhalation studies. Aerosol Sci Technol 43(9):881–890 Hamaker HC (1937) The London—van der Waals attraction be- Conflict of interest The authors declare that they have no con- tween spherical particles. Physica 4(10):1058–1072 flict of interest. Han SG, Andrews R, Gairola CG (2010) Acute pulmonary re- sponse of mice to multi-wall carbon nanotubes. Inhal Toxicol 22(4):340–347 Open Access This article is distributed under the terms of the Harris G, Maguire B (1968) A gravimetric dust sampling instru- Creative Commons Attribution 4.0 International License (http:// ment (simpeds): preliminary results. Ann Occup Hyg 11(3): creativecommons.org/licenses/by/4.0/), which permits unrestrict- 195–201 ed use, distribution, and reproduction in any medium, provided Jennerjohn N, Eiguren-Fernandez A, Prikhodko S, Fung DC, you give appropriate credit to the original author(s) and the source, Hirakawa KS, Zavala-Mendez JD, Hinds W, Kennedy NJ provide a link to the Creative Commons license, and indicate if (2010) Design, demonstration and performance of a versatile changes were made. electrospray aerosol generator for nanomaterial research and applications. Nanotechnology 21(25):255603 Lee S-B, Lee J-H, Bae G-N (2010) Size response of an SMPS– References APS system to commercial multi-walled carbon nanotubes. J Nanopart Res 12(2):501–512 Matsusaka S, Maruyama H, Matsuyama T, Ghadiri M (2010) Ahn K-H, Kim S-M, Yu IJ (2011) Multi-walled carbon nanotube Triboelectric charging of powders: a review. Chem Eng Sci (MWCNT) dispersion and aerosolization with hot water at- 65(22):5781–5807 omization without addition of any surfactant. Saf Health McKinney W, Chen B, Frazer D (2009) Computer controlled Work 2(1):65–69 multi-walled carbon nanotube inhalation exposure system. Asbach C, Kaminski H, Lamboy Y, Schneiderwind U, Fierz M, Inhal Toxicol 21(12):1053–1061 Todea AM (2016) Silicone sampling tubes can cause drastic Mercer R, Scabilloni J, Hubbs A, Wang L, Battelli L, McKinney artifacts in measurements with aerosol instrumentation based W, Castranova V, Porter D (2013) Extrapulmonary transport on unipolar diffusion charging. Aerosol Sci Technol 50(12): of MWCNT following inhalation exposure. Part Fibre 1375–1384 Toxicol 10(1):38 Bahk YK, Buha J, Wang J (2013) Determination of geometrical Myojo T, Oyabu T, Nishi K, Kadoya C, Tanaka I, Ono-Ogasawara length of airborne carbon nanotubes by electron microscopy, M, Sakae H, Shirai T (2009) Aerosol generation and mea- model calculation, and filtration method. Aerosol Sci surement of multi-wall carbon nanotubes. J Nanopart Res Technol 47(7):776–784 11(1):91–99 Baron PA, Deye GJ, Chen BT, Schwegler-Berry DE, Shvedova Nagai H, Okazaki Y, Chew SH, Misawa N, Yamashita Y, AA, Castranova V (2008) Aerosolization of single-walled Akatsuka S, Ishihara T, Yamashita K, Yoshikawa Y, Yasui carbon nanotubes for an inhalation study. Inhal Toxicol H, Jiang L, Ohara H, Takahashi T, Ichihara G, Kostarelos K, 20(8):751–760 Miyata Y, Shinohara H, Toyokuni S (2011) Diameter and Brockmann JE, Rader DJ (1990) APS response to nonspherical rigidity of multiwalled carbon nanotubes are critical factors in particles and experimental determination of dynamic shape mesothelial injury and carcinogenesis. Proc Natl Acad Sci factor. Aerosol Sci Technol 13(2):162–172 108(49):E1330–E1338 Broßell D, Valenti M, Bezantakos S, Schmidt-Ott A, Biskos G (2015) Oberdörster G, Castranova V, Asgharian B, Sayre P (2015) The nanoparticle mass classifier (nano-PMC): development, Inhalation exposure to carbon nanotubes (CNT) and carbon characterisation, and application for determining the mass, ap- nanofibers (CNF): methodology and dosimetry. J Toxicol parent density, and shape of particles with masses down to the Environ Health B Crit Rev 18(3–4):121–212 zeptogram range. Aerosol Sci Technol 49(7):495–507 O'Shaughnessy PT, Adamcakova-Dodd A, Altmaier R, Thorne PS Chen BT, Schwegler-Berry D, McKinney W, Stone S, Cumpston (2014) Assessment of the aerosol generation and toxicity of JL, Friend S, Porter DW, Castranova V, Frazer DG (2012) carbon nanotubes. Nanomaterials 4(2):439–453 Multi-walled carbon nanotubes: sampling criteria and aerosol Pauluhn J (2009) Subchronic 13-week inhalation exposure of rats characterization. Inhal Toxicol 24(12):798–820 to multiwalled carbon nanotubes: toxic effects are determined Chen BT, Schwegler-Berry D, Cumpston A, Cumpston J, Friend by density of agglomerate structures, not fibrillar structures. S, Stone S, Keane M (2016) Performance of a scanning Toxicol Sci:kfp247 mobility particle sizer in measuring diverse types of airborne Pauluhn J, Rosenbruch M (2015) Lung burdens and kinetics of nanoparticles: multi-walled carbon nanotubes, welding multi-walled carbon nanotubes (Baytubes) are highly depen- fumes, and titanium dioxide spray. J Occup Environ Hyg dent on the disaggregation of aerosolized MWCNT. 13(7):501–518 Nanotoxicology 9(2):242–252 Dazon C, Witschger O, Bau S, Payet R, Beugnon K, Petit G, Garin Plitzko S, Gierke E, Dziurowitz N, Broßell D (2010) Erzeugung T, Martinon L (2017) Dustiness of 14 carbon nanotubes using von CNT/CNF-Stäuben mit einem Schwingbett- the vortex shaker method. J Phys Conf Ser 838(1):012005 Aerosolgenerator und Charakterisierung der JNanopartRes (2018) 20:154 Page 17 of 17 154 Fasermorphologie mithilfe eines Thermalpräzipitators als Spurny KR, Boose C, Hochrainer D (1975) Zur Zerstäubung von Sammelsystem. Gefahrstoffe-Reinhaltung der Luft 70(1):31 Asbestfasern in einem Fließbett-Aerosolgenerator. Staub, Porter DW, Hubbs AF, Mercer RR, Wu N, Wolfarth MG, Sriram Reinhaltung der Luft 35(12):440–445 K, Leonard S, Battelli L, Schwegler-Berry D, Friend S, Stanton MF, Wrench C (1972) Mechanism of mesothelioma in- Andrew M, Chen BT, Tsuruoka S, Endo M, Castranova V duction with asbestos and fibrous glass. J Natl Cancer Inst 48: (2010) Mouse pulmonary dose- and time course-responses 797–821 induced by exposure to multi-walled carbon nanotubes. Su W-C, Cheng YS (2014) Carbon nanotubes size classification, Toxicology 269(2):136–147 characterization and nasal airway deposition. Inhal Toxicol Pott F, Friedrichs KH (1972) Tumoren der Ratte nach i.p.- 26(14):843–852 Injektion faserförmiger Stäube. Naturwissenschaften 59(7): Totaro S, Cotogno G, Rasmussen K, Pianella F, Roncaglia M, 318 Olsson H, Riego Sintes JM, Crutzen HP (2016) The JRC Rittinghausen S, Hackbarth A, Creutzenberg O, Ernst H, Heinrich nanomaterials repository: a unique facility providing repre- U, Leonhardt A, Schaudien D (2014) The carcinogenic effect sentative test materials for nanoEHS research. Regul Toxicol of various multi-walled carbon nanotubes (MWCNTs) after Pharmacol 81:334–340 intraperitoneal injection in rats. Part Fibre Toxicol 11:18 Vo E, Zhuang Z (2013) Development of a new test system to Ryman-Rasmussen JP, Tewksbury EW, Moss OR, Cesta MF, determine penetration of multi-walled carbon nanotubes Wong BA, Bonner JC (2009) Inhaled multiwalled carbon through filtering facepiece respirators. J Aerosol Sci 61:50– nanotubes potentiate airway fibrosis in murine allergic asth- 59 ma. Am J Respir Cell Mol Biol 40(3):349–358 Vo E, Zhuang Z, Birch E, Zhao Q, Horvatin M, Liu Y (2014) Seto T, Furukawa T, Otani Y, Uchida K, Endo S (2010) Filtration Measurement of mass-based carbon nanotube penetration of multi-walled carbon nanotube aerosol by fibrous filters. through filtering facepiece respirator filtering media. Ann Aerosol Sci Technol 44(9):734–740 Occup Hyg:meu005 Shvedova AA, Kisin ER, Mercer R, Murray AR, Johnson VJ, Wang J, Bahk YK, Chen S-C, Pui DY (2015) Characteristics of Potapovich AI, Tyurina YY, Gorelik O, Arepalli S, airborne fractal-like agglomerates of carbon nanotubes. Schwegler-Berry D (2005) Unusual inflammatory and Carbon 93:441–450 fibrogenic pulmonary responses to single-walled carbon WHO (1985) Reference methods for measuring man-made min- nanotubes in mice. Am J Phys Lung Cell Mol Phys 289(5): eral fibres (MMMF). Prepared by WHO/EURO Technical L698–L708 Committee for Evaluating MMMF http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Journal of Nanoparticle Research Springer Journals

Continuous dry dispersion of multi-walled carbon nanotubes to aerosols with high concentrations of individual fibers

Free
17 pages
Loading next page...
 
/lp/springer_journal/continuous-dry-dispersion-of-multi-walled-carbon-nanotubes-to-aerosols-gQeJt9nrbz
Publisher
Springer Netherlands
Copyright
Copyright © 2018 by The Author(s)
Subject
Materials Science; Nanotechnology; Inorganic Chemistry; Characterization and Evaluation of Materials; Physical Chemistry; Optics, Lasers, Photonics, Optical Devices
ISSN
1388-0764
eISSN
1572-896X
D.O.I.
10.1007/s11051-018-4262-y
Publisher site
See Article on Publisher Site

Abstract

JNanopartRes (2018) 20:154 https://doi.org/10.1007/s11051-018-4262-y RESEARCH PAPER Continuous dry dispersion of multi-walled carbon nanotubes to aerosols with high concentrations of individual fibers Barbara Katrin Simonow & Daniela Wenzlaff & Asmus Meyer-Plath & Nico Dziurowitz & Carmen Thim & Jana Thiel & Mikolaj Jandy & Sabine Plitzko Received: 23 November 2016 /Accepted: 25 May 2018 The Author(s) 2018 Abstract The assessment of the toxicity of airborne morphological composition of the aerosol, its fiber con- nanofibers is an important task. It relies on toxicological tent as well as fiber length and diameter distributions. inhalation studies and validated exposure measurement High fractions of individual fibers of up to 34% were techniques. Both require nanofiber-containing aerosols obtained, which shows the setup to be capable of dis- of known morphological composition and controlled persing also highly tangled MWCNT agglomerates fraction of individual fibers. Here, a dry powder disper- effectively. sion method is presented that operates with mixtures of nanofibers and microscale beads. Aerosolization exper- Keywords Multi-walled carbon nanotubes Aerosol iments of mixtures of multi-walled carbon nanotubes . . generation Dry powder dispersion Aerosol (MWCNTs) and glass beads that were continuously fed morphology into a Venturi nozzle enabled high generation rates of aerosols composed of individual and agglomerate nano- fiber structures. The aerosol process achieved good Introduction stability over more than 2 h with respect to concentra- tion and aerodynamic size distribution. Its operation The unique properties of carbon nanotubes (CNTs) make duration is limited only by the reservoir volume of the them a widely studied material with promising commer- cyclone used to separate the beads from the aerosol. The cial applications. The production capacity and use of aerosol concentration can be controlled by changing the CNTs have been reported to steadily increase (De mass ratio of MWCNTs and glass beads or by adapting Volder et al. 2013). In accord to the fiber toxicology the mass feed rate to the nozzle. For two agglomerated paradigm (Pott and Friedrichs 1972;Stanton and MWCNT materials, aerosol concentrations ranged from Wrench 1972), their fibrous morphology and 1700 to 64,000 nano-objects per cm .Comprehensive biodurability have raised concerns about potential lung scanning electron microscope analysis of filter samples carcinogenicity especially for multi-walled CNTs was performed to categorize and determine the (MWCNTs). While highly tangled MWCNT agglomer- ates with granular morphology have been reported to Electronic supplementary material The online version of this show no fiber-toxic effect (Pauluhn 2009), inhalation article (https://doi.org/10.1007/s11051-018-4262-y) contains supplementary material, which is available to authorized users. exposure studies of well-dispersed MWCNTs have ob- served asbestos-like carcinogenicity (Oberdörster et al. : : : B. K. Simonow D. Wenzlaff A. Meyer-Plath (*) 2015). Mercer et al. (2013) have shown that individual : : : : N. Dziurowitz C. Thim J. Thiel M. Jandy S. Plitzko and short MWCNT structures may translocate to sec- Federal Institute of Occupational Safety and Health (BAuA), ondary organs like the liver. Also mesothelial injury, Nöldnerstraße 40 - 42, 10317 Berlin, Germany e-mail: meyer-plath.asmus@baua.bund.de characteristic for asbestos, was observed for MWCNTs 154 Page 2 of 17 J Nanopart Res (2018) 20:154 after intraperitoneal injection of well-dispersed CNT A challenging and time-consuming step for liquid suspensions (Nagai et al. 2011; Rittinghausen et al. CNT dispersions however is the preparation of homog- 2014). The inhalative exposure to individual and fibrous enous and temporally stable CNT suspensions needed CNT structures is therefore considered a health risk, in for these methods. To obtain sufficient wettability of particular for individual MWCNTs and CNT bundles CNTs in water and other polar liquids and to promote with a geometry according to the World Health Organi- their dispersibility and stability in the solvent, signifi- zation’s (WHO) fiber definition of a length exceeding cant amounts of surfactants have to be added (Ryman- 5 μm, a diameter below 3 μm, and an aspect ratio larger Rasmussen et al. 2009; Pauluhn and Rosenbruch 2015). than 3:1 (WHO 1985). In addition, high energy liquid dispersion techniques However, the assessment of such nanoscale fibrous generally must be applied, including ultrasonication objects in air is challenging. None of the direct-reading (Shvedova et al. 2005; Ahn et al. 2011), high shear force measurement instruments currently used to assess air- mixing, or mechanical milling (Ryman-Rasmussen et al. borne nano-object exposures is capable of differentiat- 2009;Setoet al. 2010). High energy dispersion tech- ing granular and fibrous morphologies. In addition, their niques may give rise to significant shortening of fibers. response to nanofibers and low-density CNT agglomer- Chemical functionalization of CNTs (Han et al. 2010; ates is mostly unknown. It has been shown that in-depth Bahk et al. 2013;Wanget al. 2015) may increase the interpretation of scanning mobility particle sizer degree and stability of CNT suspensions. Such pre- (SMPS) and aerodynamic particle sizer (APS) data re- treatments for liquid dispersion may have considerable quires good knowledge of the aerosol morphology when effects on the physicochemical material properties. As dealing with fiber-containing aerosol (Brockmann and additional drawback, they generally result in aerosols Rader 1990;Chen et al. 2016). Systematic challenging not of pristine but surfactant-coated objects. Therefore, of such instruments with well-characterized fiber-con- aerosol generation techniques that can start from pris- taining aerosols is therefore required to study principle tine, chemically and morphologically unmodified CNT limitations and response characteristics. material appear desirable. For such inhalation toxicological and instrument re- Dry powder dispersion techniques can avoid the sponse studies, purely fiber-containing aerosols with wetting and stabilization problem of liquid-powder pro- stable concentrations of morphologically uniform and cessing. They have been successfully applied for the fully de-agglomerated nanofibers are desirable but cur- generation aerosols from pristine CNTs. Batch tech- rently not available. Considerable work is required for niques use a single batch of material that is agitated to the characterization of fiber-containing aerosols. They transfer energy to CNT agglomerates in order to release need to be sampled and analyzed microscopically at fibers, fiber fragments, and agglomerates into an air high resolution (Chen et al. 2012). flow. Powder agitation can be applied by slow or rapid Various methods for CNT aerosol generation have oscillating motion using loudspeakers (McKinney et al. been developed and reported in the past. They are either 2009; Porter et al. 2010) and so-called vortex (Lee et al. based on dry dispersion of CNT powders (Fujitani et al. 2010; Dazon et al. 2017) or linear shakers (Spurny et al. 2009;McKinney et al. 2009; Myojo et al. 2009;Lee 1975; Fujitani et al. 2009; Plitzko et al. 2010). They et al. 2010; Plitzko et al. 2010;Chen et al. 2012;Voand have been reported to be capable of releasing micro- and Zhuang 2013;O’Shaughnessy et al. 2014; Vo et al. nanoscale MWCNT structures and individual fibers. 2014) or on the spraying of liquid CNT suspensions Although stable aerosol concentration was reported to (Jennerjohn et al. 2010; Seto et al. 2010; Ahn et al. have been generated with some of these batch-style 2011; Bahk et al. 2013; Su and Cheng 2014; Wang systems (Fujitani et al. 2009;McKinney et al. 2009), et al. 2015). Pneumatic spray systems were successfully perpetual agitation of a powder batch gradually trans- used to produce individual and well-dispersed MWCNT forms the morphology of the powder. The applied ener- aerosols in high and stable concentrations (Lee et al. gy may effect both loosening and compactifying CNTs 2010; Ahn et al. 2011; Su and Cheng 2014). and their agglomerates by agglomerate abrasion and breakup or agglomerate interlocking and fiber entangle- ment processes, respectively. Batch-style dry techniques 1 may therefore require careful process optimization in The term Bobject^ will be used in the following when both granular and fibrous particles are addressed. order to maintain a perpetually agitated CNT material JNanopartRes (2018) 20:154 Page 3 of 17 154 in stable dusting condition. For some CNT material Meanwhile, the production of Baytubes has been batches, such conditions may not be achieved. Especial- discontinued. The material NM04003a from the JRC ly loosely agglomerated MWCNTs may initially release Nanomaterial Repository might serve as a substitute high aerosol concentrations. They however may exhibit (Totaro et al. 2016). SEM images of the starting mate- a tendency to interlock and compactify with progressing rials are shown in Fig. 1.TEM images of thetwo agitation duration, which would result in changes in MWCNT materials are included in the supplementary aerosol concentration and morphology. Both parameters information. are however important to assess the temporal stability of Swarcoflex glass beads (GB) (diameter 400–600 μm, aerosol generators used for instrument challenging and roundness ≥ 80%, SWARCO M. Swarovski GmbH) exposure studies. were used as received. They were mixed with the This work presents a dry dispersion technique and MWCNTs in order to transform the MWCNT powder comprehensive aerosol characterization approach to into a free-flowing powder of reduced MWCNT mass provide test aerosols that contain morphologically concentration. Prior to CNT aerosolization experiments, well-characterized high concentrations of individual the aerosol particle concentration generated by aerosol- and agglomerated fibrous structures. The method ization of different microscale glass bead types was operates with a continuous flow of fresh CNT material. studied. It does not suffer from aging effects occurring in batch For the preparation of mixtures with glass beads, agitating aerosol generators. The flow of fresh material MWCNT powder of a specific mass was added to is provided from mixtures of pristine CNTs and micro- 1 kg of glass beads in a cylindrical glass bottle of 1 l scale beads. The beads not only dilute the CNT material volume. In this work, the MWCNT concentration of the to the desired mass concentration but also transform it to mixture ranged from 10 mg/kg to 4 g/kg. The MWCNT powder was then mixed with the glass beads by slow a free-flowing powder mixture. This facilitates con- trolled continuous feeding to the aerosolization unit. revolution of the closed bottle around its axis at about Here, a Venturi nozzle generates high rates of aerosols 10–15 rpm. After typically 5 to 15 min of mixing, a from the mixture. In the following, the performance of visually homogenous, gray colored mixture was the technique is evaluated for two MWCNT materials of obtained. different agglomerate and fiber morphology. The gener- ated aerosols were injected into a home-built exposure Aerosol generation chamber that was designed with focus on high spatial aerosol uniformity and a volume flow large enough to For MWCNT aerosol generation, a dry dispersion sys- supply multiple measurement or exposure devices si- tem was developed that disperses mixtures of MWCNTs multaneously with CNT-containing aerosols. and glass beads by means of a Venturi nozzle. The mixtures were fed volumetrically into a notch in a horizontally rotating turntable. The turntable transported Materials and methods the mixture to the Venturi side inlet, where it was sucked in and was blown through the Venturi nozzle into the Materials feed line. In contrast to shaker-based vibration systems that operate on perpetually agitated batches, the system In this study, two different types of MWCNT materials presented here avoids morphological powder aging ef- were tested: ARIGM001 (industrial grade; outer diame- fects from perpetual agitation by continuously supply- ter(OD)10–30 nm; length < 15 μm; purity > 80%; Arry ing pristine CNT material to the nozzle. A downstream International Group Ltd.) and Baytubes C150P (95% two-stage cyclone system separated airborne glass carbon, OD 13 nm; median length > 1 μm; provided by beads and larger MWCNT agglomerates from the aero- Bayer Material Science AG), denoted as Baytubes in the sol fraction that was then introduced to the exposure following. Both were used without prior purification or chamber (see Fig. 2). additional treatment. Baytubes were chosen, as they are A modified SAG 410/U control unit (Topas GmbH, known to form powders of highly tangled, multi micron- Dresden, Germany) was used to control the rotational sized granular structure (Pauluhn 2009) and therefore speed of the turntable in the range from 8 to 12 rpm as tend to exhibit a low dust release propensity (cf. Fig. 1). well as the pressure of dry and particle free air that was fed 154 Page 4 of 17 J Nanopart Res (2018) 20:154 Fig. 1 Scanning electron microscope images of the starting material ARIGM001 (left) and Baytubes C150P (right). The lower right image shows the surface of a large agglomerate into the primary Venturi nozzle inlet at 4–5bar. The During operation, the feeding rate of the MWCNT/ MWCNT/GB mixture was poured from the reservoir bot- GB mixture into the nozzle was controlled via the speed tle into a stainless steel funnel. Its opening was adjusted of the turntable. Continuous weighting of the first cy- directly above the notch of 10-mm width and 5-mm depth, clone’s metal container enabled mass-controlled feed- cut into the upper surface of the turntable. The MWCNT/ ing. For the present work, feeding rates of 10–13 g/min GB mixture flowing into this notch was continuously of MWCNT/GB mixtures were used. transported towards the side inlet of a Venturi nozzle The use of several bottles and larger storage containers (ISO 5011) with an inlet nozzle diameter of 0.7 mm for the CNT/GB mixture enables long-term operation and an outlet flow of 15–20 L/min. The aerosolized mix- of the generator. For the experiments presented ture was fed into a home-built two-stage cyclone system here, for materials safety reasons, the metal container consisting of a metal container of 5 L volume, 8.5 cm of the first cyclone had no revolving door outlet as is diameter, and 22.8 cm height as first cyclone stage. A commonly used for industrial cyclones and allows for modified FSP 2 sampling head (DEHA Haan & Wittmer emptying the cyclone during operation. Therefore, the GmbH, Heimsheim, Germany) with integrated SIMPEDS operation time of our experiments was limited to about cyclone (Harris and Maguire 1968) served as second 3 h to prevent overload of the first cyclone at a mass of cyclone stage. A 300 μm stainless steel filter mesh in the about 2.5 kg glass beads. outlet of the first cyclone prevented coarser objects to enter the second stage filter in case of cyclone overload. The Instrumentation tangential inlet of 8 mm diameter and the central riser pipe outlet of 20 mm diameter and 60 mm length of the first Experimental details of the chamber used for the aero- cyclone very effectively removed the majority of glass solization and aerosol homogeneity studies are given in beads and larger CNT agglomerates from the aerosolized the Supplementary Information. powder mixture. The secondary cyclone served to further The particle number concentration in the exposure narrow down the object size distribution. chamber was monitored with a Grimm CPC model JNanopartRes (2018) 20:154 Page 5 of 17 154 Fig. 2 Schematic diagram illustrating the components and work- cyclone system separates the airborne fraction from the glass beads ing principle of the dry dispersion aerosol generator and its con- and larger MWCNT agglomerates. The aerosol composition in the nection to the exposure chamber. A mixture of glass beads and exposure chamber was monitored and sampled with a set of gas MWCNT material is continuously transported to a Venturi noz- lances, see Supplementary Information zle that aerosoloizes the mixture. Downstream, a two-stage 5.403 (Grimm Aerosol GmbH, Ainring, Germany), op- Depending on aerosol concentration, sampling times erating with a sampling flow rate of 0.3 L/min. Mea- of 5–60 min and flow rates of 2–3 L/min were applied. surement of the electrical mobility size distribution in All samplers and measurement instruments were the range of 10 to 1,100 nm was performed with a connected to stainless steel sampling gas lances of the Grimm Scanning Mobility Particle Sizer (SMPS) chamber using antistatic silicone tubes supplied by consisting of a CPC model 5.403 and an electrostatic Grimm Aerosol GmbH (Asbach et al. 2016). differential mobility analyzer (DMA) of BVienna^ type model L-DMA. The SMPS was likewise operating with Experimental procedure a sampling flow rate of 0.3 L/min at a scanning cycle time of 7 min. SMPS data analysis was done with the Before the start of an experiment, the exposure chamber software provided by the manufacturer (Grimm Soft- was flushed with dried and HEPA-14-filtered com- ware 5.477). Large objects with aerodynamic diameters pressed air until a number concentration of 0–60 parti- from 0.5–20 μm were detected with an Aerodynamic cles/cm and a relative humidity of 20–26% was Particle Sizer (APS) model TSI 3321 (TSI GmbH, Aa- reached inside of the chamber. Next, the aerosol gener- chen, Germany) equipped with a 100:1 diluting stage ator was connected to the aerosol inlet of the chamber model TSI 3302A. The APS size distribution samples and the air flow to the Venturi nozzle inlet was adjusted. were taken with scanning times of 59 s at a sampling After the start of the turntable, the MWCNT/GB mixture flow rate of 5 L/min. was poured into the funnel of the aerosol generator to For studies of morphology distributions, aerosols initiate aerosol generation (t = 0). For the present setup, were sampled on track etch membrane filters (gold- a maximum of 2.5 kg MWCNT/GB could be aerosol- coated polycarbonate with 37-mm diameter and a pore ized continuously. At the standard air flow rate of 20 L/ size of nominal 200 nm supplied by APC GmbH, min, completely flushing the 400 L exposure chamber Eschborn, Germany, using PGP sampler units (DEHA with aerosol required about 20 min. This is reflected in Haan & Wittmer GmbH, Heimsheim, Germany). the temporal development visible in Fig. 6. Plateau 154 Page 6 of 17 J Nanopart Res (2018) 20:154 concentration values were not reached before about central area of the filter was imaged at a magnification 30 min of operation. All mean concentration values of ×3000, an accelerating voltage of 3 kVand a working given in the following were obtained at minimum distance of 6.1 nm resulting in a minimum feature 30 min after start of operation. detection size of 8.3 nm (edge size of a pixel). For each All silicone tubes used for chamber and instrument sample, images with a standard resolution of 5120 × connections were cleaned or replaced after approximate- 3840 pixels and 1344 μm area were acquired at 10 to ly 3–5 experiments. Prior to injecting a new type of 15 randomly chosen filter positions and were subjected MWCNT material, the inner surfaces of the exposure to subsequent morphological characterization. chamber and all measurement lances were cleaned and all All objects imaged by SEM were categorized visual- silicone tubes were replaced to avoid cross contamination. ly according to their shape, structure, and degree of agglomeration into one of the seven object categories Determination of the system's background shown in Fig. 4. These categories do not cover all possible aerosol morphologies. They were chosen ac- concentration cording to the subject of the present study that addresses the dispersion state of fibrous and agglomerated fibrous The background concentration arising from the use of glass beads for the aerosolization process was deter- materials as generated by our setup. We therefore dif- mined for each delivered glass bead lot of about 50 kg ferentiated between objects with aspect ratios greater mass. For this purpose, 3 kg of pure glass beads was (high aspect ratio, HAR) or smaller (low aspect ratio, aerosolized with the aerosol generator at a feeding rate of LAR) than 3:1 as well as between objects with or 12 to 13 g/min. The resulting particle number concen- without visual fibrous structures. For objects with fi- tration in the exposure chamber was monitored over brous structures, we further distinguished between indi- time. After typically 30 min, a concentration plateau vidual fibers, weakly bounded agglomerates with a was reached and maintained as long as the glass beads countable number of elements, named clusters, and highly agglomerated structures with an uncountable were aerosolized. For the present study, the glass bead lot No. 058 was used. It reached a median plateau number of elements, named agglomerates in the follow- ing. Particulate, non-fibrous objects were not distin- concentration of (1143 ± 185) particles/cm ,according to CPC monitoring, averaged from minute 30 to 120. guished with respect to their agglomeration state. Fibers As shown in Fig. 3, the background particles were non- and fiber-containing agglomerates attached to granular fibrous, compact particles of relatively low SEM image objects were categorized as fibrous objects since contrast. They were easily distinguishable from fibrous attaching catalyst and catalyst support particles are a MWCNT particles during morphological analysis. common phenomenon for industrial grade CNT materials. The image manipulation software GIMP (version Electron microscopic analysis of sampled aerosols 2.8.6, GNU Image Manipulation Program) was used to visually detect and count all objects found on SEM Analyses of aerosols sampled on track etch membrane filters were performed with a scanning electron micro- images in accordance with their category. GIMP’spath tool together with an in-house developed Bscript-fu^ scope (SEM, Hitachi SU8030, Hitachi High- Technologies Europe GmbH, Krefeld, Germany). A plugin was used to measure the geometrical length of Fig. 3 Images of background particles originating from glass bead lot No. 058 during dispersion of mixtures with ARIGM001. The particles were collected on a silicon wafer by electrostatic precipitation. The white scale bar has a length of 500 nm JNanopartRes (2018) 20:154 Page 7 of 17 154 Objects free of Un-countable Individual Countable Fibres Fibers Fibres Objects HAR Agglomerates HAR Fibre HAR Fibre Cluster Individual Fibres Agglomerates LAR Particles are not distinguished LAR Fibre LAR Particles and LAR Fibre from LAR Agglomerates Cluster LAR Agglomerates Agglomerates Fig. 4 Illustration of the seven object categories used to categorize the SEM imaged objects by their visual morphology, their structure, and their degree of agglomeration individual fibers by drawing a path along the fiber and MWCNT agglomerates. They survived even if rotation- converting its length in pixels to nanometers using the al mixing durations exceeded 1 h. This suggests that ball image resolution. milling effects of the glass beads did not effectively micronize all MWCNT agglomerates. This was espe- cially the case for Baytubes MWCNTs, which were Results and discussion synthesized in highly tangled form and large agglomer- ates (see Fig. 1). Such local inhomogeneities could Mixing of MWCNTs with glass beads introduce short-term fluctuations of the aerosol concen- tration that might not be significant in large exposure To test the performance of the dry dispersion aerosoli- chambers. zation system, mixtures were prepared of glass beads The miscibility and transport behavior of MWCNTs and glass beads may be promoted by at- with either pristine ARIGM001 or pristine Baytubes MWCNTs in different mass ratios (see Table 1). Hori- tractive interaction between glass and MWCNTs. zontal rotations of the cylindrical mixing bottle around Such attraction may result from van-der-Waals inter- its rotational axis for a few minutes achieved an appar- action (dispersion forces) and from electrostatic ently homogenous distribution of the MWCNT material forces. Dispersion forces are governed by the contact in the glass beads. However, as the variations of the area between two object surfaces and are material inde- temporal stability of the aerosols generated from a con- pendent (Hamaker 1937), whereas triboelectrical charge tinuous material in Fig. 6 suggest, the homogeneity separation can lead to strong electrostatic interaction achieved by such a mixing approach cannot be assessed depending on material pairing (Matsusaka et al. with high accuracy by visual inspection alone. In some 2010). If at least one partner is an electrical insulator, batches, MWCNT-enriched or depleted zones appear to separated surface charges of opposite sign cannot re- have survived in the mixture. In future experiments, the combine easily if created by friction (tribocharging) use of alternative, e.g., wobble mixing techniques, and between to two materials of differing electronegativ- of lifting blades inside of the mixing bottle should be ity. Glass is known to exhibit a strong triboelectrically studied to further improve the mixing homogeneity. electropositive character, whereas the CNTs may Close-up visual inspection also revealed small-scale serve as electron donor partner. SEM analysis of a local inhomogeneities in the mixture caused by larger glass bead mixture with ARIGM001 showed that High aspect ratio Low aspect ratio > 3:1 < 3:1 154 Page 8 of 17 J Nanopart Res (2018) 20:154 Table 1 Overview of the prepared mixtures of MWCNTs and glass beads together with their experimental parameters for aerosolization Experiment MWCNT material MWCNT mass fraction Experimental parameters for aerosolization [g/kg] Nozzle feeding rate Nozzle outlet flow Mean object number b 3 [g/min] [L/min] concentration [#/cm ] A10 ARIGM001 0.01 11.0 16.9 4000 ± 320 A11 ARIGM001 0.01 12.0 19.8 3600 ± 130 A12 ARIGM001 0.01 13.9 19.9 2900 ± 175 A20 ARIGM001 0.06 14.7 19.9 6600 ± 580 A21 ARIGM001 0.06 10.3 19.8 8200 ± 90 A30 ARIGM001 0.20 12.4 19.8 52,000 ± 2300 A31 ARIGM001 0.20 9.7 19.8 46,000 ± 2200 A40 ARIGM001 0.33 11.0 16.9 64,000 ± 8200 B10 Baytubes 0.40 10.5 19.6 1700 ± 130 B20 Baytubes 4.00 10.5 19.6 7500 ± 1300 MWCNT mass per glass bead mass in the mixture Determined 30 min after start. Standard derivation given single MWCNT fibers and even agglomerates adhere of interaction forces and the optimum material pairing to glass bead surfaces (see Fig. 5). Other bead mate- are presently unclear. Excessive adherence of individ- rials with different triboelectrical potential, surface ual fibers should be avoided since it may lower the roughness, or polar and dispersive surface energy concentration of fiber morphologies in the aerosol. components may show different nanotube adherence A benefit of dry dispersion of MWCNTs in glass performance. Other interesting bead materials could beads is the long-term storability of dry powder mix- be, e.g., stainless steel, zirconia, or polymers like tures, since, different to liquid dispersions, no sedimen- polystyrene or polytetrafluoroethylene. Higher sur- tation of MWCNTs among glass beads occurs. Using a face roughness beads may promote not only adher- protective inert gas atmosphere may reduce possible ence to the bead but also CNT abrasion from agglom- oxidative material aging during mixing and storage in erate surfaces during mixing. The optimum strength a gas-tight bottle. Fig. 5 SEM image of MWCNTs attaching to a glass bead surface MWCNT aggregate found in a mixture of ARIGM001 and glass beads. The sample was coated with an iridium layer of 4-nm thickness prior to SEM analysis MWCNT fiber Glass bead surface A JNanopartRes (2018) 20:154 Page 9 of 17 154 Temporal development of the aerosol concentration 1.5× the approximate filling time, sufficiently stable concentration levels were observed for most tested The aerosol concentration generated with the studied MWCNT/GB mixtures. In dependence of the studied aerosolization system was monitored with a CPC over MWCNT material and mass ratio in the mixture, mean time. Figure 6 shows the results for the experiments object number concentrations from (1,723 ± 134) to A11, A12, A13, A40, B10, and B20, which were con- (64,033 ± 8,209) #/cm were measured for the plateau taining mixtures of glass beads and MWCNTs either of region. These concentration levels were maintained as type pristine ARIGM001 (BA^ experiments) or of type long as the MWCNT/GB mixture was continuously Baytubes (BB^ experiments) in concentrations as speci- transported to the Venturi nozzle. In our experiments, fied in Table 1. Once the feeding of mixtures was started we were able to produce MWCNT aerosols for up to (t ≔ 0), the number concentration in the chamber was 140 min, before aerosolization had to be stopped to increasing on a time scale of about 20 min, correspond- empty the collecting reservoir of the first cyclone stage. ing to the filling time of a volume of 400 L at an aerosol Longer aerosolization times are possible if larger flow of 20 L/min. After about 30 min, which is about reservoir volumes are used. Fig. 6 Temporal development of the concentration of MWCNT confidence level of the data for integration over the total experi- aerosols, as measured by CPC, during dry dispersion aerosoliza- ment time starting from minute 30. The relative errors for this tion experiments, cf. Table 1. Gray areas indicate the 1-sigma timespan are given as δ values in the diagram 154 Page 10 of 17 J Nanopart Res (2018) 20:154 At comparable mixture mass feed rate of about 11 g/ Interestingly, the APS size distributions in the range of min, the concentration of MWCNT material in the glass 500 to 1100 nm were significant lower than those mea- bead mixture had a significant influence on the object sured with SMPS. A comparable disparity in measure- number concentration in the resulting aerosols, which ments was reported from Baron et al. (2008)for can be seen by comparing the results for the mixtures SWCNT aerosols. The authors assumed that the partly A40 and A10, containing ARIGM001 with 0.33 and uncertain charging and light absorption behavior of 0.01 g/kg, respectively. The higher concentrated mixture these low-density, open structured, and conductive ob- A40 produced about 15 times higher aerosol concentra- jects may be the reason for such an observation. tion than mixture A10. A similar trend can be seen for Within the studied time ranges, no significant chang- the high and the low concentrated Baytubes mixtures es in the size distributions of the MWCNTaerosols were B10 and B20, respectively, even though the difference monitored, as can be seen in Fig. 7(A2–D2) showing the in object number concentration is less pronounced. SMPS size distributions as a function of time. Although It became evident that the output concentration was small fluctuations in the total number concentration had strongly dependent on the type of MWCNT material: an influence on the signal intensities, geometrical mean Mixtures with Baytubes formed significantly lower diameters as well as the distribution width were found to aerosol concentrations than those with ARIGM001 with lie in similar ranges for the compared aerosols. This is respect to the specific MWCNT mass feed rate. particularly apparent for the unimodal size distribution Baytubes were synthesized in large agglomerates of of the higher concentrated ARIGM001 aerosol in highly tangled structure (see Fig. 1). They most likely Fig. 7(B2). Here, the SMPS size distribution was mon- require higher energies for break up and detangling into itored for 70 min without showing significant changes in aerosolizable fragments than the smaller and more distribution width and mean aerodynamic diameter. loosely agglomerated material structure of ARIGM001. The dry dispersion aerosolization system studied The results show that our aerosolization method al- here has the potential to produce nanofiber aerosols of lows controlling the object number output concentra- highly stable output with respect to both concentration tions by changing the MWCNT mass fraction in the and size distribution for arbitrary times of operation. We glass bead mixture. This is comparable to atomizing believe that the powder mixture feeding system is the differently concentrated liquid CNT suspensions. key to long-term aerosol stability. It continuously sup- plies fresh nanofiber material to the Venturi nozzle, Temporal development of the aerodynamic size which prevents aging effects occurring in perpetually distribution agitated powder beds, but requires the mixture of nano- fibers and beads to be sufficiently homogenous. A feed- Parallel to number concentration monitoring with CPC, back loop that monitors the concentration in the aerosol supply line and controls the feeding rate via the turntable size distributions of generated aerosols were monitored with SMPS and APS over time. Figure 7(A1–D1) shows speed could correct for nanofiber concentration inho- the mean log-normal SMPS and APS size distribution of mogeneities in the bead mixture. each aerosol, averaged over several scans and measured in time ranges of sufficiently stable output concentra- Morphological characterization of the MWCNT tion. For ARIGM001 aerosols, unimodal size distribu- aerosols by SEM analysis tions with comparable geometrical mean diameters of 126 nm and respectively 122 nm were measured The generated aerosols were sampled on tracketch (Fig. 7(A1–B1)). A broader multimodal size distribution membrane filters. SEM images of the filters, as shown was found for both Baytubes aerosols (Fig. 7(C1–D1)). representatively in Fig. 8 for an ARIGM001 and a The distributions of Baytubes aerosols exhibit a tail Baytubes aerosol, were used to categorize all observed towards larger aerodynamic diameters. It indicates that, aerosol objects into one of the seven object categories of especially at this high mass concentration of 4 g/kg, Fig. 4. For each sample, between 1000 and 3000 objects Baytubes were far less effectively dispersed than were analyzed to characterize the aerosol morphology ARIGM001 mixtures. SEM images of the sampled aero- and to determine the concentration of individual fibers. sol in Fig. 8 confirm the presence of more microscale The results of this categorization are summarized in agglomerates for Baytubes than for ARIGM001. Table 2. For both ARIGM001 samples (A10 and A40) JNanopartRes (2018) 20:154 Page 11 of 17 154 Fig. 7 Mean log-normal SMPS and APS size distributions of the concentrations (left,A1–D1). The x-y-z plots on the right show MWCNT aerosols from mixture A10, A40 and B10, B20 as the SMPS size distributions for the same time ranges of all aerosols measured during time ranges of sufficiently stable number as function of time (right,A2–D2) 154 Page 12 of 17 J Nanopart Res (2018) 20:154 B1 A1 ARIGM001 Baytubes® B1 S8 A2 S2 S1 S3 S5 S4 S7 S6 A2 B2 A3 B3 A3 B3 350 nm 350 nm S8 S1 S2 S3 S4 S5 S6 S7 Fig. 8 SEM images of collected ARIGM001 (left)and Baytubes exhibit a mean tube diameter of about 14 nm. Therefore individual (right) aerosol with magnified areas A1–A3 and B1–B3. Both nanofibers are not easily identified in the upper right group of aerosols were well dispersed and contained of a high fraction out images B1–B3 of individual fibers. The insets S1–S7 were added since Baytubes as well as for the Baytubes sample B20, more than 30% designed to count and characterize low aspect ratio of the collected aerosol objects were individual particles, it appeared therefore important to determine MWCNT fibers. Moreover, almost 50% of all objects the amount of individual nanofibers and HAR objects in on these samples had a high aspect ratio equal or greater a test aerosol for the interpretation of instrument re- than 3:1. Only for experiment B10 lower ratios of HAR sponses. Possible limitations of the data obtained from objects and individual fibers were found. This was most the categorization approach applied here may arise from probably caused by the ratio of CNT-related and glass missing information on the diameter and size of the bead-related contributions to the aerosol. At the low categorized objects. This prevents estimating mass con- aerosol concentration of experiment B10, the finite glass centrations from SEM images. In addition, the thin and bead background became significant: LAR objects in- rather short Baytubes fibers found on samples B10 and creased to about 60%. These findings reveal the neces- B20 were categorized the same way as much thicker and sity to further reduce the background contribution from longer ARIGM001 fibers on sample A10 and A40. the glass beads. However, the toxicological relevance of the two fiber The morphological distinction between HAR and morphologies may be quite different to micron-sized pow- LAR objects was introduced here since the aerosol der structures (Pauluhn 2009). Occupational hygiene con- generation method is intended to be used for systematic trol may therefore require detecting and distinguishing in- laboratory studies on nanoparticle instrument response dividual and agglomerate fiber morphologies. It will be a to CNT-containing aerosols of known morphological challenge of future research and development to substitute distribution. As most nanoparticle instruments are the very laborious task of detecting and categorizing JNanopartRes (2018) 20:154 Page 13 of 17 154 Table 2 Results of the morphological categorization for the samples A10, A40 and B10, B20. All SEM imaged particles were categorized by their visual shape, structure and estimated degree of agglomeration into 7 particle categories Percentage of particle category for the sample A10 [%] N = 3355 A40 [%] N = 3327 B10 [%] N = 1110 B20 [%] N =2147 Total number of the categorized particles HAR particle agglomerates 3.7 2.8 1.3 1.8 HAR fiber agglomerates 4.4 5.2 1.8 6.6 HAR fiber clusters 6.6 9.0 4.1 4.8 Individual fibers 32.1 30.9 21.2 34.5 LAR particle agglomerates and individual particle objects 49.5 44.6 62.1 40.1 LAR fibers agglomerates 3.1 5.1 7.0 9.0 LAR fiber clusters 0.7 2.5 2.5 3.2 Total HAR objects 46.7 47.8 28.4 47.7 Total LAR objects 53.3 52.2 71.6 52.3 nanofibers on aerosol filter samples, which was performed aerosol to be depending on the MWCNT material. Fur- here, by an online fiber detection technique. ther experiments with other types of CNT materials are The approach followed in the present work gave recommended to study weather this aerosolization tech- valuable insights into the HAR object release propensity nique is suited for the generation of narrower length of two different MWCNTs types under studied disper- or size distributions. Also variations of the air flow rate sion conditions. The very similar individual fiber release through the Venturi nozzle that governs dispersive ratio came quite unexpected and shows that the applied forces should be studied in future. dispersion energies were high enough to break individ- The aerosolization efficiency (aerosol mass released ual fibers off the surface of agglomerates and to split up vs. CNT mass used) of our dry dispersion technique larger agglomerates. Such breaking of tangled fibers could not be determined here. The mass concentration during dispersion shifts the fiber length distribution to of the generated nanoscale aerosols was too low to be smaller values. However, due to the highly tangled state determined by weighing. A second approach to estimat- of the starting materials, the length distributions before ing the aerosol mass based on particle size distribution and after dispersion cannot be compared, since the data obtained from SMPS and APS is not reliable for lengths of tangled nanotubes cannot be measured reli- fibrous aerosols. Firstly, since the main peak(s) of the ably in the agglomerated state. SMPS size distributions in our experiments resulted For further aerosol characterization of the sampled from individual fibers but do not provide fiber length MWCNT aerosols, the geometrical length and diameter information and therefore do not provide mass informa- of all imaged and categorized fibers on sample A10 and tion. Secondly, for A10, A40, and B10, the contribution B20 were determined and plotted as scatter plot and of microscale agglomerates in the right hand tail of the histograms in Fig. 9. Log-normal peak fitting of the SMPS and APS size distributions is insignificant in our geometric fiber length distributions gave an average data especially for ARIGM001, see insets in Fig. 7(A1– length of around 350 nm for ARIGM001 (parameter w D1). It is nonetheless expected to dominate the total in Fig. 9), whereas Baytubes had a shorter length of aerosol mass since microscale agglomerates contain about 200 nm. The majority of analyzed MWCNTs (> many tangled CNTs. For experiment B20, the strong 90%) exhibited a length below 5 μm. Especially discrepancy between the SMPS and APS reading gives Baytubes formed predominantly short fibers below rise to uncertainties on the shape of the size distribution of agglomerates. A third possible approach to estimating 1.5 μm. This small length is believed to be related to the large agglomerate size and high degree of tangle- the aerosol mass could rely on measuring the volume of ment of Baytubes (cf. Fig. 1). Tangled and interlocked agglomerates and the length and diameter of individual fibers must to be broken to be released from the fiber fibers imaged with SEM. This is a very laborious task network and to become observable as individual fibers. and requires assuming agglomerate densities. Such den- The results show the fiber length distribution in the sity assumption is necessary for any mass estimation 154 Page 14 of 17 J Nanopart Res (2018) 20:154 Fig. 9 Scatter plots and histograms of the geometrical fiber diam- for Baytubes (right, N = 118) with log-normal (length) and Gauss- eter and length data pairs of individual fibers found in the aerosol ian (diameter) peak fitting sample A10 for ARIGM001 (left, N = 176) and aerosol sample B20 approach based on agglomerate geometric or mobility Analogies between dry and liquid aerosolization diameter data. For partially dispersed CNT materials, these densities are unknown. Future experiments will It is instructive to review principle analogies between aim at characterizing aerosols using a so-called nano- the dry dispersion concept presented here and atomizing particle mass classifier (nano-PMC) that was developed concepts using liquid dispersions. It shows that similar recently (Broßell et al. 2015). In combination with a components and process steps are required both for dry differential mass analyzer, it allows measuring the mass and liquid dispersion. All of them require individual and density of aerosol particles. optimization for best aerosol performance. Themicroscalebeadsusedherecorrespond to a cheap and non-toxic solvent in liquid suspensions. Like Reproducibility of the aerosolization technique a solvent, the beads allow nanofiber material dilution in replicate experiments and concentration control. For many material variants, The aerosolization technique developed here was also studied with respect to reproducibility of the generated concentration. For this comparison, all ARIGM001 mix- tures were aerosolized under similar experimental con- ditions. The resulting aerosol number concentrations measured from minute 30 to 60 after start of powder feeding are compared statistically using box-whisker plotting in Fig. 10. For the evaluated time span, the aerosols of each group of MWCNT/GB mixture with 0.01, 0.06, and 0.20 g/kg in Fig. 10 showed a fair to good degree of reproducibility. For mixtures with equal MWCNT amount, aerosols with mean object number concentra- tions in the same size range resulted. A maximum standard derivation of 25% in mean object number was found. A feedback loop that monitors the concen- Fig. 10 Box-whisker plot for the temporal development of the concentrations generated from ARIGM001 and glass bead mix- tration in the aerosol supply line and controls the feeding tures during minute 30 to 60 after start of feeding (box 25/75%; rate via the turntable speed could correct for CNT inho- whisker 1.5; mean: diamond, median: notch). The vertical separa- mogeneities in the glass bead mixture and further im- tors group the three compared mixtures of 0.01, 0.06, or 0.20 g/kg prove temporal stability and reproducibility. mass ratio MWCNT/GB JNanopartRes (2018) 20:154 Page 15 of 17 154 spherical beads can achieve transforming sticky and a flow rate of 20 L/min. The obtained MWCNTaerosols clogging low-density nanofiber materials into free- contained a high fraction of individual fibers of up to flowing powder mixtures. Their liquid-like character fa- 34%. The toxicological relevance of the obtained fiber cilitates continuous feeding to the atomizing unit, e.g., a aerosols was assessed by quantifying to the content of Venturi nozzle, and controlling the mass feed rate. fibers of WHO geometry. The thicker and more loosely Obtaining homogenous mixtures of nanofiber mate- agglomerated material ARIGM001 resulted in a higher rials in beads may require optimization of mixing move- content of WHO fibers in the aerosol than Baytubes, ments which are generally accompanied by ball milling both in relative and absolute concentration. effects on nanofiber agglomerates. The resulting energy Compared to liquid dispersion and aerosolization of transfer to the nanofiber materials is in analogy with, CNTs, dry mixing with glass bead exhibits a number of e.g., ultrasonication of nanofibers in liquid suspension. benefits. It requires neither chemical nor mechanical Successful and temporally stable mixing results of nano- pre-treatment of the CNTs. Since dry mixtures do not fibers and beads however do not require pre-treating of sediment, no possibly toxic organic solvents and surfac- the nanofibers with chemical functionalization or sur- tants are required to achieve stable dispersions in glass factant coating as in liquid dispersion. beads. Glass beads are an inexpensive dilution material Even the polarity of a solvent, which is highly critical and may be re-used. Similar to dryer columns for liquid for preparing liquid suspensions, bears analogies to dry atomization, also our dry dispersion technique requires mixing: Optimum triboelectrical pairing of powder an additional cyclone stage for downstream separation components might result in an adhering nanofiber corona of glass beads and CNTs. The dry aerosolization process on the bead surface and further stabilize nanofibers in was found to allow high aerosol stability. Observed the bead mixture via attractive electrostatic forces concentration fluctuations were attributed to partially (Matsusaka et al. 2010). inhomogeneous mixing of CNTs and glass beads, a Finally, the bead-filtering cyclone stage used here is process step that requires further improvement. an analogue of a drying unit necessary for obtaining dry Future work should be devoted to aerosols from atomized liquid dispersion. & Wind-sifting glass beads prior to use in order to reduce the glass fragment concentration. Conclusion and perspective & Studying triboelectric charge separation during mixing and its role in attaching nanoscale fragments Providing fiber aerosols of controlled individual fiber to beads. concentration for fiber toxicological inhalation studies & Optimizing dry mixing to further improve the intended as well as for assessing aerosol measurement perfor- mixing with glass bead and to control ball milling effects of MWCNTs. mance and exposure limit control strategies is an impor- tant task. & Improving process stability via a feedback loop of Here, a dry dispersion aerosolization technique was the CPC concentration to the powder feeding rate. developed that uses mixtures of nanofiber materials and & Estimating the energy transfer necessary to breakup microscale beads that are fed continuously through a or disentangle CNT agglomerates. Venturi nozzle to generate nanofiber-containing aero- & Varying the air flow rate through the Venturi nozzle sols. The technique was studied for two morphological- that governs dispersive forces. ly very different MWCNT materials in glass bead mix- & Further narrowing of the morphological distribution tures of varied mass ratio. Aerosols were characterized of the aerosol by the use of less tangled materials. by CPC, SMPS, APS, and morphological analysis of filter samples with SEM to determine and categorize the morphological composition of the generated aerosols, Acknowledgments The authors thank Guillermo Orts-Gil for relative and absolute concentrations of individual fibers TEM analysis and Nils Kujath for the technical assistance. as well as fiber length and diameter distributions. Funding This work was funded by the German Federal Ministry The generator allows controlling the aerosol concen- of Education and Research (BMBF), FKZ 03X0127B, and was tration in an extremely wide range. Aerosols of up to conducted in the frame of the SIINN ERA-Net project NanoIndEx 10 3 2×10 individual nanofibers per m were generated at in cooperation with other project partners who funded by the 154 Page 16 of 17 J Nanopart Res (2018) 20:154 French National Funding Agency for Research (ANR), the British De Volder MF, Tawfick SH, Baughman RH, Hart AJ (2013) Technology Strategy Board (TBS), and the Swiss TEMAS AG. Carbon nanotubes: present and future commercial applica- tions. Science 339(6119):535–539 Compliance with ethical standards All relevant ethical stan- Fujitani Y, Furuyama A, Hirano S (2009) Generation of airborne dards were satisfied. multi-walled carbon nanotubes for inhalation studies. Aerosol Sci Technol 43(9):881–890 Hamaker HC (1937) The London—van der Waals attraction be- Conflict of interest The authors declare that they have no con- tween spherical particles. Physica 4(10):1058–1072 flict of interest. Han SG, Andrews R, Gairola CG (2010) Acute pulmonary re- sponse of mice to multi-wall carbon nanotubes. Inhal Toxicol 22(4):340–347 Open Access This article is distributed under the terms of the Harris G, Maguire B (1968) A gravimetric dust sampling instru- Creative Commons Attribution 4.0 International License (http:// ment (simpeds): preliminary results. Ann Occup Hyg 11(3): creativecommons.org/licenses/by/4.0/), which permits unrestrict- 195–201 ed use, distribution, and reproduction in any medium, provided Jennerjohn N, Eiguren-Fernandez A, Prikhodko S, Fung DC, you give appropriate credit to the original author(s) and the source, Hirakawa KS, Zavala-Mendez JD, Hinds W, Kennedy NJ provide a link to the Creative Commons license, and indicate if (2010) Design, demonstration and performance of a versatile changes were made. electrospray aerosol generator for nanomaterial research and applications. Nanotechnology 21(25):255603 Lee S-B, Lee J-H, Bae G-N (2010) Size response of an SMPS– References APS system to commercial multi-walled carbon nanotubes. J Nanopart Res 12(2):501–512 Matsusaka S, Maruyama H, Matsuyama T, Ghadiri M (2010) Ahn K-H, Kim S-M, Yu IJ (2011) Multi-walled carbon nanotube Triboelectric charging of powders: a review. Chem Eng Sci (MWCNT) dispersion and aerosolization with hot water at- 65(22):5781–5807 omization without addition of any surfactant. Saf Health McKinney W, Chen B, Frazer D (2009) Computer controlled Work 2(1):65–69 multi-walled carbon nanotube inhalation exposure system. Asbach C, Kaminski H, Lamboy Y, Schneiderwind U, Fierz M, Inhal Toxicol 21(12):1053–1061 Todea AM (2016) Silicone sampling tubes can cause drastic Mercer R, Scabilloni J, Hubbs A, Wang L, Battelli L, McKinney artifacts in measurements with aerosol instrumentation based W, Castranova V, Porter D (2013) Extrapulmonary transport on unipolar diffusion charging. Aerosol Sci Technol 50(12): of MWCNT following inhalation exposure. Part Fibre 1375–1384 Toxicol 10(1):38 Bahk YK, Buha J, Wang J (2013) Determination of geometrical Myojo T, Oyabu T, Nishi K, Kadoya C, Tanaka I, Ono-Ogasawara length of airborne carbon nanotubes by electron microscopy, M, Sakae H, Shirai T (2009) Aerosol generation and mea- model calculation, and filtration method. Aerosol Sci surement of multi-wall carbon nanotubes. J Nanopart Res Technol 47(7):776–784 11(1):91–99 Baron PA, Deye GJ, Chen BT, Schwegler-Berry DE, Shvedova Nagai H, Okazaki Y, Chew SH, Misawa N, Yamashita Y, AA, Castranova V (2008) Aerosolization of single-walled Akatsuka S, Ishihara T, Yamashita K, Yoshikawa Y, Yasui carbon nanotubes for an inhalation study. Inhal Toxicol H, Jiang L, Ohara H, Takahashi T, Ichihara G, Kostarelos K, 20(8):751–760 Miyata Y, Shinohara H, Toyokuni S (2011) Diameter and Brockmann JE, Rader DJ (1990) APS response to nonspherical rigidity of multiwalled carbon nanotubes are critical factors in particles and experimental determination of dynamic shape mesothelial injury and carcinogenesis. Proc Natl Acad Sci factor. Aerosol Sci Technol 13(2):162–172 108(49):E1330–E1338 Broßell D, Valenti M, Bezantakos S, Schmidt-Ott A, Biskos G (2015) Oberdörster G, Castranova V, Asgharian B, Sayre P (2015) The nanoparticle mass classifier (nano-PMC): development, Inhalation exposure to carbon nanotubes (CNT) and carbon characterisation, and application for determining the mass, ap- nanofibers (CNF): methodology and dosimetry. J Toxicol parent density, and shape of particles with masses down to the Environ Health B Crit Rev 18(3–4):121–212 zeptogram range. Aerosol Sci Technol 49(7):495–507 O'Shaughnessy PT, Adamcakova-Dodd A, Altmaier R, Thorne PS Chen BT, Schwegler-Berry D, McKinney W, Stone S, Cumpston (2014) Assessment of the aerosol generation and toxicity of JL, Friend S, Porter DW, Castranova V, Frazer DG (2012) carbon nanotubes. Nanomaterials 4(2):439–453 Multi-walled carbon nanotubes: sampling criteria and aerosol Pauluhn J (2009) Subchronic 13-week inhalation exposure of rats characterization. Inhal Toxicol 24(12):798–820 to multiwalled carbon nanotubes: toxic effects are determined Chen BT, Schwegler-Berry D, Cumpston A, Cumpston J, Friend by density of agglomerate structures, not fibrillar structures. S, Stone S, Keane M (2016) Performance of a scanning Toxicol Sci:kfp247 mobility particle sizer in measuring diverse types of airborne Pauluhn J, Rosenbruch M (2015) Lung burdens and kinetics of nanoparticles: multi-walled carbon nanotubes, welding multi-walled carbon nanotubes (Baytubes) are highly depen- fumes, and titanium dioxide spray. J Occup Environ Hyg dent on the disaggregation of aerosolized MWCNT. 13(7):501–518 Nanotoxicology 9(2):242–252 Dazon C, Witschger O, Bau S, Payet R, Beugnon K, Petit G, Garin Plitzko S, Gierke E, Dziurowitz N, Broßell D (2010) Erzeugung T, Martinon L (2017) Dustiness of 14 carbon nanotubes using von CNT/CNF-Stäuben mit einem Schwingbett- the vortex shaker method. J Phys Conf Ser 838(1):012005 Aerosolgenerator und Charakterisierung der JNanopartRes (2018) 20:154 Page 17 of 17 154 Fasermorphologie mithilfe eines Thermalpräzipitators als Spurny KR, Boose C, Hochrainer D (1975) Zur Zerstäubung von Sammelsystem. Gefahrstoffe-Reinhaltung der Luft 70(1):31 Asbestfasern in einem Fließbett-Aerosolgenerator. Staub, Porter DW, Hubbs AF, Mercer RR, Wu N, Wolfarth MG, Sriram Reinhaltung der Luft 35(12):440–445 K, Leonard S, Battelli L, Schwegler-Berry D, Friend S, Stanton MF, Wrench C (1972) Mechanism of mesothelioma in- Andrew M, Chen BT, Tsuruoka S, Endo M, Castranova V duction with asbestos and fibrous glass. J Natl Cancer Inst 48: (2010) Mouse pulmonary dose- and time course-responses 797–821 induced by exposure to multi-walled carbon nanotubes. Su W-C, Cheng YS (2014) Carbon nanotubes size classification, Toxicology 269(2):136–147 characterization and nasal airway deposition. Inhal Toxicol Pott F, Friedrichs KH (1972) Tumoren der Ratte nach i.p.- 26(14):843–852 Injektion faserförmiger Stäube. Naturwissenschaften 59(7): Totaro S, Cotogno G, Rasmussen K, Pianella F, Roncaglia M, 318 Olsson H, Riego Sintes JM, Crutzen HP (2016) The JRC Rittinghausen S, Hackbarth A, Creutzenberg O, Ernst H, Heinrich nanomaterials repository: a unique facility providing repre- U, Leonhardt A, Schaudien D (2014) The carcinogenic effect sentative test materials for nanoEHS research. Regul Toxicol of various multi-walled carbon nanotubes (MWCNTs) after Pharmacol 81:334–340 intraperitoneal injection in rats. Part Fibre Toxicol 11:18 Vo E, Zhuang Z (2013) Development of a new test system to Ryman-Rasmussen JP, Tewksbury EW, Moss OR, Cesta MF, determine penetration of multi-walled carbon nanotubes Wong BA, Bonner JC (2009) Inhaled multiwalled carbon through filtering facepiece respirators. J Aerosol Sci 61:50– nanotubes potentiate airway fibrosis in murine allergic asth- 59 ma. Am J Respir Cell Mol Biol 40(3):349–358 Vo E, Zhuang Z, Birch E, Zhao Q, Horvatin M, Liu Y (2014) Seto T, Furukawa T, Otani Y, Uchida K, Endo S (2010) Filtration Measurement of mass-based carbon nanotube penetration of multi-walled carbon nanotube aerosol by fibrous filters. through filtering facepiece respirator filtering media. Ann Aerosol Sci Technol 44(9):734–740 Occup Hyg:meu005 Shvedova AA, Kisin ER, Mercer R, Murray AR, Johnson VJ, Wang J, Bahk YK, Chen S-C, Pui DY (2015) Characteristics of Potapovich AI, Tyurina YY, Gorelik O, Arepalli S, airborne fractal-like agglomerates of carbon nanotubes. Schwegler-Berry D (2005) Unusual inflammatory and Carbon 93:441–450 fibrogenic pulmonary responses to single-walled carbon WHO (1985) Reference methods for measuring man-made min- nanotubes in mice. Am J Phys Lung Cell Mol Phys 289(5): eral fibres (MMMF). Prepared by WHO/EURO Technical L698–L708 Committee for Evaluating MMMF

Journal

Journal of Nanoparticle ResearchSpringer Journals

Published: Jun 2, 2018

References

You’re reading a free preview. Subscribe to read the entire article.


DeepDyve is your
personal research library

It’s your single place to instantly
discover and read the research
that matters to you.

Enjoy affordable access to
over 18 million articles from more than
15,000 peer-reviewed journals.

All for just $49/month

Explore the DeepDyve Library

Search

Query the DeepDyve database, plus search all of PubMed and Google Scholar seamlessly

Organize

Save any article or search result from DeepDyve, PubMed, and Google Scholar... all in one place.

Access

Get unlimited, online access to over 18 million full-text articles from more than 15,000 scientific journals.

Your journals are on DeepDyve

Read from thousands of the leading scholarly journals from SpringerNature, Elsevier, Wiley-Blackwell, Oxford University Press and more.

All the latest content is available, no embargo periods.

See the journals in your area

DeepDyve

Freelancer

DeepDyve

Pro

Price

FREE

$49/month
$360/year

Save searches from
Google Scholar,
PubMed

Create lists to
organize your research

Export lists, citations

Read DeepDyve articles

Abstract access only

Unlimited access to over
18 million full-text articles

Print

20 pages / month

PDF Discount

20% off