Assessment of emissions and exposure in 3D printing workplaces in Taiwan

Abstract

Three-dimensional (3D) printing is an emerging and booming industry in Taiwan. Compared to traditional manufacturing, 3D printing has various advantages, such as advanced customization, additive manufacturing, reduced mold opening time, and reduced consumption of precursors. In this study, the real-time monitoring of particulate matter (PM) and total volatile organic compound (TVOC) emissions from various filaments is investigated using fused deposition modeling with material extrusion technology, a liquid-crystal display, a stereolithography apparatus based on vat photopolymerization technology, and binder jetting for occupational settings. An exposure assessment for nearby workers using the 3D printing process was performed, and improvement measures were recommended. Nine 3D printing fields were measured. The generation rate of ultrafine particles ranged from 1.19 × 10¹⁰ to 4.90 × 10¹² #/min, and the geometric mean particle size ranged from 30.91 to 55.50 nm. The average concentration of ultrafine particles ranged from 2.31 × 10³ to 7.36 × 10⁴ #/cm³, and the PM_2.5 and PM₁₀ concentrations in each field ranged from 0.74 ± 0.27 to 12.46 ± 5.61 μg/m³ and from 2.39 ± 0.60 to 30.65 ± 21.26 μg/m³, respectively. The TVOC concentration ranged from 0.127 ± 0.012 to 1.567 ± 0.172 ppm. The respiratory deposition (RD_UFPs) dose ranged from 2.02 × 10¹³ to 5.54 × 10¹⁴ nm²/day. Depending on the operating conditions, appropriate control and protective measures should be employed to protect workers’ health.

Keywords:

• 3D printing-induced emissions
• particulate matter
• occupational assessment
• volatile organic compounds
• respiratory deposition dose

Introduction

Recently, three-dimensional (3D) printing has become an emerging and booming industry in Taiwan. Unlike traditional manufacturing, 3D printing has many advantages, such as reducing material costs, simplifying the production process, customizing small-scale production, reducing material use, and improving efficiency (Kwon et al. 2017). According to the American Society for Testing and Materials (ASTM), 3D printing is a material bonding process. Computer-aided design (CAD) software is used to process 3D model data, and the materials are printed layer by layer to produce three-dimensional objects. The seven fundamental 3D printing technologies are vat photopolymerization (VP), material extrusion (ME), binder jetting (BJ), material jetting (MJ), powder bed fusion (PBF), direct energy deposition (DED), and sheet lamination (SL) (Stefaniak et al. 2017; Chen et al. 2020; Min et al. 2021; Stefaniak et al. 2021). The categories of domestic 3D printing technologies are shown in Figure S1 (Supplementary material).

A brief introduction to 3D printing technologies is as follows: vat photopolymerization (VP) uses ultraviolet laser light to irradiate liquid photocurable resins via a polymerization process. After irradiation, the resins change from liquid to solid and are stacked layer upon layer. VP can be divided into three categories: stereolithography apparatus (SLA), digital light processing (DLP), and liquid-crystal display (LCD) (Nugroho et al. 2021; Zhang et al. 2021). Material extrusion (ME) involves extrusion through a nozzle by applying pressure and heat. It is associated with a uniform cross-section and the desired stroke diameter. There are two common ME technologies, fused deposition modeling (FDM) and fused filament fabrication (FFF), which use high temperatures (200–240 °C) to melt filament materials, extrude the filament through the printing nozzle, and deposit the melted filament material upon heating of the build platform (Zisook et al. 2020; Hegab et al. 2023). Binder jetting (BJ) is usually performed on a powder in a machine, and then a single-layer pattern is printed by the nozzle. The binder is used to adhere to the powder layer. The materials fabricated by this technology include ceramics (glass or plaster) and metals (stainless steel). In material jetting (MJ), a liquid drop of photopolymer ink resin is commonly continuously injected via a print head nozzle onto the platform. The ink-like polymer resin is cured by ultraviolet illumination after the deposition process. The process is repeated layer upon layer to develop the products (Tan et al. 2020; Tyagi et al. 2022). Powder bed fusion (PBF) uses thermal energy to selectively fuze regions of a powder bed. This technique has two types: selective laser sintering (SLS) and selective laser melting (SLM). SLM requires the application of appropriate energy to the feedstock material to join layers together, while SLS operates by applying less energy to the feedstock material and fusing the layers on a molecular level (Singh et al. 2020; Stefaniak et al. 2021). Direct energy deposition (DED) uses an electron or laser beam as a heat source to melt the feedstock material, which is then deposited and fused on the melted substrate. Finally, the movement of the laser beam or an electron solidifies the materials (Pinheiro et al. 2021). Sheet lamination (SL) involves building materials and cutting methods. SL is classified into two types: “bond-then-form” and “form-then-bond” processes. It uses a polymer-based adhesive to bond the sheet materials (Nugroho et al. 2021).

Recently, the emission of hazardous pollutants (HPs) from 3D printing has gained attention due to potential health risks. According to previous studies, all 3D printing technologies except for DED and SL generate HPs, such as particulate matter (PM) and various types of volatile organic compounds (VOCs) (Stefaniak et al. 2017; Zhang et al. 2017). Various reports have suggested that PM emission rates increase with increasing nozzle temperature (Azimi et al. 2016; Steinle 2016; Jeon et al. 2020). High-temperature nozzles also lead to a significant change in concentration at the beginning of the boost (dramatic increase in emission) (Floyd et al. 2017; Zontek et al. 2017; Hayes et al. 2021). Commonly used filaments for 3D printing include polymers (thermoplastic), such as acrylonitrile butadiene styrene (ABS), polylactic acid (PLA), and polyethylene terephthalate glycol, with different composites and additives (Zhang et al. 2023). Previous research has shown that a printer using ABS filaments has a much higher PM emission rate than a printer using polylactic acid (PLA) feedstock, particularly ultrafine particles (UFPs) with an aerodynamic diameter less than 100 nm (Stephens et al. 2013; Azimi et al. 2016; Kwon et al. 2017; Zhang et al. 2019).

These nanoparticles can enter the alveoli of the human body and penetrate the alveoli into the blood circulation system, causing systemic health effects. PMs can cause allergic rhinitis, cough, asthma, breathing difficulties, cardiovascular disease (CVD), emphysema, damage to the alveoli, and lung malfunction (Warheit 2004; Oberdörster et al. 2005; Sioutas et al. 2005; Mokhtar et al. 2013; Zhang et al. 2017; Ding et al. 2019). In addition, 3D printing-induced PM emissions are affected by the environment of operation and parameters such as filling density, nozzle temperature, filament color, residence time, printing thickness, and printing speed (Azimi et al. 2016; Yi et al. 2016; Stabile et al. 2017; Manoj et al. 2021). Furthermore, most PM emitted during 3D printing is smaller than 100 nm in diameter (Vance et al. 2017; Zontek et al. 2017; Youn et al. 2019).

The VOCs generated from ABS filaments include styrene, 1,3-butadiene, and acrylonitrile, whereas polylactic acid and lactide come from PLA filaments (Stefaniak et al. 2017; Potter et al. 2019). Other VOCs are potentially produced from light-curable resins, which contain different VOC components due to additives and composites. These VOCs include benzaldehyde, acetone, 4-oxopentanal, and other VOCs (Du Preez et al. 2018; Davis et al. 2019; Stefaniak et al. 2019; Zisook et al. 2020). However, some studies have shown that other factors could affect the VOC emissions from the printing process. Examples include plasticizers, flame retardants, antioxidants, epoxy silanes, and other substances. In addition, filaments pyrolyzed during heating will generate semivolatile organic compounds, which could attach to the surface of particles (Gu et al. 2019; Stefaniak et al. 2019). Other studies have reported that the concentration of VOCs increased during the posttreatment process (Dunn et al. 2020; Viitanen et al. 2021); the main reason for this increase is most likely related to the emission of acetone and isopropanol, which are used in the posttreatment process for surface cleaning and smoothening (Manoj et al. 2021). The VOCs emitted from a 3D printer can potentially accumulate inside the casing and achieve a high concentration. (Stephens et al. 2013; Davis et al. 2019; Zhang et al. 2019). VOCs potentially cause respiratory damage, eye irritation, liver and kidney damage, reproductive system and cardiovascular damage, and cancer risks (Cristescu et al. 2011; Chan et al. 2018). Fortunately, there are some 3-D printers equipped with ventilation design to vent pollutants, but not all printers. However, the designs should be applied to all 3D printers (Viitanen et al. 2021).

Currently, there is no standard sampling method for 3D printing workplaces in Taiwan, and there is no on-site measurement data. Therefore, this study focuses on the research gaps in the exposure to PM and VOCs in 3D printing workplaces in Taiwan. This research aims to measure the exposure in real workplaces (on-site sampling), investigate the 3D printing pollution sources and their concentrations, and estimate their respiratory deposition dose. Finally, we suggest improvement strategies to reduce exposure and promote the health of 3D printing workers.

FDM, LCD, and SLA are gaining popularity among the public in Taiwan and are the most mainstream 3D printing technologies due to their convenient operation. Therefore, this study selected these three 3D printing technologies for investigation. The potential health risks of these hazardous pollutants to workers are a concern because 3D printing is now pervasive in everyday life and might pose health risks to consumers. Therefore, this study focused mainly on FDM, a type of ME technology; LCD and SLA, types of VP technology; and BJ. BJ was considered due to its use in domestic academic research institutions and the automobile, locomotive, and aerospace industries. Relevant information about the release of harmful substances during 3D printing operations was collected by examining the characteristics of 3D printing machines and the on-site operating environments in Taiwanese 3D printing industries. Pollutant exposure profiles and concentrations were measured in workplaces. The purpose of this study was to assess the exposure of workers in 3D printing workplaces in Taiwan.

Materials and methods

Sampling areas and their characteristics

Since ME and VP are the mainstream technologies on the market (Balletti et al. 2017), they are the focus of this study. The sampling areas of this study are five manufacturers (their products include survival game armor, lamp decoration, dolls, customized plastic objects, precision metal objects, etc.) and two academic institutes (teaching purposes) in Taiwan, and the conditions are established at various sites for real-time monitoring of the concentration changes during the printing process. The room volumes of the sampling areas ranged from 90 to 191 m³. The indoor temperatures were between 25.34 and 32.49 °C, and the relative humidity (RH) ranged from 47.1 to 64.8%. The specifications of the studied 3D printers and print setting parameters are summarized in Table 1, and the floor plan of one of the sampling areas (FDM02) among the 3D printing workplaces is shown in Figure 1 (the numbers from 1 to 5 are the sampling points for real-time monitoring of 3D printing emissions).

Figure 1. Floor plan of the FDM02 3D printing workplace.

The gray circles and the triangles represent the different types of 3D printers. The light-gray shapes represent operating printers, and the dark-gray shapes represent nonoperating printers. The numbers 1 to 5 indicate the sampling points. The sampling time for sampling point 1 was 2 hr, and those for sampling points 2, 3, 4, and 5 were 12 min. The timelines for the sampling points are shown in Figure 3.

Table 1. Specifications of the studied 3-D printers and print conditions.

		This study										BJ
Source code-Filament		FDM01-PLA	FDM05-PLA	FDM02-PETG	FDM03-ABS	FDM04-ABS	LCD01	SLA01	FDM06-PLA	BJ01	BJ01 (Post-processing)	Field/Source
Model		ME-FDM (Delta)	ME-FDM	ME-FDM (Delta)	ME-FDM	ME-FDM	VP-LCD	VP-SLA	ME-FDM& VP-LCD	BJ	BJ	Model
Color		Gray	Glue Pink Green (2)(2) $$C_{\mathit{\text{air}}}={\mathrm{C}}_1\times \frac{\mathrm{E}}{{\mathrm{Q}}_{\mathrm{F}}\times \text{t}}\times \frac{273}{273+\mathrm{T}}\times \frac{\text{Pa}}{101325}$$(2)	Transparent	Orange	Orange	Gray	Gray	Gray	Gray	Gray	Color
Description		Ventilator (172 m^3)	Classroom (109 m^3)	Exhaust fan 2 unit (172m^3)	Exhaust fan 1 unit classroom (109 m^3)	Cover classroom (109 m^3)	Cover (191 m^3)	Cover classroom (109 m^3)	Not ventilated Office (163 m^3)	Vacuum cleaner 1 unit Office (90 m^3)	Vacuum cleaner 1 unit Office (90 m^3)	Description
Air exchange rate [h^-1]		2.15	0.27	13.54	23.94	–	21.88	10.63	0.17	0.18	–	Air exchange rate [h^-1]
Number of machines		Small 48 unit	4	Large 3 unit Small 13 unit	1	1	Large 7 unit Small 4 unit	2	FDM 2 unit LCD 3 unit	1	1	Number of machines
Print conditions [°C]	Extruder	215	200	235	260	260	–	–	FDM (190 & 215) LCD (None)	70,000	–	Nozzle number
	Build plate	60	65	80	90	90	–	–	FDM (60) LCD(None)	17-4PH SUS-630	17-4PH SUS-630	Composite powder
Measure time[min]		120	180	120	180	180	120	180	180	210	36	Measure time[min]
Printing speed [mm s^−1]		40	60	40	60	60	80	80	60 ∼ 80	800 cm^3/hr (Fast speed)	–	Printing speed [cm^3 hr^−1]
Glue		White glue	No glue	White glue	No glue	No glue	No glue	No glue	No glue	Ethylene glycol & Water	–	Glue

**ME-FDM: material extrusion-fused deposition modeling; VP-LCD: vapor photopolymerization-liquid-crystal display; VP-SLA: vapor photopolymerization-stereo lithography apparatus; BJ: binder jetting; PLA: polylactic acid; ABS: acrylonitrile butadiene styrene; PETG: polyethylene terephthalate glycol. LCD and SLA01 3-D printers use light-curing resin as the raw material (Filament).

Emission measurement procedures

The particle concentration and size distribution in the 3D printing workplaces were monitored using a scanning mobility particle sizer (SMPS, Model 3936, TSI Inc., Shoreview, MN, USA) and optical particle counter (OPC, Model 3016 IAQ, Lighthouse Inc. USA). The VOC sampling method refers to the National Institute of Environmental Analysis of Taiwan “A71911C” and “A72170B,” and a constant-flow air sampling pump was used to draw the air samples to the activated carbon sampling tube (coconut shell charcoal, 100 mg/50 mg, 226–01–SKC, USA). The VOCs in the air samples adsorbed by the activated carbon sampling tube were extracted with an appropriate amount of organic solvent and a vibrator. Next, the extracts were injected into a gas chromatography–mass spectrometry (GC–MS) instrument to determine the concentrations of VOCs in the sampling areas. The analysis was conducted at the Environmental Technology Research Center, Yuan Ze University. The sampling probes of all the monitoring instruments were close to the nozzle of the 3D printers (approximately 1 m). This distance is suitable for operating 3D printers and observing printing situations, allowing better simulation of the exposure scenario of 3D-printed workers. The sampling procedure had three stages: (1)(1) $$G=Ｖ\frac{\mathit{\text{dC}}}{\mathit{\text{dt}}}-Q{C}_{\mathit{\text{input}}}+\mathit{\text{QC}}$$(1) background measurement (9–60 min); (2)(2) $$C_{\mathit{\text{air}}}={\mathrm{C}}_1\times \frac{\mathrm{E}}{{\mathrm{Q}}_{\mathrm{F}}\times \text{t}}\times \frac{273}{273+\mathrm{T}}\times \frac{\text{Pa}}{101325}$$(2) the first point printing period (120 min to monitor the variation in emissions during the different 3D printing processes); and (3)(3) $${\text{RD}}_{\mathit{\text{UFPs}}}={\text{PM}}_{\text{surface}}\times \text{IR}\times \text{DF}$$(3) four additional points for spatial sampling (15 min for each sampling point). According to the questionnaire (provided in the supplementary material), the average operating time for 3D printing employees was approximately 3 hr, which is also their exposure time. Therefore, the overall sampling time was set as 3 h. The experimental setup is depicted in Figure 2 (Afshar-Mohajer et al. 2015).

Figure 2. Experimental setup for 3D printing in occupational settings.

Estimations of the emission rates

Emission of airborne particles

The airborne particle emission rate was calculated based on the mass-balance equation to estimate the number of particles generated per minute during the 3D printing process, and the influential factors included the on-site ventilation rate and field size. Assuming uniform indoor air mixing, the particle emission rate was estimated using Equation (1)(1) $$G=Ｖ\frac{\mathit{\text{dC}}}{\mathit{\text{dt}}}-Q{C}_{\mathit{\text{input}}}+\mathit{\text{QC}}$$(1) . (1) $$G=Ｖ\frac{\mathit{\text{dC}}}{\mathit{\text{dt}}}-Q{C}_{\mathit{\text{input}}}+\mathit{\text{QC}}$$(1) where:

G = the particulate emission rate (#/min)

V = the room volume of each field (cm³)

Q = the ventilation flow (cm³/min)

C_input = the particle number background concentration (#/cm³)

C = the particle number concentration (#/cm³).

Emission of TVOCs

The concentration of the VOCs in the solvent extract after the VOC fields were desorbed from the activated carbon adsorption tube was reported in units of mg/L. The VOC concentration unit was converted to mg/m³ using Equation (2)(2) $$C_{\mathit{\text{air}}}={\mathrm{C}}_1\times \frac{\mathrm{E}}{{\mathrm{Q}}_{\mathrm{F}}\times \text{t}}\times \frac{273}{273+\mathrm{T}}\times \frac{\text{Pa}}{101325}$$(2) . (2) $$C_{\mathit{\text{air}}}={\mathrm{C}}_1\times \frac{\mathrm{E}}{{\mathrm{Q}}_{\mathrm{F}}\times \text{t}}\times \frac{273}{273+\mathrm{T}}\times \frac{\text{Pa}}{101325}$$(2) where:

C_air = the VOC concentration in the air (mg/m³)

C₁ = the VOC concentration of the extract obtained from the activated carbon adsorption tube (mg/L)

E = the volume of the extract (0.001 L)

Q_F = the sampling flow (m³/min)

t = the sampling time of the VOC adsorption tube (approximately 180 min)

T = the average indoor temperature in each field (°C)

Pa = the atmospheric pressure.

Exposure and health risk assessment

UFP inhalation exposure assessment

The assessment of UFP inhalation exposure was based on three hundred questionnaires completed by 3D printing staff. The correlation between the surface area of the UFPs and the health effect is better than that between the mass concentration of the UFPs and the health effect (Yu et al. 2015). The following formula was used to convert the particle surface area concentration into respiratory deposition (RD) (Sze-To et al. 2012). The respiratory deposition dose of the UFPs (RD_UFPs) considered that most particles emitted from 3D printing had aerodynamic diameters less than 100 nm, and the particle size range was focused on particles <100 nm in this study by applying Equation (3)(3) $${\text{RD}}_{\mathit{\text{UFPs}}}={\text{PM}}_{\text{surface}}\times \text{IR}\times \text{DF}$$(3) . (3) $${\text{RD}}_{\mathit{\text{UFPs}}}={\text{PM}}_{\text{surface}}\times \text{IR}\times \text{DF}$$(3) where:

PM_surface = the average particle surface area concentration (nm²/m³) measured by the scanning mobility particle sizer (SMPS) in the 3D printing workplaces

IR = the inhalation rate (m³/day)

DF = the deposition fraction.

DF was estimated by summing the head airways (HA), tracheobronchial (TB), and alveolar regions (AL), as shown in the following equation: (4) $$\text{DF}={\text{DF}}_{\text{HA}}+{\text{DF}}_{\text{TB}}+{\text{DF}}_{\text{AL}}$$(4) where

DF_HA = the deposition fraction in head airways, calculated as follows: (5) $$\text{DFHA}=\hspace{0.17em}\text{IF}(\frac{1}{\text{1 + exp}\mathrm{(}\mathrm{6.84\; +\; 1.1831n}\mathit{\text{dp}})}\text{+}\frac{1}{\text{1 + exp}\mathrm{(}\mathrm{0.924‐1.8851n}\mathit{\text{dp}})})$$(5) where:

d_p = the particle diameter in µm

IF = the inhalable fraction is given by Equation (6)(6) $$\text{IF = 1}-0.5\left(1-\frac{1}{\text{1 + 0.0076}d{p}^{2.8}}\right)$$(6) (Sharma and Balasubramanian 2020): (6) $$\text{IF = 1}-0.5\left(1-\frac{1}{\text{1 + 0.0076}d{p}^{2.8}}\right)$$(6)

DF_TB = the deposition fraction in the tracheobronchial region, calculated as follows (Hinds and Zhu 2022): (7) $$\text{DFTB = IF}\left(\frac{0.00352}{\mathit{\text{dp}}}\right)[\mathrm{\text{exp}}\hspace{0.17em}(-0.234{(1\mathit{\text{ndp}}+3.40)}^2)+6.39\hspace{0.17em}\mathrm{\text{exp}}\hspace{0.17em}(-0.819{(1\mathit{\text{ndp}}-1.61)}^2)]$$(7)

DF_AL = the deposition fraction in the alveolar region, calculated as follows (Hinds and Zhu 2022): (8) $$\text{DFAL = IF}\left(\frac{0.0155}{\mathit{\text{dp}}}\right)[\mathrm{\text{exp}}\hspace{0.17em}(-0.416{(1\mathit{\text{ndp}}+2.84)}^2)+19.11\hspace{0.17em}\mathrm{\text{exp}}\hspace{0.17em}(-0.482{(1\mathit{\text{ndp}}-1.362)}^2)]$$(8)

TVOC carcinogenic risk assessment

The increase in inhalation unit risk [inhalation unit risk (per mg/m³)] was calculated by multiplying by the average concentration of inhaled carcinogenic air pollutants (C_air mg/m³). C_air is the 180-min average concentration at the first point of sampling. The parameters for TVOC carcinogenic risk assessment were based on the results obtained from three hundred questionnaires completed by 3D printing staff. The slope factors for each possible exposure to contaminants were calculated using Equations (9) and (10)(9) $$\text{Risk}=\text{CDI}\times \text{SF}\times \frac{\text{ED}\times \text{EF}}{\text{AT}}$$(9) . The carcinogenic risk reference concentrations and carcinogenic risk factors for pollutants that can potentially be detected in 3D printing workplaces are summarized in Table S2 (Supplementary material).

(9) $$\text{Risk}=\text{CDI}\times \text{SF}\times \frac{\text{ED}\times \text{EF}}{\text{AT}}$$(9)

Or (10) $$\text{Risk}=C_{\mathit{\text{air}}}\times \text{Inhalation Unit Risk}\times \frac{\text{ED}\times \text{EF}}{\text{AT}}$$(10) where:

C_air = the VOC concentration (mg/m³) measured in the 3D printing workplaces.

CDI = the chronic daily intake or average daily inhalation dose over 70 years (mg/kg-day) based on the questionnaires completed by 300 3D printing workers.

SF = the slope factor, mainly referring to past literature and suitable for expressing the slope of the dose–response relationship (mg/kg-day)⁻¹

ED = exposure duration (18 to 70 years, 53 years)

EF = exposure frequency (5 days of work per week, 52 wk per year) (day/year)

AT = average time to exposure (70 years = 25,567 days).

TVOC noncarcinogenic risk assessment

The reference concentration of chemical substances [mg/m³-day]⁻¹ (RfC) was calculated using the following equation: (11) $$\text{RfC}=\frac{\text{NOAEL or LOAEL}}{\text{UF}\times \text{MF}}$$(11) where:

NOAEL = the no observed adverse effect level

LOAEL = the lowest observed adverse effect level

UF = the uncertainty factor

MF = the modifying factor.

The hazard quotient (HQ) is the ratio of exposure to a chemical substance to the reference concentration of the chemical substance and is assessed in terms of the noncarcinogenic risk.

(12) $$\text{HQ}=\frac{\text{CDI}}{\text{RfC}}$$(12) where: CDI = the long-term daily intake dose, exposure to chemical substances, or exposure dose (mg/kg-day) estimated based on the measurements performed in the 3D printing workplace and the results of three hundred questionnaires (supporting information) completed by 3D printing staff.

If there are multiple exposure routes for noncarcinogenic substances, the hazard index (HI) can be used to quantify the risk of noncarcinogenic substances: (13) $$\text{HI}={\sum }^{\mathrm{}}\text{HQ}={\sum }^{\mathrm{}}\frac{\text{CDI}}{\text{RfC}}$$(13)

The hazard risk is acceptable when the HI is less than 1. If the relative HI value is greater than 1, the hazard risk is unacceptable. A greater hazard index HI corresponds to greater noncarcinogenic harm to the human body.

Results and discussion

PM and TVOC profiles during 3D printing activities

The estimated PM and VOC concentrations during 3D printing activities in occupational settings are listed in Table 2. The emission rate and RD_UFPS were determined considering that the emission characteristics of 3D printing are less than 100 nm. When FDM is used, the filament emission pollution shows the same trend with ABS > polyethylene terephthalate glycol (PETG) >PLA. Both LCD and SLA methods using photocuring technology (Quan et al. 2020) indicate high VOC concentrations due to the solvent used in the resin and printing process. The sampling area in which the BJ printer is used has more particle control equipment in place, such as box sealers, high-efficiency particulate air (HEPA) filters, filtered vacuum cleaners, and dust removers. As a result, the particle number concentration for the BJ printer was the lowest among all areas, but there was no control equipment for VOCs.

Table 2. Estimation of particulate matter and TVOCs during 3D printing in occupational workplaces.

Field	3D printer	Diameter (nm) (GMD ± GSD)	Emission rate (#/min, <100 nm)	SMPS num. conc. (#/cm^3)	SMPS Background num. conc. (#/cm^3)	OPC-PM2.5 (µg/m^3)	OPC-PM10 (µg/m^3)	Particle Density (g/cm^3)	RDUFPS (nm^2/day)	TVOC (µg/min)	Air exchange rate (h^-1)
FDM01	ME-FDM(Delta)	52.12 ± 2.55	2.61 × 10^10	2.31 × 10^3	1.99 × 10^3	1.69 ± 0.36	4.53 ± 2.12	1.138	2.02 × 10^13	4.96 × 10^3	2.15
FDM05	ME-FDM	30.91 ± 1.93	2.84 × 10^11	1.47 × 10^4	1.85 × 10^3	0.74 ± 0.27	2.39 ± 0.60	1.138	2.01 × 10^14	4.32 × 10^2	0.27
FDM02	ME-FDM(Delta)	37.52 ± 1.70	1.39 × 10^11	2.18 × 10^4	1.99 × 10^3	4.53 ± 0.56	14.05 ± 2.21	1.145	2.18 × 10^14	1.05 × 10^4	13.54
FDM03	ME-FDM	34.14 ± 1.78	4.25 × 10^11	2.65 × 10^4	1.56 × 10^4	4.68 ± 0.24	12.20 ± 1.35	1.159	5.54 × 10^14	3.83 × 10^4	23.94
FDM04	ME-FDM	43.88 ± 1.88	4.90 × 10^12	7.36 × 10^4	1.56 × 10^4	6.64 ± 0.89	17.84 ± 7.49	1.159	1.65 × 10^14	5.04 × 10^4	–
LCD01	VP-LCD	42.85 ± 2.05	2.33 × 10^10	2.12 × 10^4	1.94 × 10^4	12.46 ± 5.61	30.10 ± 19.26	1	2.20 × 10^13	1.63 × 10^5	21.88
SLA01	VP-SLA	50.21 ± 1.81	2.07 × 10^11	1.70 × 10^4	5.09 × 10^3	4.88 ± 0.30	12.89 ± 1.93	1	1.95 × 10^14	1.58 × 10^4	10.63
FDM06	ME-FDM & VP-LCD	55.50 ± 2.20	1.15 × 10^10	2.48 × 10^3	1.87 × 10^3	2.20 ± 0.29	5.17 ± 2.84	1.138	2.09 × 10^14	1.61 × 10^3	0.17
BJ01	BJ	36.41 ± 1.77	1.19 × 10^10	7.37 × 10^3	7.25 × 10^3	8.87 ± 13.89	21.54 ± 53.88	7.76	6.47 × 10^13	5.98 × 10^2	0.18
BJ01 (Post-processing)	BJ	42.97 ± 2.00	2.55 × 10^10	5.85 × 10^3	7.25 × 10^3	8.51 ± 3.42	30.65 ± 21.26	–	5.25 × 10^13	5.96 × 10^2	–

ME-FDM: material extrusion-fused deposition modeling; VP-LCD: vapor photopolymerization-liquid-crystal display; VP-SLA: vapor photopolymerization-stereo lithography apparatus; BJ: binder jetting.

Temporal particle size distribution

The monitoring data of the temporal particle size distribution of submicron particles for various sampling areas (FDM01, FDM02, FDM03, FDM04, FDM05, and FDM06) at their occupational settings are shown in Figures 3 and 4 and Figures S1, S4, S2, and S3 (Supplementary material), respectively. The emission rates of nanoparticles in these fields are shown in Table 2. Before the beginning of printing, the number concentration and particle size distribution in all areas were relatively stable and had no significant changes. During the printing process, the printed objects were all stretched, and the thermoplastic filament melted at a high temperature, which generated hazardous pollutants.

Figure 3. Temporal submicron-particle size distribution for FDM01.

Figure 4. Temporal submicron-particle size distribution for FDM03.

The overall average particle number concentrations in the sampling areas (FDM01, FDM02, FDM03, FDM04, FDM05, and FDM06) are shown in Table 2. Compared to the background particle number concentration (1.99 × 10³ #/cm³), all the areas showed an increasing trend. Compared to reported literature values for FDM01, FDM03, and FDM06, the particle number concentrations were lower, which might be due to the filament materials used, on-site control measures, and the distance from the printers to the exhaust fan (Floyd et al. 2017; Mendes et al. 2017; Stefaniak et al. 2021). In addition, the number concentration of particles produced by the ABS filament in the FDM03 area was 2.65 × 10⁴ #/cm³, which was approximately ten times higher than that in the FDM01 and FDM02 areas. The high-temperature nozzle was in direct contact with a filament at room temperature; this contact potentially caused thermal cracking of the filament, which generated nanoparticles. For example, the UFP concentration increased sharply after approximately 85 min (10:57). This was speculated to be due to the increased temperature setting of the printers at that time point, which made the molten filament fail to cool in time after extrusion, resulting in the generation of particles. In addition, at the beginning of printing (FDM03-01), it was found that the number of particles increased abruptly, and the particle size range was concentrated below 40 nm, as shown in Figure 4. When printing the bottom support of an object, the temperature and printing density of the printer were increased to fix the object, and these settings resulted in higher UFP emissions. In addition, at 11:35, it was found that the particle concentration was concentrated below 20 nm, and the number concentration showed an increasing trend (FDM03-01). The possible reasons for this finding were that there were some problems in the extrusion process (e.g., the filament was stuck due to an inappropriate nozzle temperature) and more UFPs were generated, so the extruded materials were not completely bonded, which resulted in defects in the printed object, as shown in Figure 4. Similarly, in FMD04, the nozzle extruded the molten filament in a suspended state due to misalignment of the printing position during the printing process, as shown in Figure S4 (Supplementary material).

In addition, the number concentration of nanoparticles increased with the sampling time at the FDM05 area due to an instantaneous high-voltage laser light source cutting wood (it is often used in a Maker workplace) nearby (Supplementary material, Figure S2).

Similarly, the number concentration of the particles increased with the sampling time in the FDM06 area due to its low ventilation rate. Thus, the ventilation conditions of each workplace should be considered due to the potential health risks for workers. In all areas, the particle sizes were uniformly distributed below 300 nm, and the particle size was concentrated below 100 nm. The geometric mean diameter (GMD) and particle size range in all the areas are illustrated in Table 2 and Figures S15–S20 (Supplementary material). Similar GMD results for nanoparticles have been reported by Gu et al. (2019) and (Floyd et al. 2017) for the FDM02 and FDM03 areas, respectively, indicating that the FDM process emits nanoparticles with similar size distributions.

Similarly, the temporal particle size distributions of submicron particles in the LCD01, SLA01, and BJ01 areas are shown in Figures 5–7, respectively. The emission rates of the nanoparticles were 2.33 × 10¹⁰ #/min, 2.07 × 10¹¹ #/min and 1.19 × 10¹⁰ #/min in the LCD01, SLA01 and BJ01 areas, respectively. Compared to the background particle number concentration (7.25 × 10³ #/cm³) in other 3D printer workplaces, much lower submicron-particle pollution was generated in the BJ01 area; BJ01 and BJ01 postprocessing was performed via a vacuum cleaner connected to a HEPA filter and positive-pressure personal respiratory protective equipment (PRPE) (Mostafaei et al. 2021), respectively, which were used to prevent the escape of most submicron particles and reduce the exposure to SUS-630 particles (a kind of stainless steel particle for BJ 3D printing). However, the vacuum cleaner also created a strong airflow that might produce internal turbulence during powder removal, causing indoor dust resuspension.

Figure 5. Temporal submicron-particle size distribution for LCD01.

Figure 6. Temporal submicron-particle size distribution for SLA01.

Figure 7. Temporal submicron-particle size distribution for BJ01.

The average particle number concentration is shown in Table 2. The nanoparticle concentration for the LCD01 area was increased in different sampling time durations. This was potentially due to the isopropanol used in the posttreatment process and the overloading of the power supply on-site. The isopropanol was heated to remove the unformed resin on the 3D-printed object’s surface; this process formed aerosol particles due to condensation and resulted in a higher particle concentration and an increase in particle size. Emitted particles >300 nm were unable to be collected in the SLA01 area due to the nucleation and condensation of evaporated liquid light-curing resin, solvent, and additives. The GMD and particle size ranges in the areas are shown in Table 2 and Figures S21–S23 (Supplementary material).

Instantaneous suspended particulate mass concentration

The temporal concentrations of PM in the various areas (FDM01, FDM02, FDM03, FDM04, FDM05, FDM06, LCD01, SLA01, and BJ01) at the workplace are displayed in Figures 8–16. In all areas, the mass concentration of PM with a diameter below 2.5 µm was relatively stable, while the concentrations of PM_2.5, PM_5, and PM₁₀ fluctuated significantly, which could be related to the resuspension of ground dust caused by the activities of on-site monitoring personnel.

Figure 8. Temporal PM concentration variation for FDM01 at the workplace.

Figure 9. Temporal PM concentration variation for FDM02 at the workplace.

Figure 10. Temporal PM concentration variation for FDM03 at the workplace.

Figure 11. Temporal PM concentration variation for FDM04 at the workplace.

Figure 12. Temporal PM concentration variation for FDM05 at the workplace.

Figure 13. Temporal PM concentration variation for FDMO6 at the workplace.

Figure 14. Temporal PM concentration variation for LCD01 at the workplace.

Figure 15. Temporal PM concentration variation for SLA01 at the workplace.

Figure 16. Temporal PM concentration variation for BJ01 at the workplace.

Emission rate of TVOCs

The temporal variations in TVOC concentrations in the FDM01, FDM02, FDM03, FDM04, FDM05, FDM06, LCD01, SLA01, and BJ01 areas are shown in Figures S5–S13 (Supplementary material). The average TVOC concentrations in these occupational settings were 372, 127.2, 407.4, 546, 402, 1,567, 1,434, 378, and 1,018 ppb for the FDM01, FDM02, FDM03, FDM04, FDM05, FDM06, LCD01, SLA01, and BJ01 areas, respectively. The TVOC concentrations in the FDM06, LCD01, and BJ01 areas were higher than those in the other fields. A possible reason is poor ventilation and the prefilling of LCD raw materials, including solvents and resins. The individual VOCs of all 3D printers are presented in Table 3, and none of them exceeded the “Standard of Permissible Exposure Limit at Job Site” (Taiwan, https://law.moj.gov.tw/index.aspx).

Table 3. Detection of VOCs during 3D printing in occupational settings.

SC	S	EB	1-B	P,M-X	T	B	DCM	IPB	1,3-DCP	1,1-DCE	1,2,3-TCB	1,1-DCE	1,2,4-TCB	Cis-1,2-DCE	1,2-DCB
FDM03	–	–	–	–	–	–	–	–	–	–	0.19	0.05	–	–	–
FDM04	N.D.	N.D.	N.D.	N.D.	N.D.	N.D.	N.D.	–	–	N.D.	–	–	N.D.	–	–
FDM05	0.01	0.01	0.00	0.04	0.06	0.03	0.01	–	–	–	–	–	–	–	–
FDM06	0.07	0.01	N.D.	0.15	0.08	0.05	0.01	0.06	0.06	0.03	–	0.05	–	0.07	0.02
SLA01	0.05	–	–	–	0.02	–	–	–	–	–	–	–	0.40	–	–
BJ01	0.02	0.01	–	0.01	0.03	0.01	0.00	–	–	–	–	–	–	–	–
PEL (mg/m^3)	213.00	434.00	303.00	434.00	376.00	3.20	174.00	246.00	–	–	–	405.00	37.00	793.00	301.00

**N. D: not detected; N/A: not applicable; "–”: there are no such data; PB: propyl benzene; S: styrene; EB: ethyl benzene; 1-B: 1-butanol; P, M-X: p-, m-xylene; T: toluene; B: benzene; DCM: dichloromethane; IBP: isopropyl benzene; 1,3-DCP: 1,3-dichloropropane; 1,1-DCE: 1,1-dichloroethylene; 1,2,3-TCB: 1,2,3-trichlorobenzene; 1,1-DCE: 1,1-dichloroethane; 1,2,4-TCB: 1,2,4-trichlorobenzene; Cis-1,2-DCE: Cis-1,2-dichloroethylene; 1,2-DCB: 1,2-dichlorobenzene; PEL: permissible exposure limits.

Exposure assessment and risk estimation

Previous studies suggested that nanoparticle toxicity should be correlated with the surface area concentration (Oberdörster 2000), and this study calculated RD_UFPs using formula conversion. In addition, most particles generated during 3D printing are nanoparticles. Therefore, the surface area concentration of the particles is more suitable for estimating the incremental lifetime carcinogenic risk (ILCR_UFPs), and the assessment results are summarized in Table 2.

The estimated carcinogenic risk of VOCs in each area is summarized in Table S4 (Supplementary material). The risk results were obtained using Equations (2)(2) $$C_{\mathit{\text{air}}}={\mathrm{C}}_1\times \frac{\mathrm{E}}{{\mathrm{Q}}_{\mathrm{F}}\times \text{t}}\times \frac{273}{273+\mathrm{T}}\times \frac{\text{Pa}}{101325}$$(2) and (10)(9) $$\text{Risk}=\text{CDI}\times \text{SF}\times \frac{\text{ED}\times \text{EF}}{\text{AT}}$$(9) . All the results were below the acceptable level (<10⁻⁶) (Mohammadi et al. 2019). However, increasing the air exchange rates of the FDM05, FDM06, and BJ01 areas could reduce carcinogenic risk because the risk in these workplaces was higher than in other measurement sites.

The estimated noncarcinogenic risk of VOCs in each area is summarized in Table S5 (Supplementary material). Most areas investigated in this study had a noncarcinogenic hazard quotient (HQ) due to a VOC exposure of less than 1, which is within an acceptable range. However, the indoor air exchange rates of the FDM05, FDM06, and BJ01 areas were too low, and thus, improving the air exchange rate is recommended. For the FDM06 area, the HI (1.98) was greater than 1, indicating an unacceptable noncarcinogenic hazard to the occupants. Therefore, improving the working environment and reducing the exposure time is necessary to decrease the noncarcinogenic risk.

Limitations

Because this study was conducted during the level 3 alert for COVID-19 in Taiwan, the overall sampling willingness was not high, and some of the production processes were confidential. In addition, there were many potential confounding factors, such as nearby pollution sources other than 3D printers, affecting the measurement results of PM and VOC emissions during 3D printing. Thus, further research will be needed to consider potential confounding factors near 3D printers. Moreover, filaments for 3D printers are continuously being developed, and numerous new filament materials are created every year. Therefore, the emissions for the filaments used in this study are not standard for all technologies.

Furthermore, epidemiological studies could be conducted to validate the potential health hazards associated with 3D printing work.

Conclusion

In this study, we investigated the emission characteristics of PM and VOCs based on real-time monitoring during the 3D printing process and assessed the exposure of workers in 3D printing workplaces in Taiwan. The formation pathways of emission from 3D printing were thermal decomposition, aerosolization, and the use of organic solvents during feedstock deposition and posttreatment processes. Most nanoparticles produced by the 3D printer had similar GMDs and emission rates as observed in previous studies conducted in the laboratory, verifying that the calculation model could be applied to predict the UFP concentration in the 3D printing workplace. The TVOC concentrations and subsequent GC–MS analysis results for VOCs in 3D printing workplaces were affected by the ventilation rate but did not exceed the occupational exposure limit. The PM_2.5 and PM₁₀ concentrations measured by OPC in each field were lower than the indoor air quality guidelines and regulations. The RD_UFPs ranged from 2.02 × 10¹³ to 5.54 × 10¹⁴ nm²/day, which could be used as an estimated ILCR_UFPs value in the future. 3D printer users should be cautious because the hazardous pollutant emission rate during printing and postprocessing varies greatly under different conditions. Precautionary measures are advised during the 3D printing process, such as reducing exposure time, increasing the ventilation rate, and wearing appropriate personal respiratory protective equipment (PRPE).

Acknowledgments

The authors would like to thank all the 3D printing companies that participated in this study.

Data availability

Data will be provided by the authors upon request.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Additional information

Funding

This study was financially supported by the Institute of Labor, Occupational Safety and Health, Ministry of Labor (Grant project number: ILOSH110-H310).