Skip Navigation


Annals of Occupational Hygiene Advance Access originally published online on May 2, 2008
Annals of Occupational Hygiene 2008 52(4):287-295; doi:10.1093/annhyg/men016
This Article
Right arrow Abstract Freely available
Right arrow FREE Full Text (PDF) Freely available
Right arrow All Versions of this Article:
52/4/287    most recent
men016v1
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Email this article to a friend
Right arrow Similar articles in this journal
Right arrow Similar articles in ISI Web of Science
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Add to My Personal Archive
Right arrow Download to citation manager
Right arrowRequest Permissions
Google Scholar
Right arrow Articles by Stacey, P.
Right arrow Articles by Butler, O.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Stacey, P.
Right arrow Articles by Butler, O.
Social Bookmarking
Add to CiteULike   Add to Connotea   Add to Del.icio.us  
What's this?


© Crown Copyright 2008. Published by Oxford University Press on behalf of the British Occupational Hygiene Society

Performance of Laboratories Analysing Welding Fume on Filter Samples: Results from the WASP Proficiency Testing Scheme

Peter Stacey* and Owen Butler

The Health and Safety Laboratory, Harpur Hill, Buxton, SK17 9JN, UK

* Author to whom correspondence should be addressed. Tel: +44 (0) 114 2892645; fax: +44 (0) 114 2892544; e-mail: peter.stacey{at}hsl.gov.uk


   ABSTRACT
TOP
 ABSTRACT
INTRODUCTION
 EXPERIMENTAL
 RESULTS
 DISCUSSION
 CONCLUSIONS
 REFERENCES
 
This paper emphasizes the need for occupational hygiene professionals to require evidence of the quality of welding fume data from analytical laboratories. The measurement of metals in welding fume using atomic spectrometric techniques is a complex analysis often requiring specialist digestion procedures. The results from a trial programme testing the proficiency of laboratories in the Workplace Analysis Scheme for Proficiency (WASP) to measure potentially harmful metals in several different types of welding fume showed that most laboratories underestimated the mass of analyte on the filters. The average recovery was 70–80% of the target value and >20% of reported recoveries for some of the more difficult welding fume matrices were <50%. This level of under-reporting has significant implications for any health or hygiene studies of the exposure of welders to toxic metals for the types of fumes included in this study. Good laboratories' performance measuring spiked WASP filter samples containing soluble metal salts did not guarantee good performance when measuring the more complex welding fume trial filter samples. Consistent rather than erratic error predominated, suggesting that the main analytical factor contributing to the differences between the target values and results was the effectiveness of the sample preparation procedures used by participating laboratories. It is concluded that, with practice and regular participation in WASP, performance can improve over time.

Keywords: air sampling • analysis • fume • proficiency testing • quality of data • WASP • welding


   INTRODUCTION
TOP
 ABSTRACT
 INTRODUCTION
EXPERIMENTAL
 RESULTS
 DISCUSSION
 CONCLUSIONS
 REFERENCES
 
Historically, welders have shown an elevated incidence of lung cancer, and the inhalation of metals in welding fume, especially fume containing nickel or chromium (VI), is thought to be one factor that may cause such lung cancer or provoke occupational asthma. In the UK, the exposure of workers to potentially harmful substances is regulated by the Control of Substances Hazardous to Health regulations (Stationery Office Ltd, 2004). These regulations require the employer to control the exposure of their employees to potentially harmful substances, and limits for the airborne concentration of specific metal elements in air are stipulated in EH40 (Health and Safety Executive, 2005). Occupational hygienists often take air samples to assess the exposure of workers and the effectiveness of controls. Samples are returned to laboratories for analysis in order to assess the exposure to metals in the fume. These measurements need to be accurate to obtain a reasonable indication of the workers' exposure.

Proficiency testing (PT) schemes allow a laboratory to compare its analytical performance against its peers and help laboratories demonstrate to their customers that they can obtain reliable results. However, few PT programmes for the analysis of air samples include test samples with ‘realistic’ type matrices. The performance of laboratories analysing simulated (spiked) test filters has been described by Stacey (2006); however, this work is based on results from participants analysing relatively soluble matrices and is not representative of all dust and fume found in the workplace. Since 2002, the Workplace Analysis Scheme for Proficiency (WASP) PT scheme, administered by the Health and Safety Laboratory (HSL), has distributed to laboratories replicate test samples of complex welding fume on filters. This paper compares the performance of laboratories analysing metals in manual metal arc (MMA), metal inert gas (MIG) and flux core arc (FCAW) welding fume collected on filters over 13 rounds of the scheme. For brevity, the paper will concentrate on the results of laboratories analysing chromium in welding fume samples, although iron, manganese and nickel measurements were also performed. Chromium can form spinel-type oxides that are not easily soluble (Butler and Howe, 1999), so this metal represents the most challenging of metals found in welding fume. Therefore, if a laboratory's performance with chromium is satisfactory, then their performance analysing other metals should also be satisfactory.


   EXPERIMENTAL
TOP
 ABSTRACT
 INTRODUCTION
 EXPERIMENTAL
RESULTS
 DISCUSSION
 CONCLUSIONS
 REFERENCES
 
Generation of test filter samples containing welding fume
Air filter samples were collected in batches of 114 using a multi-port sampler device. Similar samplers have been described elsewhere (Akanes et al., 1990, 1992, 1995; Anglov et al., 1993). In summary, three-part 25-mm open-faced, asbestos-type sampling cassettes with 2'' cowl (SKC Ltd, Dorset) were used for sampling and fume was collected onto 0.8-µm mixed cellulose ester filters. Flow rates, at a nominal of 1.8 l/min through each filter, were controlled using critical orifices and matched to within 1% (1 standard deviation (SD)) of each other. The fume was generated under controlled conditions at The Welding Institute near Cambridge using a welding fume box, as specified in BS 7384 (BSI, 1991) (Fig. 1a). Emitted fume was sampled into the multi-port sampler through a length of flexible plastic hosing connected to the exhaust stack of this welding fume box (Fig. 1b).


Figure 1
View larger version (48K):
[in this window]
[in a new window]
[Download PowerPoint slide]
  		Fig. 1. (a) Welding fume generation with a fume box [215 x 208 mm (96 x 96 DPI)]. (b) Schematic diagram of the multi-port sampler [121 x 150 mm (96 x 96 DPI)].

 
Filter sample characterization
This multi-port sampler design is capable of producing near-identical air filter samples. However, slight differences in individual flow rates of the critical orifices and the annular shape of the multi-port sampler give rise to slight variations in mass loading from filter to filter.

Other users of such multi-port sampling devices (Akanes et al., 1990; Anglov et al., 1993) have attempted to normalize this variation by applying correction factors either by using flow rate measurements and/or by using gravimetric analysis of the filter samples.

An alternative approach is used at the Health and Safety Laboratory (HSL). Individual filters are removed from their sampling cassettes and analysed non-destructively by a highly precise wavelength dispersive X-ray fluorescence spectrometric (XRFS) technique. Details regarding the use of XRFS for the analysis of air filter samples are more fully described in MDHS 91 (Health and Safety Executive, 1998). Using XRFS, it is possible to measure elemental variations from filter to filter, something that cannot be achieved by alternative flow rate and/or gravimetric measurements. Data accumulated at HSL suggest that the intra-batch elemental variability, when sampling welding fume, is typically better than 3% (95th percentile), provided that obvious outliers are excluded, for example, samples collected that subsequently fail due to partially blocked critical orifices or samplers that have failed leak tests. In summary, therefore, it is possible to collect a batch size of ~100 filters that can be considered identical.

Target values
PT can be considered as a benchmarking exercise where individual laboratories are compared with their peers. To this end, so-called nominal or target values for test samples are required. The WASP scheme, as do many other PT schemes, uses consensus values derived from the participant's data as the target values. The scheme uses a robust mean value calculated from pooled participant data after exclusion of extreme values. However, with a new PT product, where the performance of participants is initially unknown, and may have a high degree of scatter, it is prudent to derive the target values using an alternative approach. In this work, a representative number of welding fume filters, 6–10 filters from each multi-port sampler collection, were selected at random and were analysed at HSL using inductively coupled plasma–atomic emission spectrometry (ICP-AES). A high-performance closed vessel microwave assisted digestion procedure, involving the use of a mixture of nitric, hydrochloric and hydrofluoric acids, was used to dissolve the filter samples prior to analysis. The analytical procedure is described in the International Standard 15202, part 2 for sample preparation (ISO 15202-2, 2001) and part 3 for analysis (ISO 15202-3, 2004), which was developed in an International Organization for Standardization working group using validation data generated at HSL. The method was validated by performing repeat measurements on homogenized bulk reference material derived from the workplace and welding fume samples and by comparing the results obtained against a variety of other digestion techniques including fusion procedures. The organizers of the only other PT programme that offers similar welding fume samples, the National Institute of Occupational Health (NIOH) in Norway, also employs a similar robust digestion procedure for the establishment of target values. HSL can verify the performance of this procedure through its previous good performance in the PT scheme managed by NIOH. Good agreement between these two laboratories can be seen through HSL's chromium analysis performance in six rounds of the Norwegian NIOH PT scheme (HSL result = 1.039 x (NIOH reference value) + 0.3; r2 = 0.994). The equation for the relationship shows a small positive difference between the HSL results and the reference values from NIOH.

Filter packing and transportation
Filter samples can be damaged in transit. In the WASP scheme, welding fume air filter samples were placed into individually labelled 49-mm plastic petri dishes (Gelman Sciences). Prior to shipment, each membrane filter sample is spiked with ~400 µl of propan-1-ol. This had the effect of slightly dissolving the mixed cellulose ester filter, thus encapsulating the welding fume deposit, and preventing loss of particles from the filter in transit.

Sample distribution strategy
In the initial six rounds of the welding fume scheme, two near-identical, filter samples were distributed to the 18 laboratories that had requested the welding fume as an additional sample to their regular WASP spiked filter samples. Such samples were prepared by pipetting a known volume of soluble metal salt onto each filter. While they can provide a useful test of the capability of a laboratory to undertake filter analysis, the samples are not representative of complex matrices such as welding fume, which often contain spinel-type oxides that can require vigorous procedures for dissolution (Butler and Howe, 1999). In the subsequent seven PT rounds, to encourage greater participation, HSL distributed a single welding fume sample to all those WASP subscribers requiring the spiked filter product. On average, 41 laboratories participated in these rounds. Summary details of the welding fume filter samples supplied are tabulated in Table 1.


View this table:
[in this window]
[in a new window]

  		Table 1. Types of fume- and HSL-derived target values (mean and 2 SDa) used in the WASP PT rounds

 

   RESULTS
TOP
 ABSTRACT
 INTRODUCTION
 EXPERIMENTAL
 RESULTS
DISCUSSION
 CONCLUSIONS
 REFERENCES
 
The performance of participants was benchmarked against the target values determined by HSL. The distributions of results for chromium submitted by participants for the various types of welding fume used in this pilot study are shown as box and whisker plots in Fig. 2. The extent of the boxes, for each data set, shows the position of the upper and lower quartile values. The line within the box is the position of the median (50%) value. The whiskers extending from the boxes are limited by adding a factor of 1.5 times the range of the box to the upper or lower quartile value or by the maximum value (whichever is the lower). Results outside these whiskers are not considered as belonging to the same distribution. Similar distributions of results were obtained for the other analytes in the welding fume such as iron, manganese and nickel. Table 2 gives the range of median percentage recoveries for each element and fume type and Fig. 3 shows the cumulative percentage recovery for all chromium analyses.


View this table:
[in this window]
[in a new window]

  		Table 2. Range of median percentage recoveries

 


Figure 2
View larger version (33K):
[in this window]
[in a new window]
[Download PowerPoint slide]
  		Fig. 2. Distribution of results for chromium in welding fume derived from (a) a manual metal arc process and (b) an MIG process [309 x 418 mm (96 x 96 DPI)] and (c) an FCAW process [287 x 204 mm (96 x 96 DPI)].

 


Figure 3
View larger version (7K):
[in this window]
[in a new window]
[Download PowerPoint slide]
  		Fig. 3. Cumulative percentage of chromium results [516 x 316 mm (96 x 96 DPI)].

 
Between-laboratory precision
Figure 4 plots the relationship between standard deviation of results obtained by laboratories for chromium and filter mass loading. Data from the analysis of both the spiked filter and welding fume samples for the WASP PT programme are presented. For comparative purposes, participant welding fume filter data from the NIOH PT scheme, over the period 1996–2005, has also been plotted.


Figure 4
View larger version (22K):
[in this window]
[in a new window]
[Download PowerPoint slide]
  		Fig. 4. Comparison between spiked and welding fume results for chromium [307 x 201 mm (96 x 96 DPI)].

 

   DISCUSSION
TOP
 ABSTRACT
 INTRODUCTION
 EXPERIMENTAL
 RESULTS
 DISCUSSION
CONCLUSIONS
 REFERENCES
 
Between-laboratory precision
The slope of the best-fit line for the WASP welding fume data (Fig. 4) is significantly different to the other two slopes, thus indicating a higher degree of scatter in results submitted from participating laboratories analysing chromium.

The between-laboratory precision for the WASP welding fume participants on a round-to-round basis is ~20%. The variability in welding fume filter production, including the subsequent XRFS measurement precision, is better than 3% (95th percentile). A recently published performance standard for PT schemes (ISO 13528, 2005) suggests that test sample homogeneity should be ≤0.3 of the anticipated PT round precision to ensure that differences from sample to sample do not statistically impinge upon a laboratory's performance, indicating that the samples were ‘fit-for-purpose’ at the time they were distributed.

Difference from target values
The majority of results submitted to HSL under-reported the mass of analyte on the filters when compared with the target values determined at HSL. Of the 447 results submitted for chromium, 83% were >10% and 60% were >20% from the target values (Fig. 3). On average, the results reported by participants were 20–30% lower than the target values, suggesting that the majority of results from welding fume reported to occupational hygienists underestimate the ‘true’ exposure of workers. Typically, the poorest performers may report a result between 20% and 50% of the target value. In contrast, the majority of laboratories perform well in the WASP spiked filter PT scheme. Approximately 80% report a result for chromium within ±8% of the respective target value (WASP satisfactory performance criteria -category 2) with 50% of these reporting results within ±4% (WASP category 1 performance). Correlation analysis was performed to ascertain whether performance with the spiked filter samples mimics performance with the welding fume samples. For example, a Pearson correlation value of –0.15 (95% confidence level = –0.51 to 0.26) was obtained for laboratories that concurrently analysed a spiked filter sample with 90 µg of chromium with a welding fume sample containing 91.6 µg of chromium. Full correlation between the two samples would give a value of 1.00. This indicates that there is no correlation between a laboratory's performance analysing a simulated filter spiked with metals and a real sample (r2 = 0.05). As the majority of the laboratories that took part in the welding fume exercise could successfully analyse the regular WASP samples in the same round to within ±8% of the target values, it suggests that digestion efficiency, rather than other method parameters, is the cause of under-recovery.

Youden plots to graphically illustrate bias and precision effects
In the initial six PT rounds, two near-identical filter samples were distributed to participants. Youden plots (Fig. 5) are a useful means of graphically illustrating bias and precision effects and can be constructed using a scatter plot of two near-identical samples. The two samples must have a similar elemental concentration and possess a similar matrix. Results grouped closely around the cross-hairs indicate low bias and good precision. Results grouped in the lower left quadrant or upper right quadrant, along the plotted 1:1 line, would indicate systematic under- or over-recovery. Results found in the upper left and lower right quadrant would indicate random measurement error or a significant difference between the two samples. The data plotted in Fig. 4 is grouped on the 1:1 line in the left-hand quadrant indicating systematic under-recovering in the welding fume results. The most likely explanation of this is incomplete elemental recovery during the sample digestion step indicating that an inadequate digestion procedure is being used or that a suitable procedure is inadequately applied. The vast majority of participants use atomic spectrometry techniques such as flame atomic absorption spectroscopy or ICP-AES where a dissolution procedure is a prerequisite. Interestingly, two participating laboratories used an XRFS technique that requires no substantial sample preparation and obtained results for chromium that were generally closer to the target values (on average within 8% of the target value) further supporting the hypothesis that the dissolution procedure is the most likely explanation of under-recovery.


Figure 5
View larger version (14K):
[in this window]
[in a new window]
[Download PowerPoint slide]
  		Fig. 5. Youden plots showing consistent error [308 x 416 mm (96 x 96 DPI)].

 
Digestion procedure explanation
Two factors dictate the efficiency of a digestion step, namely the type and combination of mineral acids used and the subsequent digestion temperature employed. For example, mixtures of nitric and perchloric acid or nitric and sulphuric acid exhibit a higher potential to oxidize a sample than nitric acid on its own. An increase in digestion temperature can increase reaction efficiency. Perchloric and sulphuric acids have higher boiling points than nitric acid and so it is possible to digest filter samples at higher temperatures before the onset of volatilization of the digestion acids. Closed vessel microwave-assisted digestion procedures, increasingly being used in laboratories, offer the potential to undertake digestions at elevated pressures, that is, within sealed vessels. This enables the use of acids, such as nitric acid, at temperatures above their nominal, atmospheric pressure boiling points with a resultant increase in digestion efficiency. Digestions, at atmospheric pressure, are often carried out in open glass beakers on a hot plate with a fume hood. As such, the skill of an experienced analyst is required to ensure that this step is consistently employed from sample to sample. Samples can be undercooked, that is, insufficient digestion time, and overcooked, that is, loss of acid with potential for loss of volatile elements of interest or complete loss of acid resulting in baking the sample to the interior of the glass vessel. Microwave systems, in contrast, are automated, reducing the involvement of the analyst and offering the potential for more consistent performance, compared with hot-plate procedures.

Not all laboratories provided detailed information about their analytical protocols so only general observations can be made from the participants' data. The most complete set of information on the digestion procedures used was provided by laboratories supplying results for the last five rounds of the trial (Table 1) which involved FCAW fume samples (Elga cromacore 308L) and MIG welding fume samples (Metrode supermig 309 L on stainless steel). A summary of the details given by laboratories of their digestion procedures and their relative performance, against fume composition, is given in box and whisker plots in Fig. 6. Given the limited numbers for some of the box and whisker groups in Fig. 6, it is difficult to obtain firm conclusions from the information available. Generally, results for laboratories using an aqua regia digestion (hydrochloric and nitric acid) obtained the lowest recoveries (<50%), especially when analysing MIG type fume (5 from 11 results), suggesting that the aqua regia digestion on a hot plate is not an effective procedure, given its lower oxidation potential. A greater proportion of results from those laboratories using microwave methods were grouped within 20% of the target values. In time, with more PT data, a further statistical analysis is envisaged.


Figure 6
View larger version (28K):
[in this window]
[in a new window]
[Download PowerPoint slide]
  		Fig. 6. Summary of digestion procedures and their performance [289 x 422 mm (96 x 96 DPI)].

 

   CONCLUSIONS
TOP
 ABSTRACT
 INTRODUCTION
 EXPERIMENTAL
 RESULTS
 DISCUSSION
 CONCLUSIONS
REFERENCES
 
  • Most laboratories reported results that underestimated that mass of analyte on filters from welding fume and the magnitude of this under-reporting from some laboratories would be sufficient to seriously affect the interpretation of data in occupational health studies.
  • A laboratory achieving a category 1 (good performance <4%) in the WASP spiked filter PT programme will not necessarily achieve a good performance when analysing welding fume samples and reliance on a laboratory's performance with this type of sample is not advised.
  • Statistical evidence from analysis of variance experiments (not reported here) indicates that the errors are not random but systematic and this is supported visually by the Youden plots of paired filter samples.
  • The digestion step is suspected as the prime cause of the under-reporting.
  • The authors recommend that to ensure the consistent reliability of exposure data from welding fume operations, occupational health professionals should require information from laboratories about their participation and performance in a suitable proficiency-testing programme. Evidence of a consistently reliable performance in a PT programme and/or accurate analysis of a suitable reference material is critical in ensuring that occupational hygienists can have confidence in analytical data supplied by laboratories.

Received December 14, 2007; in final form March 13, 2008


   REFERENCES
TOP
 ABSTRACT
 INTRODUCTION
 EXPERIMENTAL
 RESULTS
 DISCUSSION
 CONCLUSIONS
 REFERENCES
 

Akanes R, Hetland S, Thomassen Y. Quality assurance of working environment analyses: inter-laboratory (1990) Oslo, Norway: Occupational Hygiene Section, Norwegian National Institute for the Working Environment.

Akanes R, Hetland S, Thomassen Y. Quality assurance of working environment analyses: inter-laboratory (1992) Oslo, Norway: Occupational Hygiene Section, Norwegian National Institute for the Working Environment.

Akanes R, Hetland S, Thomassen Y. Quality assurance of working environment analyses: inter-laboratory (1995) Oslo, Norway: Occupational Hygiene Section, Norwegian National Institute for the Working Environment.

Anglov J, Holst S, et al. Iron, manganese, copper and titanium in welding fume dust on filters for internal and external quality assurance Fresenius. J Anal Chem (1993) 345:335–9.[CrossRef]

Butler OT, Howe AM. Development of an international standard for the determination of metals and metalloids in workplace air using ICP-AES: evaluation of sample dissolution procedures through an interlaboratory trial. J Environ Monit (1999) 1:23–32.[CrossRef][Web of Science][Medline]

BSI. Laboratory methods for sampling and analysis of particulate matter generated by arc welding consumables. BS 7384 (1991) London: British Standards Institution.

Health and Safety Executive. EH40/2005, Workplace Exposure Limits, Guidance Note (2005) Sunbury, UK: HSE Books.

Health and Safety Executive. Methods for the determination of hazardous substance (MDHS) 91: metals and metalloids in workplace air by x-ray fluorescence (1998) Sudbury, Suffolk, UK: HSE Books.

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 (2001) Geneva, Switzerland: International Organization for Standardization.

ISO 15202-3. Workplace air—determination of metals and metalloids in airborne particulate matter by inductively coupled plasma atomic emission spectrometry—Part 3: analysis (2004) Geneva, Switzerland: International Organization for Standardization.

ISO 13528. Statistical methods for use in proficiency testing by interlaboratory comparisons (2005) Geneva, Switzerland: International Organization for Standardization.

Stacey P. The performance of laboratories analysing heavy metals in the workplace analysis scheme for proficiency (WASP). Ann Occup Hyg (2006) 50:417–25.[Abstract/Free Full Text]

Stationery Office Ltd. Statutory Instrument No. 3386, Control of substances hazardous to Health Regulations (Amendment) 2004 (2004) London.


Add to CiteULike CiteULike   Add to Connotea Connotea   Add to Del.icio.us Del.icio.us    What's this?



This Article
Right arrow Abstract Freely available
Right arrow FREE Full Text (PDF) Freely available
Right arrow All Versions of this Article:
52/4/287    most recent
men016v1
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Email this article to a friend
Right arrow Similar articles in this journal
Right arrow Similar articles in ISI Web of Science
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Add to My Personal Archive
Right arrow Download to citation manager
Right arrowRequest Permissions
Google Scholar
Right arrow Articles by Stacey, P.
Right arrow Articles by Butler, O.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Stacey, P.
Right arrow Articles by Butler, O.
Social Bookmarking
Add to CiteULike   Add to Connotea   Add to Del.icio.us  
What's this?