Characterization of 3D printing filaments containing metal additives and their particulate emissions

Abstract

Polylactic acid (PLA) filaments are widely used in fused filament fabrication (FFF) processes (3D printing). Filament additives such as metallic particles incorporated into PLA to modify functional and aesthetic features of print objects are becoming increasingly popular. However, the identities and concentrations of low percentage and trace metals in these filaments have not been well described in either the literature or product safety information included with the product. We report the structures and concentrations of metals in selected Copperfill, Bronzefill and Steelfill filaments. We also report size-weighted number concentrations and size-weighted mass concentrations of particulate emissions as a function of print temperature for each filament. Particulate emissions were heterogenous in shape and size with airborne particles below 50 nm diameter dominating the size-weighted particle concentrations and larger particles (approximately 300 nm) dominating the mass weighted particle concentration. Results indicate that potential exposure to particles in the nano-size range increase when using print temperatures above 200° C. Because inhalation exposure to nanoparticles has been linked to adverse health outcomes, we suggest that using lower print temperatures for specific metal-fill filaments may reduce their operational hazard.

Graphical abstract

1. Introduction

Fused filament fabrication (FFF) processes are becoming more widely practiced in households, schools, and offices, partly due to the availability of desktop 3D printers at a relatively lower cost and wide accessibility of various types of feedstocks in the market (Lopes et al., 2020). The filament market has also expanded to meet practical and aesthetic requirements for print objects. Commercial feedstocks may contain different types of additives such as metals, ceramics, and carbon nanomaterials (Park and Fu, 2021; Tumer and Erbil, 2021).

Physical and chemical properties of additives will vary depending on the intended function of printed objects. Metal-fill additives typically contain metal particles (up to 80 % by weight) of which the diameter can range from <1 μm to over 50 μm (Laureto et al., 2017). Metal particles in poly lactic acid (PLA)-Copperfill and PLA-Bronzefill filaments have been reported as spherical, while PLA-stainless steel and PLA-magnetic iron filaments have been reported to contain flake like particles (Laureto et al., 2017).

The characteristics of fillers influence the amount and type of emissions generated during 3D printing processes. Gas-phase emissions generated during printing process are higher in filaments with metal fillers compared to those observed in filler-free filaments (Potter et al., 2021). For example, PLA filament with stainless steel fillers was reported to emit three-fold increase in volatile organic compounds compared to filaments with Copperfill and Bronzefill (Potter et al., 2021). Feedstocks with metal additives, Acrylonitrile Butadiene Styrene-Tungsten (ABS-W) and PLA-Cu, had higher mean particulate emission rates when printed at the same temperature as their respective neat thermoplastics (Alberts et al., 2021). Further, particles emitted from ABS-W and PLA-Cu were reported to have lower mean diameter compared to emissions from corresponding non-metal-fill polymers (Alberts et al., 2021). In addition to the increased release of particles, emissions from copper-containing filaments increased stress, cell death and metabolic changes in cell-based models (Singh et al., 2021).

As a result of increasing consumer demand for varied print object aesthetics, commercially available filaments that contain additive such as metals are becoming more common. In addition, the effect that these additives have on particulate emissions generated during 3D printing has not been widely reported (Poikkimaki et al., 2019; Tedla et al., 2021). While particulate and gas-phase emissions generated during 3D printing of selected filaments containing high proportion metals have been reported (Alberts et al., 2021; Laureto et al., 2017; Potter et al., 2021), less is known about additive metal types, forms, and concentrations, as well as the effect that they have on emissions during the 3D printing process. In addition, the ability to vary print temperature in many available printers has been shown to change particulate emission characteristics (Alberts et al., 2021; Laureto et al., 2017; Potter et al., 2021; Jeon et al., 2019; Byrley et al., 2019).

In this study, we explore qualitative and quantitative analyses of metals in original and printed PLA-Copperfill, PLA-Bronzefill and PLA-Steelfill filaments and characterize particulate emissions generated during 3D printing of these filaments as a function of additives and print temperature with a focus on particulate size, morphology, number, zeta potential and mass.

2. Materials and methods

2.1. Feedstocks and printer setup

Colorfabb’s Bronzefill containing approximately 80 w% bronze powder, Copperfill with 80 w% copper powders, and Steelfill with 80 w% stainless steel powders all blended in thermoplastic PLA, were purchased from MatterHackers, Lake Forest, CA. The printer used for these experiments was LulzBot TAZ6 3D printer (LulzBoat, Fargo, ND).

The printer was placed inside a 2000-L Hazelton chamber (2 m³). The print-nozzle was stationary and sealed at the mouth of a 1000 ml Erlenmeyer flask into which the printed filament was extruded. An outlet at one side of the neck was attached to a 300 nm HEPA filter to draw air from the chamber and particulate emissions were collected through a second outlet on the opposite side of the neck of the Erlenmeyer flask. In separate experiments, particulate emissions were drawn from the flask into an SMPS-CPC for particle analysis, a bio-sampler for hydrodynamic characterization and SEM analysis or a thermophilic sampler for SEM analysis (Fig. S1). Although emissions from the print head were released into the Erlenmeyer flask, the printer was placed in a larger chamber due to the particles possibly released from the biosampler exhaust. The chamber air exchange rate was set to one air exchange per hour throughout the measurement. Chamber type and set up are explained in greater detail in Byrley et al. (Byrley et al., 2021; Byrley et al., 2020).

2.2. Filament characterization

2.2.1. X-ray Fluorescence Spectroscopy (XRF)

XRF data were collected using a Vanta X-ray Fluorescence Analyzer DMTA-10072–01EN. Original and printed filaments were measured using the Vanta workstation. The instrument provides the total relative metal composition for measured elements. The restriction of hazard substances (RoHS) method automatically executed a test to determine whether the sample is in alloy, polymer or mixed, based on a criteria recommended by the manufacturer (Olympus, 2021a). The thickness of the original filament was 2 mm, whereas the thickness of the printed filament was 0.5 mm. However, as suggested by the manufacturer, the measuring window was completely covered for both measurements. The print filament measurements were made at intermediate segments of the print cycle. A NIST metal in polymer standard (SRM 2659) were used to confirm the accuracy of the metal measurements within the instrument specifications.

2.2.2. Scanning Electron Microscopy with Energy Dispersive Spectroscopy (SEM-EDS)

Filaments, original and printed, were cut in section using ceramic knife and sputter coated with gold at a thickness of 20 nm. Analyses were done using a Zeiss EVO, MA, scanning electron microscope (Carl Zeiss SMT Ltd., Cambridge, UK) and imaged using the secondary electron detector (SED2) mode. For EDS analysis, the accelerating voltage was varied based on the excitation energy of each element, up to three times the corresponding Kα spectral line value. For instance, for copper (Kα, 8.040), the accelerating voltage was increased up to 24KV. This accelerating voltage was used for most of the elements as it provides the necessary volage and current to efficiently excite the K and L x-ray lines for most of the elements that were detected using XRF screening (Small, 2002). Elements were mapped using Bruker XFlash energy dispersive detector monitored with ESPRIT 2 analytical software suite.

2.2.3. Inductively Coupled Plasma Mass Spectroscopy (ICP-MS)

To determine metals in the original filaments, pieces were cut using ceramic knife and the original and printed filaments were prepared using a microwave-assisted acid digestion method similar to EPA Method 3051A. The filaments were disolved in 20 ml concentrated (70 %) nitric acid. Each sample was diluted 1:1 with concentrated nitric acid and digested using MARS 6 microwave digestion system at 180°, 800 psi and 900–1800 watt. Sample analyses were conducted using Perkin Elmer NexIon 300D ICP-MS.

2.3. Emission characterization

2.3.1. Size and morphology, SEM

Emissions were captured in 20 ml nanopure water using a biosampler at air flow rate of 12.5 L/min, particles were imaged after the samples were pipetted onto a 100 nm pore size polycarbonate filters and dried overnight. Finally, emissions were collected on gold and nickel Transmission Electron Microscopy (TEM) mesh grids using a thermophoretic sampler. The thermophoretic sampler (TPS 100, RJ Lee Group, monroeville, PA) was set to 115 °C hot temperature, 35 °C cold temperature, 34 % relative humidity and ambient temperature of 24 °C. Emissions were collected every minute during 1 h printing period and for 1 h after the printer was turned off. Filters and TEM grids were sputter coated with gold and inserted into SEM for imaging. Particle size frequency distributions of original filaments were determined using ImageJ software wih 90 measurements for each filament type.

2.3.2. Hydrodynamic size and concentration

Nanoparticle Tracking Analyses (NTA) measurements of emissions collected in the biosampler were performed with a NANoSight NS500 instrument equiped with a green laser (Rogers et al., 2018). Instrument performance was checked using 100 nm polystyrene spheres before measuring the hydrodynamic size and particles concentrations of the samples. Size and concentration of undiluted solutions of particles captured in biosampler and background air (i.e chamber air passing through the biosampler for equal time period prior to printing) were measured using NTA software version 3.4. Instrument temperature was set to room temperature (20–21 °C) and the viscosity of water was set to 0.973–0.976 cP for all the measurements. Capture and analyses settings were the same for background and sample measurements.

2.3.3. Hydrodynamic size and zeta potential, dynamic Light Scattering (DLS)

The hydrodynamic diameter and zeta potential were determined, in part, to better understand the particle behavior at the biological interfaces for in vitro cellular modeling toxicology. DLS and NTA instrumental methods are considered to be quite robust for particulate measurements between 10 nm to 1000 nm. DLS is most effective for measurements of monodisperse particle suspensions whereas NTA is more effective for polydisperse particles. The hydrodynamic diameter and zeta potential of particulate emissions captured in the biosampler were measured using Malvern Instruments (Westborough, MA, USA) zeta sizer nano in 173° backscatter mode (MacCuspie et al., 2011; Rogers et al., 2018). Instrument performance was regularly checked using polystyrene spheres whose size and zeta potential values are known. The hydrodynamic size and zeta potential of bio-sampler particle suspensions were measured in nanopure water at pH 6.5 and data were exported for further processing in Excel.

2.3.4. Airborne size and concentration

Particulate data were collected using a Scanning Mobility Particle Sizer (SMPS) containing classifier model 3080 and differential mobility analyzer (DMA) model 3081 integrated with a condensation particle counter (CPC) model 3785 (TSI, Shoreview, MN). The TSI system in the sizing mode can measure particles in the 10–1000 nm range. A static sampling tube was connected from the Erlenmeyer flask directly to 0.071 cm nozzle size impactor connected to the SMPS-CPC system. The printer was run for a total of one hour extruding Steelfill and Bronzefill each at 200, 210, 220 or 225 °C with extrusion rates of 50 mm/min. For the Copperfill filament, the printing process was only conducted at 200 and 210 °C with a 50 mm/min extrusion rate as the filament melts into liquid when heated beyond 210 °C. The temperature settings used in these experiments are within the range of recommended values for the filament manufacturers and the printer manufacturer. Data from the aerosol instrument manager software was exported as comma separated values and further analyzed using Python and Excel. The normalized mass concentration was calculated assuming spherical particles. However, since the density values of those emissions were uncertain, we chose to normalize and present it as it is indicated in Fig. 5.

Fig. 5.

SMPS-CPC: Size weighted particulate mass concentration of emissions generated during 3D printing.

3. Results and discussion

3.1. Filament characterization

3.1.1. Elemental analyses, X-ray fluorescence

The main purpose of XRF analyses was to conduct a screening which serves a basis for further quantitative and qualitative detection of metals using ICP-MS. Approximate limit of detection for the Vanta handheld XRF analyzer is 5 ppm (0.0005 %) for most metallic elements based on a 2 min testing time and detection confidence of 3SD (99.7 % confidence)(Olympus, 2021b). Approximate instrument detection limits for Ti and V were 25 ppm (0.0025 %) and for Cr was 10 ppm (Olympus, 2021b), however, the percentage values are based on the total metals detected. Elements Cu, Sn, Zn, Ni and Ti were detected in the Bronzefill original filament (Table S1). Elements including Fe, Cr, Cu, Mn, Co, Ni, V, Sn, Mo were detected in Steelfill original filament (Table S1) and Cu, Zn, Sn and Ni in Copperfill original filament (Table S1).

Comparison of the printed and original filaments was complicated due to combined effects of detection limits and sampling. Some of the elements were not detected in printed filaments. For instance, in the Bronzefill printed filament, Zn is not detected, and Cu showed lower concentration (Table S2) than the original filament. Similarly, Cr and Cu were at a lower concentration in the printed material compared to the original in Steelfill filament (Table S1). It is likely that, some of the elements detected on the surface of the original filament were more or less embedded into the interior of the printed material (Vakharia et al., 2021).

3.1.2. Element analyses, SEM-EDS

The high throughput Silicon drift detector for EDS enables trace metal measurement with a concentration limit of 100 to 500 ppm (0.01 to 0.05 mass fraction) (Newbury et al., 2017). These values depend on element species, matrix composition and spectrum peak interferences. Nevertheless, one of the advantages of SEM-EDS over XRF is that it provides information on the localization of detected elements with respect to particle fillers using a mapping option. The most prominent elements were localized on the particles as shown on Fig. 1. SEM-EDS data confirm the presence of metals with the greatest relative contributions associated with the spherical structures in Bronzefill, Steelfill and Copperfill filaments as observed using XRF (Fig. 1, Table S2).

Fig. 1.

SEM-EDS spectra and elemental maps of original filaments: A) Bronzefill, B) Steelfill and C) Copperfill (scale bar = 100 μm applies to all micrographs, see also Fig.S2).

Cu and Sn were associated with the spherical structures in the Bronzefill filament by EDS (Fig. 1). Because the images of filament cross-sections cut using ceramic knife, the metal objects appear hemispherical. Fe and Cr in the spheres were discernible in Steelfill filament by EDS (Fig. 1). Similarly, Cu spheres were visible in the SEM-EDS map of the Copperfill filament (Fig. 1).

In addition to the differences in metal composition of microspheres, the size distribution of microparticles for each of the three filaments showed significant heterogeneity both within each filament and among filaments (Fig. 1, Figs. S2-S5). These size differences can also be seen in representative scanning electron micrographs of the surfaces and cross-section surfaces of the original and printed Bronzefill (Fig. S2), Steelfill (Fig. S3) and Copperfill (Fig. S4) filaments. The metal particles observed by SEM range in size from 1 to 50 μm (Fig. S5). It was also observed for the Bronzefill filament that the metal particles seem to be loosely contained in the polymetric matrix and so as to be easily dislodged from the filament (Fig. S4A). In addition to Fig. S2A, we have observed several incidences where a cut or abraded filament placed onto carbon tape showed numerous free metal microspheres in the immediate vicinity of the damaged surface (images not shown). Independent of the relative size, the metal particles in each of the metal-fill filaments were similar in shape (Figs. S2-S4). However, the particles in the Steelfill filament tended to be smaller than the Bronzefill which tended to be smaller than for the particles in the Copperfill filaments (Figs. S2-S5).

3.1.3. Element analyses, ICP-MS

Each of the original and printed metal-fill filaments was analyzed by ICP-MS. For the Copperfill filament, a high concentration of Cu (83 %), and lower concentrations of Ni (28.54, 23.91), Ag (5.97, 4.34), Sn (311.50, 281.81) and Pb (19.29, 18.90) ppm, were detected in original and printed filaments, respectively (Table 1). Although not identical, these observations are consistent with XRF data where Ni, and Sn were detected in trace amounts in original as well as in printed Copperfill filaments. Similarly, for original and printed Bronzefill filaments respectively, the following elements were detected: Cu (74.58, 70.25 %), Sn (51,150.15, 31,407.93 ppm), Ni (363.46, 343.38 ppm), Zn (36.16, 38.36 ppm), Co (0.52, 0.25 ppm), Ag (6.80, 5.88 ppm), and Pb (32.07, 32.02 ppm) (Table 1). The herein reported metal content values for the original Bronzefill filament were similar to those reported by Zhang et al. (2023) who also used a metal-fill filament containing 80 % bronze particles. Values reported by these authors included Cu (7.08 × 10⁵ ppm), Sn (8.98 × 10⁴ ppm), Ni (261 ppm), Zn (23.6 ppm), Pb (24.1 ppm) and Co (1.97 ppm) (for comparison see Table 1). For the original and printed Steelfill filaments, respectively, elemental compositions were Fe (327.83, 344.82), Cr (46,652.64, 46,674.16), Ni (2298.18, 2250.63), Co (129.62, 122.23), V (91.03, 90.08), Mn (172.30, 180.23) and Sn (11.31, 84.72) ppm (Table 1). Some of the elements that were below detection limit by XRF (Table S1), or SEM-EDS (Fig. 2, Table S2) were detected at lower concentrations using ICP-MS (Table 1). On average, for Copperfill and Bronzefill filaments the metal content for the original filaments is slightly higher than for the extruded filaments. Because of the relatively large size of the metallic inclusions (> 1um), and the observation of free metal particles separating from the filaments (Fig. S2A), we suggest that a small quantity of the particles may separate from the printed filament during the extrusion process and fall to the bottom of the chamber. By contrast, for the Steelfill filament we observed the Sn, Cu, and Zn were higher in the printed filament than in the original. Because brass (composed primarily of Cu and Zn) is the most common material found in 3D printer nozzles, we suggest that the steel particles in the filament may be “scrapping” some of the brass from the print nozzle.

Table 1.

Elements detected in original and printed filaments using ICP-MS.

Element	Originala Copperfill	Printeda Copperfill	Originala Bronzefill	Printeda Bronzefill	Originala Steelfill	Printeda Steelfill
Ni 58	28.54 ± 0.22	23.91 ± 1.32	363.46 ± 4.97	343.38 ± 4.40	2298.18 ± 51.44	2250.63 ± 42.65
Co 59	ND	ND	0.52 ± 0.10	0.25 ± 0.01	129.62 ± 2.30	122.23 ± 2.22
V 51	ND	ND	ND	ND	91.03 ± 2.52	90.08 ± 1.31
Mn 55	ND	ND	ND	ND	172.309 ± 8.80	180.233 ± 4.22
Ag 107	5.97 ± 0.17	4.34 ± 0.20	6.80 ± 0.29	5.88 ± 0.09	ND	ND
Sn 118	311.50 ± 4.97	281.81 ± 7.52	51,150.15 ± 10,206.87	31,407.93 ± 1677.96	11.31 ± 0.77	84.72 ± 1.31
Pb 208	19.29 ± 0.28	18.90 ± 0.44	32.07 ± 0.18	32.02 ± 0.33	ND	ND
Ti 47	12.70 ± 1.18	14.09 ± 0.90	ND	ND	ND	ND
~Cu63	NM	(83.60 ± 1.11) %	(74.58 ± 0.77) %	(70.25 ± 1.16) %	438.03 ± 322.13	1007.13 ± 9.12
Cr52	ND	ND	ND	ND	46,652.64 ± 125	46,674.16 ± 29
Zn66	ND	ND	36.16 ± 1.72	38.36 ± 11.50	33.43 ± 1.88	61.55 ± 30.25
Fe156	ND	ND	ND	ND	327.83 ± 15.71	344.82 ± 7.03

ND- not detected; NM- not measured.

^a.ppm or % where specified, n = 3, ± SD.

Fig. 2.

SEM images of emissions generated during 3D printing of Bronzefill, Copperfill and Steelfill filaments (top to bottom). Emissions were collected in nano pure water inside a biosampler. Samples were pipetted on 25 nm pore size filters, covered inside a Petri dish, and dried overnight inside a fume hood. Particles were imaged after being sputter coated with gold. Note, magnification differs across images.

3.2. Particulate emissions characterization

3.2.1. Particles hydrodynamic size and concentration, NTA

Particulate emissions captured in the biosampler from the Bronzefill, Copperfill and Steelfill filaments yielded concentrations of 1.4 × 10⁹, 1.15 × 10⁹, 7.34 × 10⁸ particles/ml respectively (Table 2). Mean hydrodynamic diameters measured by NTA from emisions were 123.8, 188.5, 170.0 nm, respectively, for Bronzefill, Copperfill and Steelfill filaments at the same printing conditions (Table 2). Size distributions of emissions generated during Copperfill and Steelfill printing processes were more polydisperse than observed in Bronzefill emissions (Table 2). Capture efficiencies typically reported for liquid based samplers such as the biosampler type that we used depend on sample flow rate, medium composition and volume (Hogan et al., 2005; Wang et al., 2015). We used a flow rate of 12.5 L/min within the optimal range reported by Hogan et al., 2005 and a collection duration of 45 min. In general the efficiency of liquid based samplers is higher for capturing larger particles and decreases for smaller particles since they lack the inertia to be collected in the media. For example, efficiencies of gas impinger-30 (AGI-30), SKC Biosampler and fritt bubbler were reported to be below 10 % for 30–100 nm particles and increasing by 40 % for particles above 200 nm (Hogan et al., 2005). Keeping these sampling limitations in mind, it is likely that the number of particles generated during printing was higher than reported for the biosampler.

Table 2.

NTA and DLS: Particle concentrations and mean size of emissions generated during 3D printing process at 210 °C. Emissions were collected in nano pure water using a biosampler set at 12.5 lpm air flow rate.

NTA
Filament	Concentration (particles/ml)	Mean size (nm)	Mode (nm)
Bronzefill	1.40e+09 ± 7.41e+07	123.8 ± 1.5	102.2 ± 3.8
Copperfill	1.15e+09 ± 7.73e+07	188.5 ± 6.3	72.7 ± 14.8
Steelfill	7.34e+08 ± 3.32e+07	170.0 ± 2.8	143.9 ± 5.6
DLS
Filament	Z-Average (d. nm) n = 6	PDI n = 6	Zeta potential (mv) n = 3, 25 °C
Bronzefill	231.5 ± 30.4	0.5 ± 0.1	−12.7 ± 1.3
Copperfill	219.0 ± 6	0.2 ± 0.0	−20.8 ± 8.4
Steelfill	230.8 ± 24.6	0.4 ± 0.1	−22.2 ± 1.8

± SD, Polydispersity Index (PDI).

3.2.2. Particles hydrodynamic size and zeta potential, DLS

Average hydrodynamic dimeter values for particulate emissions generated during printing Bronzefill, Copperfill and Steelfill filaments measured by DLS were 231.5, 219.0, 230.8 nm, respectively (Table 2). Particulate emissions generated and captured during the printing processes for Copperfill, Steelfill and Bronzefill were moderately polydisperse (Table 2). Overall, DLS measurements indicated larger average particle sizes compared to NTA measurements. It has been reported that for polydisperse populations, DLS measurements are likely to be unduly influenced by fewer larger-size particles. In this regard, size and concentration measurements done using NTA are reported to be more reliable (Bhattacharjee, 2016) than DLS.

Zeta potential values of the particulate emissions generated during the printing processes for Bronzefill, Copperfill and Steelfill were −12.7, −20.7 and −22.2, respectively (Table 2). The lower value observed in zeta potential for the Bronzefill emissions, may be due to the metal particle content or polymer additives. All the emissions have less than an absolute value of 30 mv immediately after being collected in nano pure water, suggesting that the particles are likely agglomerating during sampling processes or during storage due to the absence of a stabilizing agent (Malvern, 2015).

3.2.3. Particulate emissions size and morphology, SEM

Particulate emissions generated from printing each of the filaments were collected in the biosampler, captured on polycarbonate filters and analyzed by SEM. Representative particles were hetrogeneous in size (~50 nm to several mircometers, Fig. 2). We also observed this size range when particles were collected using a thermophoretic sampler (Fig. S6, S7). As observed by SEM, the micrometer size particles seem to be an aggregation of particles smaller than 100 nm (Fig. 2, and Fig. S6, S7). While the smaller particles seemed to have spherical shape, larger particles are irregularly shaped and the morphology of larger particles seems rough, consistent with a nonuniform assembly of smaller particles. Although, the SEM resolution is less clear for imaging smaller nanoscale structures, the smaller particles apear smooth at the scale indicated (Fig. 2, S6, S7). Some of the particles that were also identified on the TEM grid from the thermophoretic sampler appeared to be aggregated. This might be expected given that the particles form multiple layers as they accumulate on the surface in the deposition process. Although a direct comparison between drop-cast and thermophoretic methods (other than observational) may not be possible, some aggregation presumably due to evaporation artifacts was also observed for the drop cast method. Although we did not observe metals by SEM/EDS or ICPMS in the emissions generated from each of the feeedstocks, due to low recovery of emitted particles, we can not excluded the presence of trace metals in these particles.

3.2.4. Airborne particle size distributions; particle number and particle mass concentrations

The particle number concentration over the print duration for each of the metal-fill filaments varied significantly with print temperature (Fig. 3). For the Bronzefill filament, the particle number concentration was highest at a print temperature of 225 °C, increasing at the initiation of printing and dropping quickly at the end of the print cycle (Fig. 3A). At a print temperature of 210 °C, the number concentration increased rapidly then decreased by a factor of two over the print cycle then dropped at the end. The number concentration for the Bronzefill filament at a print temperature of 200 °C also increased at the beginning of the cycle but to a significantly lower maximum value than for 210 °C or 225 °C and then deceased to the baseline at the end of the print. The particle number concentrations for the Steelfill filament were also dependent on the print temperature with the highest emission level at 220 °C, increasing abruptly at the be beginning of the print, then decreasing at the end (Fig. 3B). The emission patterns for the lower temperatures of 210 °C and 200 °C showed similar patterns to the bronze filament in that the lower print temperatures showed significantly lower maximum particulate emissions. Similarly, for the Copperfill filament, the particle number concentrations also rose quickly at the beginning of the print, were higher at a print temperature of 210 °C compared to 200 °C and dropped rapidly after the end of the filament extrusion (Fig. 3C). When comparing the three metal-fill filament types at a print temperature of 210 °C, the Copperfill produced the greatest number concentration and increased emissions over the course of the printing cycle compared to the Bronzefill and Steelfill (Fig. 3D).

Fig. 3.

SMPS-CPC: Number concentration of particles emitted during printing of Bronzefill, Steelfill and Copperfill feedstocks at different temperatures. Emission concentration recorded at the same printing temperature (210 °C) is given at the lower right column. Data in the lower right panel represent separate experimental runs. On each of the curves the extrusion start time begins where the concentration increases and stop time is where the concentration decreases. Only minimal emissions were observed during warming up or after extrusion stopped.

The normalized size weighted particulate concentrations for the three metal-fill filaments over the course of the printing cycle are shown in Fig. 4. Consistent with the results shown in Fig. 3, the relative particle number values were highest for the Copperfill filament. Particle concentrations are greatest for print temperatures of 210 °C compared to 200 °C. The highest size weighed particle concentration for each filament was approximately 50 nm and most of the particles were in the nanoparticle size range (<100 nm) (Fig. 4 and Fig. S8). Fig. 4 is similar to Fig. S8 with that the exception that the number-weighted concentration scales are adjusted for each filament and concentration to allow for better maximum size determination rather than comparisons among all conditions. The size weighted mass concentrations for each of the metal-fill filaments is shown in Fig. 5 and Fig. S9. Again, the Copperfill filament showed the greatest relative mass emissions, and the highest emissions were greatest among all filaments at temperatures of 210 °C as compared to 200 °C. Compared to the size weighted particle number concertation, the size weighted particle mass concentration showed particle size distribution that was larger with a maximum between 300 and 400 nm. The small fluctuations in the emission concentration observed during the printing process could be a result of the heterogeneity of the size and localization of the metal particles within the metal-fill filaments as observed on the SEM images of original filaments (Fig. 2, Fig. S2-S4). We are not certain of the effect of our small chamber size on the particle agglomeration. Studies of 3D printers in test chambers have shown that median particle size does shift over the time of the experiment (tens of minutes to hours) to become larger, likely as a result of agglomeration, as vapors condense onto nucleation sites and particles cool and move away from the print head. In contrast, particles released directly from the print head without time to agglomerate tend to be in the sub-micron size range (Vance et al., 2017). That being said, because of the high concentration of particles within the flask, increases in particle size and particle agglomeration are still likely to occur to some degree due to contact between the particles and vapors that are present.

Fig. 4.

SMPS-CPC: Size weighted particulate number concentration of emissions generated during 3D printing.

Particulate emissions from Copperfill filaments were reported in Alberts et al. (2021) who also observed that particles smaller than 50 nm particles dominated the emissions. Other studies also reported a strong dependence of emission concentrations on temperature in copper containing filaments (Poikkimaki et al., 2019; Stabile et al., 2017; Vance et al., 2017). Our results and those from Alberts et al. (2021) are different from Vance et al. (2017) where results in the later indicated larger proportion of particles >100 nm particles in copper infused filaments (Alberts et al., 2021; Vance et al., 2017).

In our system configuration, samples were collected at close proximity, i.e., < 5 cm from the extruder’s nozzle and the printed filament was collected in a 1 L chamber (Erlenmeyer flask). The Biosampler and thermophoretic sampler were within 20 cm of the nozzle while for the SMPS samples were drawn through a tube that extends about one meter from the instrument. Consequently, because samples were not drawn from a larger chamber volume as is typically reported (Byrley et al., 2021), caution should be taken in comparisons to other studies. Although the nozzle did not contact a heated print surface, the total particle counts as well as the size-weighted particle number counts and size-weighted particle mass concentrations are within the ranges reported for similar types of filaments at similar print temperatures (Floyd et al., 2017; Seeger et al., 2018; Singh et al., 2021; Stabile et al., 2016; Vance et al., 2017; Alberts et al., 2021). In 3D printer processes that create complex geometries, the rate of extrusion may change along with the starting and stopping of extrusion as the 3D printer head moves across the object to account for infill density and different object patterns. However, for the creation of simple geometries, a constant extrusion with minimal movement may suffice. Emissions recorded from our setup do not represent the printing of all 3D printing geometries, especially those more complex; however, it does closely align with a scenario where a 3D printer operator’s head would be stationed close to the nozzle and a simple geometric figure at a constant rate is printed. Although particulate emissions from 3D printing have been reported for filaments containing copper, bronze, and steel (Floyd et al., 2017; Seeger et al., 2018; Singh et al., 2021; Stabile et al., 2016; Vance et al., 2017; Alberts et al., 2021), we report metals characterization of the filaments, hydrodynamic characterization of the particles and comparative particle emissions as a function of printing temperature.

Aerosol monitoring of the particulate emissions from printing each of the metal-fill filaments showed a size distribution where the particle number was dominated by smaller sized particles and particle mass was dominated by a population of larger particles. As suggested by SEM micrographs of particles captured by the thermophoretic sampler, the larger particles (micron-sized) appear to be aggregates of the smaller (< 100 nm) particles. This result might be expected in that thermophoretically collected particles were impacted and accumulated onto the surface of the TEM grid (Figs. S6–7).

3.3. Exposure implications

Due, in part, to their low cost and increasing availability, 3D printers are being used in a variety of home and school settings. Karwasz et al. (2022) suggest that over half of the university students surveyed who use 3D printers use them in rooms where they spend most of their waking and sleeping time. Because extended “printing jobs” may require many hours to complete, particles may remain airborne or accumulate on surfaces in these potentially poorly ventilated sleeping rooms. Modeling results suggest that nanoscale particulate emissions from 3D printing processes not only penetrate pulmonary lung passages but may also accumulate with regular printer operation, particularly for teens and young adults (Byrley et al., 2021).

Recent reports indicate that metal-fill filaments used primarily for print product aesthetics produce greater particulate emissions than non-metal filaments. Our results indicate that for the three metal-fill filaments tested, increasing print temperatures increased particulate emissions with the Copperfill resulting in the highest emissions. These results also suggest that metallic particles in the sub-micron to tens of a micron range may be separated during printing or as a result of cutting or abrasion. Because sanding and polishing is suggested by manufacturers as a common post printing practice it may be possible for these particles be ingested or accumulate in home use settings. As a result of our results as well as the results of others, we suggest additional precautions be used when printing with metal-fill filaments.

4. Summary and conclusions

Given the wide accessibility of fused filament fabrication 3D printers, it is anticipated that the market for 3D printer filaments loaded with additive such as metal fillers will likely grow in the coming years. We report the characterization of metal-fill filaments, printer extruded filaments, and particulate emissions generated during a 3D printing process.

Lower level and trace metals that were not specifically listed in safety data sheets were detected in original and printed filaments. We observed that metal particles are easily dislodged from exposed surfaces of original and printed filaments (Fig. S2A). In addition, we suggest that potential inhalation, ocular, dermal or oral exposures may occur during handling and post-print sanding or polishing. Additional studies are required, however, to assess possible metals leaching into various media particularly those such as gastric and intestinal fluids associated with ingestion exposure.

Particulate emissions observed during the printing process were heterogeneous in shape, size, and morphology. The diameter of particles was largely <50 nm, the mean hydrodynamic diameter of particulate emissions was 124–189 nm by NTA, and 219–232 nm by DLS measurements indicating possible aggregation of particles in a water matrix. In agreement to previous reports, particulate concentrations increased with printing temperature. Copperfill filament seems to generate more emissions compared to Steelfill and Bronzefill at 210 °C, possibly due to a higher thermal conductivity of copper compared to the other alloys. The particulate emission concentration in this report is relatively higher for copper-based PLA composite, compared to previous reports. This may be partly due to confinement of the print head to a smaller volume which enabled a maximum emision capture.

Due to its unique capabilities, 3D printing processes are finding applications in areas such as industrial prototyping, cottage manufacturing, education, and hobbyists. The wide range of polymers and additives, including nanoparticles, nanofibers, flame retardants, organic inorganic colorants and metal micro and nano particles has broadly expanded the variety of print product characteristics. The broad range of additives, however, has also created uncertainties in the potential for exposures to particulate emissions as well as metal particle exposure from post-processing of print products. Given the ability to vary printing temperatures for many 3D printers, our results indicating increased particulate emissions at higher print temperatures for metal-fill filaments may impact human exposure. In addition, considering their nanoscale size, these particulate emissions could be particularly significant for printer operations by children and teens in environments with low ventilation such as bedrooms, classrooms, and garages.

Data availability

Data will be made available on request.