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† Owen T. Butler and Alan M. Howe Inorganic Exposure Assessment Section, Health and Safety Laboratory, Broad Lane, SheYeld, UK S37HQ Received 28th September 1998, Accepted 3rd December 1998 Inductively coupled plasma atomic emission spectrometry (ICP-AES) is rapidly overtaking atomic absorption spectrometry (AAS) as the method of choice for the determination of toxic metals in workplace air.However, the few ICP-AES methods that have been published are not well characterised in terms of the eVectiveness of the sample dissolution procedures described and their validation status. The International Standards Organization (ISO) is currently engaged in developing ISO 15202, which will describe a generic method for the determination of metals and metalloids in airborne particulate matter by ICP-AES.One part of the proposed standard deals with dissolution procedures. The ISO work has been supported by a project carried out in the authors’ laboratory to identify, develop and validate sample dissolution procedures for inclusion in the proposed standard.This paper describes an interlaboratory comparison carried out to assess the performance of selected procedures using samples of airborne particulate matter collected on filters with a multiport sampler. Five dissolution procedures were tested. These included an ultrasonic agitation procedure, two hot-plate procedures (based upon NIOSH 7300 and OSHA ID 125G) and two microwave-assisted procedures (based upon EPA 3052).It was shown that the dissolution procedures selected for use in the trial and used internally at HSL generally gave equivalent performance. As expected, a wider spread of results was obtained by participants in the trial. More specifically, there exists some reservation regarding the ability of the ultrasonic and hot-plate procedures to attack fully on a consistent basis some resistant materials, e.g., chromium containing particulate matter.Above all, the trial demonstrated the usefulness of microwave-assisted dissolution procedures in a modern laboratory. comparison carried out to assess the performance of selected Introduction procedures using samples of airborne particulate matter col- The health of workers in many industries is at risk through lected on filters with a multiport sampler.exposure by inhalation to toxic metals and metalloids. Occupational hygienists need to determine the eVectiveness of Experimental measures taken to control workers’ exposure, and this is generally achieved by making personal exposure measure- Procedures tested in the interlaboratory comparison were ments, the accuracy of which relies heavily upon the availability selected on the basis of preliminary work carried out at the and use of validated measurement methods.Health and Safety Laboratory (HSL). A number of potentially Inductively coupled plasma atomic emission spectrometry suitable dissolution procedures were first identified in a litera- (ICP-AES) is rapidly overtaking atomic absorption spectrometry ture search.1 From this list, procedures to be included in the (AAS) as the method of choice for the determination of toxic proposed standard were selected, based not only on their likely metals in workplace air.However, the few ICP-AES methods performance, but also on the need for them to gain acceptance that have been published are not well characterised in terms of from the international community.Procedures already advothe eVectiveness of the sample dissolution procedures described cated by regulatory bodies were therefore favoured. and their validation status. The International Standards An in-house assessment of the selected dissolution pro- Organization (ISO) is currently engaged in developing ISO15202, cedures was then carried out using certified bulk reference which will describe a generic method for the determination of materials and other well characterised bulk materials taken metals and metalloids in airborne particulate matter by ICP- from the working environment.2–4 The results obtained showed AES.The existence of an International Standard will be of benefit that the selected procedures, although not universally applito agencies concerned with health and safety at work, industrial cable, were eVective for a wide range of elements in a wide hygienists, analytical laboratories and industrial users of metals range of materials. and metalloids and their workers. It was therefore decided to proceed with an interlaboratory One part of the proposed standard deals with dissolution comparison to obtain qualitative information on the rugprocedures.The ISO work has been supported by a project gedness of the procedures selected for inclusion in the proposed carried out in the authors’ laboratory to identify, develop and standard. validate sample dissolution procedures for inclusion in the Procedures evaluated proposed standard.This paper describes an interlaboratory Nitric acid–hydrofluoric acid procedure using ultrasonic agitation. Sample dissolution procedures using ultrasonic †©Crown copyright. J. Environ. Monit., 1999, 1, 23–32 23Table 1 Participant details bath and ultrasound for 1 h. Dilute to 25 ml with water in a calibrated flask.Technique used Hot-plate procedure using nitric acid–perchloric acid. Hot- Country ICP-AES AA ICP-MS Total plate procedures using a mixture of nitric and perchloric acid Australia 2 — — 2 have been recommended over many years for the dissolution Belgium — 1 — 1 of air filter samples, e.g., in National Institute for Occupational Canada 1 — — 1 Safety and Health (NIOSH) Method 7300, for elements in Denmark 1 — — 1 airborne particulate matter using ICP-AES8 and in British Germany 2 1 — 3 Standard BS 6691,9 for fume from welding of alloyed steel.Finland — 1 — 1 France 4 2 — 6 The procedure is also commonly used at the HSL for the Italy 1 — — 1 dissolution of air filter samples. A brief summary of the New Zealand — — 1 1 procedure is as follows: Norway 1 — — 1 Spain — 3 — 3 Place the filter sample in a 50 ml beaker and add 5 ml of nitric UK 12 4 1 17 acid and 1 ml of perchloric acid.Place on a hot-plate and heat USA 9 — — 9 until dense white fumes of perchloric acid are evolved. Remove Total 33 12 2 47 from hot-plate and allow to cool (if chromium is required, 1 ml of hydrogen peroxide is used to reduce chromium to the trivalent state).Add 5 ml of hydrochloric acid and heat the sample until near boiling. Allow to cool and dilute to 25 ml agitation are currently much in vogue, as can be seen in the with water in a calibrated flask. open literature, e.g., the determination of lead in environmental matrices5 and the analysis of ambient airborne particulate Hot-plate procedure using hydrogen peroxide–sulfuric acid.matter.6 In addition, procedures using ultrasonic agitation are A hot-plate procedure using a mixture of hydrogen peroxide advocated by INRS (France) and the EPA (USA) for the and sulfuric acid is used in a generic procedure developed by analysis of airborne particulate matter. It was therefore decided the Occupational Safety and Health Administration.10 This to investigate the performance of an ultrasonic procedure procedure, ID-125G, describes the use of ICP-AES for the using a mixture of nitric and hydrofluoric acid.This was analysis of airborne particulate matter. A brief summary of similar to a procedure used by INRS and the French Social the procedure is as follows: Security regional laboratories (CRAMs) for the analysis of air samples collected on glass or quartz fibre filters.7 A brief Place the filter sample in a 50 ml beaker and add 2 ml of summary of the procedure is as follows: sulfuric acid (1+1), followed by a few drops of hydrogen peroxide.Place on a hot-plate and heat until dense white Place the filter sample in a disposable graduated 30 or 50 ml centrifuge tube, add 3 ml of hydrofluoric acid and 2 ml of fumes of sulfuric trioxide are evolved.Remove from the hotplate and allow to cool. Add 5 ml of hydrochloric acid and nitric acid and cap. Place in a pre-heated (60±2 °C) ultrasonic Table 2 Digestion procedures tested Procedure Acid mixture Procedure based upon No. of data sets returned Ultrasonic HNO3–HF INRS/CRAM method 8+1a Hot-plate HNO3–HClO4–HCl NIOSH 7300 11+1a Hot-plate HNO3–H2SO4–HCl OSHA 125G 14+1a Microwave-assisted HNO3 EPA 3052 9+1a Microwave-assisted HNO3HF (+HCl) EPA 3052 6+1a Microwave-assisted HNO3–HClO4–HF (+HCl) EPA 3052 1a Total 48+6a aData returned by HSL.Table 3 Filter sample details Elements present for purpose of trial Filter type Fume/dust source Sample ID Ag Cd Co Cr Cu Fe Mn Mo Ni Sn W Zn Welding fume Corofil 309L flux core arc F1/A/XXX m m m m welding F1/B/XXX Welding fume Nimrod 182 manual metal arc F2/A/XXX m m m m welding F2/B/XXX Welding fume Cobstel 6 flux core arc F3/A/XXX m m m m welding F3/B/XXX Welding fume Nimrod C276 flux core arc F4/A/XXX m m m m m welding F4/B/XXX Silver solder TIG welding using Sif silver F5/A/XXX m m m m fume solder No. 42 F5/B/XXX Ore dust IGS 26 dust, sieved, F6/A/XXX m m m m m wet-milled and dried.Dust F6/B/XXX generated using Wright dust feeder 24 J. Environ. Monit., 1999, 1, 23–32Table 4 Data obtained at HSL (mg per filter) Dissolution with HNO3–HClO4–HF in a Ultrasonic agitation microwave oven: nominal data Filter Element with HNO3–HF: procedural data mean±s (n=4) Filter F1/A/XXX Cr 25.5, 24.9 38.2±1.4 Fe 53.9, 52.7 56.9±2.0 Mn 53.4, 52.6 58.3±2.0 Ni 7.1, 6.8 8.1±0.1 Filter F1/B/XXX Cr 59.8, 58.9 74.1±2.6 Fe 96.4, 97.8 99.0±2.4 Mn 99.8, 101 107±3 Ni 13.3, 13.5 15.0±0.7 Filter F2/A/XXX Cr 8.1, 8.0 8.8±0.4 Fe 9.4, 8.9 8.7±0.5 Mn 20.7, 20.3 21.6±0.5 Ni 12.2, 12.4 13.0±0.3 Filter F2/B/XXX Cr 21.6, 22.4 22.9±1.0 Fe 6.6, 6.6 5.5±0.7 Mn 58.3, 58.1 57.3±0.9 Ni 32.2, 32.6 32.4±0.4 Filter F3/A/XXX Co 104, 104 115±1 Cr 39.6, 39.1 55.6±0.9 Fe 25.7, 25.8 25.8±0.9 Mn 15.8, 15.8 18.1±0.6 Filter F3/A/XXX Co 224, 224 243±3 Cr 132, 133 153±3 Fe 40.2, 37.3 40.8±1.3 Mn 38.0, 38.1 41.4±0.5 Filter F4/A/XXX Cr 31.5, 33.1 45.7±1.0 Fe 13.9, 14.7 15.5±0.5 Mn 13.0, 12.3 15.1±11.1 Mo — 8.3±0.2 Ni 26.2, 27.5 32.8±0.2 Filter F4/B/XXX Cr 84.4, 83.1 101±3 Fe 27.1, 27.8 28.3±1.2 Mn 33.6, 33.4 37.4±1.6 Mo — 20.3±0.3 Ni 50.0, 50.0 57.8±1.1 Filter F5/A/XXX Cd 991, 989 992±16 Cu 15.9, 15.5 14.9±0.1 Zn 453, 451 459±6 Filter F5/B/XXX Cd 405 407±4 Cu 9.7 9.5±0.4 Zn 185 189 ±2 Filter F6/A/XXX Cu 32.7, 34.2 33.4±0.6 Fe 164, 171 179±2 Mn 16.1, 16.7 17.7±0.4 W — 165±5 Filter F6/A/XXX Cu 13.5, 13.5 12.9±0.8 Fe 66.3, 66.8 68.2±0.8 Mn 6.8, 5.0 6.9±0.8 W — 59.8±1.7 heat the sample until near boiling.Allow to cool and dilute standard.Therefore, for the purpose of this interlaboratory comparison, procedures using nitric acid and a combination to 25 ml with water in a calibrated flask. of nitric and hydrofluoric acid were tested. A brief summary of the procedure is as follows: Microwave-assisted closed vessel procedure. With the increasing sophistication of analytical instrumentation, sample prep- Place the filter sample in a digestion vessel, add 5 ml of nitric aration has become the limiting step in the overall analysis acid or 4 ml of nitric acid and 1 ml of hydrofluoric acid and scheme.The use of microwave-assisted digestions is an attractcap. Place in the microwave oven and set a programme to ive proposition, since sample dissolution times can be shortoperate the following temperature profile: to reach 180±5 °C ened considerably.In addition, laboratory microwave ovens in less than 10 min and to hold at 180±5 °C for 15 min. Allow are mostly microprocessor-controlled, allowing automated to cool and, if necessary, add 5 ml of hydrochloric acid and operation. Microwave digestion techniques are becoming repeat the digestion programme.Transfer and dilute to 25 ml increasingly popular in analytical laboratories. In the occuwith water in a calibrated flask. pational hygiene field, a number of procedures suitable for the analysis of airborne particulate matter have been published, e.g., in Health and Safety Executive (HSE) methods MDHS Other dissolution procedures. It is proposed that ISO 15202 will contain a number of other dissolution procedures, includ- 12 and 4211 and papers by Paudyn and Smith12 and Bettinelli et al.13 Along similar lines, Method 3052 developed by the ing a leach procedure for water soluble metals and metalloids, a nitric–hydrochloric acid hot-plate procedure and a micro- Environmental Protection Agency14 is a total digestion procedure for various environmental matrices.It is planned to wave procedure involving the use of perchloric acid. Additionally, plans are in hand to investigate and, it is hoped, incorporate a procedure based upon EPA 3052 in the proposed J. Environ. Monit., 1999, 1, 23–32 25Table 5 Data obtained at HSL (mg per filter) Dissolution with Dissolution with Dissolution with Dissolution with Dissolution with HNO3–HClO4–HCl H2O2–H2SO4–HCl HNO3 HNO3–HF HNO3–HClO4–HF Filter Element on a hot-platea on a hot-platea in a microwave ovena in a microwave ovena in a microwave ovenb F1/A/XXX Cr 37.6 32.4 37.5 37.7 38.1±1.1 Fe 58.8 55.9 58.9 59.5 58.9±2.2 Mn 57.6 56.5 57.8 58.0 57.6±1.4 Ni 8.0 7.1 7.7 7.5 8.2±0.2 F1/B/XXX Cr 74.0 63.5 74.2 74.8 75.2±1.7 Fe 101 96.9 102 101 102±2 Mn 107 100 109 108 108±0.2 Ni 15.1 14.1 14.8 14.0 15.1±0.2 F2/A/XXX Cr — — 8.7 8.9 9.2±0.4 Fe — — 10.5 11.7 10.8±0.7 Mn — — 21.6 20.7 21.5±0.5 Ni — — 12.5 12.1 12.8±0.3 F2/B/XXX Cr 24.5 20.9 22.1 22.9 23.2±0.8 Fe 9.1 8.6 6.5 6.8 6.8±0.5 Mn 60.6 56.9 57.8 58.4 58.4±2.2 Ni 32.8 31.2 32.2 30.9 32.5±0.3 F3/A/XXX Co 109 110 115 114 116±2 Cr 50.4 46.8 52.9 51.1 55.2±1.2 Fe 27.6 26.8 27.5 27.5 26.9±1.0 Mn 17.4 17.2 17.8 17.9 17.9±0.6 F3/B/XXX Co 226 237 236 242 241±6 Cr 138 142 145 147 153±4 Fe 40.4 39.9 39.4 40.5 40.4±1.2 Mn 38.9 40.1 39.3 40.2 41.2±0.5 F4/A/XXX Cr 42.0 36.5 44.4 44.6 46.0±0.9 Fe 15.7 15.5 16.9 17.0 17.0±0.3 Mn 12.7 12.8 14.1 14.0 15.0±0.5 Ni 32.6 28.2 31.9 33.3 33.3±0.9 F4/B/XXX Cr 95.7 92.4 99.6 102 102±2 Fe 29.8 31.8 30.3 29.7 29.1±0.3 Mn 35.9 36.1 37.2 37.5 37.6±0.6 Ni 58.0 56.5 58.0 56.9 58.8±1.0 F4/A/XXX Mo — — — — 7.3±0.2 F4/B/XXX Mo — — — — 18.6±0.6 F5A/XXX Cd 991 985 984 971 969±19 Cu 14.4 14.5 15.6 14.3 14.9±0.7 Zn 426 417 448 442 447 F5B/XXX Cd — — 410 398 401±9 Cu — — 9.4 8.7 9.5±0.3 Zn 188 188 187±4 F6A/XXX Cu 32.1 32.7 34.6 32.6 32.7±1.0 Fe 176 179 173 177 177 ±4 Mn 17.4 16.3 17.4 17.4 17.2±0.4 W — — — — 168±6 F6B/XXX Cu 11.9 12.8 13.2 12.6 12.8±0.1 Fe 63.2 81.9 67.2 68.2 70.2±3.1 Mn 6.2 7.1 6.9 6.7 6.6±0.5 W — — — — 64.7±3.7 aProcedural data.bNominal data, mean±s (n=6). incorporate a procedure using an open focused-microwave the use of blank filters, quality control and spike solutions. Participants also received a customised result sheet to report digestion system. their data. Laboratories involved Samples In order to recruit participants for the interlaboratory comparison, a mail shot was sent to approximately 450 laboratories Each participant received air filter samples (described below), worldwide.Laboratories invited to take part included partici- associated blank filters, ready to run matrix matched quality pants in NIOSH’s PAT and HSE’s WASP proficiency testing control solutions and spike solutions. Quality control solutions schemes, agencies concerned with health and safety at work were supplied in order to verify instrument calibrations.and university and other research laboratories with interests Sample manipulation and work-up procedures were assessed in atomic spectrometric techniques. Sixty-seven laboratories using the spike solutions, i.e.the spike solutions were used to signed up for the trial, with 47 laboratories returning data at perform recovery tests. the time of writing. Pertinent data regarding these laboratories can be found in Tables 1 and 2. Air filter samples Prior to the start of the trial, participants received drafts of the dissolution procedures. Further instructions were sent out Each participant received six diVerent air filter samples, each at two loadings (A and B).Participants were supplied with with the samples. These included information regarding the elemental composition of the samples and solutions, indicative filters loaded with welding fume, solder fume and ore dust. Such samples represent only a small fraction of the types of concentration ranges and procedural information regarding 26 J.Environ. Monit., 1999, 1, 23–32Fig. 1 Participants’ data using ultrasonic agitation procedure. dust which can be found in workplace air, but they did contain Filter sample characterisation a fair range of metals and metalloids which have occupational Sample to sample variability. The multiport sampler was exposure limits. Further details regarding the air filter samples capable of producing near identical air filter samples.However, can be found in Table 3. slight diVerences in individual flow rates and the annular shape Participants evaluating the HNO3–HF ultrasonic agitation of the multiport sampler gave rise to variations in mass loading procedure were supplied with samples collected on 25 mm from filter to filter. Other users of such multiport sampling diameter (1.2 mm pore size) quartz fibre (QF) filters.devices15,16 have attempted to minimise this variation by Participants evaluating the other procedures were supplied applying corrections either using flow rate measurements with samples collected on 25 mm diameter (0.8 mm pore size) and/or by using gravimetric analysis. mixed cellulose ester (MCE) membrane filters.At HSL, individual filters were removed from their sampling cassettes and analysed by wavelength dispersive X-ray fluorescence spectrometry (XRFS). Details regarding the use of Air filter sample production XRFS for the analysis of air filter samples is more fully Air filter samples were collected in batches of 114 using a described in MDHS 91.11 Using XRFS, it was possible to multiport sampler.Similar samplers have been described else- measure elemental variations from filter to filter, something where.15,16 Three-part 25 mm sampling cassettes with 2 in cowl that cannot be achieved with flow rate and/or gravimetric (SKC, Blandford Forum, Dorset, UK) were used to collect measurements. The data thus obtained were then used to the samples. A welding fume box, as described in BS 7384,17 correct the participants’ results for within-batch variability on was used to collect welding fume produced under controlled an element by element basis.Results from the interlaboratory conditions. The fume was sampled into the multiport sampler comparison and from other similar multiport sampling exerthrough a length of flexible plastic hosing connected to the cises18 suggest that this elemental variability is typically less exhaust stack of the welding fume box.The refinery ore filter than 3% (one standard deviation), provided that obvious samples were produced in a two stage fashion. Initially a outliers are excluded, e.g., samples collected with partially certified bulk reference material of a geological origin (IGS blocked critical orifices.standard) was reduced in particle size using a combination of sieving and wet micromilling. Dried dust was subsequently Nominal loadings. A number of filters from each batch were analysed at HSL to provide nominal loading values. These packed into a Wright dust generator, and this was used to produce a dust cloud inside a dust box. Once again, the filters were subjected to microwave dissolution, using a procedure similar to that described above, and the resultant resultant dust cloud was sampled into the multiport sampler using a length of flexible plastic hosing.solutions were analysed using ICP-AES. A 3+1+1 mixture J. Environ. Monit., 1999, 1, 23–32 27Fig. 2 Participants’ data using hot-plate dissolution procedure.of nitric, perchloric and hydrofluoric acid was used to digest involving the use of homogeneous bulk samples. Instead the following procedure was used to process the returned data: the the filters. The performance of this procedure had previously been verified in-house.4 eVect of sample to sample variability was minimised using normalised XRFS data; blank subtraction was carried out in cases where participants reported blank values; outlier data were Sample packing and transportation rejected using quality control and spike recovery data; elemen- Air filter samples were placed in individually labelled 49 mm tary statistics were performed on the accepted data; and participlastic Petri dishes (Gelman Sciences, Ann Arbor, MJ, USA).pants’ data were compared with data obtained at HSL.Prior to shipment, each membrane filter sample was spiked with approximately 400 ml of propan-1-ol (HPLC grade). This Minimisation of the eVects of within-batch sample variability had the eVect of slightly dissolving the mixed cellulose ester Correction factors, based upon XRFS analysis, were applied filter, thus encapsulating the deposit and preventing loss of to participants’ results to minimise within-batch sample varia- sample from the filter in transit.Participating laboratories also bility. These factors were typically in the range 0.97–1.03. received a spiked blank filter to assess potential impurity levels in the propan-1-ol. Quartz fibre filters, being depth filters, did Blank subtraction not require this treatment, because the deposit is embedded in the fibres. Quality control and spike solutions were supplied Participants who noted contributions from filter blanks and/or in prewashed 30 ml Nalgene bottles (LDPE). Samples were reagents were asked to submit these data.Where appropriate, shipped to participating laboratories in Nalgene Passport UN these blanks were subtracted from the participant’s sample combination boxes by specialised courier. These combination results.boxes met the required regulatory standards for shipment of chemicals, e.g., UN transportation regulations and air ship- Rejection of outliers ment regulations. Participants’ quality control and spike recovery data were used to verify instrumental calibration and to ascertain the ability Results of the analyst to perform quantitative manipulations during the sample dissolution stage, respectively.Data returned from participants (filter results, quality control and spike recovery data) were entered into a customised Excel In order for sample data to be accepted: quality control data were required to be within ±10% of the participants’ spreadsheet. Once entered, the data set was double checked for potential transcription errors.Although established protocols mean value and spiked recovery data were required to be in the range 90–110% (and with an RSD of <10%, if replicate for statistical analysis of data returned from interlaboratory trials exist, e.g., ISO 5725,19 they are more suitable for trials data were supplied). 28 J. Environ. Monit., 1999, 1, 23–32Fig. 3 Participants’ data using hot-plate dissolution procedure. If these conditions were not met, data supplied for the randomly selected from each batch of samples were analysed. samples was closely examined and the following action was Procedural data represent the data obtained at HSL using taken: if a systematic bias was noted for quality control, spike each of the selected dissolution procedures, i.e., by HSL acting recovery and sample data, then the sample data were rejected; as a participant in the trial.These data were used as a if a systematic bias was noted for the quality control and spike benchmark to demonstrate what could be achieved using the recovery data, but data for the samples showed no systematic selected dissolution procedures. A shortage of filter samples bias, i.e., the data fell within the range obtained by other meant that only two replicate quartz filter samples were laboratories, then the data were accepted; and if high filter analysed using the ultrasonic procedure.An acute shortage of results from participants were noted, these were rejected if membrane filters meant that only single samples could be they fell outside the HSL nominal value +3s.analysed using the hot-plate and microwave sample dissolution procedures. Accepted data sets Elementary statistics were performed on the accepted data. Data presentation Mean, median, standard deviation and RSD were calculated using a customised spreadsheet for each of the twelve filter The results obtained are presented in two formats.Data samples analysed by the five dissolution procedures. obtained in-house at HSL (procedural and nominal ) are given No attempt was made to evaluate results for silver and tin in Tables 4 and 5. Participants’ data are reproduced in the as few participants returned data for these elements. The form of standardised box and whisker plots (Fig. 1–5). For quality of returned data for these elements was also poor, illustrative purposes, these plots represent all the data received, which was not unexpected.This was attributable to the use of i.e., no outliers removed. an inappropriate acid mixture and/or incompatibility of the Each figure represents participants’ data obtained using one elements concerned with certain acids. of the selected dissolution procedures.Each figure is broken down into six plots representing data for each of the six filter Comparison with HSL data types. Each individual plot shows elemental data for a particular filter type at two mass loadings ( labelled A and B). The Participants’ data were compared with two sets of data box and whisker plots follow the standard convention, i.e., obtained at HSL, termed nominal and procedural.Nominal four quartiles and median are shown. Participants’ data have data represent the data obtained at HSL using the high been normalised against the HSL nominal data with an upper performance microwave dissolution procedure (outlined and lower limit set at one standard deviation (represented by above). The data represent an estimation of the true elemental loading on each filter.Four quartz and six membrane filters the three vertical dashed lines). The relationship between the J. Environ. Monit., 1999, 1, 23–32 29Fig. 4 Participants’ data using microwave assisted dissolution procedure. X and Y variables is represented by the following equation: method was satisfactory (good agreement between participants’ data, HSL’s procedural and nominal data) for elements Y=6 [(X-nominal )/(nominal+s)-(nominal-s)] such as cadmium and zinc.This is consistent with XRD data, which showed that these elements existed as simple oxides, Participants obtained these data in a diVerent format. Data were supplied in a numerical format. For graphical comparisons, i.e., acid soluble. Poorer quality data were obtained for elements such as chromium and nickel.On occasions, results the same data were supplied in the form of bar graphs (A3 colour). Box and whiskers plots were used here in an attempt to were up to 30% lower than those obtained using the microwave assisted dissolution procedure. Preliminary work carried out produce a concise graphical representation of the data. at HSL4 investigated the eVectiveness of this procedure using milligram amounts of bulk reference materials.This showed Discussion similar trends for these more ‘refractory’ elements. Airborne particulate matter collected from the workplace atmosphere Sample morphology often contains chromium and nickel in the form of spinel Six diVerent sample matrices were examined: four welding oxides, which can be diYcult to digest.fume matrices, a silver solder fume matrix and a geological The data obtained at HSL using this ultrasonic procedure ore based matrix. The welding fume matrices were derived were slightly inferior to those obtained by the participating from flux core arc and manual metal arc welding processes. laboratories. This may in part be explained by the fact that Qualitative X-ray diVraction (XRD) identified a spinel type the majority of laboratories involved in testing this procedure oxide as the dominant crystalline phase in each of these were French and consequently, as mentioned above, experimatrices. These spinel type oxides can generally be represented enced in its use. Additionally, whilst samples digested at HSL by the formula AB2O4 (where A=Co, Fe, Mg, Mn and Zn were allowed to stand prior to analysis, thus allowing sedimenand B=Al, Cr, Co and Fe).The silver solder fume showed tation of undissolved residues, it is known that some of the crystalline phases consistent with Ag2O, CdO and ZnO. The participating laboratories homogenised their samples prior to diVraction pattern also suggested the presence of phases con- nebulisation, thus allowing the possibility that fine undissolved sistent with CdAg alloy and Ag metal.The geological ore particulates were also analysed, i.e., slurry nebulisation. based matrix, prepared from a certified reference material, IGS 26 Tin–Tungsten Ore, showed diVraction patterns consistent with the presence of SnO2, CaWO4, FeWO4 and Fe2O3. Hot-plate digestion procedures The quality of data obtained using the hot-plate procedures was Ultrasonic agitation procedure found to be good when applied at HSL.There was good agreement for most elements between HSL’s procedural data and the The quality of data obtained using this procedure was found to vary depending upon the element. The performance of the nominal data. One obvious exception was chromium.Chromium 30 J. Environ. Monit., 1999, 1, 23–32Fig. 5 Participants’ data using microwave assisted dissolution procedure. containing spinel type oxides can be diYcult to digest fully, closed microwave-assisted dissolution procedures than using other procedures, although it has been shown that equivalent especially using hot-plate digestion procedures. This is consistent with previous investigations carried out in-house.3 performance can be obtained using hot-plate procedures for many elements in the matrices examined.The ultrasonic The performance of participating laboratories in applying the hot-plate procedures was slightly inferior. Both hot-plate agitation procedure can be a rapid and eVective method for certain elements and matrices.The methods themselves were procedures require the analyst to heat samples to a ‘fuming’ end-point. Experience is important in such circumstances, as adjudged by the participants to be straightforward to follow, although other factors may hinder their usability, e.g., cost insuYcient ‘fuming’ will lower the digestion eYciency, and insuYcient/excessive ‘fuming’ will introduce an acid matrix (e.g., expense of microwave systems), safety (e.g., use of perchloric and hydrofluoric acid) and compatibility (e.g., mis-match and hence a measurement bias.It is fair to say that many laboratories analysing environmental samples using hot- requirement for hydrofluoric acid resistant nebuliser systems). plate digestions avoid using higher boiling-point acids such as perchloric acid and sulfuric acid, and this might in part account for their inferior performance.Future work A small amount of further practical work is necessary. An Closed vessel microwave-assisted digestion procedures evaluation of an open focused microwave assisted dissolution procedure is planned, and parallel investigations will The quality of data obtained using the microwave procedures was found to be good when applied at HSL.For elements re-evaluate the ultrasonic procedure with specific reference to its performance in digesting chromium containing samples. such as chromium and nickel, bound in more refractory phases, the use of perchloric and hydrofluoric acid is at times Further refinement of the current draft of the proposed International Standard will take place, to take account of required to ensure complete recoveries. The performance of participating laboratories in applying the microwave pro- comments received in the public enquiry and to ensure consistency of language between the diVerent methods.Consideration cedures was inferior to that of HSL, although the range of data (± one standard deviation) overlapped with the HSL might also be given to providing the user with guidance to help decide which is the most appropriate dissolution nominal data for many elements. procedure for a particular application.However, drafting a selection guide will not be facile, as this inherently depends Conclusions upon many factors, such as the nature of the sample matrix, equipment availability, experience of staV and safety The overall performance of participating laboratories was creditable. As expected, better performance was obtained using considerations.J. Environ. Monit., 1999, 1, 23–32 318 US National Institute for Occupational Safety and Health, Acknowledgements NIOSH Manual of Analytical Methods, US Government Printing OYce, Washington, 4th edn., 1994 DHHS Publication The authors thank S.D. Bradley (HSL), G. J. Carter (TWI), No. 94–113. K. Y. K. Chung (HSL), J. E. Chisholm (HSL), R. D. Foster 9 British Standards Institution, Sampling and Analysis of Welding (HSL) and P. R. Stacey (HSL), who assisted in practical Fume, BS 6691: Part 1, BSI, London, 1986. aspects of this work. Clerical support from P. Twigg, 10 US Occupational Safety and Health Administration, OSHA M.Turner and D. Hoyland is acknowledged. Analytical Methods Manual, USDOL/OSHA, Salt Lake City, 2nd edn., 1991. Acknowledgement of the contributions of the working group 11 Health and Safety Executive, Methods for the Determination of involved in developing ISO 15202 is also noted. Many thanks Hazardous Substances. MDHS 12/2, Chromium and Inorganic are due to the immense contributions of the participating Compounds of Chromium in Air, 1996.MDHS 42/2, Nickel and laboratories. This work was funded and carried out on behalf Inorganic Compounds of Nickel in Air (Except Nickel Carbonyl), of the Directorate of Science and Technology, the Field 1996. MDHS 91, Metals and Metalloids in Workplace Air by Operations Division and the Health Directorate of the Health X-Ray Fluorescence Spectrometry, 1998.HSE Books, London. 12 A. M. Paudyn and R. G. Smith, Can. J. Appl. Spectrosc., 1992, and Safety Executive. 37, 230. 13 M. Bettinelli, U. Baroni and N. Pastorelli, J. Anal. At. Spectrom., References 1987, 2, 485. 14 Environmental Protection Agency (US EPA) OYce of Solid 1 O. T. Butler and A. M. Howe, ICP-AES Method for Metals in Air: Waste, Microwave Assisted Acid Digestion of Siliceous and a Critical Review of Published Dissolution Procedures, HSL Organically Based Matrices, Method 3052, in Test Methods for Internal Report IR/IS/96/02, Health and Safety Laboratory, Evaluating Solid Waste Physical/Chemical Methods, SW-846, SheYeld, report available from the authors. National Technical Information (NTIS) Service, Springfield, 2 O.T. Butler, S. D. Bradley and A. M. Howe, A Revised Dissolution NTIS Order No. PB97-501928INQ, 1995. Procedure for the Determination of Nickel and Its Inorganic 15 R. Aksnes, S. Hetland and Y. Thomassen, Quality Assurance of Compounds Using Flame Atomic Absorption Spectrometry, HSL Working Environment Analyses: Interlaboratory Trials—Elements, Internal Report IR/L/IS/94/04, Health and Safety Laboratory, Norwegian National Institute for the Working Environment, SheYeld, report available from the authors. Occupational Hygiene Section, Oslo, 1990, 1992 and 1995 (in 3 O. T. Butler, S. D. Bradley and A. M. Howe, A Revised Dissolution Norwegian, except for 1990 report, HSE Translation Procedure for the Determination of Chromium and Its Inorganic No. 14280F). Compounds Using Flame Atomic Absorption Spectrometry, HSL 16 J. Anglov, E. Holst, S. Dyg and J. M. Christensen, Fresenius’ Internal Report IR/L/IS/95/12, Health and Safety Laboratory, J. Anal. Chem., 1993, 345, 335. SheYeld, report available from the authors. 17 British Standards Institution, Laboratory Methods for Sampling 4 O. T. Butler, S. D. Bradley and A. M. Howe, ICP-AES Method and Analysis of Particulate Matter Generated by Arc Welding for Metals in Air: Preliminary Tests on Bulk Reference Materials Consumables, BS 7384, BSI, London, 1991. to Test the EVectiveness of the Proposed Sample Dissolution 18 O. T. Butler, unpublished data, 1997. Procedures, HSL Internal Report IR/IS/96/04, Health and Safety 19 International Standards Organization (ISO), Precision of Test Laboratory, SheYeld, report available from the authors. Methods—Determination of Repeatability and Reproducibility for 5 K. Ashley, Electroanalysis, 1995, 7, 1. a Standard Test Method by Inter-laboratory Tests, ISO 5725, 6 L. Jalkanen and E. Hasanen, J. Anal. At. Spectrom., 1996, 11, 365. Geneva, 2nd edn., 1986. 7 Metanal Technique Operatoire Standardise�e INRS/CRAM, Technique No. 1B, Institut National de Recherche et de Securite, Vandoeuvre, 1993. Paper 8/07526C 32 J. Environ. Monit., 1999, 1, 23–