Embed Size (px)
Citation preview
RESEARCH ARTICLE
An exposure assessment ofdesktop 3D printing
Tracy[22_TD$DIFF] L.Carolinae-mail: z
Burton RNC 2872
John T. JRidge, T
Scott M.Oak Rid
1871-5532
http://dx.do
A preliminary hazard analysis of 3D printing included process monitoring in two working environments; asmall well ventilated materials development laboratory with a Makerbot printer (polylactic acid filament)and a poorly ventilated lab, home-like in terms of room size and ventilation with a Da Vinci XYZ printer(acrylonitrile- [8_TD$DIFF]butadiene-styrene). Particle number, size and mass concentration were measured within theprinter enclosures, breathing zone, and room simultaneously. Number concentrations were elevated abovebackground typically in the [9_TD$DIFF]103
[7_TD$DIFF]– [10_TD$DIFF]105 particles/cm3 range. During printing >99% of the aerosol numberconcentration was within the ultrafine particulate (UFP) and nanoscale size range. Condensed aerosolemissions from the Da Vinci XYZ printer was examined by Fourier infra-red spectroscopy and suggestedisocyanic acid and n-decane as two possible chemical components. Light microscopy and transmissionelectron microscopy with energy dispersive analysis by X-ray identified individual and aggregated particleshighly suggestive of combustion, accompanied by a variety of metallic elements. Adverse health effectsassociated with 3D printing related to chemical vapor off-gassing in well ventilated space appears to be low.At this point the significance of ultrafine particle emission is under growing suspicion in its relationship toinflammatory, pulmonary, and cardiovascular effects. Preliminary recommendations for particulate controldeveloped from this analysis are based on good industrial hygiene practice rather than compelling adversehealth effects.
By Tracy L. Zontek,Burton R. Ogle,John T. Jankovic,Scott M. Hollenbeck
INTRODUCTION
The availability of low cost desktop sizethree dimensional (3D) printers hasincreased due to easy availability andtheir ability to customize object printingto exact specifications based on CADdrawings. This availability for use innon-industrial settings (offices andhomes) may create an environmentwhere poor ventilation and limited
Zontek, PhD is affiliated with thUniversity, Cullowhee, NC 2872
. Ogle, PhD is affiliated with the W3, United States.
ankovic, MSPH is affiliated with thN 37831, United States.
Hollenbeck, MSPH is affiliated wige, TN 37831, United States.
i.org/10.1016/j.jchas.2016.05.008
� 20
health and safety controls exist; thus,there may be an increased risk for ad-verse health effects, particularly if theuser is in close proximity to emissions,the zone of highest exposure. While 3Dprinters are becoming more accessibleand widely used in professional andpersonal applications, studies depictingthe potential health effects and indoorair quality implications are still emerg-ing.
When reviewing regulatory limits onparticle exposure, the United States Oc-cupational Safety & Health Adminis-tration (OSHA) defines nuisance dust,Particulates Not Otherwise Regulated(PNOR), into two categories: total dust(Permissible Exposure Limit (PEL),
e School of Health Sciences, Western3, United States (Tel.: 828 227 2146;
estern Carolina University, Cullowhee,
e Oak Ridge National Laboratory, Oak
th the Oak Ridge National Laboratory,
16 Division of Chemical Health and Safety of the American
15 mg/m3[11_TD$DIFF] as 8-hour TWA) and respira-
ble dust (50% cut point of 4 mm) (PEL5 mg/m3 as 8-hour TWA).1 These clas-sifications are mass based and apply toparticulates that do not have a specificOSHA regulation. The United StatesEnvironmental Protection Agency haspromulgated the National Ambient AirQuality Standards for PM2.5 (35 mg/m3
[1_TD$DIFF]
of particulate matter less than 2.5 mm,measured as a 24 hour average).2
Initial studies of 3D printers classi-fied particulate emissions primarily inthe ultrafine range.3–5 Ultrafine parti-cles (UFD) are defined as those with adiameter less than 0.1 mm. Ultrafineparticles contribute negligibly toPM2.5 mass but contribute significantlyto the particle number concentration.6
Therefore mass based measurementsmay not be appropriate to measureexposure, and subsequent exposure as-sessment to 3D printer emissions.
Ultrafine particulate air pollution isassociated with a variety of adversehealth effects in the scientific litera-ture.7,8 Ultrafine particles cause morerespiratory system inflammation thanlarger particles in rodent studies andUFP surface properties greatly influ-ence toxicity.8 Mass based measure-ments do not consider number count
Chemical Society. Published by Elsevier Inc.
All rights reserved.
15
or surface properties of particulates.Further, epidemiological studies haveassociated increased particulate expo-sure and adverse health effects onthose people with pre-existing respira-tory and cardiovascular diseases.8 Dueto the primary particulate emission of3D printers in the UFP range with theincreasingly associated health effects,this size range is significant to charac-terize in desktop 3D printers.
The printers used in this study useadditive manufacturing where a fila-ment of acrylonitrile-butadiene-styrene(ABS) or polylactic acid (PLA) is fed ata constant rate, extruded at a set tem-perature, and builds the selected objectlayer by layer, on a heated plate. Ther-mal oxidative degradation of filamentmaterial during heating and extrusionare known to release a variety of irri-tants and systemic toxins in lowamounts.9 It is not clear what physicalform these low level combustion deg-radation products may take. One studynoted that emissions consisted of vola-tile droplets with little solid matter.4
The purpose of this study was tocharacterize particle emissions fromtwo widely used 3D printers whoseprice point makes them attractive tobusinesses and consumers.
METHODS
The 3D printers chosen were based onavailability at Oak Ridge National Lab-oratory. Multiple research groups usevarious 3D printers; the Maker Bot andDa Vinci XYZ were chosen accordingto their common use, price point, andavailability for testing purposes. Cur-rent literature depiction of 3D printeremissions in the UFP and nanoscalerange led investigators to use the U.S.Department of Energy (DOE) Ap-proach to Nanomaterial ES&H indus-trial hygiene sampling protocol.10 Theinvestigators of this study developed theinitial DOE industrial hygiene samplingprotocol and have made multiple revi-sions (not yet available to public) basedon new air sampling equipment that isbetter able to resolve UFP and nano-scale materials. The DOE protocol fo-cuses on collecting size and numberconcentration of particulates to charac-terize the aerosol exposure as well asmicroscopy to provide morphology and
16
elemental analysis. NIOSH identifiedtools for exposure assessment that aresimilar to the DOE protocol; howeverair sampling equipment to measureUFP and nanoscale aerosols have im-proved tremendously since these pro-tocols were published.11 NIOSH alsohas a protocol for carbon nanotubesand carbon nanofibers that is massbased; however, it does not cover othernanoscale materials.12 The methodolo-gy used in this study was based on theseprotocols and new aerosol monitoringequipment.
Two different 3D printers were usedin testing. The Makerbot Replicator 2Xprinter was run at 180–230 8C, feedrate of 40 mm/s with a polylactic acid(PLA) filament. This singular run, ap-proximately 60 min, printed multipleobjects. The Makerbot data was in-cluded as the pilot study to refine asampling strategy and protocol.Makerbot data is included for refer-ence, particularly to demonstrate theconcentration maps in a well ventilat-ed space.
The primary ‘‘test’’ printer was a DaVinci XYZ model 1.03D. Initially theprinter was run with different objectsand run times to identify a standardobject that could be printed consistent-ly (keychain was chosen). The testprinter was operated at a printing tem-perature of 213 8C using an acryloni-trile-butadiene-styrene (ABS) filament.These are fixed parameters for this par-ticular printer. The resolution was set atbest quality (Fine 0.1 mm) resulting inthe longest printing times. Filamentfeed rate is a variable function of thepart being printed, as is the amount offilament used. The test printer did notprovide the precise amount of filamentused. Therefore, the same part wasprinted for all of the trial measurements(n = 10), thereby keeping filamentquantity and print time constant.
Instrumentation used to character-ize particle count and size in this studyincluded condensation particle coun-ters (CPC) (measured #/cc at variouslocations for Makerbot and test print-er); a TSI scanning mobility particlesizer (SMPS) 3080 equipped with ei-ther a long differential mobility analyz-er (DMA) (TSI 3081) or nano DMA(TSI 3085) (measured #/cc and mg/m3
from 2 nm to 300 nm at various
Journal of Chem
locations for test printer only); a TSIoptical particle sizer (OPS) (measured#/cc and mg/m3 from 300 nm to 10 mmat various locations for test printeronly); and the TSI Mim2 software tomerge data from the SMPS and OPS.In order to characterize particulateemissions from 3D printers, instru-ment data was used to develop thefollowing: particle size and concentra-tion comparisons during the heatingand printing process and concentra-tion maps. Figure 1 provides an over-view of equipment, data andsubsequent analysis. Since no singlemethod of analysis is competent toevaluation UFP and nanoscale materi-als, microscopy was also used.13
Particle characteristics and mapping
Particle concentration measurements(#/cc) were performed with three TSImodel 3007 condensation particlecounters (CPC) for both the MakerBotand test printer. Time series measure-ments were collected inside and outsidethe test printer enclosure. The monitor-ing position outside the enclosure waslocated at a point representative of thebreathing zone (BZ) of a person sittingat the table where the printer was lo-cated. Background particulate concen-trations were determined prior tofilament heating and extrusion. Roomconcentration maps during printing(based on initial testing that this pro-duced highest number concentration)using 3DField mapping freeware14 anda handheld CPC were developed.
For the test printer runs only, parti-cle number and mass concentration bysize was determined using a combina-tion of scanning mobility particle ana-lyzer, including a TSI SMPS 3080equipped with either a long differentialmobility analyzer (DMA) (TSI 3081)or nano DMA (TSI 3085). The longDMA and nano DMA were alternatedin order to obtain particle sizes from2 nm to 300 nm. A TSI optical particlesizer (OPS) was simultaneously run tomeasure particle size information from300 nm to 10 mm. TSI’s merging soft-ware (Mim2) was used to combine thesmall fraction mobility diameters withthe large fraction optical diameters in-to one normalized particle size distri-bution as either DN/Dlog dp or DM/Dlog dp. These instruments also
ical Health & Safety, March/April 2017
[(Figure_1)TD$FIG]
Figure 1. Summary of equipment, data collection, and analysis results.
provided corresponding number andmass concentrations using TSI algo-rithms provided in their internalMim2 software. These instrumentswere run inside and outside the testprinter enclosure (n = 10) andreported data averaged.
In summary, the CPC concentra-tions were used to develop numberconcentration data to establish expo-sure assessment. The measurementsfrom the SMPS and OPS were usedto develop size selective number andmass concentration data for exposureassessment (e.g. breathing zone, insidevs. outside test printer enclosure).
Microscopy and chemical analysis
Methods to characterize the morphol-ogy and nature of the particles emittedfrom the printing process followed De-partment of Energy and NIOSH guide-lines.10,11,13
[12_TD$DIFF] Large particles werecollected on glass coverslips using asingle stage inertial impactor14 and ex-amined under a light microscope at 40,100, and 400� to view physical char-acteristics such as size, shape, and li-quidity. These particles were alsoexamined by attenuated total reflec-tance Fourier transform infrared spec-troscopy (ATR-FTIR) with a BrukerVertex 70/OPUS spectral library fortentative substance identification;
Journal of Chemical Health & Safety, March
additional analysis was not performedat this time to confirm substance iden-tification. Aerosol was also collectedon transmission electron microscope(TEM) grids (nickel mesh with carbonfilm) using a thermophoretic sampler,inside the printer enclosure, developedby RJ Lee, to obtain high resolutionimages of the smaller particles, anddetermine elemental composition byenergy dispersive analysis of X-rays(EDAX) as part of the particulate char-acterization process.15
[13_TD$DIFF]
A direct reading photoionization de-tector (PID) with a particulate filterwas located inside the printer enclo-sure to detect any gaseous emissionsduring printing with test printer(detection limit nominally 0.01 ppmas isobutylene). A direct reading non-dispersive infra-red analyzer with par-ticulate filter (Foxboro Sapphire) wasalso used to probe the test printer en-closure during printing.
RESULTS
Particle characteristics and mapping
The Makerbot pilot study used theCPCs to measure particle number con-centration throughout the printingprocess (Figure 2). During the timedepicted, several different ‘‘runs’’
/April 2017
occurred due to filament issues. Spikesrepresent initial heating and printing ofdifferent items. In Figure 2a, a timestudy of all monitoring using CPCs isexpressed such that the process CPCwas placed inside the MakerBot enclo-sure during printing; the computerCPC represents the breathing zone,and the cart CPC represents conditionsin lab away from MakerBot printer. InFigure 2a, the number concentrationin the breathing zone and lab condi-tions is difficult to discern due to highconcentration inside the printer enclo-sure. For better resolution, Figure 2bonly represents the breathing zone andlab conditions. During filament heat-ing, particle concentration exceeded[14_TD$DIFF]250,000 #/cc inside the enclosureand 4,000 #/cc outside the enclosure(breathing zone and lab). Higher print-ing temperatures resulted in highernumber particle concentrations. TheTSI CPC 3007 measured everythingfrom 10 nm to >1 mm but there is nosize distribution available with theseinstruments.
Particle number concentration mapswere developed for each printer duringthe print phase. The number of airchanges/hour (AC/hr) was calculatedfor each room to demonstrate ‘‘poor’’and ‘‘good’’ ventilation.16 Whilethe number of air changes/hour is
17
[(Figure_2)TD$FIG]
0
50000
100000
150000
200000
250000
300000
350000
09:09:18
09:16:18
09:23:18
09:30:18
09:37:18
09:44:18
09:51:18
09:58:18
10:05:18
10:12:18
10:19:18
10:26:18
10:33:18
10:40:18
10:47:18
10:54:18
11:01:18
11:08:18
#/cc
Time
Process
Computer
Cart
a
0500
10001500200025003000350040004500
09:09:18
09:16:18
09:23:18
09:30:18
09:37:18
09:44:18
09:51:18
09:58:18
10:05:18
10:12:18
10:19:18
10:26:18
10:33:18
10:40:18
10:47:18
10:54:18
11:01:18
11:08:18
#/cc
Time
Computer
Cart
b
Figure 2. MakerBot printer using PLA. Spikes at 9:40 and 10:24 indicate initialheating to 180 and 230 8C, respectively. (a) contains: process depicts concentra-tion inside the MakerBot enclosure; computer represents breathing zone and cartrepresents general lab area. (b) contains an inset of just the computer (breathingzone) and cart (general area), excluding process concentration with different scale.
[(Figure_3)TD$FIG]
Figure 3. Makerbot printing, good ventilation.
18 Journal of Chem
considered a poor basis for industrialventilation due to its limited impact oncontaminant control, it can be used asan easy method of comparison, espe-cially for non-industrial or indoor airquality.17
Breathing zone number concentra-tion maps from a well-ventilated labo-ratory (10 m � 10 m � 6 m, 20 AC/hr)during a routine Makerbot printer op-eration using polylactic acid filament(PLA), printer top removed, were de-veloped during the background (beforeprinter was turned on) and printingstages (Figure 3). Air concentrationsreached a nominal 3,000 [16_TD$DIFF]#/cc closeto the printer, but fell off rapidly withdistance from the printer. Approxi-mately 3/4 of the room maintainedparticle concentrations at or nearbackground levels.
The test printer was located in apoorly ventilated storage room(3 m � 9 m � 6 m, 1.8 AC/hr), andused acrylonitrile-butadiene-styrenefilament (ABS) with printer fullyclosed (as manufacturer designed),but not air tight. Rapidly, particles builtup in the breathing zone near the print-er to 104
[15_TD$DIFF] [3_TD$DIFF]#/cc [4_TD$DIFF] (Figure 4). With in-creased printing time (middle to endof the 60 min printing cycle), the sur-rounding room concentration reached104 particles/cc as well.
The remaining results apply to theDa Vinci XYZ test printer. Particle sizeand concentration numerical data ispresented in Table 1 depicting use oftest printer (ABS) in a poorly ventilat-ed space.
Table 1 depicts the specific aerosolparameters obtained from the SMPSinside and outside the test printer en-closure, providing an enclosure reduc-tion factor. During heating andprinting, the highest number and massconcentration, the enclosure (keeping3D printer door closed) provided ap-proximately 95% reduction in thenumber and mass concentration, thusreducing exposure to individual usingthe printer and others in the area.
Figure 5 graphically depicts the num-ber concentration over a variety ofprinting conditions (warm-up, initialprinting, continued printing, andcool-down). Each line represents a dif-ferent run of all printing conditions(n = 10). Figure 5 depicts size variation[18_TD$DIFF]
ical Health & Safety, March/April 2017
[(Figure_4)TD$FIG]
Figure 4. Test printer, poor ventilation.
Table 1. Aerosol Parameters Measured Inside and Outside Test Printer Enclosure During Printing.
PrinterFunction
Parameter Inside Printer Enclosure Outside Printer Enclosure in BZ
dp50
(nm)Mode(nm)
GSD #/cc mg/m3 dp50
(nm)Mode(nm)
GSD #/cc(reduction)
mg/m3
(reduction)
Heating 24.2 16.4 20.7 522,000 83.3 14.7 9.3 1.79 1,860 (0.996)a 0.0237 (0.999)a
3.5 2.4 0.34 582,258 70.2 2.9 2.5 0.56 640 0.0410Printing 30.5 34.1 1.95 71,450 7.6 16.3 9.4 1.71 3,780 (0.947)a 0.0013 (0.999)a
4.6 13.2 0.12 45,068 6.2 0.5 2.1 0.01 283 0.0019Cooling 16.3 14.3 1.50 66,308 0.4
1.1 1.0 0.05 38,234 0.2Door open 25.6 19.2 2.17 834 0.2
3.3 4.7 0.21 33 0.1
a Values inside parentheses represents the enclosure reduction factor.n = 10 for both sample sets.[(Figure_5)TD$FIG]
Figure 5. Inside test printer enclosure; variability in size and cumulative frequency distributions from SMPS with Nano DMA (3–100 nm size resoultion).
over an entire run (all printing condi-tions) using the Nano DMA (3–100 nmsize resolution). The x-axis depicts themobility median diameter (dp50) from 1to 100 nm, the y-axis depicts particleconcentration (#/cc) and cumulativepercentage respectively. As depictedin Figure 5, the dp50 diameter of ap-proximately 10 nm has the highest con-centration (107
[17_TD$DIFF] #/cc) and particleswith a dp50 diameter �10 nm accountfor approximately 40% of particle num-ber count. Note all particles in Figure 5
Journal of Chemical Health & Safety, March
are in the respirable region and havethe potential to travel to the alveolarregion of the respiratory system; theyalso can be classified as UFP and/ornanoscale.
Figure 6 used the TSI SMPS 3080and TSI OPS, along with Mim2 soft-ware to characterize aerosol from 1 to10,000 nm in theoretical operatorbreathing zone. Figure 6 further con-firms that majority of particulates, mea-sured both as number count and mass,are in the UFP range. In Figure 6a,
/April 2017
particle mobility diameter is almost en-tirely (99%) within the UFP classifica-tion. Particles collected near the printerin the theoretical breathing zone ofsomeone monitoring the operationare also presented as a normalizednumber distribution (Figure 6b); theaerosol had a bimodal distribution withmodes at approximately 7 and 15 nm.This was typical of BZ values for mostprinting runs in the early stages of ex-truder heating and printing. As time ofprinting progressed, often the particles
19
[(Figure_6)TD$FIG]
Figure 6. Test printer using ABS. Representative cumulative number frequency (a, b) and representative cumulative massfrequency size (c, d) in theoretical breathing zone.[(Figure_7)TD$FIG]
Figure 7. Inside and outside test printer mass and number concentration timeseries (ABS).
grew in size to between 20 and 30 nm(not depicted).
Sixty-eight percent of the particlemass was present as UFP (Figure 6c).The normalized size distribution ofparticle mass presented as bimodal
20
with the largest mode at about 90 nmand a second, smaller mode centeredat about 230 nm (Figure 6d).
Figure 7 (test printer with ABS)demonstrated concentrations at vari-ous stages in the process, as well as
Journal of Chem
between inside and outside the printerenclosure. It also demonstrated smallmass concentrations in contrast tonumber concentrations.
Microscopy and chemical analysis
Light microscopy at 400� was used toexamine the large particulate fractionto confirm the suspected liquid natureof the aerosol (Figure 8).
Transmission electron microscopy(TEM) with energy dispersive analysisby X-ray was used to examine the smal-lest particles of the aerosol as well as toidentify elemental composition (Figures9 and 10). The TEM provided insightinto the morphology and agglomera-tion of the particulates. Table 2 liststhe elements identified by EDAX. InTable 2, the number 1 indicates themost abundant element and 5 the leastabundant element found. EDAXdepicts elements and their relative im-portance; in Figure 10, EDAX identi-fied titanium (previously used in thisarea but not part of PLA filament.)
ABS materials identified by ATRFTIR and its reference spectral librarywere: cyclohexane, n-decane, ethylene-propylene-diene terpolymer, 1-decanol,and isocyanic acid (Table 3). A PID
ical Health & Safety, March/April 2017
[(Figure_8)TD$FIG]
Figure 8. ABS filament impactor parti-cles from test printer.
inside the printer enclosure failed todetect any gaseous emissions duringprinting. The Foxboro Sapphire wasalso used to probe the enclosure during[(Figure_9)TD$FIG]
[(Figure_10)TD$FIG]
Fig
Fig
Journal of Chemical Health & Safety, March
printing and also failed to detect anygaseous emissions.
DISCUSSION
Particle characteristics and mapping
In both the Makerbot and test printerruns, particle concentration increasedin the room (Figures 3 and 4) withrespect to background. During the 3Dprinting process particle concentrationand size are seen to vary (Table 1 andFigures 2, 5–7) with the operation/timein the printing process. Note in partic-ular the differences in the mode anddp50 particle sizes in Table 1. Since themode is less affected by extremes in thesize distribution, it may be a better
ure 10. PLA TEM/EDAX particle examples.
ure 9. ABS TEM/EDAX particle examples.
/April 2017
indicator of the size of the particlesproduced; the geometric mean andstandard deviation being relevant asan overall description of the aerosol.In either case the majority of the par-ticles fall into the ultrafine category, anarea under study for potential healtheffects. Note the bi-modal distributionsfor both number and mass (Figure 6).The test printer initially heats to 220 8Cto clear the nozzle prior to printing at anominal 213 8C. The change in numberconcentration between heating andprinting is believed to be due to a com-bination of the location of the samplecollection point, the rapid movement ofthe printer head (stirring) which is sta-tionary until printing begins, and theintermittent nature of filament feeding
21
Table 2. Elements Found in or as UFP Particulate by EDAX (Number 1 is Most Abundant Element; Number 5 is Least AbundantElement.).
Element C O Na Si S K Cu Mg Ti Ca Al
ABS 1 2 3 3 4 3 4 – – 4 4PLA 1 2 5 4 – – 3 5 5a – 5
a Present as a contaminant from a previous specialty filament used.
Table 3. List of Liquid Aerosol Components Identified by ATR-FTIR.
Chemical Agent Exposure Limit Health Effects
Cyclohexane 1,050 mg/m3 Eye irritationCNS depression (high conc.)
n-Decane None Eye irritationSkin irritationCNS depression (high conc.)
Ethylene-propylene-diene terpolymer None None reportedLow density polyethylene None None reported1-Decanol None Eye irritation
Skin irritationIsocyanic acid 18 mg/m3 (Sweden) Part of the biochemical pathway
linked with cataracts and inflammationthat can trigger cardiovascular diseaseand rheumatoid arthritis.Skin blistering (liquid contact)
as the part is constructed. Among allthe variability, a large enclosure effectseems evident and consistent over theprocess.
Table 1 also demonstrates the effec-tiveness of the enclosure reducing theparticle count in the BZ. Concentra-tions in the BZ consistently reached105 #/cc in close proximity to the testprinter; while averages were somewhatlower (Figure 7 and Table 1). Particlemapping demonstrated concentrationbuild up throughout the poorly venti-lated room while staying localizedaround the printer in the highly venti-lated laboratory (Figures 3 and 4).These demonstrated the importanceof the enclosure around the 3D printerand having adequate general dilutionventilation to reduce particulate con-centrations.
Microscopy and chemical analysis
A variety of materials in the aerosolparticulate phase were identified (Fig-ures 9 and 10; Table 2). The aerosolmass concentrations estimated withthe SMPS and OPS, boiling points ofputative chemicals, and the particulatephotomicrographs collectively suggesta liquid and solid aerosol structure(Figures 8–10). As part of the exposure
22
assessment, possible chemical emis-sions in the gas phase were evaluated.In current literature, none of the che-micals released by thermal oxidativedecomposition were found in measur-able quantities by our techniques atprinter temperatures used in experi-mental conditions.9[19_TD$DIFF]Direct reading sur-vey instruments (PID and IR)monitoring of the test printing consis-tently found no measurable gas/vaporemissions (detection limit �0.1 ppm).
The particles themselves containedindividual and aggregated particleshighly suggestive of combustion (Fig-ures 9 and 10 and Table 2) accompa-nied by a variety of metallic elements.This is consistent with low level ther-mal degradation of carbon-based mate-rials. Therefore, the low gas/vaporconcentrations (demonstrated withPID) and high particulate loading sug-gest the exposure assessment for the 3Dprinting could focus on UFP particu-lates and draw on the health effectsassociated with UFP found primarilyin indoor and outdoor air pollutionstudies related to combustion process-es. Of the many reviewed papers, onlyone stipulated a no observed adverseeffect level (NOAEL) of 10 mg/m3 forhealthy non-smoking individuals; this
Journal of Chem
did not include populations at risk,such as asthmatics.26 The mass concen-tration average in the BZ while printingis lower than the stipulated NOAEL forcombustion type particulate.
Another study, presented by Wright,reported very low concentrations ofstyrene, ethylbenzene, formaldehyde,and acetaldehyde for a small printingprocess (255 8C) in relation to occupa-tional exposure limits.18 Figure 11a andb contains Wright’s reported concen-trations for styrene and acetaldehydeon health effects graphs. In both casespossible exposures were below any ad-verse effect levels presented as Cn � Tcurves. However, several previously un-reported substances were identified inthe liquid phase particulate collected byimpaction19 and identified by FTIRspectroscopy in this study. Of these,cyclohexane and isocyanic acid(Figure 11c and d) were selected forfurther hazard analysis because of theirpresumed toxicological relevance.Since these materials were not collect-ed in such a way as to be quantifiable, itwas assumed the total mass concentra-tion estimate of 2 mg/m3 determined forthe printing process, was composed en-tirely of one substance to the exclusionof all others. These values were then
ical Health & Safety, March/April 2017
[(Figure_11)TD$FIG]
Figure 11. Health effects and reported printer emissions for styrene (a) and acetaldehyde (b), estimated for cyclohexane (c) andisocyanic acid (d).
compared with standards or other tox-icological data. This is conservativesince it is unlikely that any single sub-stance composed the entire aerosolmakeup. In the case of cyclohexanethe potential exposure is orders ofmagnitude (<0.35 mg/m3 vs. OSHA1,050 mg/m3) below the occupationalexposure limit established by the Occu-pational Safety and Health Administra-tion. Although not enumerated here,this was also the case for the reportedconcentrations of styrene, formalde-hyde, acetaldehyde, and ethyl ben-zene.18 Our finding of the possiblepresence of isocyanic acid is consistentwith other reports of low temperaturecombustion byproducts.20,21 Isocyanicacid may be of special concern becauseof the low levels suggested as potential-ly detrimental. Exposures >1 ppb(1.76 mg/m3) isocyanic acid and cya-nate ion (NCO�) have been associatedwith atherosclerosis, cataracts, andrheumatoid arthritis in humans.21
Isocyanic acid is also under consider-ation as a possible respiratory irritantand sensitizer.22 Again if the total
Journal of Chemical Health & Safety, March
mass printing emission estimate weredue to isocyanic acid, the Swedish ex-posure limit of 18 mg/m3 established onthe basis of di-isocyanate sensitizationwould not be exceeded (Figure 11b).23
Another compound found by FTIRanalysis was n-decane. Researchers ex-perienced mild eye irritation accompa-nied by a chemical odor during the testprinter investigation. Anecdotally thesesame two signs accompany most ABSfilament printing operations. This effectis consistent with a finding of eye irri-tation bio-markers in human exposuretrials with n-decane.24
The limitations of this study includeissues related to lack of standard pro-tocol and toxicology information.While the purpose of this study wasto develop an exposure assessment,there are no standard industrial hy-giene protocols (e.g. NIOSH analyticalmethods) to ensure all emissions arecollected in a uniform manner for easycomparison between studies. This isfurther exacerbated by the aerosol airmonitoring equipment availability andcost. In the test printer design in this
/April 2017
study, a TSI SMPS (mobility diameter)and TSI OPS (optical diameter) weremeasured and merged in the Mim2software. The accuracy and effective-ness of the software merging processhas not been experiementally testedoutside of the manufacturer. Further,since particle emissions are in the UFPand nanoscale, mass is decreasinglyimportant, number count and otherproperties (surface area, surfacecharge) become more critical to antic-piation of health effects. There is in-creasing data of potential health effectsof UFP and nanoscale particulates, butfew studies have been completed fo-cusing on 3D printers. Further, the useof 3D printers is increasing, yet thereare no guidelines on sound industrialhygiene controls. Isolation is an effec-tive control, if printer has an enclosureand consumer uses it regularly(Table 1). General dilution ventilationcan also decrease emissions if printeris used in an area with ‘‘good’’ventilatilon (Figures 3 and 4); the ef-fectiveness of ventilation is not likelyknown by a consumer. In this study
23
one run was completed each day; thisstudy does not address potential buildup of aerosol if multiple objects areprinted consecutively without ade-quate ventilation. This study did notdemonstrate any exposure over regu-latory limits (OSHA, EPA) or healtheffects (Figure 11); however, the printtime and frequency was limited andmay not be generalizable to other 3Dprinters and their subsequent use.Based on these limitations and thenumber of replications performed(n = 10 for test printer) in this study,the literature is ripe for additionalinvestigations. Additional printingtrials at different extrusion tempera-tures, feed rates, filament sizes andtypes would aid in providing betterestimates of both mass and particlegeneration rates for future modeling.
When developing future studies for3D printing, chemical exposure assess-ment may also be important to explore.Literature to date shows that chemicalexposures do not appear to be of sig-nificant concern, with the possible ex-ception of isocyanic acid found in thisstudy. New information continues toemerge as the analytical methods fordetection improve. Studies to confirmor reject the presence of isocyanic acidshould be undertaken. If confirmed,the presence of isocyanic acid or n-decane may provide an explanationfor the odor and mild eye irritationexperienced during ABS filamentprinting. One study indicated that anindicator for ABS was styrene and anindicator for PLA was methyl-methac-rylate (MMA).4 Further studies maytest the relevance of using these che-micals as indicators, as well as con-ducting industrial hygiene monitoringto determine concentrations of cyclo-hexane and n-decane.
The cancer risk associated withprinter particulate emission seems rea-sonably low. However, carbon agglom-erates, along with some of the metalsidentified, suggest an aerosol capableof generating reactive oxygen spe-cies.26 This is another area that shouldbe studied further as a mechanism ofproducing inflammation. Risk of ad-verse health effects based on mass con-centration levels appears to be lowfrom 3D printing when compared tothe NOAEL for ultrafine combustion
24
particulate. The Rochester PM Centerprovides an excellent compilation ofSource-Specific Health Effects of Ul-trafine/Fine Particles; however, riskbased on number concentrationsremains difficult to define.6,27 The Na-tional Institute for Occupational Safe-ty and Health set a recommendedexposure limit carbon nanotubes andcarbon nanofibers of 1 mg/m3 as an 8-hour time weighted average (TWA)measured as elemental carbon in therespirable size fraction.12 Again, this isa mass based limit and does not ac-count for number count, an area ripefor future study.
Initial particle size is difficult to knowbeyond the fact that with a nano DMA(2 nm size limit) particles in the 5–7 nmrange were observed, but not with theSMPS (10 nm size limit). Even with thelarger size detection limit devices; therewas evidence of rapid agglomeration(Figures 9 and 10). Printers with non-air tight enclosures reduce particleemissions substantially, but do not en-tirely eliminate UFP emissions. Rapidagglomeration may help explain theprinter enclosure effectiveness in re-ducing number concentrations outsidethe enclosure. Regardless, it is recom-mended that printers without enclo-sures be restricted to use in large,highly ventilated spaces. Before useby asthmatic or otherwise atopic indi-viduals, consideration should be givento incorporating local exhaust ventila-tion to further reduce emissions. Sincethe UFP fraction is essentially airborne(low sedimentation and inertia) capturevelocity is less important than main-taining the printer enclosure underslight negative pressure with respectto the surroundings.
Any addition to the UFP aerosol inthe home or work environment from3D printers, somewhat similar to com-bustion aerosols, should be undertak-en only after serious consideration.Both indoor and outdoor air pollutionstudies of UFP consistently associate avariety of adverse health outcomes toincreases in UFP.27,28
[20_TD$DIFF]
The size detection limit and maxi-mum concentration limit before signif-icant coincidence effects becomeprominent in many available directreading particle counters may lead toan underestimate of the number
Journal of Chem
concentration, and further research isneeded to address these conditions.However, these instruments are super-ior to mass measurement devices forUFP aerosols which typically have de-tection limits above the mass concen-trations of these small size particles.
CONCLUSION
The purpose of this study was to per-form an exposure assessment of con-sumer[21_TD$DIFF] desktop 3D printers. The resultsindicate that 3D printing generateshigh number concentrations of particu-lates in the UFP and nanoscale, further;this is an area with limited standardanalytical techniques, toxicologicalimplications, and regulatory guidance.The measurement of number and massconcentration, use of microscopy andsubsequent data analysis (time seriesgraphs, concentration maps, reductionfactors) are useful in exposure assess-ments. Future study should target thedevelopment of analytical techniques,ventilation recommendations, andestablishing suitable printer locationswith respect to occupied locations[5_TD$DIFF].
ACKNOWLEDGEMENTSSpecial thanks to Shelby Clark for herassistance with data collection. Thismanuscript has been authored byUT-Battelle, LLC under Contract No.DE-AC05-00OR22725 with the U.S.Department of Energy. The UnitedStates Government retains and thepublisher, by accepting the article forpublication, acknowledges that theUnited States Government retains anon-exclusive, paid-up, irrevocable,world-wide license to publish or repro-duce the published form of this manu-script, or allow others to do so, forUnited States Government purposes.The Department of Energy will providepublic access to these results of feder-ally sponsored research in accordancewith the DOE Public Access Plan(http://energy.gov/downloads/doe-public-access-plan).
REFERENCES1. United States Department of Labor,
Occupational Health & Safety Admin-istration. Salt Lake Technical Center,Chemical Sampling Information, n.d.
ical Health & Safety, March/April 2017
Available at: https://www.osha.gov/dts/
chemicalsampling/toc/toc_chemsamp.html.
2. United States Environmental Protec-tion Agency. Particulate Matter. 2016.Available at: https://www3.epa.gov/airquality/particlepollution/.
3. Stephens, B.; Azimi, P.; El Orch, Z.;Ramos, T. Ultrafine particle emissionsfrom desktop 3D printers. Atmos En-viron, 2013, 79, 334–339.
4. Steinle, P. Characterization of emis-sions from a desktop 3D printer andindoor air measurements in office set-tings. J Occup Environ Hyg, 2016,13(2), 121–132, http://dx.doi.org/10.1080/15459624.2015.1090957.
5. Zhou, Y.; Kong, X.; Chen, A.; Cao, S.Investigation of ultrafine particle emis-sions of desktop 3D printers in theclean room. Procedia Eng, 2015, 121,506–512.
6. Koehler, K. A.; Peters, T. New methodsfor personal exposure monitoring forairborne particles. Curr Environ HealthRep, 2015, 2(4), 399–411, http://dx.doi.org/10.1007/s40572-015-0070-z.
7. Manufacturers of Emission ControlsAssociation (MECA). Health Effectsof Particulate Air Pollution; MECA:2200 Wilson Boulevard, Suite 310,Arlington, VA, 2014, October 14.
8. Oberdorster, G. Pulmonary effects ofinhaled ultrafine particles. Int ArchOccup Environ Health, 2001, 74(1),1–8.
9. Rutkowski, J. V.; Levint, B. C. Acrylo-nitrile-butadiene-styrene copolymers(ABS): pyrolysis and combustion pro-ducts and their toxicity – a review of theliterature. Fire Mater, 1986, 10, 93–105.
10. U.S. Department of Energy. Approachto Nanomaterial ES&H. 2008. Avail-able at: http://science.energy.gov/�/media/bes/pdf/doe_nsrc_approach_to_nanomaterial_esh.pdf [accessed20.09.15].
11. United States Department of Healthand Human Services. Approaches to
Journal of Chemical Health & Safety, March
Safe Nanotechnology: Managing theHealth and Safety Concerns Asso-ciated with Engineered Nanomaterials.DHHS (NIOSH) Publication No.2009-125. 2009. Available at: http://www.cdc.gov/niosh/docs/2009-125/pdfs/2009-125.pdf.
12. United States Department of Healthand Human Services. Current Intelli-gence Bulletin 65 Occupational Expo-sure to Carbon Nanotubes andNanofibers, DHHS (NIOSH) Publica-tion No. 2013-145. 2013. Available at:http://www.cdc.gov/niosh/docs/2013-145/pdfs/2013-145.pdf.
13. Levy, B. S. Occupational and Environ-mental Health: Recognizing and Pre-venting Disease and Injury, 6th ed.Oxford University Press: USA, 2011.
14. Galouchko, V. 3DField. Available at:http://3dfmaps.com/ [accessed 20.09.15].
15. Leith, D.; Miller-Lionberga, D.; Casuc-cio, G.; Lersch, T.; Lentz, H.; March-esec, A.; Volckensac, J. Developmentof a transfer function for a personal,thermophoretic nanoparticle sampler.Aerosol Sci Technol, 2014, 48(1),81–89.
16. Jankovic, J. T.; Ihle, R.; Vick, D. O.Occupant generated carbon dioxide asa measure of dilution ventilation effi-ciency. Am Ind Hyg Assoc J, 1996,57(8), 756–759.
17. American Conference of Governmen-tal Industrial Hygienists (ACGIH). In-dustrial Ventilation, A Manual ofRecommended Practice, 21st ed. AC-GIH: Cincinnati, OH, 1992.
18. Jones, W.; Jankovic, J.; Baron, P. De-sign, construction and evaluation of amulti-stage cassette impactor. Am IndHyg Assoc J, 1983, 44(6), 409–418.
19. Wright, S. Clearing the air about 3Dprinting emissions: Volatile organiccompound and nanoparticle emissionduring fused filament fabrication,2014. Presentation at 3D Printshow,London, Sept. 4–6, 2014.
/April 2017
20. Hansson, K. M.; Samuelsson, J.; Tullin,C.; Amand, L. E. Formation of HNCO,HCN, and NH3 from the pyrolysis ofbark and nitrogen-containing modelcompounds. Combust Flame, 2004,137(3), 265–327.
21. Roberts, J. M.; Veres, P. R.; Cochran, A.K.; et al. Isocyanic acid in the atmo-sphere and its possible link to smoke-related health effects. Proc Natl AcadSci USA, 2011, 108(22), 8966–8971.
22. Health and Safety Executive – WorkingGroup on Action to Control Chemicals(WATCH). Assessment of the potentialfor isocyanic acid and other monoiso-cyanates to cause respiratory irritationand sensitization. 2008. Available at:http://www.hse.gov.uk/aboutus/meetings/iacs/acts/watch/170608/p4ann2.pdf [accessed 20.09.15].
23. National Board of Occupational Safetyand Health (NBOSH). Occupationalexposure limits 2000:3; NBOSH:Stockholm, Sweden, 2000.
24. Kjaergaard, S.; Molhave, L.; Pedersen,O. F. Human reactions to indoor airpollution: n-decane. Environ Int, 1989,15(1–6), 473–482.
26. Nicole, A. H.; Janssen, P. H.; Fischer,G.; et al. Contrasts in oxidative poten-tial and other particulate matter char-acteristics collected near major streetsand background locations. EnvironHealth Perspect, 2012, 120(2), 185–191.
27. Oberdorster, G.; Beckett, W.; Cass, G.;et al. Final Report: Ultrafine Particles:Characterization, Health Effects andPathophysiological Mechanisms. 2006.EPA grant, Available at: http://cfpub.epa.gov/ncer_abstracts/index.cfm/fuseaction/display.abstractDetail/abstract/1098/report/F.
28. Ibald-Mulli, A.; Wichmann, H. E.;Kreyling, W.; Peters, A. Epidemiologi-cal evidence on health effects of ultra-fine particles. J Aerosol Med, 2002,15(2), 189–201.
25
