Annals of Occupational Hygiene Advance Access originally published online on February 5, 2009
Annals of Occupational Hygiene 2009 53(2):99-116; doi:10.1093/annhyg/mep001
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
A Headset-Mounted Mini Sampler for Measuring Exposure to Welding Aerosol in the Breathing Zone
1 Department of Applied Environmental Science, Stockholm University, SE-106 91 Stockholm, Sweden
2 Swedish Work Environment Authority, Division of Chemical, Microbiological and Physical Factors, SE-112 79, Stockholm, Sweden
* Author to whom correspondence should be addressed. e-mail: goran.liden{at}itm.su.se
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONCONCLUSIONSFUNDINGACKNOWLEDGEMENTSREFERENCES |
|---|
There is a need for a small personal aerosol sampler for measuring occupational exposure to airborne particles in the breathing zone. Existing aerosol samplers are too large to be mounted inside modern welder's protective equipment without disturbing the worker. A headset-mounted mini sampler has been developed to fill this gap with focus on manganese exposure. This mini sampler is easy to use and can be mounted inside modern, slimline welder's face shield. The mini sampler is based on a commercially available 13-mm filter holder that has been modified to incorporate an inlet nozzle made of aluminium. The nominal flow rate of the mini sampler is 0.75 l min–1. The mini sampler is to be worn mounted on a headset, modified from professional microphone headsets. Several aspects related to using the mini sampler have been tested by personal and static sampling at five workplaces and in the laboratory. Four headset models were tested for their suitability as a sampler holder, of which three models were accepted by the welders. The sampling bias of the mini sampler versus the IOM sampler and the open-face 25-mm filter holder, respectively, depends on the size distribution of the sampled aerosol. At the lowest encountered mass concentration ratio of the open-face 25-mm filter holder to the IOM sampler (0.65), the sampling bias of the mini sampler versus the IOM sampler is approximately –26% and versus the open-face 25-mm filter holder is approximately +12%. For manganese, the negative root mean square (RMS) sampling bias of the mini sampler versus the IOM sampler is –0.046 and versus the open-face 25-mm filter holder is non-significant. Both these biases are statistically non-significant. The mini sampler can therefore be employed for determining welders’ occupational exposure to manganese. The pressure drop across the filter can become excessive due to the small filtration area. Compared to the Casella Apex pump, the SKC AirChek2000 pump was generally found to be able to keep its flow rate constant within ±5% at higher concentrations and for longer sampling times. Our results indicate that the inhalable fraction of the welding aerosol mass at the visited plants only consisted of 25–55% welding fume particles (agglomerates of coagulated particles generated by nucleation/condensation). The rest of the mass is made up of particles from spattering and grinding. More than 65% of manganese is generally found in the fume particles. The weighing precision of 13-mm filters is 2.2 µg. The RMS sample loss due to transport when loaded samples are shipped by mail in padded envelopes is 6 µg. Both figures are very low in comparison to the mass expected to be collected by personal sampling, generally exceeding 200 µg. The headset-mounted mini sampler is user-friendly, easy to adjust individually, does not disturb the welder during sampling and allows sampling inside personal protective equipment. The headset mounting arrangement improves personal sampling as it maintains the sampler close to the nose/mouth during the whole sampling period. This study shows that the developed headset-mounted mini sampler is suitable for assessing exposure to manganese in welding aerosol.
Keywords: aerosol • exposure • fume • manganese • sampling • welding
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONCONCLUSIONSFUNDINGACKNOWLEDGEMENTSREFERENCES |
|---|
The sampling position plays a key role when measuring a welder's occupational exposure to airborne particulates as the concentration inside the plume is 10–100 times higher than outside. This paper describes the development of a sampling method that should be more representative of the welder's exposure than traditional sampling on the lapel, especially for manganese.
The breathing zone
A central concept in occupational exposure assessment related to personal sampling for chemical agents is the ‘breathing zone’. This concept has many definitions and they are all similar in content. For example, Occupational Safety and Health Administration (OSHA) defines it as
Air that would most nearly represent that inhaled by an employee. (OSHA, 1999)
A reference book on occupational hygiene by the American Industrial Hygiene Association (AIHA) defines it as
The volume surrounding a worker's nose and mouth from which he or she draws breathing air over the course of a work period. This zone can be pictured by inscribing a sphere with a radius of10 inches centred at the worker's nose. (Dinardi, 1997)
The European Committee for Standardization (CEN) defines the breathing zone very explicitly as
The space around the worker's face from where he takes his breath. For technical purposes a more precise definition is as follows: hemisphere (generally accepted to be 0.3 m in radius) extending in front of the human face, centred on the mid point of a line joining the ears; the base of the hemisphere is the plane through this line, the top of the head and the larynx. The definition is not applicable when respiratory protective equipment is used. (CEN, 1998)
These (and other) definitions are all technical definitions, and as such not unreasonable. The definitions are not based on science, i.e. a demarcated space in front of the worker inside which the concentration has been shown to be constant. The purpose of these definitions is instead to somewhat harmonize how far away from the nose/mouth of a worker samplers may be mounted and still be considered to be at a valid position. One of the objects of the definition of the breathing zone has been to give a rationale for mounting samplers on the lapel/collar, which today is the by far preferred sampler mounting position for personal sampling.
However, according to current knowledge, the concentration is not constant within the breathing zone when the worker is very close to the source or handles the source. In experiments with particles of various sizes, gases and vapours, performed both at workplaces and in laboratories, with real persons or (breathing) mannequins, several studies (Martinelli et al., 1983; Bull et al., 1987; Parker et al., 1990; Steel and Maican, 1994; Brohus, 1997; Rosén et al., 1997; Malek et al., 1999; Welling et al., 2000; Guffey et al., 2001) have shown that the difference between the concentration measured close to the nose/mouth and on the lapel or chest could be anything from insignificant to two to four times larger when the worker is in the near-field of the source. The situation is different when the worker is in the far-field from the source. For this situation, Cohen et al. (1982) and Vinson et al. (2007) have shown that the concentrations of the respirable fraction measured simultaneously at the nose and the lapel are identical, within the uncertainty of the measurements. Additionally, Bohne and Cohen (1985) and Cohen and Positano (1986) have shown that resuspension of dust from work clothing may be due to the workers movement and thus systematically cause the concentration measured at the lapel to be higher than at the nose/mouth. This scientific knowledge has demonstrated that workplaces exist where one could expect to find very significant differences between what a worker is exposed to and what is sampled by a personal sampler mounted on the lapel or chest. Occupational hygienists will presumably anyhow continue to use the concept of the breathing zone when it comes to mounting samplers. The main reason for this is that so far nothing better has been proposed. Possibly future revisions of the concept of breathing zone may decrease its range from the present radius of 25–30 cm. However, none of the samplers suggested to date for sampling closer to the nose/mouth (see below) has become so widely used that a reduction in the size of the breathing zone has been discussed within the occupational hygiene community.
Some previous suggestions for mounting samplers close to the nose/mouth
Bloor et al. (1968) proposed that an HASL (the Health and Safety Laboratory of the US Atomic Energy Commission) half-inch cyclone with a 47-mm filter should be mounted on the forehead and held in place with a headband. ter Kuile (1982) designed a special helmet in which an air stream was aspirated by a fan into the opening between the worker's forehead/top of the head and the helmet. Into this air stream was inserted a tube leading to a filter holder on top of the helmet. Chien (1992) proposed an ear-mounted tube with sampling close to the nose using a lightweight sampling medium. None of these three mounting designs seems to have been practical and user-friendly enough to have led to any extended use.
Allen et al. (1981) designed a mounting system consisting of a (closed-face) 13-mm Swinnex filter holder mounted on a bended tube attached to the headband of a worker's helmet. This design was later modified by the UK Health and Safety Laboratory for its sampling method for resin acids in solder fume (Pengelly et al., 1994; HSE, 2006). In this method, the Swinnex filter holder was attached to the sidearm of a pair of safety spectacles. In a later investigation, Simpson (2005) compared two different mounting positions and found that a filter holder mounted on spectacles sampled 50% more than a filter holder mounted on the microphone support arm of a telephone headset.
Characteristics of the welding aerosol
The particle size distribution of aerosols generated by a welder performing both welding and grinding consists of three modes. The smallest mode, with a mass median aerodynamic diameter (MMAD) usually <1 µm, consists of agglomerates of original particles made up of nucleated oxidized metal vapours (evaporated at the weld) that have grown by diffusion. This is the welding ‘fume’ from a physics point of view, though an international standard on the sampling of airborne particles during welding (ISO 10882-1) specifically defines ‘welding fume’ as ‘(the) airborne particles generated during welding’ (2001). Due to the lower boiling point of manganese (1962°C) compared to iron (2750°C), manganese is four to six times more abundant in the fume than in the filler material (Worobiec et al., 2007). Manganese is mainly found in the divalent and trivalent forms, partly combined (Mn3O4), and sometimes in combination with iron as a spinel (MnFe2O4) (Jenkins and Eagar, 2005; Marcy and Drake, 2007). The next largest size mode consists of spatter particles. These are ‘spherical’ particles consisting of solidified molten metal droplets ejected out of the weld, which therefore contain a high fraction of non-oxidized metals (Jenkins and Eagar, 2005; Worobiec et al., 2007). Jankovic et al. (1999), Zimmer et al. (2002) and Stephenson et al. (2003) have measured the size distribution of the spatter particles generated during tungsten inert gas welding, gas metal arc weldingand shielded metal arc welding, respectively. Recalculation of the measured number particle size distributions resulted in modes of 13, 9 and 6 µm, respectively, for the spatter particle mode. If grinding or slagging is performed at the workplace, a third mode consists of grinding/slag particles. These particles are mechanically generated and half is made up from (non-oxidized) material of the weld and half from the grinding wheel (Koponen et al., 1981; Kalliomäki et al., 1982). Chung et al. (1999) measured the size distribution of grinding aerosol at a simulated workplace with a cascade impactor and found an MMAD of 17 µm. In addition, individual spatter particles may be up to several hundred micrometres in diameter.
Breathing zone sampling during welding
For a welder using a welder's face shield, both air inside the face shield and in front of the lapel/chest is within the range of the technical definition of the breathing zone as, for example, stated by OSHA (1999) and CEN (1998) above. From a formal point of view, mounting a sampler on the lapel/chest should be considered to be in accordance with the definition of the breathing zone. However, in order to decrease the amount of welding fume inhaled by welders, modern close-fitting welder's face shields allow inhaled air to enter preferentially from the side of the welder's head, i.e. further away from the plume of welding fume. Therefore, the general definition of the breathing zone (as exemplified above) is not applicable to welders using modern face shields with the visor in the down position. An international standard on the sampling of airborne particles during welding, ISO (2001a) explicitly defines the breathing zone as being behind the welder's face shield, when worn. This standard presents some examples on how to mount 37-mm filter holders and an HSE modification of the IOM sampler (Chung et al., 1997) behind the old, large version of a welder's face shield (with ample space between it and the head).
There are only a few published reports (since the mid-seventies) presenting data for comparing concentrations inside and outside of a welder's face shield. Goller and Paik (1985) sampled welding aerosol iron oxide with closed-face 37-mm filter holders at four positions within the breathing zone of eight welders in a locomotive manufacture shop. One position was inside the welder's face shield, while the three others were outside. The concentrations sampled at the positions outside the welder's face shield were all significantly higher (on average 33–40%) than the concentrations inside. Liu et al. (1995) sampled with 37-mm filter holders at two positions (inside the welder's face shield and on the collar near the lower edge of the face shield) on 16 welders. The experiment was performed in a simulated workplace with good ventilation. No statistically significant difference was found between the two mounting positions. Harris et al. (2005) sampled with 37-mm filter holders at two positions (inside the welder's face shield and on the collar) on one welder who in a booth welded with three types of electrodes in five ventilation rates. The fume concentration outside the face shield was significantly higher (13%) than inside. Possibly the different findings regarding whether the concentration sampled behind the welder's face shield is smaller or similar to the concentration sampled outside it were influenced by the different experimental set-ups, ranging from a construction hall over a welding booth to an exposure chamber.
Chung et al. (1999) investigated the relationship between the welding aerosol concentrations determined when sampling with five samplers (a UK open-face 37-mm filter holder, a closed-face 37-mm filter holder, a closed-face 37-mm filter holder with the entry nozzle enlarged for an entry speed of 1.25 m s–1, an HSE modification of the IOM sampler with a polyurethane foam insert and a GSP sampler) mounted behind a welders face shield and the concentration of welding aerosol inhaled by a breathing (inhalation only) mannequin. The GSP was mounted under the chin, whereas for the others, one sampler was mounted on each side of the face. Tests were performed for five kinds of welding and the results showed a slight oversampling by the samplers relative the inhaling mannequin: The average ratios of the sampler concentration to the concentration inhaled by the mannequin ranged 1.07–1.13, with standard deviations in the range 0.10–0.16. The standard deviation of an individual concentration ratio ranged 0.10–0.23, slightly less than the difference between the right and the left side of the head. This implies that in the confined space behind the welder's face shield, the concentrations of welding aerosol determined by all samplers were (on average) the same, but the position of the sampler will influence the actually measured concentration.
Swedish halving of the Occupational Exposure Limits for manganese
In Sweden, the occupational exposure limits (OELs) for manganese, both the fractions sampled as and as respirable dust, were decreased by 50% and the new OELs became effective 1 January 2007 (SWEA, 2005). Welders were expected to be the only section of the workforce who might be significantly exposed above the new OELs. The reduction in the new OELs was considered to be a major reduction possibly necessitating expensive technological improvements in order to control exposure to manganese during welding. This was the motivation for developing a sampling method that leads to more representative measurement of the concentration of welding aerosol and, in particular, manganese to which a welder is exposed, i.e. sampling behind the welder's face shield, closer to the mouth/nose.
The new sampling method is based on a modified 13-mm Swinnex filter holder mounted on a headset with a beam—termed a headset-mounted mini sampler for welding aerosol. The headsets concerned are commercially available professional headset microphones with the microphone and electronics stripped away. This paper presents the workplace and laboratory tests performed to evaluate the performance of a modified 13-mm Swinnex filter holder and the experience gained from using it for sampling from behind a welder's face shield during personal sampling at workplaces. Additionally, portable X-ray fluorescence spectrometers have been calibrated and used for measurement of manganese and iron in welding aerosol sampled on 13-mm filters. This work will be presented elsewhere (G. Lidén and L. Lundgren, in preparation), as will a pump performance test for the sampling of welding fume (G. Lidén and S. Lundström, in preparation).
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONCONCLUSIONSFUNDINGACKNOWLEDGEMENTSREFERENCES |
|---|
The collection efficiency of the mini sampler was determined by static parallel sampling with the IOM sampler and an open-face 25-mm filter holder at five welding plants. The practical usability of the headset-mounted mini sampler was evaluated by personal sampling at three welding plants.
The mini sampler
The mini sampler is based on the Millipore (Billerica, MA, USA) Swinnex filter holder made from polypropylene for 13-mm filters. The filter holder was modified from being a closed-face filter holder into being an open-face filter holder by incorporating an entry nozzle in aluminium with a diameter of 10 mm and a length of 9 mm, of which 7 mm protrudes out of the front of the filter holder. The Swinnex silicone gasket could not be used with mixed cellulose filters as filter damage was found to often occur. The problem was solved by replacing it with a PTFE-coated O-ring (Fig. 1). In the work reported here, Millipore 13-mm mixed cellulose ester (MCE) filters with a nominal pore size of 8 µm (SCWP) have been used. The diameter of the filter deposit is 10.4 mm. The nominal flow rate was 0.75 l min–1, generally as low as the pumps used were able to operate for 6–8 h. For flow rate measurement an adapter, consisting of a piece of latex tubing with 10 mm inner diameter, was used.
|
The separation efficiency of the filter was estimated from the data by Liu et al. (1983). At the nominal flow rate, at which the filtration velocity is
15 cm s–1, the lowest separation efficiency of the SCWP filter was estimated to be 99% for 100 nm particles. The filtration efficiency will successively increase when welding aerosol collected on the filter builds up a so-called ‘filter cake’.
Headset mounting systems
Four different commercial headsets with professional microphones, intended for use by musicians, TV commentators, etc. were bought; IsoMax HH (Countryman), WH20QTR (Shure), HT2 (AUDIX) and C420IIIL (AKG) (The AKG C420IIIL headset is no longer commercially available at the time of writing.) These models were all hung over the ears with a headband behind the neck, but not over the top of the head, and with a beam holding the microphone close to the nose/mouth. The headband behind the neck was needed to increase the vertical and lateral stability of the beam. The selected headset models had different means to adjust the headset to fit the size and shape of the head and ears of the wearer and to angle the beam close to the nose/mouth. The IsoMax HH and the C420IIIL were easiest to fit to a wearer. The width of the headband behind the neck could be changed for both of these headsets. Additionally, both the angle and height of the earpiece could be changed on the C420IIIL. The behind-the-neck headband of the WH20QTR was rigid and difficult to adjust to fit the individuals well. The HT2 had a thick lining of rubber around its mounting which made it a little too thick to fit between the ear and the head. All headsets, except the WH20QTR, had the microphone beam on the left side.
The headsets were stripped of the microphone and the electrical cables. Polyvinyl chloride (PVC) tubing (inner diameter 3 mm) was fastened with shrinking tubing onto the beam of each headset. This tubing fits well with the male Luer Slip connector of the Swinnex filter holder. The other end of the PVC tubing was connected to the latex tubing ordinarily used in workplace aerosol sampling. As an example of the headsets that were tested by welders, the AKG C420IIIL modified headset is shown in Fig. 2.
|
Mounting the headset
Sampling for welding aerosol was performed with the mini sampler without using any harness for mounting the sampler, tubing and pump. Before the headset was mounted on the welder, he/she was informed how to mount, wear and dismount it, as well as how to adjust it, so that he/she could do it single-handed if necessary. Once the headset with the mini sampler was mounted and the welder's face shield donned, the welder was asked to raise the protective glass of the visor in order to see the mini sampler in position relative to the mouth/nose. It was important that the mini sampler was not mounted directly in front of the mouth/nose, where the welder might easily blow air direct into the sampler, but instead mounted to the side of and slightly behind the mouth/nose. If a welder's hood was used to protect the skin on the head and neck against ultraviolet radiation, the welder was especially shown how to put it on and off so that the mini sampler would not be wrongly positioned (inside the hood). Specially made clips were used to fix the rubber tubing to the welder's clothes. An ordinary waistbelt was always used to mount the sampling pump, as the belts of the fan used with air-fed welder's face shields were too wide for the sampling pump clip.
Visited plants
The mini sampler was evaluated at five manufacturing companies carrying out welding operations. The five plants were: A, a small shipyard where static sampling was performed close to a welder welding ladder rails; B, a manufacturing workshop where static sampling was performed close to a welder welding girders for a building construction; C, a manufacturing workshop for chill moulds for steelworks and subcontract manufacture of minor construction components, where static sampling was performed close to the subcontract work; D, a manufacturer of scoops for excavators, where static sampling was performed in several welding tents and E, a manufacturer of lorry trailers, where static sampling was performed in the vicinity of the primary steps in the construction of the trailers.
At plants C–E, personal sampling was also performed on a selection of the welders. Metal Inert Gas (MIG) welding and Metal Active Gas (MAG) welding were the main types of welding carried out at the plants, with very few and short-time stick welding episodes. The MIG/MAG welding consumables used contained 1.0–1.5% manganese by weight.
Aerosol sampling
The following types of aerosol sampling were performed:
- Static sampling in a rig, in order to compare the mini sampler with the IOM sampler (SKC, Eighty Four, PA, USA) and three versions of the open-face filter holder; a 25-mm filter holder in polystyrene (FH25OF), an electrically conducting 25-mm filter holder in graphite-filled polypropylene and a 37-mm filter holder in polystyrene (FH37OF).
- Static rig sampling with SIMPEDS cyclones (Casella, Bedford, UK) at the flow rates of 2.0 and 5.0 l min–1, in order to obtain information on the particle size distribution of the inhalable fraction in the range 2–5 µm.
- Separate static sampling with mini samplers, in order to obtain samples with different loading of welding aerosol.
- Personal sampling with headset-mounted mini samplers for welding aerosol.
For all samplers, except the mini sampler, MCE filters (Millipore) with a nominal pore size of 5 µm were used. For the IOM sampler, the previous version of the black sampling cassette was used and for the SIMPEDS cyclone, 37-mm filters in SIMPEDS plastic filter holders were used.
Static parallel sampling
Rig sampling.
Static rig sampling was performed on horizontal rods mounted on a vertical rod. The horizontal rods were mounted in two levels 20–25 cm apart, with three to four rods per level and the lower level 125 cm above the floor. On each horizontal rod, two samplers were mounted on each side of the vertical rod. The rig was not continuously rotated but at some workplaces it was turned at regular intervals. The total number of samplers of the different types mounted in the rig varied between plants. The samplers were distributed over the rig as uniformly as possible. At three plants, static sampling was carried out for 8 h in order to test the capacity of the pumps, whereas at other plants it was only carried out for 4 h in order to provide more samples.
Six different personal pumps were used for the static sampling, an AirChek2000 (AC2000), a PocketPump (PP), a AirChek52 (AC52), a Leyland Legacy (LL), a 224-PCXR (all by SKC) and a 400 Personal Sampling Pump (400PSP) (BGI, Waltham, MA, USA). The AC2000 was used for all static sampling with the mini sampler. The PocketPump was used for low-flow sampling with the mini sampler. The flow rates were measured with a triCal (BGI) every 2 h.
All filter samples were weighed and selected samples were additionally analysed by inductively coupled plasma–mass spectrometry (ICP–MS) and/or by portable X-ray fluorescence spectroscopy (XRFS). For each sampler type, the average and the relative standard deviation (the ratio of the standard deviation to the average) of the aerosol concentration were calculated. Individual extreme values outside the range of the other samples collected with the same sampler type for that run and possibly caused by observed and noted sampling or analytical errors (e.g. pump stop, loose tube, dropped filter, etc.) were considered to be outliers. If a pump failure occurred during sampling, but it did not result in an extreme value for the measured concentration, the sample was not considered to be an outlier. With at least four samples collected in parallel (4 or 5 IOM samplers, 5 or 6 FH25OF and 5, 6 or 12 mini samplers), the variability of the concentration across the rig was considered to be well gauged. The main object of the static sampling was to compare the average concentration of the samplers. Therefore, individual extreme samples, even though being without any associated known error, but for which the standard deviation was reduced by at least 30% upon exclusion, were also treated as outliers, even if the determined concentration at the location of the sampler might have been unbiased. All outliers were left out of the subsequent statistical analyses, based on averages per run.
Statistical Analysis of bias
The bias of one sampler (x) versus another sampler (y) was calculated according to the formula
 | (1) |
The concentration measured by the mini sampler is modelled by regression from the concentrations measured by the IOM sampler and FH25OF
 | (2) |
 | (3) |
 | (4) |
 | (5) |
For the welding aerosol mass concentration data of this project, it will be shown below that (1 –aIOM)/bFH25OH is very close to unity and that
 | (6) |
Based on this result, the corresponding model for the bias of the mini sampler versus FH25OF will be
 | (7) |
All statistical analyses were carried out with JMP 6.0 (SAS Institute, Cary, NC, USA). The determined value for the regression coefficient will be presented as value ± standard deviation.
Estimates of the size distribution of the welding aerosol sampled by static sampling.
The particle size distribution of the sampled aerosol was estimated using the three ratios of the concentrations measured by the SIMPEDS cyclone (at 2 and 5 l min–1) and the 25-mm open-face filter holder (FH25OF) to the concentration measured by the IOM sampler. For example, we know the selection curve of the cyclone at 5 l min–1, and also the selection curve in calm air of the IOM sampler, so we can find an aerodynamic diameter at which the concentration measured by the cyclone would be half the concentration measured by the IOM sampler. We can designate this as a characteristic D50 for the cyclone (at 5 l min–1)–IOM sampler pair and assign the concentration ratio of those two samplers to that D50 This is analogous to assigning to the D50 of an impactor stage the ratio of the concentration measured by that stage to the unselected concentration, although it is less exact because of the shallower selection curves.
In this way, the three concentration ratios SIMPEDS (2)/IOM, SIMPEDS (5)/IOM and FH25OF were assigned to be 1.75, 5 and 20 µm, respectively. The first two sizes were calculated from the results of Lidén and Kenny (1991), Bartley et al. (1994) and Kenny et al. (1999); no data on the aspiration efficiency of the FH25OF in calm air were available in the literature, and a ‘guesstimate’ of 20 µm was used as a representative smallest particle size associated with the third ratio. Estimating a particle size distribution in this way is similar to the parallel cyclone systems that have previously been described by Blachman and Lippmann (1974) and Chang (1974).
Thus the particle mass size distribution was estimated for all rig runs, based on all accepted samples. Filter samples from plants A-E (except for the IOM sampler, from which filter samples from plants C-E) were additionally analysed for manganese and a similar manganese size distributions were estimated for these runs. In contrast to the gravimetric analysis, the ICP–MS analysis was only performed on two filters per run and sampler. For the IOM sampler, only the filter was analysed for manganese by ICP–MS, but the estimated representative particle sizes were kept the same as for the mass size distributions. As the largest fraction of manganese is for fume particles rather than the spatter particles, it can be anticipated that the amount of manganese in the (large) particles that would deposit in the IOM inlet tube would be small in relation to the amount deposited on the filter of the IOM sampler.
Minor parallel sampling studies carried out with the rig.
The aims of three parallel exercises were to investigate the effect of three potential determinants of the sampling bias of the IOM sampler versus the open-face 25-mm filter holder, namely, electrical charge on the open-face filter holder, the size of the open-face filter holder and the possible occurrence of passive sampling by the IOM sampler: Two versions of the open-face 25-mm filter holder, manufactured from electrically insulating polystyrene and electrically conducting polypropylene, respectively, were used in parallel at plant A; 25- and 37-mm open-face filter holders manufactured in polystyrene were used in parallel at plant E and parallel sampling using IOM samplers with and without drawing air (2 l min–1) through the filter was carried out at plant B in order to study whether any significant passive sampling occurred for the IOM sampler as a result of particles being carried into it by means other than the air flow, e.g. ultralarge ‘projectiles’ by their own momentum or smaller particles by turbulent diffusion.
Personal sampling
Personal sampling with the mini sampler mounted in a headset was carried out at plants C–E in 2005–2006, i.e. before the new lower OELs for manganese were in force. In total, 43 personal samples were taken on 23 welders over 34 work shifts. No parallel sampling was carried out versus a sampler mounted on the lapel, as this would have required the welders to wear two pumps. Additionally, this sampling position was no longer considered to be unbiased and therefore the new sampling position should anyway not be compared against it.
Apex IS pumps (Casella) were used for personal sampling due to the limited number of AC2000 pumps available. During personal sampling with these pumps, the nominal flow rate was set to 0.85 l min–1, as this pump had grave problems working closer to the lower limit of its working range without stopping. A consequence of the 12% higher flow rate is that the Apex pump was working at a 25% higher pressure drop because a higher flow rate yields a thicker filter cake. The aspiration efficiency of an aerosol sampler without internal inertial separation would only be marginally influenced by a flow rate change of 12%. The flow rates were measured at approximately every 2 h with a DryCal DC-Lite (BIOS, Butler, NJ, USA). All samples were analysed by weighing (see below) and by XRFS (see G. Lidén and L. Lundgren, in preparation).
Laboratory analyses
Filter weighing.
The filters and the IOM sampling cassettes (including filters) were weighed before and after sampling in an air-conditioned room using a MT5 balance (Mettler-Toledo, Greifensee, Switzerland) at the Department of Applied Environmental Science, Stockholm University. The temperature was controlled to 21 ± 0.5°C and the relative humidity to 50 ± 2%. When handled, the filters were held using a pair of metal tweezers and electrically neutralized by 
-rays (210Po). All sample weights were blank corrected.
The pooled standard deviations for the blanks to the 25- and 37-mm filters and the IOM sampling cassette were 10 µg or less, except for the IOM cassettes used in plants C and D. Here the pooled standard deviation was inexplicably high, 35 and 55 µg, respectively. The pooled standard deviation for the blank 13-mm filters was 3.5 µg. The collected samples exceeded the limit of quantification, taken as 10 times the pooled standard deviation of the blanks, in all cases, except for the IOM sampling cassette at plants C and D. However, as only the average of the samplers in the static sampling rig runs was employed in the further evaluation and the IOM samples at plants C and D exceeded 10 times the pooled standard deviation of the average of the blanks, these samples were considered valid.
Tests were performed to determine the uncertainty of weighing 13-mm MCE filters (8 µm pore size) that had been mounted in a mini sampler for 2–3, 10 and 15 weeks. In these experiments, spread out over a period of 3 years, nine batches of ten 13-mm filters were used. The data from these tests were pooled with the data for the 13-mm blanks used with the sampling at plants A–E.
Analyses for iron and manganese by ICP–MS or XRFS.
For each run with the rig, one to two samples per cyclone flow rate and one to two filter samples each for the IOM sampler and the FH25OF (plus blanks) were selected for analysis by ICP–MS. However, for three runs up to three or five filter samples from the FH25OF and the IOM sampler were analysed by ICP–MS. No IOM filters from plants A and B were analysed by ICP–MS. From each run with the rig, one to five mini sampler filter samples were analysed by ICP. The analyses were carried out at the Department of Applied Environmental Science, Stockholm University. The filters were dissolved using a microwave oven according to Annex G of ISO (2001b) and then analysed according to ISO (2003). The blank values for manganese (<0.02 µg) and iron (<2 µg) were low and therefore blank subtraction was not applied in the subsequent analysis. The performance of the ICP–MS instrument was tested against a certified water sample (SLRS-4 Riverine water, Canadian National Research Council) and it was found to exhibit a combined measurement uncertainty for manganese and iron of 6.7 and 5.7%, respectively.
Some of the personal samples were analysed using a portable XRFS instrument, Niton XLt-700 (Thermo Scientific, Boston, MA, USA) for the purpose of calibrating this analyzer for 13-mm filter samples. The calibration of the XRFS instrument for 13-mm filters is described in detail by G. Lidén and L. Lundgren (in preparation).
Pressure drop across 13-mm filters with deposited welding aerosol.
The pressure drop across 13-mm MCE filters with 8-µm pore size and the filter cake was determined as a function of flow rate, amount of welding aerosol deposited and aerosol coarseness for each static sampling run. The pressure drop across one or two filters per run were measured with a SITRANS P (Siemens, Munich, Germany) with a precision of 5 Pa.
The total pressure drop across a loaded filter (with a specified filtration area), 
P, can be modelled as the sum of the pressure drop across the clean filter, P0,and the pressure drop across the filter cake
 | (8) |
 | (9) |
Test for transport loss of loaded 13-mm filters, mounted in mini samplers.
The loss of the deposited mass of the sampled aerosol from filters mounted in mini samplers during transport by ordinary mail was determined. Twenty samples each from plants C–E, were collected by static sampling with four different nominal flow rates (0.20, 0.75, 2.0 and 5.0 l min–1), i.e. five samples per flow rate. After sampling the mini samplers (including blanks) were transported to the laboratory, weighed and mounted in the original mini samplers again. Then the samplers were packed in a padded envelope and sent by ordinary mail a distance of 500 km and back again. Upon return, the filters were reweighed and the transport loss was determined as the change in (blank-corrected) sample weight between the first and the second weighing.
Determination of the amount of mass deposited in the inlet nozzle of the mini sampler.
The external surface of the inlet nozzle of four mini samplers used in two of the static sampling runs at plant E were cleaned, and for each run, the nozzles were put in a beaker with ethanol in an ultrasonic bath. After ultrasonification, the liquid was filtered through pre-weighed MCE filters (Millipore) and the amount of welding aerosol deposited in the inlet nozzle was determined gravimetrically by reweighing the filters after drying under an IR lamp. The values were corrected both for blanks and the part of these welding particles dissolving in ethanol, determined by weighing a part of a filter cake before and after treating it with ethanol in an ultrasonic bath.
Determination of the amount of iron and manganese deposited in the inlet nozzle of the IOM sampling cassette.
The internal surface of the inlet nozzle of all the IOM sampling cassettes used in two static sampling runs at plant E were wiped with damp MCE filters. All five wipe filters per run were put in one test tube and analysed by ICP–MS for manganese and iron. Additionally, sets of five blank damp MCE filters were also analysed.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONCONCLUSIONSFUNDINGACKNOWLEDGEMENTSREFERENCES |
|---|
Pump capacity
Of 114 samples collected with the mini sampler by static sampling using in total 24 AC2000 pump individuals, for 4.4% of the samples the flow rate changed by >5% from the initial value over the sampling period without the pump stopping and for 8.8% of the samples the pumps did stop. Three of the pumps stopped two to three times and one pump individually exhibited a reduction in flow rate exceeding 5% on four occasions. These four pumps were presumably bad individuals in need of service. Three pumps either stopped only once or only once exhibited a reduced flow rate of >5%. Generally, but not always, the AC2000 pump can withstand the encountered pressure drop across the filter of the mini sampler. All the pumps worked without problems when sampling with the FH25OF, IOM sampler or the SIMPEDS cyclone at 2 l min–1. Both the 400PSP and the LL pumps had problems sustaining a flow rate of 5 l min–1 during the whole sampling period, although for the LL pump the flow reduction was considerably smaller than for the 400PSP pump.
The Apex IS pumps did not maintain its flow rate within ±5% of the initial value for 56% of the personal samples. There was no correlation between flow rate decrease and welding aerosol mass sampled if the three samples with the highest mass loading were excluded (2.0–2.5 mg versus <1.5 mg for the other personal samples). The root mean square (RMS) flow rate deviation (excluding the three samples with the highest loading) was 6.9%.
Static sampling
Descriptive statistics of measured concentrations, after exclusion of outliers, for the IOM sampler, FH25OF, the mini sampler and the SIMPEDS cyclone at 2 l min–1 from the static sampling at five plants are presented in Table 1. During the first run at plant B, the ventilation system was down and the concentration was very high, 30 mg m–3. This caused almost all pumps to stop within 1 h and the results are considered invalid and will not be presented. Excluding this run, the measured inhalable welding aerosol concentrations are in the range 0.7–3.4 mg m–3. The standard deviations for the IOM sampler, the FH25OF and the mini sampler are generally in the range 2–10%, apart from the first run at plant A. It is unknown whether this was caused by an exceptionally highly inhomogeneous concentration at this workplace on this day or because this workplace was the first at which all the equipment was used. The relative standard deviation of the SIMPEDS cyclone is generally much lower than for the other samplers.
|
The data presented in Table 1 excludes 18 samples classified as outliers: six from the IOM sampler (10.3%), none from FH25OF, nine from the mini sampler (7.6%) and three SIMPEDS samples (12.5%). Observed reasons for the outliers were pump stoppages (two each of IOM samples and mini samples), loose plastic tubing (two IOM samples and one mini sample), slightly incorrectly adjusted initial flow rate (three SIMPEDS samples) and samples inadvertently dropped or defective after sampling (three mini samples). Five samples (three IOM samples and two mini samples) caused a considerably higher standard deviation and were deleted as outliers, which changed the measured averages by an amount in the range 0.5–8%.
The concentration ratios mini sampler/IOM, mini sampler/FH25OF, FH25OF/IOM and SIMPEDS/FH25OF, respectively, for welding aerosol mass and manganese are presented in Tables 2–3, respectively. The range of the standard deviations of the concentration ratios is similar for all ratios for both welding aerosol mass and manganese, 0.05–0.10. The main difference between the two tables is that in each case, the sampling bias is twice as large for welding aerosol mass as for manganese, except for the ratio SIMPEDS/FH25OF, for which the sampling bias is four times as large for welding aerosol mass as for manganese. This illustrates that the manganese is mainly found in the respirable thoracic fraction (fume and spatter particles), which the three samplers, the IOM sampler, the FH25OF and the mini sampler, all collect with high efficiency.
|
|
Size distribution of the inhalable fraction at static sampling.
The measured size distributions of the inhalable welding aerosol mass determined from concentration ratios from the static runs at plants A–E, according to the method presented in the section ‘Estimates of the size distribution ...’ above, are shown in Fig. 3. Welding fume (submicron particles or at least particles <2 µm) at these workplaces did not constitute >25–55% of the inhalable fraction, the rest was particles generated by spatter, grinding or slagging.
|
The measured size distributions of the inhalable manganese determined from concentration ratios from the static runs at plants C–E, are presented in Fig. 4. Manganese fume at these workplaces in most cases constituted >75% of the inhalable manganese, with only a minor fraction generated by spatter, grinding and slagging. For manganese in all runs, the aerosol coarseness ratio was <0.12. (The non-negative slope in Fig. 4 for some runs is presumably due to too few samples being analysed by ICP–MS.)
|
Minor parallel rig sampling studies.
These results are presented in Table 4. At plant A, the aerosol coarseness ratio
[see equation (4)] was 0.34, indicating that there was a significant fraction of grinding dust in the sampled aerosol. The obtained concentration ratio of 0.995–1.035 shows that there is no evidence of a sampling bias between the electrically insulating FH25OF and the conducting IOM sampler caused by electric forces. At plant E, the aerosol coarseness ratio was 0.14, which indicates a minor fraction of grinding dust in the sampled aerosol. In this case, the concentration ratio FH37OF/FH25OF was 0.93, which suggests that differences between these two samplers can be disregarded for a welding aerosol with a minor fraction of grinding dust. At plant B, the coarseness ratio was 0.10, indicating that the sampled aerosol was constituted mainly of welding fume particles and very little of grinding particles. This coarseness ratio corresponds to a 0.10 higher sampling efficiency of IOM sampler compared to the FH25OF. Despite this, the mass fraction deposited in the IOM sampler by other means than by drawing air through it only constitutes 0.025 of the mass sampled by the IOM sampler operated by drawing air through it. This only constitutes 25% of the sampling bias of the FH25OF versus the IOM sampler. ‘Passive’ (non-suction) sampling can therefore be ruled out as the cause of the sampling bias of the FH25OF versus the IOM sampler.
|
Laboratory evaluations of the mini sampler
Pressure drop across a 13-mm filter, loaded with welding aerosol, mounted in a mini sampler.
Figure 5 shows the relative pressure drop across the filter cake for a 13-mm filter mounted in a mini sampler for one to two samples per static run from plants A–E. Regression analysis shows that the slope is statistically significant, with Student's t being 2.73, which exceeds the critical value of 2.145 at 14 degrees of freedom. The figure shows that the relative pressure drop increases with a higher fraction of small particles in the sampled aerosol. The highest value but one (which for our data constitutes the upper 91st percentile) is 17.1 kPa/(mg l min–1). This value, rounded off to two digits, can be taken as a representative upper limit of the relative pressure drop. Table 5 presents the pressure drop corresponding to at a sampling flow rate of 0.75 l min–1, three concentrations and three sampling times based on this value. The table shows that 8 h sampling at a concentration of 5 mg m–3 would give a pressure drop of
24 kPa, whereas 4 h sampling at a concentration of 3 mg m–3 leads to a much lower pressure drop, 8 kPa.
|
|
Precision of 13-mm filter weighing.
The weighing experiments gave a pooled standard deviation of 2.2 µg which corresponds to a limit of quantification of 22 µg. The standard deviation did not increase with longer storage time before weighing.
Test for transport loss of loaded 13-mm filters, mounted in mini samplers.
The maximum filter loadings obtained in this experiment were 0.9, 1.5 and 0.8 mg, for plants C–E, respectively, and the lowest loading was 10 µg (plant E). The transport losses are shown in Fig. 6. The average transport loss is 3.7 µg (with a standard error of 0.6 µg). The minimum and maximum values of the transport loss are –5 µg and +19 µg, respectively. The median loss is 3 µg and only for 10 samples did the loss exceed 10 µg. Disregarding any correction for transport losses the RMS transport loss is 6.0 µg. Figure 6 shows a weak, but insignificant, tendency for the losses to increase with increasing sample mass. The average relative transport loss is 0.024. Only for sample masses <60 µg did the relative transport loss become considerable. For sample masses exceeding 60 µg, the relative transport loss exceeded 0.05 for only 3.9% of the samples.
|
Sampling efficiency of the mini sampler
Sampling bias of the mini sampler versus the IOM sampler.
The sampling bias for welding aerosol mass was determined according to the model in equation (2). The iron concentration ratio for the first run at plant C was considered a potential outlier and thus disregarded from the regression. This model is significant with a probability <0.001, and the residual standard deviation is 0.043. The values of the regression coefficients are aIOM = 0.230 ± 0.103 and bFH25OF = 0.782 ± 0.127, which gives a ratio (1–aIOM)/bFH25OF = 0.98 ± 0.13. As a consequence, the model in equation (6) is used for modelling the bias of the mini sampler versus the IOM sampler for welding aerosol mass. This model gives almost identical estimates, kmS–IOM = 0.732 ± 0.055 with the residual standard deviation equal to 0.042.
The dependence of the concentration ratio of the mini sampler to the IOM sampler for welding aerosol mass, iron and manganese on the aerosol coarseness ratio for each substance is shown in Fig. 7. The mini sampler undersamples for all substances versus the IOM sampler. The absolute value of the (negative) bias increases as the coarseness ratio, , of the sampled substance increases, i.e. as the aerosol becomes coarser. The plotted regression line is determined according to the model in equation (6), based on all three substances, aerosol mass, manganese and iron. For this data set, the obtained values for the regression coefficient (kmS–IOM) was 0.750 ± 0.057 with the residual standard deviation equal to 0.051. The results of the regression analysis were virtually independent of regression model and data set, and the value kmS–IOM = 0.750 will be retained to characterize the bias of the mini sampler versus the IOM sampler for welding aerosol mass, manganese and iron, respectively.
|
Both the concentration ratio mini sampler/IOM sampler and the aerosol coarseness ratio are independent of the concentration measured by the IOM sampler.
Sampling bias of the mini sampler versus the open-face 25-mm filter holder.
The dependence of the concentration ratio of the mini sampler to the FH25OF for welding aerosol mass, iron and manganese on the aerosol coarseness ratio for each substance is shown in Fig. 8. No significant dependence on aerosol coarseness is distinguishable for iron or manganese. Only for welding aerosol mass does the mini sampler significantly oversample relative to the FH25OF. The bias increases as the coarseness ratio, , of the sampled substance increases. The iron concentration ratio for the first run at plant C was considered a potential outlier and thus disregarded from the regression. The plotted regression line was determined for the welding aerosol mass, iron and manganese data according to the model in equation (7). The obtained values for the regression coefficient (kmS–FH25OF) was 0.341 ± 0.069 with the residual standard deviation equal to 0.062.
|
Wall deposit of welding aerosol on the mini sampler and the IOM sampler (plant E).
The sampling bias of the mini sampler is approximately twice as large (but with opposite sign) versus the IOM sampler as versus the open-face 25-mm filter holder. At an aerosol coarseness ratio of 0.35, the biases were –0.27 and +0.13, respectively. A possible partial explanation of the bias versus the IOM sampler could be how the sample is defined for the two samplers. For the IOM sampler, the cassette is weighed, i.e. also incorporating the particles deposited on the internal surface of the inlet nozzle, whether by sedimentation, impaction or turbulent diffusion, whereas for the mini sampler only the filter is analysed. Possibly therefore, the bias of the mini sampler versus the IOM sampler could be explained by the fraction collected on the inlet internal surface of the nozzle of the mini sampler. The aerosol coarseness ratio was
0.14, considerably lower than the highest value encountered. Obtained fractions of the amount of welding aerosol mass deposited in the inlet nozzle of the mini sampler were 0.017 and 0.035, respectively, for the two runs. For both runs, the obtained fractions of the amounts of iron and manganese deposited in the inlet nozzle of the sampling cassette of the IOM sampler were 0.004 and 0.001, respectively. Both the fraction of aerosol mass deposited in the inlet nozzle of the mini sampler and the fractions of iron and manganese deposited in inlet nozzle of the IOM sampling cassette can therefore be disregarded compared to the corresponding amounts on the filter. At this plant, the fraction of aerosol mass deposited in the inlet nozzle of the mini sampler (0.026) constitutes less than one-fourth of the bias of the mini sampler versus the IOM sampler (0.12, see Table 2).
Personal sampling
Samplers mounted on a modified headset were accepted by the welders. No big problems were encountered by the welders, even though they on rare occasions had to take them off or on by themselves, for example when going to the lavatory or at a coffee break. None of the headsets broke during this project. The headset model C420IIIL was best received by the welders. The headset model WH20QTR was not accepted by the welders as it pressed too hard onto the head bone behind the temples and was almost impossible to adjust individually. The headset model HT2 was accepted only after ripping away the thick rubber lining from the mounting, so it became much thinner. The PVC tubing used to connect the mini sampler on the headset with pump was somewhat too rigid and thus uncomfortable to the welder when moving his head.
Welding aerosol mass, iron and manganese in personal samples were all approximately log-normally distributed. The limits of quantification for mass (gravimetric) and iron and manganese (XRFS) were 0.022, 0.006 and 0.007 mg, respectively. Result ranges and geometric means (geometric standard deviations) for welding aerosol mass, iron and manganese were 0.15–2.5 and 0.71 mg (1.9), 0.05–1.4 and 0.30 mg (2.1) and 0.003–0.20 and 0.025 mg (2.4), respectively. On average, manganese constituted 3.5% of the welding aerosol mass.
The personal sampling results correspond to 34 shift averages. The concentration ranges for welding aerosol mass was 1.2–8.8 mg m–3 (median 3.3), for iron 0.4–5.0 mg m–3 (median 1.3) and for manganese 0.03–0.73 mg m–3 (median 0.10), respectively. The three shifts with the highest exposure for welding aerosol mass also had the highest exposure to iron. The manganese exposure of one welder (two shifts: 0.65 and 0.73 mg m–3) was considerably higher than for the other welders (<0.29 mg m–3). This was the only welder for which the exposure exceeded the OEL for manganese valid at the day of sampling (0.4 mg m–3). For five welders, the exposure on eight of the nine sampled shifts exceeded the current OEL for manganese, 0.2 mg m–3.
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONCONCLUSIONSFUNDINGACKNOWLEDGEMENTSREFERENCES |
|---|
The headset
This study shows that aerosol sampling using a mini sampler mounted on a headset behind a welder's face shield is possible and does not unduly disturb the wearer. Of the four tested versions, three could be mounted in such a way that they were not uncomfortable to the welders. A headset-mounted sampler is preferable to the mounting arrangements shown in the 2001 edition of the EN ISO 10882-1 standard on sampling airborne particles during welding (ISO, 2001a). The advantage of the mini sampler is that it can be positioned close to the mouth/nose whether a welder's face shield or other face shield is worn or not and whether the visor is in the up or down position. Due to ergonomic issues and personal comfort, the welders participating in this investigation all removed their face shield when not welding (except during very short breaks).
Uncertainties of sample analyses versus sample transport loss
The weighing uncertainty for 13-mm filters was found to be 2.2 µg, considerably less than the sample uncertainty due to transport by ordinary mail, 6.0 µg. Both of these uncertainties are small compared to aerosol mass collected by personal sampling, which was at least 25 times larger than the transport uncertainty. For the personal samples, manganese constituted on average 3.5% of the welding aerosol mass collected. If the fraction of manganese in the amount of welding aerosol lost in transport is assumed to be identical to the fraction in the welding aerosol sampled, the ratio of the amount of manganese in the sample to the amount lost in transport is 15 and 32 times for the lowest and next-to-lowest personal manganese samples. The transport losses are therefore not important in relation to either sampled welding aerosol mass or manganese.
Capacity of the sampling pumps
Our results show that the pressure drop across the filter cake of a 13-mm filter loaded with welding aerosol can be very high. Of the pumps used in our workplace sampling, generally the AC2000 pump (SKC) was able to withstand the pressure drop increase of sampling welding aerosol for >6 h.
Particle size of welding aerosol
An aerosol coarseness ratio for welding aerosol mass exceeding 0.35 was never encountered at static sampling at the visited plants (Figure 3). Personal sampling generally yields higher concentrations than static sampling. This is mainly because the worker is close to the source and the airborne pollutants are diluted when transported from the source to a static sampler mounted some distance away. Additionally, a higher fraction of coarse particles with an aerodynamic diameter exceeding 10 µm will sediment out before being sampled by a static sampler. Therefore it is likely that at personal sampling during welding, one would encounter an aerosol consisting of coarser particles, which would thus be characterised by an aerosol coarseness ratio exceeding 0.35. In many dusty workplaces a coarseness ratio of approximately 0.5 is common (Werner et al., 1996), and this value could presumably be taken as an upper limit of what to expect at welding workplaces not entirely dominated by grinding/slagging. Data is available for grinding in stainless steel and hard metal steel, for which Koponen et al. (1981) found that the ratio of respirable to total dust was approximately 0.45 and 0.60, respectively. In regard to manganese, the aerosol coarseness ratio ranged from –0.02 to 0.12, with most values close to 0.09 (Table 3). As there is almost no manganese in the coarse aerosol fractions, 0.15 can be used as an upper limit for the aerosol coarseness ratio in personal sampling for manganese. Recalculation of the data in Table 3 into ratios SIMPEDS (2 l/min)/IOM sampler gives an average of 0.83 which compares well with the result of Berlinger et al. (2007) for static sampling, 0.86. For personal sampling they obtained an average ratio of 0.98. The particle size distributions in our tests does not seem to be less coarse than those of Berlinger et al., both in regard of the static sampling and the personal sampling.
The cumulative size distributions for welding aerosol and manganese measured in the workplaces visited (plants A–E) agree broadly with the model of three different modes outlined in the introduction. Manganese is mainly encountered in particles of size <2 µm (i.e. fume) whereas the aerosol fraction below 2 µm only constitutes 25–55% by mass of the welding aerosol. The welding aerosol fraction sampled by the IOM sampler, but not by the open-face 25-mm filter holder, presumably consists mainly of grinding aerosol.
Sampling bias of the mini sampler versus the IOM sampler and the open-face 25-mm filter holder
The bias of the open-face 25-mm filter holder versus the IOM sampler has been shown to be a function of the arithmetical complement of the concentration ratio FH25OF/IOM sampler (which in this paper has been termed the aerosol coarseness ratio, see equation (5)). The aerosol coarseness ratio is not unduly dependent on unintended parameters, as the results of a test of three of these listed in Table 4 demonstrates: it is independent of whether the filter cassette is electrically conducting or insulating, the amount of aerosol mass collected by the IOM sampler when no air is drawn through it is negligible and open-face filter holders of size 25 and 37 mm collect, within 10%, the same welding aerosol concentration. Consequently, the measured concentration ratios FH25OF/IOM sampler fundamentally expresses the differences in sampling efficiency between the two samplers, and thereby implicitly also the size distribution of the sampled aerosol.
The concentration ratios of the mini sampler to the IOM sampler (as a function of the aerosol coarseness ratio) fall along a line for all measurements of the three welding aerosol components welding: aerosol mass, iron (except one run) and manganese (Fig. 7). This line therefore describes the sampling bias of the mini sampler as a function of the size distribution of the sampled aerosol, irrespective of aerosol component. As the sampling bias of the mini sampler versus the IOM sampler cannot be explained by deposition in its inlet nozzle, it must be concluded that this bias mainly depends on other factors, e.g. differences in aspiration efficiency.
As discussed above, larger particles are expected to be encountered in personal sampling of welding aerosols compared to static rig sampling. Within a coarseness ratio range of (0.00–0.50) for the welding aerosol mass, the RMS bias of the mini sampler, based on the regression model above, would be –0.180. This bias is so large that in order to use the mini sampler for sampling inhalable dust, it would need to be optimized by changing its geometry and flow rate. This does not seem impossible in relation to the results of the sampler evaluations by Paik and Vincent (2004) and Marley (1994). Paik and Vincent found in a wind tunnel experiment similar aspiration efficiencies for both the IOM sampler and a sampler with an inlet tube diameter of 8 mm operated at 0.5 l min–1. In workplace static sampling, Marley obtained similar concentrations with the seven-hole sampler as with a sampler with an inlet tube diameter of 9 mm operated at 0.4 l min–1.
As discussed previously, in personal sampling for manganese, an aerosol coarseness ratio <0.15 might be expected. Within this aerosol coarseness ratio range, the RMS bias of the mini sampler for manganese, based on the regression model above, would be –0.056. (This value also compares well with the average bias versus the IOM sampler found in the static rig sampling, –0.049, see Table 3.) A Student's t-test on the distribution of the bias of the mini sampler versus the IOM sampler for manganese (Table 3) shows that it is non-significant at the 5% level. The bias of <10% need therefore not be further reduced by optimization of the sampler in order to use the mini sampler for collecting manganese in welding aerosols or other aerosols with a similar particle size distribution (90% of mass <20 µm). Additionally, the bias is of the same order as Chung et al. (1999) found for samplers mounted behind a welder's face shield. The performance of sampling manganese with the mini sampler and analysing the samples with portable XRFS as determined according to the standard EN 482 (CEN, 2006) is presented in a companion paper (G. Lidén and L. Lundgren, in preparation).
A Student's t-test on the distribution of the bias of the mini sampler versus the open-face 25-mm filter holder for manganese (Table 3) shows that it is non-significant at the 5% level. For manganese, the bias exhibits no systematic dependency on the aerosol coarseness ratio (see Fig. 8).
|
CONCLUSIONS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONCONCLUSIONSFUNDINGACKNOWLEDGEMENTSREFERENCES |
|---|
A headset-mounted mini sampler has been developed in this study to improve personal sampling in welder's breathing zone. It does not disturb the welder who can use his/her own face shield without discomfort, also when it is removed. Above all, the headset mounting arrangement maintains the sampler close to the nose/mouth during the whole sampling period.
Headsets for mounting a mini sampler behind a welder's visor have been modified from commercial headsets. Especially, two headset models (C420IIIL and IsoMax HH) were acceptable by the welders. These look the weakest in design, but as none broke, they are recommended for further use. The tubing clip and the waistbelt for the sampling pump made it easy to attach the sampling equipment to the welders.
The particle size distributions at the visited plants, showed that the inhalable fraction of welding aerosol mass only consisted of 25–55% welding fume. The remainder was welding spatter, grinding dust and slagging dust. More than 65% of the manganese is found in the fume particles.
For welding aerosol mass, the mini sampler exhibits a negative sampling bias versus the IOM sampler that is two to three times as large as the positive sampling bias versus the open-face 25-mm filter holder. The value of the sampling bias depends on the size distribution of the welding aerosol sampled. For a coarser welding aerosol, the absolute value of the sampling bias increases. For a welding aerosol with an aerosol coarseness ratio of 0.35, the bias of the mini sampler versus the IOM sampler is approximately –0.26 and versus the open-face 25-mm filter holder is approximately +0.12.
The weighing precision for 13-mm MCE filters was found to be 2.2 µg. The RMS transport losses of 13-mm filters mounted in mini samplers transported by mail 500 km there and back was 6.0 µg. Both of these values are small compared to the range of masses collected in 43 personal samples.
For manganese, the sampling bias of the mini sampler versus the IOM sampler was for all eight sampled welding aerosols numerically less than –0.14 and does not depend on the aerosol coarseness ratio of the sampled manganese aerosol. The average sampling bias of the mini sampler versus the IOM sampler for manganese (with a negative RMS value of –0.046) is not statistically significant. Likewise, the sampling bias of the mini sampler versus the open-face 25-mm filter holder for manganese is negligible (average <0.04) and not statistically significant. The mini sampler can therefore be used for determining welder's exposure to manganese.
The pressure drop across 13-mm MCE filters loaded with welding particles increases with the amount of sample, but seems also to be affected slightly by the coarseness of the collected aerosol. Whether a pump is able to withstand the increased pressure drop and keep the flow rate within ±5% of the initial flow rate will depend on the pump type, the welding aerosol concentration and the sampling time.
A headset-mounted mini sampler is user-friendly, easy to individually adjust, does not disturb welder during sampling and allows sampling behind the welder's face shield. The sampling bias and precision of the mini sampler has been determined for assessing occupational exposure to manganese in welding processes. The mini sampler can also be used for personal sampling of the inhalable fraction of other aerosols for which 90% of the mass size distribution is below 20 µm. For coarser aerosols, a redesigned sampler would be needed.
|
FUNDING |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONCONCLUSIONSFUNDINGACKNOWLEDGEMENTSREFERENCES |
|---|
Partially funded by Swedish Work Environment Authority under contract CTK 2005/40918.
|
ACKNOWLEDGEMENTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONCONCLUSIONSFUNDINGACKNOWLEDGEMENTSREFERENCES |
|---|
We would like to thank everybody who has helped us in this project. Leif Bäcklin has manufactured the mini sampler in several versions and rebuilt the commercial headsets for our purposes. Gunnel Sundström has weighed all filters. Jüri Waher has performed several of the laboratory tests of the mini sampler. Karin Holm has performed all analyses with ICP–MS. Villy Glas, Nils Kulle and Bengt Christensson have participated in the workplace sampling campaigns. We would also like to thank those who have lent us sampling pumps; IVL Swedish Environmental Research Institute, Swedish National Institute for Working Life (shut down in June 2007) and the Department of Occupational and Environmental Health, Stockholm Centre for Public Health. Last, but not least, we would like to thank management and the workers at the visited plants for co-operation during our workplace sampling. One of the authors (G.L.) has an interest in a proposed commercial exploitation of the method.
|
FOOTNOTES |
|---|
The free full text of this article can be found in the online version of this issue.
Received July 18, 2008; in final form December 19, 2008
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONCONCLUSIONSFUNDINGACKNOWLEDGEMENTSREFERENCES |
|---|
Allen J, Bellinger EG, Higgins RI. A full shift true breathing zone air sampler and its application to lead workers. Proc Instit Mech Eng (1981) 195:325–8.[CrossRef]
Bartley DL, Chen C-C, Song R, et al. Respirable aerosol sampler performance testing. Am Ind Hyg Assoc J (1994) 55:1036–46.[Web of Science]
Berlinger B, Náray M, Záray G. Comparison of different sampling heads applied for investigation of welding fume. Microchem J (2007) 85:25–30.[CrossRef][Web of Science]
Blachman MW, Lippmann M. Performance characteristics of the multicyclone aerosol sampler. Am Ind Hyg Assoc J (1974) 35:311–26.[Web of Science][Medline]
Bloor WA, Eardley RE, Dinsdale A. A gravimetric personal dust-sampler. Ann Occup Hyg (1968) 11:81–6.[Medline]
Bohne JE, Cohen BS. Aerosol resuspension from fabric: implications for personal monitoring in the beryllium industry. Am Ind Hyg Assoc J (1985) 46:73–9.[Web of Science][Medline]
Brohus H. Measurement of personal exposure using a breathing thermal manikin. In: Ventilation ‘97; 14–17 September 1997—Goodfellow H, Tähti E, eds. (1997) Ottawa, Canada: The Canadian Environment Industry Association. 781–91.
Bull RK, Stevens DC, Marshall M. Studies of aerosol distributions in a small laboratory and around a humanoid phantom. J Aerosol Sci (1987) 18:321–35.[CrossRef][Web of Science]
CEN. EN1540 Workplace Atmospheres Terminology (1998) Brussels, Belgium: Comité Européen de Normalisation.
CEN. EN482 Workplace atmospheres. General requirements for the performance of procedures for the measurement of chemical agents (2006) Brussels, Belgium: Comité Européen de Normalisation.
Chang H-C. A parallel multicyclone size-selective particulate sampling train. Am Ind Hyg Assoc J (1974) 35:538–45.[Web of Science][Medline]
Chien PT. A proposed sampling device for personal exposures. Am Ind Hyg Assoc J (1992) 53:527.[Web of Science]
Chung KYK, Aitken RJ, Bradley DR. Development and testing of a new sampler for welding fume. Ann Occup Hyg (1997) 41:355–72.
Chung KYK, Carter GJ, Stancliffe JD. Laboratory evaluation of a protocol for personal sampling of airborne particles in welding and allied processes. Appl Occup Environ Hyg (1999) 14:107–18.[CrossRef][Medline]
Cohen BS, Chang AE, Harley NH, et al. Exposure estimates from personal lapel monitors. Am Ind Hyg Assoc J (1982) 43:239–43.[Web of Science]
Cohen BS, Positano R. Resuspension of dust from work clothing as a source of inhalation exposure. Am Ind Hyg Assoc J (1986) 47:255–8.[Web of Science][Medline]
Dinardi SR, ed. The occupational environment: its evaluation, control and management (1997) Fairfax, VA: American Industrial Hygiene Association.
Goller JW, Paik NW. A comparison of iron oxide fume inside and outside of welding helmets. Am Ind Hyg Assoc J (1985) 46:89–93.[Web of Science][Medline]
Guffey SE, Flanagan ME, vanBelle G. Air sampling at the chest and ear as representative of the breathing zone. Am Ind Hyg Assoc J (2001) 62:416–27.
Harris MK, Ewing WM, Longo W, et al. Manganese exposures during shielded metal arc welding (SMAW) in an enclosed space. J Occup Environ Hyg (2005) 2:375–82.[CrossRef][Web of Science][Medline]
HSE. MDHS 83/2. Resin acids in rosin (colophony) solder flux fume (2006) London, UK: Health and Safety Executive.
ISO. EN ISO 10882-1. Health and safety in welding and allied processes. Sampling of airborne particles and gases in the operator's breathing zone Part 1: Sampling of airborne particles (2001a) Geneva, Switzerland: International Organization for Standardization.
ISO. ISO 15202-2. Workplace air. Determination of metals and metalloids in airborne particulate matter by inductively coupled plasma atomic emission spectrometry Part 2: Sample preparation (2001b) Geneva, Switzerland: International Organization for Standardization.
ISO. EN ISO 17294-2. Water quality. Application of inductively coupled plasma mass spectrometry (ICP-MS) Part 2: Determination of 62 elements (2003) Geneva, Switzerland: International Organization for Standardization.
Jankovic JT, Underwood WS, Goodwin GM. Exposures from thorium contained in thoriated tungsten welding electrodes. Am Ind Hyg Assoc J (1999) 60:384–9.[Web of Science][Medline]
Jenkins NT, Eagar TW. Chemical analysis of welding fume particles. Weld J (2005) 84:87s–93s.
Kalliomäki P-L, Aittoniemi K, Gustafsson T, et al. Research on industrial ferrous metal aerosols. Ann Occup Hyg (1982) 26:337–45.
Kenny LC, Aitken RJ, Baldwin PEJ, et al. The sampling efficiency of personal inhalable aerosol samplers in low air movement environments. J Aerosol Sci (1999) 30:627–38.[CrossRef][Web of Science]
Koponen M, Gustafsson T, Kalliomäki P-L, et al. Grinding dust of alloyed steel and hard steel. Ann Occup Hyg (1981) 24:191–204.
Lidén G, Kenny LC. Comparison of measured respirable dust sampler penetration curves with sampling conventions. Annals of Occupational Hygiene (1991) 35:485–504.
Liu BYH, Pui DYH, Rubow KL. Characteristics of air sampling filter media. In: Aerosols in the mining and industrial work environments—Marple VA, Liu BYH, eds. (1983) Ann Arbor, MI: Ann Arbor Science. 989–1038.
Liu D, Wong H, Quilan P, et al. Welding helmet airborne fume concentrations compared to personal breathing zone sampling. Am Ind Hyg Assoc J (1995) 56:280–3.[Web of Science][Medline]
Malek RF, Daisy JM, Cohen BS. Breathing zone concentration variations in the reinforced plastic industry—field measurements in a boat manufacturing plant. Appl Occup Environ Hyg (1999) 14:777–84.[CrossRef][Medline]
Marcy AD, Drake PL. Development of a field method for measuring manganese in welding fume. J Environ Monit (2007) 9:1199–204.[CrossRef][Web of Science][Medline]
Marley KA. Field performance of a new design of a total inhalable dust sampling head. Ann Occup Hyg (1994) 38:895–902.
Martinelli CA, Harley NH, Lippmann M, et al. Monitoring real-time aerosol distribution in the breathing zone. Am Ind Hyg Assoc J (1983) 44:280–5.[Web of Science][Medline]
OSHA. Sampling and analytical methods for MDA monitoring and measurement procedures (1999) Washington, DC: Occupational Safety and Health Administration.
Paik SY, Vincent JH. The orientation-averaged aspiration efficiency of IOM-like personal aerosol samplers mounted on bluff bodies. Ann Occup Hyg (2004) 48:3–11.
Parker RC, Bull RK, Stevens DC, et al. Studies of aerosol distributions in a small laboratory containing a heated phantom. Ann Occup Hyg (1990) 34:35–44.
Pengelly MI, Groves JA, Foster RD, et al. Development of a method for measuring exposure to resin acids in solder fume. Ann Occup Hyg (1994) 38:765–76.
Rosén G, Andersson I-M, Niemelä R, et al. Air transport phenomena in the breathing zone and its implication on exposure assessment. In: Ventilation 1997 14–17 September 1997—Goodfellow H, Tähti E, eds. (1997) Ottawa, Canada: The Canadian Environment Industry Association. 489–95.
Simpson AT. Comparison of sampling position when measuring personal exposure to solder fume. Ann Occup Hyg (2005) 49:439–42.
Steel J, Maican A. The measurement of exposure. In: Second International Occupational Hygiene Association Conference; 16–18 November 1994 (1994) Hong Kong, China: HKIOEH Co. Ltd.
Stephenson D, Seshadri G, Verandth JM. Workplace exposure to submicron particle mass and number concentrations from manual arc welding of carbon steel. Am Ind Hyg Assoc J (2003) 64:516–21.
SWEA. Occupational exposure limit values and measures against air contaminants (2005) Solna, Sweden: Swedish Work Environment Authority. Provision AFS 2005:17.
ter Kuile WM. Ontwikkeling van de totaalstof helm (in Dutch—Development of the total dust helmet) (1982) Delft, The Netherlands: IMG-TNO.
Vinson R, Volkwein J, McWilliams L. Determining the spatial variability of personal sampler inlet locations. J Occup Environ Hyg (2007) 4:708–14.[CrossRef][Web of Science][Medline]
Welling I, Andersson I-M, Rosén G, et al. Contaminant dispersion in the vicinity of a worker in a uniform velocity field. Ann Occup Hyg (2000) 44:219–25.
Werner MA, Spear TM, Vincent JH. Investigation into the impact of introducing workplace aerosol standards based on the inhalable fraction. Analyst (1996) 121:1207–14.[CrossRef][Medline]
Worobiec A, Stefaniak EA, Kiro S, et al. Comprehensive microanalytical study of welding aerosols with X-ray and Raman based methods. X-ray Spectrom (2007) 36:328–35.[CrossRef]
Zimmer AT, Baron PA, Biswas P. The influence of operating parameters on number-weighted aerosol size distribution generated from a gas metal arc welding process. J Aerosol Sci (2002) 33:519–31.[CrossRef][Web of Science]
CiteULike 
Connotea 
Del.icio.us What's this?
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||