Characterization of ultrafine particles emitted during laser-based additive manufacturing of metal parts

Particulate matter (PM) emitted during laser additive manufacturing with stainless steel powder materials has been studied in detail. Three different additive manufacturing techniques were studied: selective laser melting, direct metal deposition and laser cladding. Gas flow and temperature fields accompanying the processes were numerically modeled for understanding particle growth and oxidation. Transmission and scanning electron microscopy were used for primary particle and PM characterization. The PM collected in the atmosphere during manufacturing consisted of complex aggregates/agglomerates with fractal-like geometries. The overwhelming number of particles formed in the three processes had equivalent projected area diameters within the 4–16 nm size range, with median sizes of 8.0, 9.4 and 11.2 nm. The primary particles were spherical in shape and consisted of oxides of the main steel alloying elements. Larger primary particles (> 30 nm) were not fully oxidized, but where characterized by a metallic core and an oxidic surface shell.

Scientific Reports | (2020) 10:20989 | https://doi.org/10.1038/s41598-020-78073-z www.nature.com/scientificreports/ considered. Few studies have assessed occupational exposure and characterized the airborne particulate matter emitted during 3D metal printing. Workers exposure in 12 German enterprises to dust, Fe, manganese (Mn), chromium (Cr), Co and Ni was measured by collection of both, personal and stationary respirable and inhalable air samples 23 . This study showed that the measured workroom air concentrations, especially for Ni, were higher during cleaning than manufacturing. The higher exposure during cleaning was confirmed by 24 . It has also shown that nanometer-sized particles were present at relatively high number air concentrations in the additive manufacturing environment (Graff et al., 2016;Ljunggren, 2019) 24,25 .
Despite the fact that particulate emissions from 3D printers using different thermoplastics have been characterized (e.g., Stephens [26][27][28][29][30] , no information, to our best knowledge, regarding the primary particle characteristics of the nano-sized particulate matter formed during PBF-LB/M manufacturing is available. A fundamental prerequisite for evaluating potential adverse health effects among workers after inhalation of ultrafine particulate matter is the knowledge of relevant physical and chemical characteristics of the particles such as primary particle size distribution, agglomeration/aggregation state, shape and chemical composition (e.g., Oberdörster et al., 2005) 31 .
In the present contribution, ultrafine particulate matter (PM) emitted during laser processing of stainless steel powder materials is characterized in detail by transmission and scanning electron microscopy (TEM/SEM). The primary particle size, chemical and phase composition as well as mixing state of the particles are studied.  Table 1. Images of the instruments can be found in the electronic supplement (Figures S1).
The chemical composition of the powders used for different AM techniques as measured in this work and given by the producers (values in brackets) is shown in Table 2.
Air sampling. Bulk PM and individual particles for microscopy were collected by use of 25 mm "total dust " "open-face" aerosol cassettes (M000025A0, Millipore, Bedford, MA, USA) operated at 2.0 L min -1 air flow rate using in-house produced PS103 personal pumps (National Institute of Occupational Health, Oslo, Norway). An air pump equipped with filter cassette was placed in each of the two ventilation outlets at both sides of the EOS 270 M270 cabinet while four pumps with cassettes were placed inside along the cabinet walls of InssTek MX-Mini with each filter cassette facing about 25 cm from and 20 cm above to the laser spot. During operation of the LC-10 IPG-Photonics equipment two cassettes were mounted at a cast iron support stand with clamps facing with a distance of about 20 cm from and 20 cm above to the laser spot. Air sampling times varied between 20 and 20 min for InssTek MX-Mini and LC-10 IPG-Photonics and between 2 and 7 min for EOS 270.
The deposition pattern of the PM on the TEM grids is visualized in SEM images, Fig. 1c,d.
In case of overloading of PM during sampling this would mainly have influenced the possibility to characterize the size distribution of the generated aggregates/agglomerates. However, since the objective of this work was to characterize the primary particles formed, an overload of PM on the grids would cause negligible effects. However, the two SEM micrographs indicate that the collected individual particles were deposited without any significant interference of overload.
Elemental composition of powders. The elemental composition of the three stainless steel powders used was measured after 10 mg portions were dissolved in a mixture of 2 mL of aqua regia and 0.5 mL of hydrofluoric acid with microwave-assisted heating (MLS 1200, Teflon container SV140, 10 bar, Milestone, 556 J. Sorisole, Italy). Before digestion, 10 µg of beryllium (as chloride) was added as internal standard before dilution to 15 mL with de-ionized water. The elemental composition was measured with a Perkin-Elmer Optima Model 7300 DV inductively coupled plasma optical emission spectrometer (Perkin-Elmer Inc., Waltham, MA, USA).
After removing the TEM grids from the air filter surface, the same analytical procedure was used for measurement of the elemental content of the particulate matter collected by the filters. www.nature.com/scientificreports/ Scanning and transmission electron microscopy. The TEM grid specimens were imaged and analyzed in a Hitachi SU6600 field emission scanning electron microscope (Hitachi, Tokyo, Japan) equipped with a Bruker energy-dispersive X-ray detector (Bruker Nano GmbH, Berlin, Germany) and a NORDIF electron backscatter (EBSD) detector (NORDIF, Trondheim, Norway). For elemental analysis of particles, an acceleration voltage of 15 keV, analytical working distance of 10 mm and electron probe current 7-8 nA were used. For aggregates/agglomerates with primary particle sizes less than 20 nm, an area of approximately 100 × 100 nm was scanned for X-ray acquisition in a particle-dense area to obtain elemental spectra. For larger particles (> 30 nm) X-ray spectrums from single particles were obtained. In addition, an Auriga Crossbeam Workstation (Carl Zeiss AG, Oberkochen, Germany), equipped with INCA X-Max silicon drift detector (Oxford Instruments, Abingdon, UK) for energy dispersive X-ray microanalysis was used. The phase and elemental composition of the particles were studied by a Zeiss Libra 120 transmission electron microscope equipped with an OMEGA energy filter (Carl Zeiss AG, Oberkochen, Germany). Particle diameter measurements were conducted by statistical analysis of TEM images using the Minitab version 16 software (Minitab Statistical Software, Minitab 16; https ://www.minit ab.com).
Numerical modelling. The ANSYS FLUENT CFD software (ANSYS Inc., Canonsburg, PA, USA, https :// www.ansys .com) was used for mathematical modelling of gas-phase processes that mimics operating conditions of the EOSINT-M270 dual mode and the InssTekMX Mini techniques. Computational domain of 2 mm width consists of two regions for solid (of 0.35 mm in height) and gas phase (of 1 mm in height). Solid phase is Steel 316, and gas is nitrogen. Ambient conditions correspond to typical working atmosphere in the chamber with 101,325 Pa pressure and 310 ̊ K temperature. Laser heating is emulated via volumetric heat source sliding along with the surface of solid phase with a velocity of 0.7 m/s. Laser spot radius equals R = 0.05 mm and its penetration depth is D = 0.1 mm. Using powder absorbance of A = 0.7 (for infrared radiation source) and laser power of P = 100 W, one can compute volumetric dissipated power of that heating source as Q = A × P / (π × R 2 × D) = 89,127 GW/m 3 . Under such conditions the steel within the keyhole will immediately evaporate forcing significant convective upstream right above the heat affected zone covered by the laser spot. Metals vaporize mostly from a thin layer of few mean free paths of molecules above the heated surface, so called Knudsen layer. Recoil velocity of hot vapour can be estimated 34 by the expression: U vap = (γ × k × T vap /m) (1/2) , where γ = 1.67 is adiabatic constant, k is the Botlzmann constant, m = 56 is molar mass in a.m.u, and T vap = 3400 K is a metal boiling point. Using this formula, the vapour velocity is estimated to about 930 m/s which was added in FLUENT via the user-defined-function interface as a custom spot-wide momentum source (in the Navier-Stokes equation) moving with the laser scanning speed.

Results and discussion
Atmospheric cabinet concentrations of elements emitted as ultrafine PM during additive laser processing are given in Table 3.

Primary particle formation and size-distribution.
For all three instruments studied, the collected PM consisted of complex aggregates/agglomerates with fractal-like geometry (Fig. 1). No more than ten coarser particles with geometric projected diameters between 0.7 and 2 µm were observed on each filter. The elemental compositions were similar to the bulk material and no crystalline phases were identified. The presence of these may be due to sputtering from the melted alloy. No larger particles were seen. An equivalent projected area diameter of primary particles measured by TEM are shown in Fig. 2 and primary particle size-distribution summary statistics is presented in Table 4. The overwhelming number of particles formed in the three processes had equivalent projected area diameters within the 4-16 nm size range, with median sizes of 8.0, 9.4 and 11.2 nm for EOS M 270 dual mode, InssTek MX-Mini and LC-10 IPG-Photonics, respectively. The largest primary particles identified in the size-measurements had diameters of 50.4, 82.0 and 77.5 nm, correspondingly. Compared to previous research of laser ablation of metals where a maximum of the particle size distribution at 6-11 nm, dependent on laser intensity, were observed, the sizes of the primary particles in the laser additive processes studied in this work are similar 32 . It has previously been shown that the PM generated during manual metal arc, metal inert gas (MIG) and tungsten inert gas welding operations consists of agglomerates with primary particle diameters in the range of 5-40 nm with very few above 50 nm 33 .
Numerical modelling of gas-phase processes. To understand particle growth and oxidation it is essential to locate their trajectories in the zones of heated laser spots. Although it is difficult to visualize tracks of nano-sized particles directly because of their size, as well as gas flow dynamics in vicinity of processed zone, Table 3. Atmospheric concentrations of elements (mg/m 3 ) measured during laser additive processing (n = 2). www.nature.com/scientificreports/ it is, however, possible to perform a close-to-real-life numerical simulation of these gas-phase processes during laser surface treatment of the substrate. In Fig. 3 the presence of toroidal eddies surrounding a hot vertical jet of metal vapour is demonstrated. These vortices remain unchanged in vicinity of a laser spot during all the process of sintering and form a recirculation zone around the heat-affected region. A close look with streamlines plotted from a base of this hot up-stream (Fig. 3b) shows the recirculation zone more clearly. Nano-sized particles due to their extremely low mass (about 5 × 10 -16 ng) will exactly follow gas streamlines, finally trapping into that toroidal eddies. However, the particles do not stay in the recirculation zone permanently: they grow and drift to peripheral regions of the vortex, and finally leave it. According to estimations based on our numerical simulation, particle mean residence time in a vortex is about 0.5 ms for a particle of initial diameter of 10 nm and density of 7850 kg/m 3 . Similar behavior of particles is expected for the DED machine. To verify that, gas flow and temperature dynamics for the InssTek MX Mini machine have also been simulated. Laser power of 200 W is focused in a Gaussian beam having a spot diameter of 1 mm, whereas surface absorbance and penetration depth are the same as in the modeling of the PBF-LB/M machine. Although complete information on inner design of a nozzle mounted in that machine is not available, we primarily oriented on its general view and typical conventional flow rates used in three-stream coaxial nozzles. Computational domain is initially filled with air, and all in-nozzle inlets consist of 99.9% pure argon. We consider even this "approximate" case is still usable to estimate flow pattern near the laser spot. Although the cladding head moves horizontally with speed of 1 cm/s, the flow field is relatively steady. Vaporization-induced puff above the treated surface of steel has a height of about 1 mm. Again, streamlines sampled close to the heat-affected zone represent the gas recirculation regions which are shown in black color on top of the vector velocity field, demonstrating the occurrence of a toroidal vortex caused by the hot vertical gas stream (Fig. 4). In contrary to the PBF-LB/M process, this gas jet cross-collides with a flow moving  www.nature.com/scientificreports/ in opposite direction produced by the cooling gas. Nano-particles (NP) of partially condensed metal vapor should thus be trapped and turned back to the laser-affected zone again, but some of them will follow peripheral streamlines and slide along the treated surface. Estimated particle in-eddy residence time approximately equals 1.5 ms which is about 3 times more than in the PBF-LB/M process. The boundary zone of the vortex is located close to the mixing zone of the surrounding air (see Ar mass fraction in its mixture with air in the right part of Fig. 4, where "Ar-air" interface is marked in green color) with a likely forced oxidation of airborne particles because of their possible interaction with oxygen (O) from air.
Various fabrication methods of NP based on vapor deposition have been developed. Laser ablation is a method where very high energy is focused to a solid material for evaporation of light-absorbing materials where the vapor phase is thermodynamically unstable. Under chemical supersaturation vapor-phase atoms/molecules will rapidly and uncontrolled be condensed with a coagulation rate proportional to the square of their number concentration. At high temperatures particles coalesce faster than they coagulate; at lower temperatures loose agglomerates with rather open structures are formed 35 . In high temperature aerosol reactors NP (< 5 nm) coalesce almost takes place instantaneously even at temperatures considerably lower than the melting temperature of the bulk material due to e.g. the reduced melting temperature of NP 36 . The primary particle growth is also dependent on the metal vapour concentration. It has been shown in a simulation of fume formation mechanism in arc www.nature.com/scientificreports/ welding that particles are mainly generated by FeO nucleation with small sizes formed at low Fe concentrations/ low temperatures and larger primary particles at higher Fe concentrations which are not fully oxidized because of their lower surface to volume ratio 37 .When aggregates/agglomerates are allowed to be exposed to high temperatures the whole or part of the particle may be restructured during sintering even until a fully coalesced sphere is formed 36 . Similar particle formation mechanisms in PBF-LB/M processes are expected (Fig. 1b,c) where both restructured and fully coalesced spheres are present in particles consisting of both loosely and sintered primary particles. In our simulation of the gas-phase processes it is shown that when metal powder is rapidly heated by the laser, a mushroom-like cloud of hot metal vapour is formed just above the laser-processing zone of the metal surface. The evaporation rate in PBF-LB/M is more intensive than in DED because of smaller laser spot diameters (providing higher dissipated power density) used in PBF-LB/M. When released from the metal surface, the vapor  www.nature.com/scientificreports/ rapidly expands into the surrounding atmosphere where fast-moving gas jets with surrounding toroidal eddies are formed with a typical speed of 930 m/s. These vortices remain unchanged in the vicinity of the laser spot forming a recirculation zone around the heataffected zone. Thus, the vapor will rapidly be transported to relatively cold regions with following condensation. In the case of DED, condensed low-mass particles follow the streamlines of the toroidal vortices located on the boundary of vapor cloud. This boundary surface is located remarkably close to the argon-air interface allowing further low-temperature oxidation of particles and their subsequent growth. Once particle diameter reaches its critical value (due to continuous cooling and oxidation), the particle may release from these vortices because of inertia forces and gas flow instabilities with abortion of further growth.
Evolution of processes in gas metal arc welding fume has been investigated by Vishnyakov et al. 38,39 by numerical modeling. According to their model it was shown that the primary particle chemical composition and the particle size distribution strongly depends on the vapor-gas mixture cooling rate. Such a dependency was explained in their model by the decrease of the vapor-gas mixture cooling rate when the shielding gas temperature was increased. Therefore, duration of particles growth via vapor condensation and coalescence is increased with subsequent increase of particle size. The number based primary particle size distributions presented by 38 compare well with our results of particle size-distributions during additive processing (Table 4 and Fig. 2). This may indicate that the mechanism of formation and growth of primary particles during AM is similar to the processes occurring during arc welding.
Analysis of gas flow in a cross-section of the domain shows that there are two primary gas jets affecting the flow pattern: carrier gas-powder stream and shielding gas protecting the optics. It looks like increased flow rates of shielding gas (simultaneously maintaining the carrier gas flow unchanged) will lead to efficient removal of airborne particles out the zone of processing preventing their remelting in a circulated toroidal vortex of the carrier gas. Higher flow rates will also possibly lead to decrease of particle oxidation due to their reduced residence time in the air-argon mixture on the boundary of nozzle-produced gas streams and ambient atmosphere. This is a typical scenario of gas-flow-particle interaction. However, to establish exact relation between gas flow rates and powder removal rate a series of additional numerical simulations under various operating conditions will be done in further. Since it was not possible to quantitatively measure the elemental composition of the individual primary particles due to their small size (< 20 nm) the major element content of the aggregates was determined by scanning the electron beam over an area of approximately 100 nm × 100 nm with roughly the same number of primary particles (Fig. 8, Table 5). In addition, the larger particles (30-50 nm) from LC-10 IPG-Photonics were analyzed. Aluminum, carbon, copper and tin which were detected in all samples were excluded, as they are artifacts from the TEM grids and the substrate, respectively. Since the EDS X-ray analysis system has a default spectra for the elements, the measurement differences between these spectra used for estimating the elemental composition and our TEM grid samples, limits the possibility to obtain quantitative data because appropriate ZAF correction factors could not be obtained. Especially for O, this is critical 42 . The amount of Fe and Cr in PM, (shown as Fe/ Cr mass ratios in Table 5) quantified from SEM spectra of agglomerates and analysis of air filters is comparable to the content in stainless steel alloys used in the experiments. The ratios for Ni and Mn indicate fractionation of both elements with increased content in PM for the InssTek MX-Mini process. Especially for Mn, the fractionation is significant. For the other methods, valid elemental ratios were not obtained due to absence of usable quantitative information in the measurements.
By modeling Fe fume formation in arc welding 37 it has been shown that Fe atoms react in the vapour phase with O to form vaporous Fe monoxide (FeO) at 2500 K followed by nucleation of the FeO to liquid particles at temperatures around 2000-2500 K; remaining Fe atoms are condensing on the particle surface. Further on, the liquid particles are oxidized to Fe 3 O 4 around 2000 K with solidification around 1800 K. Fe is further oxidized to Fe 2 O 3 around 1500 K, but this is only for the smallest primary particles due to diffusion limitations. Thus, at Selected area electron diffraction (Fig. 1a,b,c) reveals that the PM consist of crystalline phases. Furthermore, the particles were investigated with EBSD in SEM which earlier has been used to characterize phase compositions of particles 41 . The EBSD patterns, shown in Fig. S2 and S3, are a Fe 3 O 4 spinel phase for the EOS M 270 dual mode and α-Fe for the LC-10 IPG-Photonics. EBSD is a single particle analysis principle and it might very well be that a mixture of the two phases are present for all three techniques. Because of the small size of the primary particles, it is expected that the EDSD signals originate from the particle core. The shell surrounding the particles are clearly an oxide layer and could be Fe 3 O 4 . These results support also the formation process suggested by Sanibondi's model 37 . Health aspects. Inhalation is the most relevant exposure route for occupational exposure to ultrafine PM.
The lung deposition characteristics, the potential toxic effect induced and the kinetic fate and possible translocation to other organs are predominantly mostly determined by the agglomeration/aggregation status of the inhaled ultrafine PM 42 . Scheckman and McMurry 43 have shown experimentally using a silicone rubber lung cast model that silica agglomerates with primary particle diameter of 10 nm deposited more efficiently than sodium chloride (NaCl) particles and oleic acid (OA) spheres with equal mobility and aerodynamic sizes in the size ranges 30-300 nm. At larger primary particle sizes the deposition pattern for the agglomerates was closer to match that of NaCl and OA. This may indicate that the ultrafine PM characterized in the present work, if inhaled, would deposit more efficient than predicted using the International Commission on Radiological Protection and/or multiple-path particle dosimetry models 44,45 .
If it is considered that deposited agglomerates/aggregates consist of about 10 nm sized primary particles as characterized in the present study, that are more or less loosely bound to each other, the crucial question for assessing possible biological effects of these particles upon inhalation is, if the agglomerates remain as agglomerates or if the agglomerates break down to smaller aggregates or even primary particles in contact with the lung surface-which may significantly influence their toxicological properties 46 .
Buckley et al. 47 exposed rats nose-only to aerosols of radioactive iridium-192 particles with sizes ranging from 10 to 75 nm demonstrating a slow lung clearance and increasing concentrations of particles with decreasing particle size in secondary target organs as liver and kidney with a translocation efficiency of max 0.5% of the  www.nature.com/scientificreports/ lung burden. If this low dose build-up of particles in other organs than lung could lead to any systemic effects is still unclear in their opinion. Animal experimental studies, in vitro and in vivo, have demonstrated the tendency of nano-sized particles to form larger size agglomerates following deposition and an increase of particle number due to disintegration of agglomerates seems not to be of high relevance 42 .
Therefore, particle sizes measured airborne in the respiratory zone of individuals seem to be a reasonable estimate of the size related properties of particles in the lungs.
To our best knowledge the primary spherical particles emitted in this work with a core composition of mainly Fe, Cr, Ni and Mn and an oxidic coated surface have not been toxicology tested. However, comparable primary particles in the same size range and chemical composition are generated during solid stainless steel wire welding 48 . When such particles have been investigated in lung cells and reporter cell lines, they showed no toxic effects to the reporter cells, no cytotoxicity, genotoxicity and no generation of reactive oxygen species 49 .

Conclusion
Our study shows that considerable amounts of ultrafine particles are emitted during the laser additive processing of stainless steel powder material. From high-resolution electron microscopy characterization and numerical modelling it has been shown that the PM emitted in the three additive processing technologies consists of aggregates/agglomerate with primary equivalent projected area diameters predominantly within the 4-16 nm size range and with median sizes of 8.0, 9.4 and 11.2 nm. The primary particles were spherical in shape and consisted of oxides of the main steel alloying elements. Larger primary particles (> 30 nm) were not fully oxidized, but where characterized by a metallic core and an oxidic surface shell. According to the simulation of the laser processing of metal powder, it can be assumed that the particles aggregate close to the thermal heating zone, where secondary melting of ultrafine particles also take place. Detailed examination of primary particles indicated that they are sintered to each other leading to aggregates where particles are bound together by chemical and/or sinter forces. The primary particles are usually oxide spheres with a pronounced core shell structure. All particles contained the main elements present in the stainless steel alloy processing powder used.
Comparable PM with primary particles in the same size range and chemical composition are generated during solid stainless steel wire welding.