This content has been downloaded from IOPscience. Please scroll down to see the full text.

Download details:

IP Address: 128.97.83.27

This content was downloaded on 14/11/2014 at 20:49

Please note that terms and conditions apply.

Design, demonstration and performance of a versatile electrospray aerosol generator for

nanomaterial research and applications

View the table of contents for this issue, or go to the journal homepage for more

2010 Nanotechnology 21 255603

(http://iopscience.iop.org/0957-4484/21/25/255603)

Home Search Collections Journals About Contact us My IOPscience

IOP PUBLISHING NANOTECHNOLOGY

Nanotechnology 21 (2010) 255603 (11pp) doi:10.1088/0957-4484/21/25/255603

Design, demonstration and performance ofa versatile electrospray aerosol generatorfor nanomaterial research andapplicationsNancy Jennerjohn1, Arantzazu Eiguren-Fernandez2,Sergey Prikhodko3, David C Fung1, Karen S Hirakawa1,Jose D Zavala-Mendez4, William Hinds1 and Nola J Kennedy1

1 Environmental Health Sciences Department, UCLA, USA2 Center for Occupational and Environmental Health, UCLA, USA3 Materials Science and Engineering Department, UCLA, USA4 Mechanical and Aeronautical Engineering Department, UC Davis, USA

E-mail: njnnrjhn@gmail.com

Received 1 September 2009, in final form 30 April 2010Published 2 June 2010Online at stacks.iop.org/Nano/21/255603

AbstractCarbon nanotubes are difficult to aerosolize in a controlled manner. We present a method forgenerating aerosols not only of carbon nanotubes, but also of many reference and proprietarymaterials including quantum dots, diesel particulate matter, urban dust, and their mixtures,using electrospraying. This method can be used as a teaching tool, or as the starting point foradvanced research, or to deliver nanomaterials in animal exposure studies. This electrospraysystem generates 180 µg of nanotubes per m3 of carrier gas, and thus aerosolizes anoccupationally relevant mass concentration of nanotubes. The efficiency achievable forsingle-walled carbon nanotubes is 9.4%. This system is simple and quick to construct usingordinary lab techniques and affordable materials. Since it is easy to replace soiled parts withclean ones, experiments on different types of nanomaterial can be performed back to backwithout contamination from previous experiments. In this paper, the design, fabrication,operation and characterization of our versatile electrospray method are presented. Also, themorphological changes that carbon nanotubes undergo as they make the transition from drypowders to aerosol particles are presented.

S Online supplementary data available from stacks.iop.org/Nano/21/255603/mmedia

(Some figures in this article are in colour only in the electronic version)

1. Introduction

Electrospraying is one of several possible methods forforming aerosols composed of or containing nanoparticlesand nanomaterials. In electrospraying, a suspension in areservoir flows without interruption toward a sprayhead (needleor capillary), forming a hanging ‘pendant’ drop. Voltage isapplied between the sprayhead and a nearby counterelectrode,forming an intense electric field in the small space betweenthem. As voltage is increased, the interaction betweenthe moving charges in the suspension and the electric field

strengthens, ultimately forcing the pendant drop to break apartinto a spray of charged aerosol droplets. Because these dropletsare charged unipolarly, mutual repulsion results in a welldispersed spray (Hartman et al 1999). The aerosol movestoward the counterelectrode and either deposits on it, or isforced to move beyond it, for example by encountering astream of carrier gas or a vacuum. If the electric field is notthe main force driving the spray (Cloupeau and Prunet-Foch1994) or if the suspension flow rate is very high, then sprayingis occurring rather than electrospraying, though both can occursimultaneously.

0957-4484/10/255603+11$30.00 © 2010 IOP Publishing Ltd Printed in the UK & the USA1

Nanotechnology 21 (2010) 255603 N Jennerjohn et al

Electrospraying requires the material to be aerosolizedto be mixed with a solvent, whereas no solvent is neededwhen aerosolizing using some mechanical methods (Baronet al 2008). Even if the material is insoluble, irreversiblealterations due to the presence of the solvent likely occur,such as absorption, dispersion, coating and chemical reaction.However, electrosprayed aerosols are still representative ofdry dispersed airborne nanomaterials that could be found inoccupational settings. In fact, professionals concerned withworker safety encourage the mixing and binding of certainnanomaterials as soon in the production process as possibleto minimize inadvertent aerosolization. Aerosols of singleagent nanomaterials will probably seldom be encounteredby workers. Instead, nanomaterials will often be foundalready mixed with binding agents and with common airborneworkplace contaminants such as dust and diesel.

Maynard et al (2004) found 0.7–52.7 µg m−3 of nanotubematerial in air during simulated nanotube transfer operations.Four years later, Han et al (2008) reported four times more,210 µg m−3, in a nanotube synthesis facility. Our electrosprayaerosol generator, a modification of a device designed byDr Joelle S Underwood (Ahmad et al 2009), produces ananotube concentration of 180 µg m−3, and thus aerosolizesoccupationally relevant mass concentrations of nanotubes. Italso aerosolizes quantum dots, diesel particulate matter, dustand their mixtures. These aerosols are characterized and thenused to challenge air-monitoring instruments. Such challengescan reveal which of the available instruments are efficaciousfor exposure assessment in the nanomaterial workplace. Air-monitoring instruments may misinform or even malfunctionwhen encountering nanoaerosols because of their atypical size,density, charging, magnetism and shape. Carbon nanotubes aredifficult to aerosolize in a controlled and robust manner. Strongvan der Waals interactions cause them to bundle. This explains,in part, the paucity of animal inhalation studies investigatingnanotube aerosols (Shvedova et al 2008, Mitchell et al 2007).

In this paper, the design criteria for our versatileelectrospray generator are presented. The fabrication,operation and characterization of our system are specified.Finally, the morphology of the aerosol particles that result fromelectrospraying five types of carbon nanotubes are presented.

2. Theory

Electrospraying is only one method for aerosolizing nanopar-ticles and nanomaterials. Other widely used methods includenebulization and mechanical agitation. The electrospraying ofnanotubes has previously been reported by Jayasinghe (2006)and Ku et al (2001), nanorods by Song et al (2005), and nano-sized PSL particles by Kim et al (2005).

Electrospraying has a minimum and maximum suspensionflow rate which depend on factors unique to each electrospraysystem including distance between the sprayhead and thecounterelectrode (‘gap length’), the type of sprayhead used,and suspension viscosity and surface tension. Such factorsdefine, for each suspension’s conductivity and flow rate, therange of voltages for which electrospraying is most stable.Also, aerosol particles generated using electrospraying may be

liquid, solid, multiphase or mixtures of all three. The structureof the electrosprayed particles depends in part on whether thematerial agglomerates or disperses in suspension while in thereservoir. The amount of solvent remaining in the final aerosolvaries depending on solvent volatility, and can be controlledby altering the volume of and residence time in the carrier gas,and by using techniques such as diffusion drying.

When rapid dripping shatters into a spray at the criticalvoltage, a persistent meniscus forms at the tip of the sprayhead.As voltage is further increased, this meniscus develops intoa cone which changes shape as voltage is altered. Theseshapes are called ‘modes’ and have been extensively studiedby electrospray researchers. The modes in order of increasingvoltage are the dripping, microdripping, pulsating jet, simplejet, cone-jet and multi-jet modes. The cone-jet mode formsa more stable electrospray and a more monodisperse aerosolthan any other mode. The cone-jet mode has been achievedusing our device but is not the mode we usually operate inbecause it does not deliver enough mass to duplicate the massconcentration of nanotubes in air in occupational settings. Ifwe had used multiple nozzles (Deng et al 2006), it mighthave been possible to operate in the cone-jet mode and deliverenough mass. Because a single nozzle, high voltage, and highsuspension flow rates were used, we operated in the multi-jetmode, and generated a polydisperse aerosol. Duby et al (2006)reported that the multi-jet mode, even though less stable andmonodisperse than the cone-jet mode, is likely the mode thatdelivers the most mass.

In an electrospray device, the carrier gas is a dielectricmaterial. If dielectric strength is exceeded, a dielectricmaterial breaks down, allowing the formation of a conductivepath through it. When using high voltages to electrospray,the dielectric breakdown voltage of the carrier gas may beexceeded in the region between the electrodes, leading to‘corona discharge’ in which the surrounding gas moleculesionize, becoming charged molecules. Corona discharge islikely occurring during electrospraying if there is visibleglowing or audible sizzling. When low level corona dischargecan be regulated, it can be used to suppress aerosol charging. Inair and at atmospheric pressure, electrical discharges can eitherbe impulsive, inducing electric field variations that destabilizethe electrospray, or continuous, inducing a constant electricfield that stabilizes the electrospray. If the discharges arecontinuous and occur during the cone-jet mode, this morestable state is sometimes referred to as the ‘cone-jet-glow’mode (Borra et al 1999b, 2004). However, if corona dischargegrows, an avalanche of free electrons may begin (Cloupeau1994) which cannot be regulated, preventing adequate controlof the electrospray system. Some researchers electrosprayfreely into a vacuum to eliminate corona discharge (Cloupeauand Prunet-Foch 1994). In our system, 100% CO2 is thecarrier gas. CO2 has high dielectric strength and is oftenadded to carrier gas to suppress corona discharge. Thehigher the applied voltage, the more likely it is that coronadischarge occurs. Using high voltage is unavoidable whenelectrospraying suspensions with high surface tension such aswater. High surface tension exerts strong capillary pressurewhich must be opposed by the suction exerted by electrostatic

2

Nanotechnology 21 (2010) 255603 N Jennerjohn et al

pressure which helps draw suspension out of the sprayhead,and electrostatic pressure is large when voltage is high.

In electrospraying, the aerosol particles that emergefrom the sprayhead are highly charged. They are alsounipolar and therefore mutually repel via Coulombic repulsion.Researchers who partially neutralize these aerosols employseveral methods. These include radioactive charge neutralizersand corona dischargers which ionize gas molecules in thesurrounding carrier gas, producing bipolar ions. These ions areattracted to oppositely charged aerosol particles, neutralizingthem to different degrees. Neither method works in a vacuumsince they require the presence of gas molecules. Anothertechnique uses the coagulation of positive and negative dropletsproduced by two electrosprays operated in opposite polarity(Borra et al 1999a). The degree of charge neutralizationachieved by any given method depends on how highly chargedthe aerosol particles are to begin with. The charge state ofthe aerosol particles exiting a neutralizer is often assumedto be the Boltzmann equilibrium charge distribution. Theamount of neutralization this distribution provides to aerosolparticles depends on the size of the individual particles andtheir residence time within the neutralizer (Hinds 1999). Thestrength of the radioactive source appropriate for neutralizationis debatable, and can be too small or too large. Ifthe Boltzmann distribution is not achieved, then those air-monitoring instruments assuming this distribution, and relyingon aerosol charging in their operating principles, will likelynot report accurate number concentrations and particle sizedistributions. On the other hand, a neutralizer that is toostrong can emit its own aerosol due to particle formation fromradiolysis of air (Leong et al 1983).

Details of a variety of electrospray systems are given intables 1 and 2 of the supplementary information (availableat stacks.iop.org/Nano/21/255603/mmedia). These detailswere extracted from publications that provided substantialinformation about their methods. Table 1 (availableat stacks.iop.org/Nano/21/255603/mmedia) specifies eachsystem’s applied voltage, carrier gas, suspension, flow rate,conductivity, surface tension and viscosity. A wide range ofparameters lead to successful electrospraying. For example,suspension flow rate can vary by six orders of magnitude, from5 nl (Kelly et al 2006) to 6 ml (Jayasinghe 2006) per minute.Conductivity can range over nine orders of magnitude, from10−9 S m−1 (Cloupeau and Prunet-Foch 1994) to 0.8 S m−1

(Chen et al 1995). Sprayhead inner diameter can range from5 (Kelly et al 2006) to 8000 µm (Hartman et al 2000).Table 2 (available at stacks.iop.org/Nano/21/255603/mmedia)specifies sprayhead type and diameter, charge neutralizationmethod, droplet and relic particle size, and instrumentsused by researchers for measuring particle size and numberconcentration.

The aerosol droplets produced at the exit of the sprayheadusually decrease in size as they travel downstream dueto evaporation. These smaller aerosol particles are called‘relics’. Some of them form in a few milliseconds priorto neutralization, and possibly affect final aerosol particlemorphology. In general, droplets achieve a smaller finalsize as the suspension’s conductivity increases and flow rate

Figure 1. Purge box. The four symmetric carrier gas ports and thesprayhead port are at the front, and the aerosol exit is at the rear.

decreases. For researchers striving to operate solely in thecone-jet mode, suspension flow rate is adjusted depending onthe suspension conductivity until this mode is achieved. Inour research, flow rate is fixed to provide a mass deliverythat best reproduces what would be encountered in anoccupational setting. When electrospraying highly viscousliquids such as glycerol, slower liquid flow rates producesmaller aerosol particles which are still in the liquid state.When electrospraying solutions instead of suspensions, wheresolids completely dissolve in the solvent, for example sugarin water, slower liquid flow rates again produce smalleraerosol particles whose size is so predictable, depending oninitial concentration in the solvent and other factors, that theycan be used for calibration (TSI 2003). However, wheninsoluble or partially soluble materials are aerosolized, thesize of the final aerosol particle may not be related to theliquid flow rate. It is possible to produce small relic aerosolparticles from sprayheads with large inner diameters. Forexample, in our system, 10 nm quantum dots have beenelectrosprayed using a 720 µm capillary. Most nanomaterialsuspensions will be able to be dispersed in gases for nano-sized aerosol dispersion studies. In this paper, we focus onthe electrospraying of droplets of liquid suspensions of carbonnanotubes, followed by the evaporation of their carrier solvent,leaving dry nanomaterial in the form of aerosol particles.

3. System description and methodology

3.1. Fabrication

Underwood’s design is a variation of the ‘cross’ (Lenggoroet al 2002) which has been used by electrospray researchersfor many years. The sprayhead and counterelectrode areenclosed in a rectangular Pomona Electronics 3850 moldedthermoplastic box (Everett, WA), hereafter called the ‘purgebox’. This box has eight ports: sprayhead, carrier gas(four symmetric ports), aerosol exit, illumination window andviewing window (figure 1).

A conductive needle and nonconductive capillary arejoined using a union body assembly (figure 2). The union bodyassembly consists of two Upchurch Scientific (Oak Harbor,WA) ‘NanoTight Kit’ capillary sleeves, a NanoTight union

3

Nanotechnology 21 (2010) 255603 N Jennerjohn et al

Figure 2. Union body assembly. It joins the needle on the left to the capillary on the right. The intervening parts left to right are a capillarysleeve, nut, ferrule, union body, ferrule, nut, and capillary sleeve.

Figure 3. Needle joined by the union body assembly to the capillarywhich extends into purge box.

body, and ‘NanoTight Fitting Long PEEK’ ferrules and nutsmatching the capillary sleeves. The union body (178 µm indiameter) may need to be increased, if blockage occurs, to600 µm, using a needle in a vise. The nonconductive capillaryis the sprayhead. The counterelectrode is a conical nut at therear of the purge box. The sprayhead is at a higher dc potentialthan the grounded counterelectrode. In this configuration,electrospraying has positive polarity, and creates positivelycharged aerosol particles. When polarity is negative, aerosolparticles are negatively charged, and the same cone modesoccur. However, corona discharge is less likely if polarity ispositive (Cloupeau and Prunet-Foch 1994).

The entire union body assembly is inserted into the centralgrommet on the front of the purge box as shown in figure 3.

3.2. Operation

A Hamilton gas-tight glass syringe with a luer lock and a PTFEplunger tip (Reno, NV) was used during method development.Suspension flow rate was regulated with a KD Scientific780100 syringe pump (Holliston, MA). Aerosolization ceaseswhen charges do not remain in motion because of interruptionscaused by gas bubbles or pump pulsations. Gas bubble

-2000

0

2000

4000

6000

8000

10000

12000

100 1000

electrical mobility diameter (nm)

1 10

Figure 5. The average (10 runs) of the particle size distribution ofthe electrical mobility diameter, measured using a long DMA (TSI3081), methanol blank-subtracted, for 0.75 mg ml−1 SWCNTs at1.0 ml h−1.

formation is suppressed by using a high quality pump anda ‘zero volume’ union. Dried, HEPA-filtered medical gradeCO2 was the carrier gas, usually used at a flow rate of 2–3 lpm (liters per minute). Our system introduced carriergas through four ports symmetrically arranged around thesprayhead. This resulted in less deposition to the walls of thepurge box. A Glassman PS/EL10P04.0 ten kV dc power supply(High Bridge, NJ) was the voltage source. Aerosol particlecharge was radioactively neutralized by passing it through anNRD Nuclecel stainless steel in-line ionizer (Grand Island,NY). Finally, the aerosol entered two DD250 Air TechniquesInternational diffusion driers (Owings Mills, MD) in series toenhance solvent evaporation. A typical experimental set-up isshown in figure 4, and full construction and operation detailscan be found by emailing the author.

3.3. Carbon nanotube specifications

We aerosolized five kinds of nanotubes, each synthesizedseveral years ago. The following information regarding

Figure 4. Electrospray system in a typical experimental set-up.

4

Nanotechnology 21 (2010) 255603 N Jennerjohn et al

a b

c d

Figure 6. (a) TEM image of P3 SWCNTs as a dry powder, (b) TEM image of the dried suspension, (c) SEM image of the dried suspension,and (d) TEM image of aerosol particles.

each type of nanotube material comes either from themanufacturer’s website or the product insert.

(1) P-3 SWCNTs, designated ‘P3’ in this paper, are ‘as-produced’ SWCNTs purified using acid. They aremanufactured by Carbon Solutions (Riverside, CA). Theircarbonaceous purity is 80–90%, metal content 5–10%,and carboxylic acid content 4–6 at%. Of the five typesof nanotubes studied, these are the least hydrophobicas evidenced by their ability to remain in suspension inmethanol or water for months, while the four other typesbegin to settle within hours or days.

(2) NTP L SWCNTs, designated ‘CC’, are manufactured byShenzhen Nanotech Port Co., Ltd (China). Their ‘mainrange of diameter’ is less than 2 nm while lengths rangefrom 5 to 15 µm. Their purity is greater than 50%, ashcontent less than or equal to 2 wt%, and amorphous carboncontent less than 5%.

(3) XNRI MWNT-7, designated ‘MW’, are manufactured byBussan Nanotech Research, Inc. (Japan). Their averagediameter ranges from 40 to 90 nm and their carbon purityis 99.5 wt%.

(4) Two grades of HiPCO SWCNTs, research and low (5),were studied. These are designated ‘EH’ and ‘CH’respectively. The low grade product (CH) was donated.The research grade product (EH) was manufacturedby Unidym (Houston, TX). The median length of the

SWCNTs in the EH sample is 620 nm, iron content5.52 wt%, and bulk density 0.228 g cm−3.

3.4. TEM and SEM imaging

We examined the five kinds of nanotubes, P3, CC, CH, EHand MW, using transmission electron microscopy (TEM) andscanning electron microscopy (SEM). We used three kinds ofsample preparation for the TEM. First, we deposited nanotubeaerosol particles directly onto TEM grids mounted in theHering LPI. Second, we placed a drop of the sonicatednanotube suspension in methanol onto a TEM grid and driedit under HEPA-filtered air. Third, we dipped TEM grids, heldwith tweezers, directly into the dry nanotubes in the containersthey arrived in from their manufacturers. The TEM grids(carbon film on copper) were examined using a JEM1200-EX(JEOL) transmission electron microscope (TEM) operated atan accelerating voltage of 80 kV and equipped with a GatanBioScan-600 W digital camera.

For the SEM study, a sample of each kind of nanotubewas prepared by placing a drop of sonicated suspension ofnanotubes in methanol on a small piece of rigorously pre-cleaned silicon wafer attached to an SEM stab, securing itwith conductive carbon paint and drying it under HEPA-filteredair. Use of a silicon wafer prevents charging effects in theSEM during sample observation and provides a flat surface forthe nanotubes to attach to. SEM images were obtained using

5

Nanotechnology 21 (2010) 255603 N Jennerjohn et al

a b

c d

Figure 7. (a) TEM image of CC SWCNTs as a dry powder, (b) TEM image of the dried suspension, (c) SEM image of the dried suspension,and (d) TEM image of aerosol particles, including a coil super-nanostructure.

a field-emission Dual-Beam Nova Nano 600 (FEI) operatedat 10 kV. Typically, samples must be coated with conductivematerials for optimal SEM imaging but none of our samplesrequired coating, suggesting that they are all conductive.

4. Results and discussion

4.1. Electrospray

In this system, electrospraying does not commence until avoltage of at least 2.2 kV is reached. Kim et al (2005) wereable to use a voltage as low as 0.57 kV because their deviceoperated at 5 kPa instead of at normal atmospheric pressure(101 kPa), and air breakdown voltage decreases as ambientpressure decreases. Meesters et al (1992) used dc voltages ashigh as 30 kV. Some researchers use ac instead of dc voltageto produce successively positive and negative droplets (Borraet al 1999b) while others use ac and dc simultaneously.

If flow rate is high and the suspension is volatile,temperature falls as solvent evaporates and ice crystals mayform from the water in the air near the counterelectrode(Cloupeau 1994). This has been observed to destabilizeour device. The electric current provided by the chargedmoving fluid increases as flow rate increases (Hartman et al2000). High flow rates may be used when high massdeposition is desired, whereas low flow rates are common

in electrospray ionization mass spectrometry (EIS-MS), theonly commercially successful application of electrospraying todate. When flow rate is low, conductivity must be high formaximum mode stability (Cloupeau and Prunet-Foch 1994).TSI advocates increasing conductivity to 0.2 S m−1 (TSI2003), unusually high compared to other conductivity values intable 1 (available at stacks.iop.org/Nano/21/255603/mmedia).However, high conductivity allows modes to be more easilydistinguished (Grace and Marijnissen 1994).

Different researchers use a variety of types of sprayheads.If sprayhead inner diameter is too small, it can become blockedby suspended solids. We used a sprayhead that was 720 µm indiameter to maximize flow rate and minimize blockage. Theopening in the union body in our system eventually had tobe tripled in diameter using a needle in a vise, from 178 to600 µm. Although the inner diameter of even a small capillaryfar exceeds the size of most nanotube bundles, the use of alarger diameter capillary, and enlargement of the opening inthe union body, virtually eliminates clogging.

4.2. Number and mass concentration

Suspension concentrations of single-walled carbon nanotubes(SWCNTs) of 0.75 mg ml−1 at liquid flow rates of 1.0 ml h−1

were routinely aerosolized using this system. We defineefficiency as the percentage of dry residual mass collected on

6

Nanotechnology 21 (2010) 255603 N Jennerjohn et al

a b

c d

Figure 8. (a) TEM image of CH SWCNTs as a dry powder, (b) TEM image of the dried suspension, (c) SEM image of the dried suspension,and (d) TEM image of aerosol particles.

a downstream filter divided by the initial mass of solids in theelectrosprayed suspension. In order to measure it, a filter ina grounded metal filter holder was placed downstream of theradioactive charge neutralizer but prior to the diffusion drier.We were able to achieve 9.4% efficiency with our system usinga suspension of SWCNTs in methanol. Thus far we haveaerosolized a maximum mass concentration of 0.18 mg m−3

SWCNTs in CO2 carrier gas when using a suspension flowrate of 0.7 ml h−1, SWCNT concentration in methanol of0.5 mg ml−1, and a carrier gas flow rate of 2 lpm. Thisconstitutes an efficiency slightly better than 6%. Particlenumber concentrations as high as 104 cm−3 have been reachedwith both SWCNTs (figure 5) and quantum dots.

We have tested the system and observed that the mostefficient condition does not deliver the most mass. Lowerconcentrations and suspension flow rates are more efficient,but deliver less mass. This relatively low efficiency can byexplained by Rayleigh disintegration (also called Rayleighexplosions) and wall deposition (electrostatic attraction) forcharged particles, and thermophoresis and diffusion for neutralparticles (van Erven et al 2005). Also, there may beelectrostatic repulsion from charged particles that have alreadydeposited on the downstream filter. To help increase efficiency,we use neutralization and also tubing, between neutralizer andmonitoring instrumentation, that is conductive. Conductivetubing suppresses the attraction of particles that remaincharged despite neutralization to the tubing walls. Ding et al

(2005) increased their efficiency by increasing carrier gasflow rate and optimizing aerosol charge neutralization. Theirfilter collected 44%, while 23% deposited on the walls ofthe electrospray chamber and 13% became trapped in thesprayhead.

Suspension concentration and flow rate, carrier gas flowrate, degree of charge neutralization and applied voltage mustbe optimized for each test material to achieve the most efficientdownstream delivery. Higher suspension flow rates in thissystem increase the mass delivered to the carrier gas, limitedby sprayhead inner diameter and maximum pumping capacity,as long as material loss to the system can be tolerated. Oursystem might be improved by adopting a method described byDuby et al (2006). They use a single capillary whose end isscored giving it a dozen sharp features that each emit its ownjet. This delivers more mass than a single jet but at the expenseof monodispersity as evidenced by large geometric standarddeviation.

4.3. Morphology

Nanotubes were suspended in methanol and sonicated for onehour prior to aerosolization using the electrospray system.The aerosol, dried by passing it through two diffusion driersin series, entered a Hering (Hering et al 1978) low pressureimpactor (LPI) with a TEM grid mounted under the nozzleon each of its eight stages. An LPI consists of a series of

7

Nanotechnology 21 (2010) 255603 N Jennerjohn et al

a b

c d

Figure 9. (a) TEM image of EH SWCNTs as a dry powder, (b) TEM image of the dried suspension, (c) SEM image of the dried suspension,and (d) TEM image of aerosol particles.

stages, each at a different pressure, on which nozzles of aspecific size have been drilled. This enables the LPI to collectaerosol particles of progressively smaller aerodynamic size. Inan LPI, pressure decreases and the particle’s Cunningham slipcorrection factor increases as the aerosol moves to sequentiallylower stages (Hinds 1999). The likelihood of each particleimpacting a particular plate depends on how large the slipcorrection factor has become at that stage. When a particle canno longer maintain its position in the carrier gas streamline, itimpacts the stage’s plate, and the TEM grid taped to it. TheLPI is a fairly efficient particle trap because, unlike many othertypes of impactors with dozens to hundreds of nozzles perstage, the Hering LPI has only one nozzle per stage. Thus, oneach of its eight stages, a single 3 mm TEM grid positioneddirectly below each nozzle receives a substantial portion ofaerosol particles. In other more typical impactors, there area large number of nozzles on each stage, arranged into an arraythat covers an area much larger than the area of a single TEMgrid.

When dry carbon nanotubes arrive in their manufacturer-supplied containers, all five types are clumped black powders.All of them fly away easily in the air stream of a fumehood, some brands more easily than others, or are stronglyattracted to nearby objects, making their transfer betweencontainers without loss of product difficult. When dispersedin solvents, carbon nanotubes are known from the literature to

form ‘super-nanostructures’: rings, coils (linear with regularlyspaced waves), hairpins, springs, lassos (overlapping rings),catenanes (interlocked chains), pseudorotaxanes (dumbbell-shaped structures) and figure-eights (Nakashima et al 2005,Smith et al 2006a). In the figures that follow, 6(a)–10(a)are TEM images of the dry powders, prior to being addedto solvent. Figures 6(b)–10(b) are TEM images of drops ofsuspension that have been dried on TEM grids. Figures 6(c)–10(c) are SEM images of drops of suspension that have driedon pieces of silicon wafer. Figures 6(d)–10(d) are TEM imagesof nanotubes after being aerosolized, deposited onto TEMgrids that were positioned in the Hering LPI.

The morphology of the dry powders as they arrive fromthe manufacturer is approximately the same as the morphologyof the dried drops of suspension except for some loosening ofbundles and ropes once they nanotubes have been in suspensionand then dried. Aerosolization introduces major morphologicalchanges in all but the MW type.

P3 dry powder (figure 6(a)) consists of nanotube bundlesbarely visible at the boundaries of larger nanotube clumps.Once P3 has been suspended in methanol, sonicated anddried (figure 6(b)), the bundles of nanotube material spreadinto straighter ropes. The SEM image (figure 6(c)) showsdried nanotube material cracking apart, displaying looselybound fibrous material. The slight motions of this materialin the electron beam of the SEM have been recorded on

8

Nanotechnology 21 (2010) 255603 N Jennerjohn et al

a b

c d

Figure 10. (a) TEM image of MWCNTs as a dry powder, (b) TEM image of the dried suspension, (c) SEM image of the dried suspension,and (d) TEM image of aerosol particles.

a video clip available in the supplementary information(available at stacks.iop.org/Nano/21/255603/mmedia). Asaerosol particles (figure 6(d)), P3 nanotubes are complex andheterogeneous, sometimes self-organizing into interlockingfigure-eight superstructures, sometimes with long taperedends.

The morphology of CC SWCNT as a dry powder(figure 7(a)) is similar to P3 except that some nanotube bundlesand ropes are frond-like. After being mixed with methanoland dried, some densely packed material forms smaller clumps(figure 7(b)). This is confirmed by the SEM image (figure 7(c))at approximately the same magnification. As an aerosol, somefrond-like nanotubes remain (figure 7(d)), but they are nowaccompanied by an array of super-nanostructures, includingfigure-eights. Note especially the 500 nm long faint ‘coil’ onthe far right, confirming the coil as a nano-superstructure thatexists as an aerosol particle.

The morphology of CH SWCNT as a dry powder(figure 8(a)) is exceedingly tangled. After being mixedwith methanol and dried, CH disentangles and swells(figure 8(b)). Post-aerosolization, CH forms a disorganizedstructure (figure 8(d)), failing to exhibit any of the well-knownnanotube superstructures. Instead, it consists primarily ofisolated tangled flowing bundles. Also, each aerosol particlehas no near neighbors, since the aerosol is sparse despite allnanotube aerosols having identical deposition conditions.

Dry EH SWCNT powder (figure 9(a)) is more sharplyfeatured than CC, and less tangled than CH. After beingsuspended in methanol, sonicated, and then dried, EH doesnot change much (figure 9(b)). EH forms open structurespost-aerosolization, and as with CH, the aerosol particledeposition is sparse. These images differ from those ofHiPCO obtained by Ku and Kulkarni (2009) in which they areore elongated fibrous materials. Ku collected electrosprayedaerosol particles using electrostatic precipitation (Ku andKulkarni 2009) instead of low pressure impaction which mayaffected particle morphology. Also, Ku’s particles weretrapped on holey TEM grids (Ku and Kulkarni 2009) whichmay have mechanically selected out longer fibrous particlesfrom the aerosol deposition.

This particular brand of MWCNTs appear the samewhether they are a dry powder (figure 10(a)), a driedsuspension (figure 10(b)), or an electrosprayed aerosol(figure 10(d)). They look strikingly different from single-walled carbon nanotubes. They are homogeneous, and clearlytrue fibers as opposed to SWCNTs which are better describedas heterogeneous fibrous materials. These MWCNTs existas singlets frequently, unlike SWCNTs which almost alwaysbundle, though Han et al (2008) observed some MWCNTsbundling. The TEM image is very different from the SEMimage. In the SEM image (figure 10(c)), they have theappearance of transparent straws, similar to chrysotile asbestos

9

Nanotechnology 21 (2010) 255603 N Jennerjohn et al

(McCrone and Delly 1973). Post-electrospraying, they haveless curvature than they exhibit as dried suspensions.

The quantification of the morphology of these nanotubeaerosol particles, especially with respect to properties relatedto potential health effects, is currently the focal point of intensestudy by our research group. These are the aerosols being usedby us to challenge air-monitoring instruments to determinewhich ones may be appropriate for exposure assessment in thenanotube workplace.

5. Summary

A versatile electrospray generator has been constructed,characterized and demonstrated. It is capable of generatingoccupationally relevant aerosol mass concentrations of carbonnanotubes. Carbon nanotubes often undergo strikingmorphological changes as they transition from dry powders toaerosol particles.

Acknowledgments

The authors would like to thank Dr Joelle S Underwood,Department of Chemistry, Loyola University, New Orleans,LA 70118 for sharing the details of her electrospray method.We would also like to thank Dr Vince Castronova of NIOSHfor the kind gift of multi-walled carbon nanotubes. We arealso grateful to the Southern California NIOSH Educationand Research Center (SCERC) Occupational Health andSafety Research Training Pilot Project for funding this workunder Grant Number 5T42OH008412, and to the SouthernCalifornia Environmental Health Sciences Center (SCEHSC)Environmental Health Research Pilot Project for funding thisproject under Grant Number 5 P30 ES07048-12. Finallywe thank Dr Sergey Ryazantsev and Ivo Atanasov, theassociate directors of UCLA’s Electron Imaging Center forNanoMachines, for their many helpful suggestions.

References

Ahmad A, Dopfer O, Harmon C, Nizkorodov S and Underwood J2009 Hygroscopic growth and deliquescence of NaClnanoparticles coated with surfactant AOT J. Phys. Chem. A113 7678–86

Baron P, Deye G, Chen B, Schwegler-Berry D, Shvedova A andCastranova V 2008 Aerosolization of single-walled carbonnanotubes for an inhalation study Inhalat. Toxicol. 20 751–60

Borra J, Camelot D, Chou K, Kooyman P, Marijnissen J andScarlett B 1999a Bipolar coagulation for powder production:micro-mixing inside droplets J. Aerosol Sci. 30 945–58

Borra J, Ehouarn P and Boulaud D 2004 Electrohydrodynamicatomisation of water stabilised by glow discharge—operatingrange and droplet properties J. Aerosol Sci. 35 1313–32

Borra J, Tombette Y and Ehouarn P 1999b Influence of electric fieldprofile and polarity on the mode of EHDA related to electricdischarge regimes J. Aerosol Sci. 30 913–25

Chen D, Pui D and Kaufman S 1995 Electrospraying of conductingliquids for monodisperse aerosol generation in the 4 nm to 1.8micrometer range J. Aerosol Sci. 26 963–77

Cloupeau M 1994 Recipes for use of EHD spraying in cone-jet modeand notes on corona discharge effects J. Aerosol Sci.25 1143–57

Cloupeau M and Prunet-Foch B 1994 Electrohydrodynamic sprayingfunctioning modes: a critical review J. Aerosol Sci. 25 1021–36

Deng W, Klemic J, Li X, Reed M and Gomez A 2006 Increase ofelectrospray throughput using multiplexed microfabricatedsources for the scalable generation of monodisperse dropletsJ. Aerosol Sci. 37 696–714

Ding L, Lee T and Wang C 2005 Fabrication of monodispersedtaxol-loaded particles using electrohydrodynamic atomizationJ. Control. Release 102 395–413

Duby M, Deng W, Kim K, Gomez T and Gomez A 2006Stabilization of monodisperse electrosprays in the multi-jetmode via electric field enhancement J. Aerosol Sci. 37 306–22

Grace J and Marijnissen J 1994 A review of liquid atomization byelectrical means J. Aerosol Sci. 25 1005–19

Han J, Lee E, Lee J, So K, Lee Y, Bae G, Lee S, Ji J, Cho M andYu I 2008 Monitoring multiwalled carbon nanotube exposure incarbon nanotube research facility Inhalat. Toxicol. 20 741–9

Hartman R, Borra J, Brunner D, Marijnissen J and Scarlett B 1999The evolution of electrohydrodynamic sprays produced in thecone-jet mode, a physical model J. Electrostat. 47 143–70

Hartman R, Brunner D, Camelot D, Marijnissen J andScarlett B 2000 Jet break-up in electrohydrodynamicatomization in the cone-jet mode J. Aerosol Sci. 31 65–95

Hering S, Flagan R and Friedlander S 1978 Design and evaluation ofnew low-pressure impactor. I ES&T Res. 12 667–73

Hinds W 1999 Aerosol Technology (New York: Wiley)Jayasinghe S 2006 A novel technique for forming self-assembled

nanotube structures Fullerenes Nanotubes Carbon Nanostruct.14 67–81

Kelly R, Page J, Luo Q, Moore R, Orton D, Tang K andSmith R 2006 Chemically etched open tubular and monolithicemitters for nanoelectrospray ionization mass spectrometryAnal. Chem. 78 7796–801

Kim J, Mulholland G, Kukuck S and Pui D 2005 Slip correctionmeasurements of certified PSL nanoparticles using a nanometerdifferential mobility analyzer (nano-DMA) for Knudsen numberfrom 0.5 to 83 J. Res. Natl Inst. Stand. Technol. 110 31–54

Ku B, Kim S, Kim Y and Lee S 2001 Direct measurement ofelectrospray droplets in submicron diameter using a freezingmethod and a TEM image processing technique J. Aerosol Sci.32 1459–77

Ku B and Kulkarni P 2009 Morphology of single-wall carbonnanotube aggregates generated by electrospray of aqueoussuspensions J. Nanopart. Res. 11 1393–1403

Lenggoro I, Xia B, Okuyama K and de la Mora J 2002 Sizing ofcolloidal nanoparticles by electrospray and differential mobilityanalyzer methods Langmuir 18 4584–91

Leong K, Hopke P, Stukel J and Wang H 1983 Radiolyticcondensation nuclei in aerosol neutralizers J. Aerosol Sci.14 23–7

Maynard A, Baron P, Foley M, Shvedova A, Kisin E andCastranova V 2004 Exposure to carbon nanotube material:aerosol release during the handling of unrefined single-walledcarbon nanotube material J. Toxicol. Environ. Health A67 87–107

McCrone W and Delly J 1973 The Particle Atlas vol III The ElectronMicroscopy Atlas (Ann Arbor: Ann Arbor Science PublishersInc.)

Meesters G, Vercoulen P, Marijnissen M and Scarlet B 1992Generation of micron-sized droplets from the Taylor coneJ. Aerosol Sci. 23 37–49

Mitchell L, Gao J, Vander Wal R, Gigliotti A, Burchiel S andMcDonald J 2007 Pulmonary and systemic immune response toinhaled multiwalled carbon nanotubes Toxicol. Sci. 100 203–14

Nakashima N, Tanaka Y, Tmonari Y, Murakami H, Kataura H,Sakaue T and Yoshikawa K 2005 Helical superstructures offullerene peapods and empty single-walled carbon nanotubesformed in water J. Phys. Chem. B 109 13076–82

10

Nanotechnology 21 (2010) 255603 N Jennerjohn et al

Shvedova A et al 2008 Inhalation vs aspiration of single-walledcarbon nanotubes in C57BL/6 mice: inflammation, fibrosis,oxidative stress, and mutagenesis Am. J. Physiol. Lung CellMol. Physiol. 295 552–65

Smith D, Lee D and Korgel B 2006a High yield multiwall carbonnanotube synthesis in supercritical fluids Chem. Mater.18 3356–64

Smith K, Alexander M and Stark J 2006b Voltage effects on thevolumetric flow rate in cone-jet mode electrospraying J. Appl.Phys. 99 064909

Song D, Lenggoro I, Hayashi Y, Okuyama K and Kim S 2005Changes in the shape and mobility of colloidal gold nanorodswith electrospray and differential mobility analyzer methodsLangmuir 21 10375–82

TSI 2003 Model 3480 Electrospray Aerosol Generator InstructionManual Revision D, October

van Erven J, Moerman R and Marijnissen J 2005 Platinumnanoparticle production by EHDA Aerosol Sci. Technol.39 941–6

11