Arsenic Exposure during Preventive Maintenance of an Ion Implanter in a Semiconductor Manufacturing Factory

Workers in the semiconductor fabrication process may be exposed to higher levels of hazardous materials, such as arsenic, during preventive maintenance (PM) tasks than during the regular operation of the fabrication process. This study investigates the exposure to arsenic and other metals during PM tasks in the ion implantation process. Airborne arsenic samples and bulk samples were obtained during various ion implanter PM tasks in a semiconductor fabrication factory. The arithmetic mean (AM) and standard deviation (SD) of airborne arsenic in personal samples were 0.64 μg m ± 0.92 μg m (n = 9), and the highest level was 2.39 μg m during medium-current ion implanter PM tasks. For area samples, the AM and SD were 0.42 μg m ± 0.69 μg m (n = 5) and the highest level was 1.79 μg m during medium-current ion implanter PM tasks. Arsenic was also found in the bulk samples of debris produced during PM tasks. Other metals (Ag, Al, Cu, Pb, Cr, Sn, Mn, Ti, Fe, and W) were found, but at low levels, prompting few health concerns compared with those of arsenic. This study found that PM workers were exposed to airborne arsenic levels that differed significantly according to the type of ion implanter used.


INTRODUCTION
The semiconductor industry involves hazards and risks in many processes, such as ion implantation, photolithography, etching, and diffusion (Bender et al., 2007;Chien et al., 2007;Ladou and Bailar 2007;Park et al., 2011;Yoon, 2012;Kim et al., 2014).Arsenic and arsenic compounds, such as arsine gas, are widely used in the semiconductor and electronics industries (IARC, 2006).Arsenic, which is used as a dopant to create functional units for semiconductor devices because of its high solubility (Ungers and Jones, 1986;Jaeger, 2002), is the main hazardous material in the semiconductor fabrication process (Bender et al., 2007).
The International Agency for Research on Cancer (IARC) has confirmed that inorganic arsenic compounds are human carcinogens (Group 1).Thus, inorganic arsenic compounds, including arsenic trioxide, arsenite, and arsenate, can cause cancer of the lung, urinary bladder, and skin (IARC, 2012).Exposure to arsenic and inorganic arsenic compounds is associated with cancer of the kidney, liver, and prostate (IARC, 2012).In addition, several studies have reported that arsenic exposure can cause respiratory irritation and nausea and affect the skin and neurological system (ATSDR, 2007;Hughes et al., 2011).Arsenic is not an essential mineral in humans, and no deficiency disorders have been reported.Some studies have reported that compared with cells of animal origin, human cells are more sensitive to arsenic (Liu et al., 2001;Romach et al., 2000).
Although most fabrication processes are automated, preventive maintenance (PM) tasks, such as cleaning the ion source and beam path, and troubleshooting of facilities, are performed manually by workers (Ungers and Jones, 1986).PM workers are assumed to have a higher exposure to hazardous agents than workers in other occupational settings (De Peyster and Silvers, 1995;Park et al., 2010;Yoon, 2012).During PM tasks, debris may be dispersed in the air and workers can be exposed through inhalation (Choi et al., 2015).Arsenic exposure is a site-specific health hazard for PM workers assigned to cleaning ion implanters because of the use of arsine gas (Ungers and Jones, 1986;De Peyster and Silvers, 1995).Few studies have investigated the level of arsenic present during PM tasks in the semiconductor industry (Park et al., 2010).Therefore, this study investigated workers' exposure to arsenic during ion implanter PM tasks.

General Characteristics of Investigated Tasks
We investigated PM tasks performed from February to September 2015 in a semiconductor manufacturing company.The sampling site was the ion implantation process in a fabrication facility for the production of 200-mm wafers.Maintenance tasks consisted of PM and breakdown maintenance (BM) tasks.PM refers to maintenance work that is regularly performed on a piece of equipment to reduce its possibility of failure.BM is unplanned maintenance performed after an equipment has failed in order to return it to a functional state following a malfunction or shutdown.All maintenance tasks were manually conducted by subcontracted workers according to a prescheduled plan.These maintenance workers operated according to a threeteam, three-shift, continuous full-day shift work system.Two workers operated as a team to conduct maintenance work on each piece of equipment, with additional workers occasionally joining the teams as assistants.A standard PM operation procedure was as follows: (1) a preparation step of pre-purging with nitrogen gas, cooling, and using a local exhaust fan to remove the remaining gas from every part of the equipment; (2) dismantling parts of the equipment for cleaning; (3) cleaning or preparing pre-cleaned parts; and 4) reassembling equipment and testing for the next run.The first stage of the PM was operated with a remote switch control system, with no workers present.Because PM workers could be exposed to arsenic dust during tasks, such as dismantling, cleaning, and reassembling, we investigated steps 2 to 4, which took approximately 2 hours.
Table 1 lists the general characteristics of the eight tasks that were investigated.PM Tasks 1-7 were performed in the service area of the ion implantation process, whereas Task 8 was conducted in the cleaning room, located away from the fabrication facility.Three types of implanter equipment were investigated: three ion implanters with a medium current (Tasks 1, 2, and 3); two ion implanters with a high current (Tasks 4 and 5); and two high energy ion implanters (Tasks 6 and 7).Two workers dismantled the equipment and cleaned it with compressed air and an isopropyl alcohol (IPA)-soaked pad for Tasks 2 and 3.After cleaning, workers reassembled the equipment.During Tasks 2 and 3, one worker performed PM tasks inside the equipment, whereas the other worked performed PM tasks around the equipment.For Task 4, workers detached various parts from the main body and cleaned solid debris from both the parts and the main body with vacuum cleaners.After cleaning, new or precleaned parts from outside were integrated into the main body of the equipment.Task 5 was monthly PM, which involved dismantling and cleaning with an IPA-soaked pad.We collected bulk and air samples during the PM tasks.We tried to compare between bulk and air samples; however, this was challenging because there were difficulties in accessing the workplace, and PM tasks were performed irregularly because the predetermined schedule changed frequently depending on the production schedule.

Bulk Sampling
Eight bulk samples were collected from five PM tasks.One of the samples was collected by aggressive sampling by using a high-volume pump for 1 minute from inside the beam line source chamber (Bulk 4-2).The other samples were collected by researchers during PM tasks from the Cleaning in the bead room inside of the chamber and around the implanter equipment.
The researchers put the samples into precleaned glass bottles.Detailed descriptions of the bulk samples are listed in Tables 1 and 2. The bulk sample from Task 1 was a used cleaning pad, and the bulk sample from Task 4 was debris collected from the beam line.During Task 4, aggressive sampling was performed.During Task 5, debris bulk samples were collected from a mat used for maintaining the beam line and from the surface of the source chamber.During Task 6, two samples were collected.One was debris from the surface of the source chamber after cleaning, and the other was a cleaning pad used during the maintenance.Graphite particles, one of the inner components of an ion implanter, were collected from the cleaning chamber with a vacuum cleaner during Task 8.In the cleaning chamber, the parts of the implanter were transferred and cleaned by PM workers.

Air Sampling
Airborne particles were sampled using 5-µm PVC (Poly vinyl chloride) filters (SKC Inc., Eighty-Four, PA, USA) in a closed-faced three-piece cassette for inductively coupled plasma-mass spectrophotometer (ICP-MS) analysis for personal and area samples and 0.4-µm PC (Polycarbonate) filters (SKC Inc.) with an open-face cassette for scanning electron microscope-energy dispersive spectrometer (SEM-EDX) analysis for Task 3. In total, 14 samples were collected.Among them, nine were personal and five were area samples.For personal sampling, the sampling media was attached within the breathing zone of workers, and pump flow rate was maintained at 2 liters per minute (L min -1 ).Area sampling was performed near the source of emission or at the height of workers' breathing zone at a flow rate of 2 L min -1 .Filters were pre-and post-equilibrated before weighing in an environmentally controlled weighing room that was maintained at a temperature of 20°C ± 1°C and a relative humidity of 50% ± 5%.After sampling, the cassettes were sealed tightly by using a silicon tape and transported in a clean box.Weighing was performed using a high-speed microbalance with a readability of 1 µg (XP6; Mettler Toledo, Columbus, OH, USA).For SEM analysis, the open-face sampling technique was used (Ham et al., 2015).The PC filter for SEM analysis was cut randomly because it was assumed that particles were evenly distributed on the filter.Sampling was performed throughout each PM task.

Chemical and Microscopic Analysis
We used two separate instruments to identify and quantify the metal composition in the air and bulk samples: an ICP-MS (NexION 350; Perkin Elmer Co., Waltham, MA, USA) and a SEM-EDX.For bulk samples, we expressed the concentration as % w/w (by weight per weight), except for the one, which was obtained by aggressive sampling (µg/sample).Airborne metals were quantitated using ICP-MS with an S10 auto sampler.The size, morphology, and elemental composition of airborne particles were determined using FE-SEM (MERIN COMPACT; Zeiss, Jena, Germany) with an EDX (NORAN system 7; Thermo Scientific, Waltham, MA, USA) and a normal SEM (JEM-6360; JEOL, Tokyo, Japan) with EDX (INCA 7582; Oxford, U.K.) for bulk samples.The filters were platinum coated for electron microscope analysis.

Statistical Analysis
Airborne arsenic concentrations between the personal and area samples were compared using a t-test, and differences among the types of implanters (medium current, high current, and high energy) were examined using analysis of variance at a significance level of 0.05.We confirmed the log-normality of airborne exposure data by using the Shapiro-Wilk test, and all data were log-transformed before the analysis.Data analysis was performed using SPSS for Windows (Version 20.0;IBM Corp, Armonk, NY, USA), and graphs were plotted using Sigmaplot (Version 12.5; Systat Software Inc., Chicago, IL, USA).

RESULTS
The arsenic concentration and chemical composition of arsenic in bulk samples during PM tasks at the ion implanter are listed in Table 2.The highest concentration (80.60%, w/w) was observed in the implanter source chamber after the cleaning process during Task 5.The debris from the implanter beamline had 18.84% w/w of arsenic during Task 4.
Fig. 1 shows the results of the SEM-EDX analysis in bulk samples from PM Tasks 4 and 5. Fig. 1(a) shows particles composed of arsenic and tungsten, and the samples from the implanter beamline surface (Bulk 4-1).Fig. 1(b) shows the debris from inside the implanter beamline, which was also composed of arsenic and tungsten (Bulk 5-1).Arsenic, tungsten, and aluminum were identified in the debris of the implanter source chamber in Fig. 1(c) (Bulk 5-2).
The ICP-MS and SEM-EDX results for airborne samples are shown in Table 3. Airborne arsenic was found in the area samples as well as in personal samples during the PM tasks.Among all the personal samples, the level of arsenic was the highest (2.39 µg m -3 ) in the medium-current ion implanter personal sample (Air 3-2) during PM Task 3. The area sample for Task 3 had a higher arsenic content (1.79 µg m -3 ) than did the other area samples.Airborne arsenic was found (2.30 µg m -3 ) in the personal sample collected during the medium-current ion implanter PM Task 2 (Air 2-2: chamber cleaning).Other elements, such as gold, aluminum, copper, lead, chromium, antimony, manganese, and titanium, were also detected; however, their airborne concentrations were much lower than their occupational exposure limits.For example, the highest airborne lead concentration during PM work was 0.34 µg m -3 in PM Task 4 (beam source cleaning and replacement) but much lower than 50 µg m -3 , which is the threshold limit value-time weighted average (TLV-TWA) of the American Conference of Governmental Industrial Hygienists (ACGIH).
The shape of the particles and their chemical composition during Task 3 were determined through SEM-EDX, with the results presented in Fig. 2.There was a range of particle sizes on the filter.Fig. 2(a) shows particles composed of tungsten.Particles were aggregated in sphere shapes with a size ranging from 0.3 µm to 1 µm.Amorphous particles were also shown.Arsenic particles were identified in personal and  The results indicate that the particles were mainly composed of arsenic, aluminum, and lead.The particle shown in Fig. 2(a) is 1.7 µm in length and 1 µm in width with an amorphous shape.In Fig. 2(c), the particle is in three parts, with the particle in the upper part of image being 6 µm long and 4.4 µm wide.Fig. 3 compares the airborne arsenic concentrations by the type of sample and implanter.The bar graph shows the geometric mean, and the error bars indicate the minimum and maximum values of the airborne arsenic concentration.The personal airborne arsenic concentration was higher than that of the area samples, but the difference was not statistically significant (p = 0.85).However, there was a significant difference between the values for the medium-current, highcurrent, and high-energy implanters (p < 0.05).
Fig. S1(a) shows a worker with a portable local exhaust ventilation system and personal protective equipment (PPE), such as goggles, attached to an airline respirator during Task 5. Fig. S1(b) shows the collection of debris from Task 5 (Bulk 5-1).A worker cleaned the equipment with a cloth (see supplementary material in the online edition).
Fig. S2(a) shows the beam source chamber and the debris remaining on the surface of the chamber (Bulk 6-1).

DISCUSSION
This study investigated the level of arsenic and other metals in airborne and bulk samples during ion implanter PM tasks using ICP-MS and SEM-EDX.High arsenic levels were found in several airborne samples, and particle characteristics, including size and morphology, were identified in both air and bulk samples.
Our study results revealed that worker exposed to arsenic during PM tasks in the ion implantation process.PM workers typically perform tasks including disintegration, cleaning, and reassembling.We found the highest level of arsenic (2.39 µg m -3 ) in airborne samples during the cleaning of the medium current ion implanter, implying that workers are exposed to arsenic during manual PM tasks.The two PM workers from Tasks 2 and 3 were exposed to arsenic levels higher than the recommended exposure limits-short term exposure limit (REL-STEL: 2 µg m -3 ) recommended by the National Institute for Occupational Safety and Health (NIOSH) but lower than the TLV-TWA (10 µg m -3 ) of the ACGIH (ACGIH 2015;NIOSH 2016).According to another study, maintenance workers have a higher potential for occupational exposure than operators in charge of routing production workers from thin-film processes in the solar cell production industry; for full-shift sampling, the maximum concentration was 0.13 µg m -3 for maintenance workers and 0.01 µg m -3 for laboratory workers (Spinazzè et al., 2015).area sample from around worker 2.

μm μm
Before ion implantation was developed in the early 1970s, a diffusion process was used to dope arsenic to silicon.However, the widespread adoption of arsenic deposition by diffusion was delayed because of safety issues.Because of the presence of arsenic, it was essential for an exhaust system to be established.Ion implantation is currently the most common technique used for arsenic doping (Jaeger, 2002).From the perspective of risk management, the elimination or substitution of arsenic by a less toxic material could be considered; however, it is not presently viable because of chip conformity requirements.Therefore, measures should be implemented, such as purging with N 2 gas before PM work; using local exhaust ventilation with a HEPA filter (i.e., a portable fume extractor with a HEPA filter); additional gloves to prohibit skin contact; installing air supply breathing apparatus; designating PM work as high risk and giving PM workers health and safety priority; and regular air and health monitoring.In addition, a wet cleaning technique can be used for PM tasks, involving a wet cloth with a low toxicity solvent (e.g., housekeeping cleaning agents).In this study, some workers used a dry cloth to remove dust and debris from the surface and floor, whereas other workers used a wet cloth with a cleaning agent.When arsenic-contaminated parts were kept wet, the geometric mean exposure level was found to be 1.4 µg m -3 compared with the 53 µg m -3 recorded during dry cleaning (Baldwin, 1988).This finding suggests that the wet cleaning technique can reduce exposure.
Other control methods include appropriate exhaust ventilation, enclosure of tasks, and wearing adequate PPE during PM tasks (NIOSH, 1981;HSE, 2013).During Task 5 [Fig.S1(a)], workers used portable local exhaust ventilation equipment and full-face respirators.However, PPE should be the final option among the various control methods.Until 2010, workers wore only cleanroom face masks rather than an appropriate respirator for breathing zone protection during PM tasks.Since 2010, more appropriate full-face air-purifying respirators have been provided.In 2013, these were replaced with clean-air-supplying respirators for main PM workers.Other workers still wear a full-face respirator equipped with multipurpose vapor cartridges or canisters.However, this cartridge-type filter is insufficient against particulate matter.Therefore, a combination of a cartridge and a filter respirator should be worn during PM tasks to prevent exposure to gaseous and particulate matter.Task 8 involved the collection of graphite particles from implanter graphite in the cleaning booth [Fig.S3(a)]; the worker ground the graphite particles with an abrasive pad, and particles were collected from a vacuum cleaner filter [Fig.S3(b)].
Arsine gas is the source of arsenic in the ion implantation process.Arsine gas is highly toxic and is easily converted to arsenic when heat is applied.The inhalation of arsine gas has separate effects on the body than arsenic and its compounds (HSE 2013).According to Park et al. (2010), arsenic is generated during the implanting process in the semiconductor industry; they reviewed that the arsenic level (weighted arithmetic mean, 7.8 µg m -3 ) among PM workers was higher than that among operating workers (1.6 µg m -3 ).However, the latest study in their review was published in 2007 and does not contain recent data, whereas the techniques and equipment in the semiconductor industry have changed considerably since then.The airborne arsenic concentrations varied significantly according to the type of implanter used.The airborne arsenic concentration during PM for a mediumcurrent implanter was higher than that for high-current and high-energy implanters (Fig. 3).
Three months before this study, the health and safety policy of the company was strengthened because a fatal accident occurred in which three workers died.A new standard operating procedure was established for each task.
For example, all PM work is now undertaken under the inspection of an in situ supervisor, after a preview of the PM procedure and designation of specific personnel to enter the PM working area.The supervisor remains in place during the PM work and has the authority to permit and inspect all PM work.This indicates that the arsenic levels reported here may be lower than those reported in the previous year or at other companies.The occupational arsenic exposure in the semiconductor industry was previously reported to be 7.8 µg m -3 during maintenance tasks (Ungers and Jones, 1986).The measurement strategy should be divided, accounting for the exposure status of operators and PM workers separately, because PM workers are likely to be exposed to higher concentrations (Yoon, 2012).
In this study, to analyze the metal components of bulk and air samples, we used both ICP-MS and SEM-EDX.Both methods have advantages and disadvantages, and these methods were used together to obtain more complete information (Haley et al., 2006;Choi and An, 2016).
As shown in Table 3, SEM-EDX could detect relatively high levels of arsenic but could not detect low levels.SEM-EDX is a relatively simple procedure for the on-site analysis of particles and is a non-destructive analytical method for elemental analysis, with a detection limit of 0.1-0.5 wt.% (Haley et al., 2006).Other studies have suggested that SEM-EDX can be used accurately and precisely for the direct analysis (physical form and chemical composition) of powder samples consisting of metal alloys and oxidation products (Einhäuser, 1997).ICP-MS has a very low limit of detection; therefore, it detected much lower levels of the metals studied here.ICP-MS can be used for both qualitative and quantitative analysis, whereas SEM-EDX can only perform a qualitative and semi-quantitative analysis because of the image resolution.
This study provides crucial scientific evidence of the potential exposure to arsenic that PM workers face in the semiconductor industry.This industry is changing rapidly.Furthermore, it is challenging to investigate arsenic exposure because of the high secrecy inherent in this industry.Moreover, PM tasks are considered an irregular job and are likely to be missed during exposure monitoring.
This study had several limitations.We investigated each PM task only once because of the irregular working schedule and duration of the tasks and because of the difficulty in accessing the work area during each task.The results are still valuable because no other study has investigated implanter PM tasks and the subsequent arsenic exposure.Nevertheless, we confirmed that the arsenic in both air and bulk samples from PM work in ion implanters were above the occupational exposure limit.
Another limitation is that we only measured air and bulk samples without obtaining biomarker measurements.Arsenic compounds may be absorbed after ingestion, inhalation, or skin contact and are excreted in urine.Biological monitoring is useful to interpret and understand exposure to arsenic.Previous biological monitoring studies have shown that As 3+ , As 5+ , the sum of inorganic arsenic (As 3+ + As 5+ ), and the sum of inorganic arsenic and monomethylarsonic acid (MMA) were higher in ion implanter PM engineers than in a non-exposed group after adjusting for smoking and seafood intake (Byun et al., 2013).Monitoring urinary arsenic by using the percentage change of MMA in total urinary inorganic arsenic metabolites as an indicator for the verification of arsenic exposure is helpful and appropriate for ion implanter PM workers (Hwang et al., 2002).A case control study showed that the mean urinary concentration of total arsenic was significantly higher in arsenic-exposed workers than in non-exposed workers (Hu et al., 2006).The levels of arsenic in the air and bulk samples, recorded together with a matched biomonitoring examination, could furnish an appropriate indicator of exposure.

CONCLUSION
This study found that PM workers may be exposed to arsenic and other metals during implanter PM tasks in the semiconductor manufacturing industry.Airborne arsenic levels were significantly higher during PM work for mediumcurrent implanters than for high-current and high-energy implanters (p < 0.05), as demonstrated by the arsenic content in bulk samples and SEM images.Other metals, including aluminum, tungsten, lead, titanium and chrome were found, but their levels were much lower than their occupational exposure limits.Appropriate control methods to prevent worker exposure to arsenic are required, and it is necessary to further investigate possible exposure during implanter PM tasks, as well as in other processes.
Fig. S2(b) shows a cleaning pad collected from the high energy implanter during the beam source cleaning task (Bulk 6-2).Fig. S3(a) shows the cleaning booth, in which a vacuum cleaner was used to clean the graphite particles generated by grinding.Fig. S3(b) shows the sample collected by the vacuum cleaner during Task 8 (Bulk 8-1).

Fig. 3 .
Fig. 3. Comparison of airborne arsenic concentrations by type of sample and implanter (The bar graph shows the geometric mean and the error bar indicates the minimum and maximum values of the airborne arsenic concentration).

Table 1 .
General characteristics of the tasks investigated in the ion implantation process.

Table 3 .
Concentration of elements in air samples.