|
11. |
Foreword. AIRMON '96, The Second International Symposium on Modern Principles of Air Monitoring |
|
Analyst,
Volume 121,
Issue 9,
1996,
Page 1151-1152
Jan Olof Levin,
Preview
|
PDF (971KB)
|
|
摘要:
Analyst, Sqtember 1996, Vol. 121 1151 Foreword AIRMON '96, The Second International Symposium on Modern Principles of Air Monitoring AIRMON '96 held in Salen, Sweden, on February 5-8, 1996, was organized by the National Institute for Working Life (NIWL), Ume& Sweden, the National Institute of Occupational Health (NTOH), Oslo, Norway and the Nordic Institute for Advanced Training in Occupational Health (NIVA). This was the second AIRMON Symposium, the first was held in Geilo, Norway in 1993. All key lectures and some poster presentations from AIRMON '93 were published in a special issue of The Analyst. The AIRMON Symposia are intended not only for scientists in academic circles working in the field of air monitoring, but also for environmental and health professionals in industry and officials from governmental and regulatory agencies.The plenary programme for AIRMON '96 was planned with a view to providing a comprehensive overview of the latest knowledge in this important field. Since the world's leading authorities in the field were present, the symposium was an excellent forum for the exchange of ideas as well as an opportunity for private informal discussions, for all those involved in method development, air sampling, regulatory issues or other areas related to air monitoring. The participants, a total of 130, came from most parts of Europe to the small mountain village of Salen, situated in the Swedish county Dalarna, about 400 km north-west of Stockholm, and 250 km north-east of Oslo. Oral presentations (24 in all) by invited speakers were given in five sessions: (1) measurement strategy and quality (four lectures); (2) measurement of gases, vapours and mixed phases (seven lectures); (3) measurement of micro-organisms and metabolites (two lectures); (4) particle measurements (seven lectures); and ( 5 ) ambient air measurements (four lectures).Each presentation was given 45 min, discussion included. The participants were generally very active in these discussions. Session 1 focused on sampling as the critical step in measurements. European quality initiatives, such as standardi- zation and the organization of sampling exercises were reviewed, as well as the US NIOSH guidelines for air sampling and analytical method development and evaluation. Session 2 focused on newly developed methodology for measurement of gaseous and particulate organics.Diffusive and denuder samplers and thermal desorption in combination with a variety of new adsorbents are the methods of choice for measurements in the future. Session 3 dealt with bioaerosols and volatile metabolities from micro-organisms. This is a new area of growing concern. Researchers in Scandinavia have been especially active in this area, and the measurement methods developed open up possibilities for dealing with the sick-building syndrome. Session 4 was devoted to particle measurements. Toxicolog- ical aspects of aerosol measurements, including speciation and biological monitoring were discussed. Most lectures focused on the important issue of size-selective particle sampling, and the implications of the new standards for existing particle exposure 1 imits.In Session 5 methodology and quality assurance in ambient air measurements were discussed. The many advantages with the use of diffusive sampling for ambient air sampling was demonstrated. Ongoing European harmonization in ambient air measurement was also reported. The poster session comprised 28 posters. The topics covered the whole spectrum of workplace, indoor and ambient air measurements, from ozone measurements in the European alps, to bioaerosol exposure in connection with household waste collection. Novel instrumentation, such as portable FTIR measurement systems, as well as new devices for personal Participants enjoying the conference dinner at AIRMON '96. From L to R: Dr. Yngvar Thomassen, Dr.Eugene Kennedy, Dr. Eileen Birch, Dr. K. J. Saunders and Dr. Richard H. Brown.1152 Analyst, September 1996, Vol. 121 monitoring of exposure, were presented. A number of posters dealt with particle measurements using various size-selective samplers. The quality of the posters was generally high. During the conference, the scientific programme was from 08.30 to 12.00 and 16.00 to 19.00, allowing the participants excellent opportunities for exhibition- and poster-viewing and outdoor activities between 12.00 and 16.00. Sunny weather helped promote downhill and cross-country skiing, snow- mobile safaris and long walks. The outdoor activities were very refreshing and sharpened the mind for the afternoon session. This conference format with mid-day outdoor activities was much appreciated by the participants. . The papers that made it through the peer review procedure of The Analyst and comprise this special issue of The Analyst are fifteen of the key-note lectures and eleven of the poster presentations, and they give a broad overview of the whole area of ‘Air Monitoring’. The next AIRMON conference will be held in Geilo in Norway on 10-14 February 1999. Reference 1 The Analyst, 1994, 119, 1-107. Jan Olof Levin National Institute for Working Life P.O. Box 7654, S-90713 Umed. Sweden Yngvar Thomassen National Institute of Occupational Health P.O. Box 8149 DEP N-0033 Oslo 1, Norway
ISSN:0003-2654
DOI:10.1039/AN9962101151
出版商:RSC
年代:1996
数据来源: RSC
|
12. |
Editorial. Airmon '96, The Second International Symposium on Modern Principles of Air Monitoring |
|
Analyst,
Volume 121,
Issue 9,
1996,
Page 1153-1154
Harp Minhas,
Preview
|
PDF (344KB)
|
|
摘要:
Analyst, September 1996, Vol. I21 1153 Editorial Airmon ‘96, The Second International Symposium on Modern Principles of Air Monitoring I consider myself privileged to have been asked by the Organizing Committee to attend and encourage publication in The Analyst of the work presented at this Second Symposium on Air Monitoring. Airmon ’96 was a well run, well thought out and well planned conference that provided just the right balance of work and relaxation to stimulate some excellent discussions and debates. These discussions were enhanced by the fact that the Symposium brought together professionals and researchers from various sections of related industries, allowing different sectors to identify not only common bonds, problems and requirements but also highlighted those areas where further work, additional guidelines and more standards are essential. The papers in this issue go some way towards providing a basis for future developments in this field and also emphasize the areas where workers require more experience and need to concentrate their efforts.The first such area was sampling strategy and statistics; this topic and variability of exposure are extremely important subjects that do not receive the attention they deserve. Many authors could have vastly improved the impact of their work by use of appropriate statistics and validation procedures. Several authors gave no measurement of uncertainty in their data, thus bringing in to question the validity of some of their results. However, after discussions at the conference itself and through the refereeing process of The Analyst most authors appear to have addressed these problems.In this issue Dr. Erik Olsen presents a good overview of how sampling contributes to measurement errors on p. 1155 and what can be done about it. The effects of storage of various types of samples/samplers are not yet fully understood. There are large differences between samplers due to oversampling and undersampling under various conditions. More information is needed on size- dependent sampling efficiencies because if the size of particles is known, it may be possible to predict the inhalation risks more meaningfully. These factors support the idea that all new samplers should be constructed to conform to accepted international guidelines. Generally, more methods and procedures need to be agreed upon and guidelines given for calibration methods. Additional new standard calibration materials are required to improve measurement accuracy.There is a need for international harmonization particularly between the NIOSH(US) and EC(Europe) based methods and guidelines. After all, air pollution/monitoring poses global problems. Dr. L. C. Kenny expands upon some of these ideas in her paper on p. 1233 providing a comprehensive review of the issues and current status of the significant progress that has been made towards international harmonization of philosophies and methods. Professor Vincent’s paper (p. 1207) also sets a firm founda- tion for the ongoing international rationalization and improve- ment of methods for measuring aerosols in the workplace.It provides information on the comparison of samplers in the workplace which is essential before appropriate conventions can be adopted and before revision of the relevant international occupational exposure limits. For a relatively ‘young’ area of analytical science, this area is growing rapidly in terms of the amount and pace of research. This is helped, of course, by the fact that the subject falls under the remit of several scientific disciplines, as well as govern- mental and regulatory agencies. For example, this AIRMON Symposium included presentations about legislation/standards/ guidelines on an international level, statistics/chemometrics, analysis, general chemistry, engineering, physics, envi- ronmental health/pollution, microbiology, toxicology, specia- tion, clinical chemistry and sensors.These subject areas, together, involve a myriad of analytical techniques in order to observe and measure the particular phenomena being studied. The quest for knowledge and cooperation between the participants of this symposium, approaching similar problems from a variety of scientific disciplines and a mixture of working backgrounds, resulted in an extremely productive and enjoyable conference that clearly revealed the true interdisciplinary nature of modern analytical science. Haip Min has Managing Editor The Analyst and Analytical CommunicationsThe Third International Conference on SPECIATION OF ELEMENTS IN BIOLOGICAL, ENVIRONMENTAL AND TOXICOLOGICAL SCIENCES The Torresian Resort Port Douglas, Queensland, Australia, September 15-1 9, 1997 INVITATION AND CALL FOR PAPERS The Organising Committee extends an invitation to all individuals involved in element research or its applications.A major goal of the symposium is to facilitate interdisciplinary and intersector discussion about all aspects of elements requiring an understanding of speciation, including: analytical chemistry; geochemistry; biochemistry; environmental sciences; essentiality and nutrition; medical uses; occupational hygiene; human toxicology; and regulatory aspects. A small number' of travel scholarships will be provided to encourage overseas graduate students to attend and participate. THE SCIENTIFIC PROGRAMME The symposium programme will comprise four days of oral presentations, posters and discussion.All presenters will be asked to focus on new developments in research. Oral presentations (invited or submitted) will be 20 or 30 mins in duration. As at previous symposia (Loen, Norway, 199 1 and 1994) posters will play a central role, after formal viewing each poster presenter will be given five minutes to present the salient features of their work to a discussion group to encourage in depth feedback. SYMPOSIUM LOCATION AND DETAILS The venue of the symposium, is The Torresian Resort of Port Douglas, Australia. This tropical Queensland location is situated near Cairns, between the Great Barrier Reef and the Daintree Rainforest. A Symposium Package rate has been arranged: AUD $155(per person, per night, twin share) and AUD $225 (single occupancy) and includes accommodation (Garden View Room) all meals and morning and afternoon teas.A limited amount of less expensive accommodation (room and board) will be available. The scheduling of this conference will allow the participants to join the XXX Colloquium Spectroscopium Internationale (21-26 September). /I CONFERENCE PROCEEDINGS As with previous Speciation Symposia (see The Analyst 117; 549-691 and 120; 29-30N and 583-763) all papers presented as posters or lectures may be submitted as full papers for publication in a special issue of The Analyst, subject to the normal review procedure of this journal. SOCIAL PROGRAMME All participants and accompanying persons are invited to the symposium reception on Monday evening, September 15, and the dinner on Friday evening, September 19.Because of the numerous attractions available ( e g , swimming, all other watersports, cruises, canoeing, hiking, horse riding etc.) no other formal social events are planned. However, please note that for each full day of scientific sessions, the period 15.30 onwards will be set aside for the enjoyment of the mentioned activities by all. Port Douglas has a comfortable, year round, tropical climate. Day tours to the outer Barrier Reef are available. REGISTRATION FEE The registration fee per delegate is AUD $480 (AUD $150 for students)and includes the cost of the symposium dinner SECRETARIAT Local (Registration) Third Speciation Symposium c/o Dr J. P. Matousek, Department of Analytical Chemistry, The University of New South Wales, Sydney, NSW 2052, Australia Tel : Fax : E-mail : +6123854713 + 61 2 4512322 (home) +6123856141 J.Matousek@unsw .edu.au THE SYMPOSIUM IS ORGANISED BY : The University of New South Wales (Sydney, Australia) The National Institute Of Occupational Health (Oslo, Norway) The Institute of Environment and Health (Universities of Toronto and McMaster, Canada) MAFF CSL Food Science Laboratory (Nonvich, UK) ORGANIZING COMMITTEES Local Programme Graeme Batley (CSIRO, Lucas Heights) Helen Crews (MAFF CSL, UK) Jarda P. Matousek (Sydney, N S W) Evert Nieboer (Hamilton, Canada) Yngvar Thomassen (Oslo, Norway) R. (Dick) Finlayson (Sydney, NSW) D. Brynn Hibbert (Sydney, NSW) Jarda P. Matousek (Sydney, NSW)
ISSN:0003-2654
DOI:10.1039/AN9962101153
出版商:RSC
年代:1996
数据来源: RSC
|
13. |
Effect of sampling on measurement errors |
|
Analyst,
Volume 121,
Issue 9,
1996,
Page 1155-1161
Erik Olsen,
Preview
|
PDF (1047KB)
|
|
摘要:
Analyst, September 1996, Vol. 121 (1155-1161) 1155 Effect of Sampling on Measurement Errors* Erik Olsen National Institute of Occupational Health Lersd Parka114 105, Copenhagen @, Denmark Often the analyst is taken as a guarantor for data quality in spite of the fact that sampling is commonly performed by others. If the analyst ignores sampling uncertainties, the money spent on quality control of analysis may sometimes be in vain. The analyst ought to be aware of the difference between controlling exposure and measuring workers’ exposure at the workplace. When controlling exposure the aim is to ensure that workers’ exposures are below the given occupational exposure limits (OELs); when measuring exposure the aim is to determine what the worker is actually exposed to, on average.In the working environment, exposure is usually controlled by measuring ‘worst case’ situations, i e . , situations where exposure is higher than average by an unknown amount. As pointed out by Eisenhart (cf. Anal. Chem., l981,53,1588A), measuring without a state of statistical control being attained cannot in any logical sense be regarded as measuring anything at all. Except for substances for which the OELs are ceiling limits that must not be exceeded, ‘worst case’ results cannot be used for documenting non-compliance or for risk assessment, epidemiology or standard setting. Measuring workers’ exposure requires estimation of the time weighted average concentration in the exposure period considered (TWACExposure period) by carrying out measurements, preferably over a series of days (TWACDay).Kromhout et al. (Ann. Occup. Hyg., 1993, 37,253) found TWACDay data to be lognormally distributed with a median geometric standard deviation of 2.5. Sampling from such distributions is shown to give very disperse results. Consequently, many measurement days are needed. A TWACExposure period estimate, therefore, is either very uncertain or has been very costly to obtain. In order to obtain more reliable results at an affordable cost, an alternative approach, called the logbook method, has recently been suggested for the estimation of TWACExposure Period. Commonly, workers considered to be similarly exposed are grouped. In contrast, the logbook method groups processes causing similar exposures. The time component of exposure is measured by workers keeping logs of their activities over a period of several weeks.Keywords: Sampling; measurement error; exposure control; exposure measurement; sump ling strategy; bias; exposure variability Introduction In recent years, considerable time and money have been spent on improving the quality of analysis, nationally as well as internationally. Accreditation of laboratories has become com- mon, that is, a laboratory must state the analytical accuracy with which it claims it is capable of achieving. The accredited * Presented at AIRMON ’96, Salen, Sweden, February 5-8, 1996. laboratories are subjected to controls to ensure that their performance is as stated. The analyst is undoubtedly responsible for the accuracy of the result obtained by analysing samples delivered to the laboratory.However, if the analyst does not take wider responsibilities, the time and money spent on quality assurance may sometimes have been in vain. Some examples will be given below, where sampling introduces errors, which are orders of magnitude larger than the errors of analysis in the laboratory. Apart from the analyst, nobody is interested in what is in a sample. Other peoples’ interest is what the sample represents. The content of protein in a sample of grain is not of interest. What is of interest is the average content of protein in the heap of grain from which the sample has been taken. Similarly, it is of no interest what is in a sample of water taken from a river or in a sample collected from the air a worker is breathing at the workplace.What is of interest is the condition of the river and the quality of the air at the workplace. In summary, measurements are performed to obtain qual- itative and quantitative information about a system. The sample collected and analysed must, therefore, be a good representation of this system. Obviously, non-representative samples cannot be compensated for, not even by the best analyst, or by the best and most well controlled analytical method. In order to obtain a sample that is representative of a heap of grain, the collector must take into consideration the variation of protein in space; when sampling from a river the collector must take into consideration variations in space as well as variations over time. When sampling from the air breathed by a worker, attention must be paid to variations in space and time, as well as the performance of the worker and his or her work pattern.Except for workers employed on an assembly-line, workers in modern industry seldom carry out the same tasks every day: they have a shifting work pattern. Fig. 1 shows the work pattern for a worker in a glass fibre reinforced polyester facility. The work pattern was derived from a logbook kept by the worker detailing when he started and stopped the different processes he performed during the log period.’ 100 h 80 > W Q 60 5 -I- : 40 .- c E 20 I& 0 16 17 18 19 20 ‘21’22 23 24 Date Deforming class cutting 0 Fining Polishing Fig. 1 Work pattern for worker No. 55 at facility 2.1156 Analyst, September 1996, Vol.121 Fig. 1 shows that the work pattern differed significantly from day-to-day. Fig. 2 shows the corresponding exposure pattern. The exposures were calculated using the logbook method. 1 l2 Further details on the logbook method are given below. Two Purposes of Measuring in the Working Environment Traditionally, workers' exposures are measured over a full shift, because air quality standards are defined for an 8 h period. That is, workers' exposures are measured over a period where they may perform more than one process.2 Controlling exposure Compliance measurements of the air breathed by workers are measurements that are performed to ensure that pollutant levels in the air are below the air quality standards, e.g., occupational exposure limits (OELs). The questions asked, when compliance measurements are performed, are: (1) 'Which substance(s) are the workers exposed to?' and (2) 'Are concentration(s) below OEL(s) on the day of measurement and on unmeasured days, for both measured and unmeasured workers?'.The principle of 'worst case' measurements is shown in Fig. 3. The sample collector aims to measure situations where the concentration is (as far as possible) above average. If, in spite of this, the result is below the OEL, it is believed that the average exposure is in compliance. If the result is above the OEL, it cannot be concluded that the average exposure is in non- 0.6 0.5 v 0.4 m 0.3 m . 6 0.2 Io.l 1 0 1 16 ' 17 ' 18 ' 19 ' 20 ' 21 ' 22 '23 ' 24 ' 25 ' 26 ' 27 t 28 JL 29 30 1 2 3 d 4 Date Deforming Class cutting APPlYina glass web Fining rm;I]I Polishing Fig.2 Exposure pattern for worker No. 55 at facility 2. II \ "Worst case" result I AM \ 1 do n! 300 200 Concentration/mg m-3 Fig. 3 Principle of measuring for a 'worst case' strategy. Compliance is documented when results are above the (unknown) true mean value (AM) but below the OEL (lognormal distribution). compliance, because the high result may have been caused by measuring a 'worst case' situation, not because the average exposure is above the OEL. Measuring exposure For use in epidemiology, risk assessment, and standard setting, and for considering substances with long half-lives in the human body and effects due to long-term exposure, the relevant measure of exposure is the arithmetic mean concentration3.4 in a well defined exposure period.An exposure period is a period in which no changes occur, which systematically alter the worker's exposure.2 When measuring workers' exposure, the first question asked is the same as when controlling exposure; however, the second question is changed to: 'What are the average concentration levels experienced by the workers in the exposure period considered?'. Commonly, for technical reasons, pollutants in the working environment cannot be collected over a full shift, but must be sampled in a series of consecutive sampling periods. The combined result of such consecutive measurements is reported as the time-weighted average concentration (TWACDa,). The TWACDay is a measure of the average dose rate [dose/(8 h) = (average concentration X 8 h)/(8 h) = average concentration in an 8 h period] received by the worker on the day of measurement.The TWAC for a single day or a few days, however, is not of much interest in itself, as one day is less than 0.5% of the working days in a year. What is of interest is the T W A C E ~ ~ ~ ~ ~ ~ ~ period, the TWAC for the exposure period, for which the TWACDay is one estimate. TWACD,, data may have measurement errors due to: (i) the analysis, including pretreat- ment of the sample(s); (ii) sampling procedure, calibration of pumps and transport of samples to the laboratory; and (iii) sampling strategy, i.e., number of days on which samples, are taken and where and when the samplings are performed. To date, considerable attention has been paid to the reduction and control of the errors of analysis; this is necessary because if the errors of analysis are not well controlled, the other components of the measurement errors cannot be determined.In general, however, the errors of an analysis are under control, provided that the analysis is routinely supervised. Less attention has been paid to sampling procedures and sampling strategies. Using modern equipment, the errors of the sampling procedures are modest.5 The choice of sampling strategy, however, may cause errors, which potentially are much greater than the errors of the sampling procedure and analysis together. Two sources of error are: (1) a potential inhomogeneity in space of the breathing zones of the workers measured and (2) a potential inhomogeneity in time of the object of measurement, viz., the worker's exposure at the workplace during the exposure period.That is, sampling may not be in statistical control. Measurement results are often used as the basis for decisions, which may have consequences for the health and safety of the workers and have economic consequences for the employer. The accuracy of measurement results is, therefore, important. If the responsibility of the analyst is limited to the accurate determination of the content of the samples delivered to the laboratory, the analyst may sometimes produce erroneous results, which may lead to incorrect decisions. The analyst, therefore, ought to know the quality of the data leaving the laboratory and not just the quality of the analytical work performed. For instance, the analyst might refuse io analyse some samples if the results of the analysis are not reliable, e.g., because too few samples have been taken to make statistically valid conclusions.This paper will focus on some sources of errors of sampling, which may be of interest to the analyst, when analysing samples taken from the outdoor as well as the working environment. The examples given are from the working environment, but similarAnalyst, September 1996, Vol. 121 1157 examples can also be found in other fields, where wide and skewed distributions are found. Methods Measurement errors can be divided into two types: unsystematic or random errors and systematic errors. The different types of measurement errors, as defined in IS0 5725,6 are given in Table 1.One additional source of error must be mentioned: gross errors, blunders or mistakes, e.g., errors due to unnoticed departure from the prescribed measurement procedure.7 Gross errors may be caused by an omission or an incorrect action due to lack of understanding or attention. Gross errors are to be distinguished from the two other types of measuremeEt errors, as gross errors cannot be quantified. Unsystematic Errors Unsystematic errors can be divided into errors due to the measurement procedure and errors due to the variability of the object of measurement: o2 = 02Measurernent Procedure 020bject ( 1 ) where o2 is the variance of the result, CJ’Measurement Procedure is the variance due to the measurement procedure and 020bJect is the variance due to the variation of the measurement object. The measured property of the object of measurement does not, in general, have a single value, but a distribution of values.The variance of the measurement procedure can be separated into two terms: (2) where 02Sampllng is the variance due to sampling, i.e., sampling procedure, pump calibration, sample transport, etc., which are commonly performed by people other than the analyst. (J2Analysis is the variance due to storage of the sample in the laboratory, pre-treatment procedures, the analysis itself, calibrations, etc. oZMeasurement Procedure can be established a priori and controlled a posteriori8 by sampling in duplicate, i.e., two samples are collected as close as possible in time and space, i.e., under repeatability conditions.When measuring, for example, the normal boiling-point of a pure volatile organic compound (VOC), 020bJect is zero, because a single true value exists for this physico-chemical property. For a heap of grain, a single true value also exists for the protein content, but all the grains in the heap are not analysed. The average concentration of a series of samples may deviate significantly from that of the heap as a whole. For a given point in a river, the concentration of a pollutant is a distribution in 02Measurernent Procedure = o2Sarnphng 4- 02Analysis Table 1 Measurement uncertainties Concept Definition Accuracy Trueness Closeness of the agreement between a test Closeness of the agreement between the result* and the true value? average value obtained from a large series of test results and the true value+ Bias Quantitative measure of trueness Precision Closeness of agreement between independent test results obtained under stipulated conditions Standard deviation Quantitative measure of precision * A test result may consist of the average of a series of measurements.+ IS0 5725’s concept: ‘accepted reference value’ has in this paper been replaced by the concept ‘true value’, i.e., the (unknown) true arithmetic mean of the worker’s exposure concentration distribution. time, because a series of temporal variables influences the concentration. Similarly, the concentration of a given pollutant in the air breathed by a worker is influenced by a series of time- dependent variables, but it also depends on the performance of the worker and his or her work pattern.When &!object is large, the result obtained may differ significantly from the true mean value of the exposure distribution, if only a few samples are taken. Systematic Errors As shown in Table 1, the systematic component of an error is measured by the bias (6). Bias may be divided into the bias of the measurement procedure and bias due to choice of measure- ment strategy: (3) 6 = 8Measurernent Procedure -I- GStratcgy beasurernent Procedure - 8Sarnplmg + GAndlyu\ where (4) GAnalysls is the sum of positive and negative biases originating from all steps in the analysis. This bias can be measured and corrected for. It is part of the analytical laboratory’s quality assurance of data. GSamplmg may, for example, be caused by the spatial inhomogeneity of the breathing zone, in which the sample is collected.aNleasurement procedure can be controlled so that it is acceptably small by sampling in duplicate and inserting control samples before and after the analysis, in combination with blank and spiked field samples. - Variability of Exposures at the Workplace Deciding on the number of samples is not trivial, because, in the working environment, the variation of the object of measure- ment is commonly high and unknown. Results from measurements of shiftlong exposures at the workplace are reported to be lognormally distrib~ted.~,~-*6 Kromhout et al.9 have, by studying a data set of 20000 data points, collected by various investigators, in five different countries, convincingly demonstrated the variability of expo- sures at the workplace.In their analysis, only samples with sampling times of more than 4 h were included; hence, workers were measured in the time domain,* because they may have performed more than one process during sampling. Kromhout et al.9 found the median geometric standard deviation (GSD) was close to 2.5 for gaseous exposures in the chemical industry. If this finding is generally true, then it is important to realize the types of data that can be expected when sampling from such a distribution. Table 2 shows five sets of simulated data. Sampling was performed from a distribution with an arithmetic mean of 167 mg m--3 and a GSD of 2.6, corresponding to a ‘Roller’ (see below for further details). In column 2 data for the ‘Roller’ are shown.As can be seen, the data obtained are very disperse, and a significant bias can often be anticipated, unless measurements are made on many days. Controlling Exposure The occupational hygienist circumvents the problem of wide, disperse exposure distributions by selecting highly exposed workers and carrying out measurements on them during periods of high exposure.17 If the ‘worst case’ result is well below the OELs in force, then the occupational hygienist can, with some confidence, infer that exposures are probably also in com- pliance on days when no measurements are made, for both measured and unmeasured workers. As can be seen from Table1158 x 25- s 8 Fr 15- 8 20- Analyst, September 1996, Vol. 121 2, inferences could be based on only a few results if the sample collector was sufficiently skilled to select the ‘worst case’, provided that the ‘worst case’ situation occurred on the measurement day. Logically, ‘worst case’ data cannot be used for demonstrating non-compliance and they cannot be used for epidemiological purposes, risk assessment or standard setting.4.J 8 Measuring Exposure Measuring exposure by sampling on many days Table 2 shows that when sampling from a wide and lognormal distribution, measurement of shiftlong exposures requires that measurements be made on a large number of days. This is because when an extremely high result is obtained a large number of low values are needed to ‘neutralize’ its effect. One possible solution to the problem of accurately measuring the arithmetic mean of distributions that are wide and skewed is to carry out measurements on many more days than is commonly done.12 Table 2 shows that the expected value of TWAC,,,; viz., E(TWACD,,) =: T W A C E ~ ~ ~ ~ ~ ~ ~ Period, depends heavily on the number of days on which measurements are made.Measuring over a period of several weeks, for example, is costly, because the hygienist should be present during all measurement days to ensure sampling quality. Measuring exposure using the logbook method Another possible solution is to use the logbook method. In brief; processes are grouped in similar emission process groups (SEPGs), i.e., processes where the same substances and tools are used on similar working objects, in a similar environment, equipped with the same engineering controls, etc.The contribu- tion of SEPG,s to workers’ exposure is estimated by measuring in the process domain, i.e., sampling is carried out during performance of only one process.2 The result obtained is called the process concentration, which is the measured concentration minus the background exposure: PC, = E(PC,!) = E(Cpj- BCj), where p indicates the pth process and j the jth day; BC is Table 2 Five sets of data simulating 10 days of shiftlong sampling from a lognormal distribution.* Values in mg m--3 Day EXP Set 1 Set 2 Set 3 Set4 Set 5 1 2 3 4 5 6 7 8 9 10 11 12 83 27 150 181 134 146 246 23 120 76 75 39 88 81 221 23 39 1597 137 43 215 37 368 486 552 61 86 484 20 107 188 106 109 108 194 82 250 158 71 37 34 I 113 130 161 37 148 194 143 400 705 37 125 200 101 886 14 67 559 96 69 106 220 33 48 262 187 86 135 420 301 501 464 AM/mgm-3 167 222 172 151 272 230 s/mg m-3 154 435 84 165 295 162 RSD (%) 93 196 49 110 109 70 GM/mgm-3 112 109 154 90 144 171 GSD 2.64 2.74 1.65 2.91 3.55 2.41 AM: Arithmetic mean.True arithmetic mean (AMT) = 167 mg m--3. EXP: Experimental data for the ‘roller’. s: Standard deviation. RSD: Relative standard deviation. GM: Geometric mean. GSD: Geometric standard deviation. True geometric standard deviation (GSDT) = 2.6. the background concentration. The contribution of SEPG,s to workers’ exposure on other days is called the APC,, (assigned process concentration). For the kth day, the APC, can be calculated by adding the background concentration of the kth day to the PC,: APCp,k = PC, + BCk. In general, when using VOCs, PC, is much larger than the background concentration.When measuring PC, values the object of measurement is more specific than that for the measurement of TWACD,,. Hence, the PC, distributions can be expected to be more narrow and more symmetrical. An even more specific measurement object would be to define the SEPG as consisting of only one worker. However, for workers in the glass fibre reinforced polyester facilities measured in this work, no differences could be seen in either the between- or within-worker variance measuring in process domain.’ If this can be generalized, SEPGs consisting of processes operated by various workers would seem to be appropriate for measuring in the working environment using the logbook method. The TWACD,, can be calculated as the product sum of the daily work pattern obtained from the logbooks and the APC, values.1 Results Homogeneity of the Breathing Zone The exposure of 14 workers, at two glass fibre reinforced polyester facilities, was measured. Duplicate sampling was carried out, simultaneously, 3 cm apart, about 10 cm below the chins of the workers. The 14 workers were all right-handed. Fig. 4 shows the relative deviation between duplicate samples, calculated as (C, - Cf)/[O.S(C, + C,)], where the subscripts c and f indicate the charcoal tubes closest to and furthest from the pollution source, respectively. As can be seen, there are large random errors compared with the random errors usually found for the analysis of VOCs. Further, a bias is seen towards higher values for the sampling tubes closest to the pollution source.Measuring the SEPG for a ‘Rolling’ Process Fig. 5 shows data for the SEPG for a process called ‘Rolling’. In performing this process, the workers press a glass-fibre web into the polyester using a stainless-steel roller. Seventeen measure- ments were carried out in the process SEPG for the ‘rolling’ process on nine workers, on two days, 9 months apart, in three different halls, in two different companies, under different 401 I I n I ”1 I I 30i I I - I Relative difference (8) Fig. 4 samples. Breathing zone homogeneity. Relative deviation between duplicateAnalyst, September 1996, Vol. 121 .999 - .50 rft. -~ .20 - .05 - .01 - .001 - - * .- 1159 (a) _1 . . . 7 , , production and weather conditions. In a normality plot, normal data will fit a straight line.It was found that exposures in the SEPG for the ‘rolling’ process are normally distributed (W-test for normality, MINI- TAB).’9 Measuring a ‘Roller’ in the Time Domain At one of the facilities measured, a worker performed a task named a ‘Roller’. In order to perform this task several processes must be carried out, including ‘rolling’, ‘glass cutting’, ‘mixing’ (polyester, styrene and catalysts) and ‘cleaning’ (of tools). Fig. 6(a) and (b) shows TWACDay data for a ‘Roller’ over a period of 3 weeks measured by the logbook method. In a normality plot, the logarithms of lognormal data will fit a straight line. As can be seen, the exposure data are better approximated by a lognormal distribution [Fig. 6(h)] than by a normal distribution [Fig.6(a)]. The GSD = exp(0.96252) = 2.6 i s close to the median finding of Kromhout et ul.9 The exposures range from 20 to 552 mg m-3. The arithmetic mean ,999 1 .99 ::: f .001 , 20 70 120 170 Concentration/mg m-3 Fig. 5 Measurement? in the SEPG for the ‘rolling’ process. Average: 89 mg m-3; s: 46 mg mP3; No. of measurements: 17. W-test for normality: R = 0.97; p-value (approximately) = > 0.100. .20 .05 3 4 5 6 Ln [Concentration/mg m-3] Fig. 6 TWACDdy data for a ‘roller’ over a period of 3 weeks measured by the logbook method. The exposure data are better approximated by a lognormal distribution (h) than by a normal distribution (a). For (a): Average: 166 mg m-?; s: 154 mg m-?; No. of measurements: 12. W-test for normality: R = 0.91; p-value (approximately) = 0.03.For (b): Average: 4.73; s: 0.96; No. of measurements: 12. W-test for normality: R = 0.9906; p-value (approximately) = > 0.1000, of the exposures of the ‘roller’ is 167 mg m--3, which is higher than the simulated data in Table 2. Disregarding three extreme values of 1597, 886 and 705 mg m-3, corresponding to 6% of the results, the range of the simulated data in Table 2 is close to that of the ‘roller’, i.e., 14-559 mg m--3. Simulating Samplings from Normal and Lognormal Distributions Let us assume that exposures in the SEPG for the ‘rolling’ process are normally distributed and that the TWACDay of the ‘roller’ in a 3 week period is lognormally distributed. What is the difference in data quality when measuring in the time and process domains? As a measure of group homogeneity, the Health and Safety Executive (HSE) in the UK has arbitrarily, but practically, defined that a worker exposed above half but below twice the group mean is considered to experience the same exposure as the group as a whole.20 In our context, the rule can be reformulated to: A result is considered to be significantly different from the true arithmetic mean (AMT) if it is less than half or more than twice the AMT.Hereafter, this will be referred to as the HSE rule. Compared with the accuracy commonly demanded from analytical laboratories, the HSE rule is not particularly strict. Table 3 shows the probability of obtaining results inside and outside the HSE rule, depending on the GSD of the worker’s exposure distribution.21 Table 4 shows the probability of obtaining a result signifi- cantly different from the true value as a function of the number of days on which measurements are made for a lognormal Table 3 Probability of obtaining a result inside or outside the limits of the HSE rule as a function of the true GSDT (lognormal distribution)* 1.5 2.0 2.5 3.0 3.5 Probability (%) of being outside the HSE rule- P { C < P{C > O.~AMT} 6.6 25.6 38.5 47.1 52.9 0.5AMT) 3.4 8.9 11.0 14.8 11.7 Probability (%) of being inside the HSE rule- P { ~ .~ A M T < C < ~ A M T ) 90.0 65.5 50.6 38.1 35.4 * AMT: True arithmetic mean. C: Measurement result. GSD.,: True geometric standard deviation. P indicates probability. Table 4 Probability of obtaining a result outside the limits of the HSE rule as a function of number of replicates (lognormal distribution)* P(C < 0.5AMT) P(2AM-r < C) RSD,,f,,,,, j (a) (%) (%I GSDof means 1 38.5 11.0 114.7 2.50 2 26.7 6.8 81.1 2.04 5 11.7 3.4 51.3 1.62 10 3.6 3.8 36.3 1.42 25 0.2 0.0 22.9 1.25 50 0.0 0.0 16.2 1.18 * AMT: True arithmetic mean (= 100). C: measurement result.RSDofmeans: Relative standard deviation of means of j days. GSDofmeanh: Geometric standard deviation of the distribution of means of j days. .j: Number of days (replicates) measured. P indicates probability.1160 Analyst, September 1996, Vol. 121 distribution with an AM of 100 and a GSD of 2.5. The latter value is very close to the GSD of 2.6 for the ‘roller’. In contrast, Table 5 shows the probability of obtaining a result inside or outside the HSE rule as a function of the number of measurements for a normal distribution with the same parameters as the SEPG for the ‘rolling’ process. The relative standard deviation is approximately 50%.Bias due to Choice of Strategy If the finding of Kromhout et al.9 holds true, there is a large risk of obtaining biased data, when measuring the ‘worst case’ situation to control exposure. Data are often stored in databases and are easily retrievable. It is, therefore, tempting to use such data, for example, for risk assessment, epidemiology or standard setting. Data in databases are mainly results from measurement activities performed by occupational hygienists measuring the ‘worst case’ situation.5 Table 6 compares results from ‘worst case’ measurements with results from ‘average case’ measurements.‘Worst case’ is defined here as measure- ments on randomly selected workers in the process domain, i.e., the workers were sampled while using VOCs.5 ‘Average case’ is defined as measuring in the time domain, on randomly selected workers, over randomly selected time periods, during which these workers performed a series of processes, some of which involved the direct handling of VOCs and some did not? Discussion The present author is aware of several analysts, who are investing considerable time and money in improving analytical accuracy in their laboratories. This involves participation in a series of proficiency tests, the purchase of expensive certified Table 5 Probability of obtaining a result outside the limits of the HSE rule as a function of number of replicates (normal distribution)* P{C < 0.5AMT) P(2AM.r < C ) RSD,~,,,,, N (%) (%I (%) 1 13.6 4.6 50.0 2 7.6 0.5 35.4 5 1.3 8 x 10-4 22.4 10 8 x 10-2 0.0 15.8 25 0.0 0.0 10.0 50 0.0 0.0 7.1 * AMT: True arithmetic mean (= 100).C: measurement result. RSD of means: Relative standard deviation of means of N measurements. N number of replicates. P indicates probability. Table 6 Bias between ‘worst case’ and ‘average case’ measurements* ‘Worst case’ ‘Average case’ Ratio Compound AM s AM s AM s Butyl acetate 70 128 24 32 2.9 4.0 Butanol 24 43 9 11 2.7 3.9 Acetone 107 365 14 26 7.6 14.0 Toluene 39 98 14 40 2.9 2.5 Xylenes 37 88 9 13 4.1 6.8 Ethanol 62 108 28 59 2.2 1.8 Ratio: (‘worst case’)/( ‘average case’). * AM: Arithmetic mean (mg m-3).s: Standard deviation (mg m-3). reference materials and the use of elaborated control charts to ensure the quality of analysis. In spite of this, why are some analysts not more interested in sampling than they appear to be? Sampling in the working environment, and probably also in the outdoor environment and other areas, potentially causes larger measurement errors than the analysis itself, as sampling can only be partly controlled. Further, even when sampling can be controlled, it is usually not. As pointed out by Eisenhart (cfi ref. 22): ‘Until a measure- ment operation has attained a state of statistical control, it cannot be regarded in any logical sense as measuring anything at all’. By statistical control it is meant that the dispersion of data is due to random events only.Accreditation bodies ensure that analyses performed by accredited laboratories are in statistical control, but the analysis is only one part of the measurement procedure. Duplicate sampling close in space and time is a low-cost solution, which will reveal any potential lack of statistical control of sampling and also some gross errors. Fig. 4 shows bias and random errors arising from the spatial inhomogeneity of the workers’ breathing zones from which the samples were taken. Homo- geneity of the workers’ breathing zones is a fundamental prerequisite for measuring exposure at the workplace; this is challenged by these results. However, the inhomogeneity of the breathing zones shown in Fig. 4 is small compared with the variability (‘inhomogeneity’) in time of the TWACDay, viz., the worker’s exposure over a working day.From Table 2 it is clear that performing measurements on single days, as is commonly done, risks introducing severe bias into the data, particularly if measure- ments are carried out on only a few days. One way commonly used by occupational hygienists to circumvent the problem of wide and skewed exposure distribu- tions is to measure the ‘worst case’ situation. However, the ‘worst case’ sampling strategy is not in statistical control, because biases of unknown sizes are deliberately introduced into the data. According to Eisenhart (cf. ref. 22), it is not measuring at all. Bias data are of minor interest if the bias introduced is insignificant; however, as shown in Table 6, biases of up to a factor of 7 have been found.Apart from substances for which the OEL is a ceiling value that must not be exceeded, ‘worst case’ data cannot be used for documenting non-compliance. Further, they cannot be used for risk assessment, epidemiology or standard setting, because of the potentially large and unknown bias in the data. If used, for example, for standard setting, the values of the standards will be excessive. The arithmetic mean exposure is a parameter that can be related to risk. Therefore, it can be used for risk assessment, epidemiology and standard setting. Because of the wide and skewed distributions often encountered in the working en- vironment, a large number of measurement days are necessary to estimate the means of such distributions, as can be seen from Tables 2 and 4.If the finding of Kromhout et al.9 holds true, an estimate of the T W A C E ~ ~ ~ ~ ~ ~ ~ period is either very uncertain or would be very costly to obtain. Measuring more specific measuring objects than TWACD,, distributions, such as processes grouped in SEPGs, potentially offers an increased accuracy. Tables 4 and 5 illustrate that many more data points are needed when sampling from a lognormal distribution than from a normal distribution. Most SEPG distributions can be expected to be approximately normally distributed, with a lower standard deviation than that reported for TWACDay data (see Fig. 5). The time component of exposure can be measured, for example, by applying the logbook method. The author thanks the workers and management in the two facilities where measurements were made for their co-operationAnalyst, September 1996, Vol.121 I161 and for completing the logbooks. T. Schneider is also thanked for useful discussions. Financial support of this study by the Commission of the European Union is acknowledged. Disclaimer The Commission of the European Union accepts no responsi- bility for the contents of this paper. References Olsen, E., and Jensen, B., Appl. Occiip. Environ. Hyg., 1994, 9, 245. Olsen, E., Appl. O t ~ x p . Eniiron. Hyg., 1994, 9, 712. Hawkins, N. C., Jayjock, M. A., and Lynch, J. R., Am. Znd. Hyg. Assoc. .I., 1991, 53, 34. Rappaport. S. M., Ann. Ou.up. Hvg., 1991, 35, 61. Olsen, E., Laursen. B., and Vinzents, P. S., Am. Ind. Hyg. Assoc. J . , 199 I, 52, 204. International Standards Organisation, International Standard. Part I . General Principles and Definitions. I S 0 5725-1 994(E), ISO, Geneva, 1994. Dybkjzr. R., Jordal, R., and Jorgensen, P. J., Scand. J . Clin. Lab. Invest., 1993, 53, Suppl. 212, 60. Heydorn, K., Mikrochim. Acta, 1991, 111, 1. Kromhout, H., Symanski, E., and Rappaport, S . M., Ann. 0ct.up. Hyg., 1993, 37, 253. 10 1 1 12 13 14 15 16 17 18 19 20 21 22 Esmen, N. A., and Hammad, Y. Y., J . Environ. Scz. Health, 1977, Buringh, E., and Lanting, R., Am. Znd. Hyg. Assoc. .I., 1991, 52 12. Rappaport, S. M., Ann. Otcup. Hyg., 1991, 35, 61. Roach, S., Health Risks ,from Hazardous SubmnceJ at Work: Ashessment, Evaluation and Control, Pergamon Press, Oxford, 1992. Kromhout, H., Symanski, E., and Rappaport, S. M., Ann. Occup. Hyg., 1993, 37, 253. Rappaport, S. M., Appl. Occup. Environ. Hyg., 199 1, 6, 448. Rappaport, S. M., Kromhout, H., and Symanski, E., Am Ind. Hyg. Assoc. J., 1993, 54, 654. EN 689, European Standard, English version: Workplace Atmos- pheres-Guidance for the Assessment o f Exposure by Inhalation to Chemical Agents for Comparison with Limit Values and Measure- ment Sti-ategy. European Committee for Standardization, Berlin, 1995. Kauppinen, T. P., Appl. Occup. Eni>iron. Hyg , 1991, 6, 482. MINITAB, Release 16, MINITAB, State College, PA. Gardiner, K., Otcup. Environ. Med., 1995, 705 MATHCAD for Windows, MathSoft, Cambridge, MA. Taylor, J. K., Anal. Chem., 1981, 53, 1588A. A12( l-2), 29. Paper 61031 45E Received Muy 7, 1996 Accepted May 31, 1996
ISSN:0003-2654
DOI:10.1039/AN9962101155
出版商:RSC
年代:1996
数据来源: RSC
|
14. |
Summary of the NIOSH guidelines for air sampling and analytical method development and evaluation |
|
Analyst,
Volume 121,
Issue 9,
1996,
Page 1163-1169
Eugene R. Kennedy,
Preview
|
PDF (1247KB)
|
|
摘要:
Analyst, September 1996, Vol. 121 (1163-1169) I163 Summary of the NIOSH Guidelines for Air Sampling and Analytical Method Development and Evaluation* Eugene R. Kennedya, Thomas J. Fischbach", Ruiguang Songb, Peter M. Ellera and Stanley A. Shulmana a US Department of Health and Human Services, US Public Health Service Centers for Disease Control and Prevention, National Institute ,for Occupational Safety and Health, 4676 Columbia Parkway, Muil Stop R-7, Cincinnati, OH 45226, USA Computer Sciences CorporationlNIOSH, Cincinnati, OH, USA Suggested guidelines for the development and evaluation of sampling and analytical methods for industrial hygiene monitoring have recently been published in a NIOSH technical report. These guidelines are based in part on various published approaches for method development and evaluation and serve as an attempt at a more unified experimental approach.This paper presents some salient features of this unified approach for method development and evaluation. The basic goal of the approach is to determine if the method under study meets the criterion to produce a result that fell within 25% of the true value 95 times out of 100 on average, although other factors of method performance are evaluated. The experiments proposed for the evaluation of method performance include determination of analytical recovery from the sampler, sampler capacity, storage stability of samples and effect of environmental factors. Evaluation criteria for the experimental data and procedures for the calculation of method bias, precision and accuracy are also included.Keywords: Precision; bias; accuracy; air sampling; NIOSH guidelines Background The Occupational Safety and Health Act of 1970 (Public Law 9 1-596)l charged the National Institute for Occupational Safety and Health (NIOSH) with the responsibility for the conduct of research relating to innovative methods for dealing with occupational safety and health problems. Under that charge in 1974, NIOSH and the Occupational Safety and Health Admini- stration (OSHA) jointly undertook the evaluation of sampling and analytical methods for airborne contaminants by contract. During this work, a protocol was developed to define the experiments and Performance criteria to be used for method evaluation.24 For each method under consideration, the objective of this protocol was to decide if the method would provide results that were within f25% of the (true) concentra- tion 95% of the time on average.The protocol developed for this contract has been used for method development and evaluation research for many years with adaptations and modifications. During method development and evaluation research, the effects of certain environmental and experimental conditions on sampling and analytical method performance have been documented and other protocols for method evaluation have * Presented at AIRMON '96, SLilen, Sweden, February 5-8, 1996. US Government Copyright. been developed.5-12 The information included in the original NIOSH protocol and other independent protocols has served as the basis for the revision of the NIOSH guideline document13 for air sampling and analytical method development.An attempt was made to include experiments that were universally covered in all of the protocols for method evaluation, so that a more harmonized approach for method evaluation might result. In particular, CEN standards and proposed standards8-9 were considered in the design of the experiments included in the guidelines. The objectives of the guidelines are the following: (1) to provide an outline for method development and evaluation work; and (2) to provide guidance and procedures to estimate the precision, bias and accuracy of a sampling and analysis method. This paper provides a synopsis of the experimental aspects of this NIOSH guideline document. Method Development In the development of a sampling and analytical method, there is a logical progression of events that range from a search of the literature to gather pertinent information to preliminary experi- mentation for selection of analysis technique and sampling medium. After the developmental experimentation has been completed, a proposed sampling and analytical method results.Since innovation is a key element in the sampling and analytical method development process, detailed experiments for the initial development of the sampling approach and optimization of the analytical procedure are better left to the discretion of the researcher. During development, it should be recognized that appropriate, statistically designed experiments will optimize the information obtained.Therefore, consultation with a statistician will be of value during this phase of the research. Several key points, including calibration, limits of detection and quantification and selection of measurement technique and sampling medium, should be addressed during these initial method development experiments. The physical state of the analyte ( i x . , gas, aerosol, vapour or combination thereof) plays an important part in the selection of an appropriate sampler and medium. Analytes that can exist in more than one physical state may require a combination of sampling media in one sampler for efficient collection.14 Where possible, commonly available and easily used samplers should be investigated initially. As the preliminary testing of a sampling method progresses, further modification in the sampling media or sampler design may be required and may affect the measurement procedure. Sampler design and medium selection should include consideration of regulations for shipment of the sampler back to a laboratory for analysis.Since industrial hygiene analytical methods are geared towards measuring personal exposure, the size, mass and1164 Analyst, September 1996, Vol. 121 convenience of the sampler are important elements in sampler design. The personal sampler should allow freedom of move- ment and should be unobtrusive, unbreakable and not prone to leakage. The pressure drop across the sampler should not be so great as to limit sample collection times to d 10 h with personal sampling pumps. For situations where only a short-term sample will be required (i.e., 15 min for ceiling determinations), this d 10 h recommendation can be reduced to d I h.The use of potentially toxic reagents for analyte collection should be avoided unless they can be used safely. Reagents used should not pose any exposure hazard to the worker wearing the sampler or to the industrial hygienist taking the samples. Recovery of the Analyte from the Medium During method development experiments, the ability to recover the analyte efficiently from the sampling medium is crucial to the success of the method. A suggested experiment to evaluate recpvery from the sampler entails the fortification of sets of six samplers with amounts of analyte equivalent to sampling concentrations of 0.1, 0.5, 1.0 and 2.0 (or higher) times the exposure limit for a minimum of 4 h at the typical sampling rate used for that type of sampler.If the analyte has a ceiling or short-term exposure limit, the amount of analyte fortified should be adjusted for the shorter sampling time required for this type of exposure limit. If the sampler has a backup section, then the same number of separate backup sections should be fortified with amounts of analyte equivalent to 25% of the amount fortified on the front sections of the samplers, since this amount has been used to characterize the breakthrough limit of useful samples.I2 Samples (and backup sections) should be prepared for analysis and analysed according to procedures defined in the proposed method. Results of these analyses should be expressed as estimated percentage recovery accord- ing to the following equation: Recovery(,,,,, (%I) = (amount of analyte found on sampler) (amount of analyte fortified on sampler) x 100 After initial analyses of the samples, the samples should be resealed and analysed on the following day, if possible. If the sample work-up procedure results in a solution of the sample, these solutions should be re-capped after the initial analysis and re-analysed on the following day using fresh standards.The recovery of the analyte should be calculated for the primary and backup media in the sampler. Although complete recovery of the analyte from the sampler is most desirable, at a minimum, the estimated recovery of the analyte from the primary collection medium should be greater than or equal to 75% for concentrations equivalent to sampling 0.1,0.5, 1 .O and 2.0 times the exposure limit.The value of 75% has typically been used as a lower limit for desorption efficien~y.~.y If the recovery varies with analyte loading but remains greater than 75%, results should be graphed as recovery versus loading during the calibration of the method, so that appropriate correction can be made to the sample results.15 If the estimated recovery does not exceed 75%, the method is not suitable for monitoring at this limit. Fig. 1 displays an example where loading affects desorption efficiency from a sorbent. The data indicate that at levels of vinyl acetate above 300 vg, the recovery exceeds the recommended lower limit of 75%, although the recovery is not linear with loading above this level.16 Estimated recovery from any backup medium should be noted so that appropriate corrections can be applied if breakthrough of the sampler has occurred during sampling.The recovery of the analyte from the medium in the backup section of a sampler may be different from that of the front section, since the backup section of a sorbent-based sampler usually contains only half of the sorbent of the primary section. If the same volume of desorption solvent is used for both the primary and backup sections of the sampler, the desorption equilibrium can be shifted, since the backup section is typically being desorbed by twice the volume (i.e., on a ml solvent/ mg sorbent basis).17 Re-analysis of the samples on the day after initial analysis indicates if immediate analysis after sample preparation is required.Often when processing many samples, preparing the samples for analysis as a batch may be necessary. In these instances, the last samples may not be analysed for up to 24 h or more after preparation owing to the time required for analysis. If samples prepared for analysis exhibit time-dependent stabil- ity after desorption, analyses must be conducted within acceptable time constraints. As a tentative guideline, analysis and re-analysis results should agree within 5% of each other. Stability of the Analyte on the Medium To evaluate the stability of analyte on the sampling medium, an extension to the experiment described above may be performed. An additional set of fortified samples at each of the four concentrations should be prepared and analysed after storage for 7 d at room temperature.The recovery should be similar to the above results, within experimental error. Discrepancies larger than those expected by experimental error suggest sample stability problems that will need correcting by additional developmental effort ( e g . , refrigerated storage). Comparison of results can be performed with statistical tests, such as an analysis of variance (AN0VA)'X test of the 'day' difference or a paired t-test19 of the means of the day 1 and day 7 storage results. Fig. 2 shows data from the combined analytical recovery experiment and the evaluation of the stability of the analyte on the medium for formaldehyde on a reagent-coated sorbent. The , , " I ' , 0 1000 2000 3000 4000 5000 Load inglpg Fig.1 Graph showing the desorption efficiency of vinyl acetate from coconut-shell charcoal vemus loading. Carbon disulfide was used as the desorbing solvent.16 Formaldehyde per sample/pg Fig. 2 Graph showing the analytical recovery of formaldehyde from a reagent-coated sampler. Data points for the day results for each level are within 5% and appear overlaid on the graph. Day 2 results represent the re- analysis of the day 1 solutions. Day 7 results are based on the analysis of duplicate samples stored at ambient temperatureyq 7 d prior to analysis. H, Day 1; A , day 2; and +, day 7. '1Analyst, September 1996, Vol. 121 1165 data show that the day 2 re-analysis results are almost the same as the day 1 results. The day 7 analysis results also closely parallel the day 1 results, indicating that the analyte will be stable on the sampler for at least 7 d when stored at room temperature.By incorporating these experiments during the method development phase of the research, potential problems regarding recovery and storage can be discovered and solved. Method Evaluation After the initial development experiments for the method have been completed and a method has been proposed, the sampling and analysis approach should be evaluated to ensure that the data collected provide precise and accurate results with low bias. Specifically, the goal of this evaluation is to determine whether, on average, over a concentration range of 0.1-2 times the exposure limit (or higher), the method can provide a result that is within f25% of the true concentration 95% of the time with the absolute bias being 10% or less.An experimental approach for collecting the data necessary for this determination is described below. A key issue in method evaluation is to attempt to have the samples collected from an environment that is as close to actual sampling conditions as possible. Usually, this entails the generation of an atmosphere containing the analyte of interest. In cases where generation is not feasible or even hazardous, the fortification of samples is an alternative approach, although this is not as rigorous an evaluation and ignores the effects of sampling on precision, bias and accuracy. Concentration ranges to be used in the evaluation of the method should be based on several factors.These ranges, at a minimum, should cover 0.1 to at least 2.0 times the exposure limit. In some instances, higher multiples of the exposure limit can be added if needed ( e . g . , 10 times the exposure limit). In situations where multiple exposure limits (i.e., from different authorities) exist for an analyte, the lowest exposure limit should be used to set the lower limit of the evaluation range (0.1 times the lowest exposure limit) and the highest limit used to calculate the upper limit of evaluation range (twice the highest exposure limit). Intermediate evaluation concentrations should be within these exposure limits. The potential toxicity of an analyte (e.g., suspected carcinogenicity) may indicate that a concentration lower than that calculated by the exposure limit should be included in the measurement and evaluation ranges.Previous monitoring information from other methods may indicate that typical concentrations of the analyte may be below or above a concentration range based on the exposure limit. In this case, this lower or upper level may be included in the concentration range. Feasibility of Analyte Generation As part of the evaluation of a method, the collection of samples from a generated atmosphere is needed to assess more adequately the performance of a method.20 22 If possible, the generated atmosphere should represent the environment en- countered when sampling for the analyte in the workplace. When attempting to generate a concentration of an analyte, the impact of environmental conditions, such as temperat~re,'~3~5 pressure,Js humidity12723 and interferences, on sampler per- formance and/or generation should be considered.The effect of elevated temperature on the collection medium of a sampler may decrease the capacity of the sampler or may decompose the analyte during generation and sampling. Reduced pressure may also reduce the capacity of a sampler. High relative humidity has usually been observed to reduce sampler capacity; l 2 in other instances it has increased sampler capacity.23 A typical interference(s) should be generated along with the analyte to approximate a typical workplace sampling environment. Generation of particulate material can be extremely com- p1ex,24,25 especially if particles of a specific size range must be generated for the evaluation of a specified sampler inlet design.The aerodynamic performance of the generator is a factor in the generation of this type of atmosphere and should be evaluated carefully. Appropriate independent methods should be available to verify particle size, if this is a critical element in the generation. The concentration of the generated atmosphere should be verified by analysis of replicate samples (if possible) by an independent method at each concentration used. A note to Appendix 1 of ref. 13 deals with this situation in more detail. In addition, a statistician may be consulted for advice on the experimental design and sample sizes necessary to accomplish this validation, depending on the variability of the independent method.Ideally, the independent method should not be biased and should provide an accurate estimate of the concentration generated, assuming error is randomly distributed around the mean. Also, the precision and bias of the independent method should be homogeneous over the concentrations investigated. In instances where the concentration of the generator can be based only on calculations using flow rates in the generator and the amount of analyte injected, the generation system should be well characterized so that analyte losses are reduced. In some instances, generation of an analyte may be difficult and even hazardous. As an alternative to direct generation in these cases, generation may be simulated by fortifying samplers with an amount of analyte expected to be sampled over a specified period at a specific flow rate.When this is necessary, fortification of the sampler by vaporization of a known amount of analyte on to the sampling medium is a more appropriate method, since this approach more closely approximates a generated atmosphere.26 The alternative of direct application of a solution of analyte on to the collection medium is less desirable but may be necessary in some instances. After fortification, air, conditioned at both high and low humidity, should be drawn through samplers at the flow rate and time period used in the calculations for the amount of analyte expected to be collected. In the method report, the fact that samples were not collected from a generated atmosphere should be discussed. Capacity ofthe Sampler and Sampling Rate To determine the applicability of the sampling method, the capacity of the sampler should be determined as a function of flow rate and sampling time by collecting samples from a generated atmosphere.This is particularly important if the analyte has both a short-term limit and a time-weighted average. At this time the effect of environmental conditions (e.g., temperature and relative humidity) on the sampler should also be studied, In the typical sampler capacity determination experiment, the sampling rates used should be appropriate for the media selected. These may range from 0.01-4 1 min-I, depending on sampler type. At extremely low flow rates (about 5 ml min-I), the effect of diffusion of the analyte into the sampler must be considered. Flow rates should be kept high enough to prevent diffusion from having a positive bias in the sampler collection efficiency.Sampling should be performed at three different flow rates covering the range discussed above for the particular sampler type, unless the sampler is designed to operate at only one flow rate. Sampling times should range from 22.5 min for short-term exposure limits to 900 min (15 h) for time-weighted averages. Shorter sampling times (e.g., 7.5-22.5 min) may be used for ceiling measurements. Flow rates should be based on accurately cdibrated sampling pumps or critical orifices. The mass of analyc collected at the lowest flow rate and shortestI166 Analyst, September 1996, Vol. 121 sampling time should be greater than the limit of quantification of the method. The generated concentration used for capacity determination should be at least twice the highest published exposure limit and verified by an independent method.Sampling should be conducted at ambient, elevated (> 35 “C) and low ( ~ 2 0 ° C ) temperatures to assess the effect of temperature on sampling. To assess the effect of humidity on capacity, sampling should be performed at both low and high humidities (<20% and >80%, since both have been observed to affect capacity.12.23 Triplicate samplers at each of three different flow rates should be included to verify the capacity at each combination of environmental conditions. For samplers that contain backup sampling media, only the front section of the sampler should be used. A means is required to determine analyte in the effluent from the sampler.This may involve the use of a backup sampler, continuous monitor or other appropriate means that can provide a measure of analyte concentration in the sampler effluent (about 1-5% of the influent concentration). If the mass of analyte found on a backup sampler totals 5% of the mass found on the front sampler or if the effluent concentration of the sampler contains 5% of the influent concentration, breakthrough has occurred and the capacity of the sampler has been exceeded. If the analyte is a particulate material and collected with a filter, the capacity of the filter is defined by the pressure drop across the sampler or by the loading of the filter. For 37 mm filter-based samplers, the pressure drop should be less than 40 in (1016 mm) of water for a total loading less than 2 mg.Larger filters may tolerate higher mass loadings. If the collection process is based primarily on adsorption, the breakthrough time should be proportional to the inverse of the flow rate.27 This relationship can be checked by plotting the 5% breakthrough time versus the inverse of the flow rate. If the resulting plot is a straight line, then this relationship should hold for all flow rates in the flow rate range studied. Fig. 3 presents capacity data which follow a linear relationship for several different sorbents and vapours.28 Some non-linearity in the plot may be noted owing to experimental variability and assump- tions made to simplify the relationship of breakthrough time and flow rate.Results from these experimental trials should provide a prediction of the capacity of the sampler at various flow rates and sampling times. If the flow rates and sampling times used in the experiment do not provide for sufficient capacity, a lower flow rate range may have to be studied and the experiment repeated. 300 II 200 E 2 3 m 100 .- . c cn c k! 0 * -t- Toluene/Carbotrap - f - Toluenenenax GR + HexaneKarbotrap - 4 - 2-Butanone/Carboxen - -x- - Methylene Chloride/Carboxen / / / / / / / / / / / A’ 0-x / 0 0.01 0.02 0.03 0.04 0.05 Flow rat e-’/m in m I-’ Fig. 3 Graph showing the relationship of the inverse of flow rate to breakthrough time for several sorbents and vapours. Data for hexane on Carbotrap and toluene on Tenax GR fall on the same line.2* With samplers that use reagents for collection of the analyte, rhe amount of the reagent in the sampler will also be a limiting factor in the capacity of the sampler, based on the stoichiometry of the reaction.Other factors, such as residence time in the sampler and kinetics of reaction between analyte and reagent, may also affect the capacity of this type of sampler. The combined temperature and humidity conditions that reduce the sampler capacity to the greatest extent should be used in all further experiments. The maximum recommended sampling time (MRST) for a specific flow rate is defined as the time at which sampler capacity was reached, multiplied by 0.667. This adds a measure of safety to this determination. The relationship of breakthrough time with flow rate and the safety factor correction can be used to adjust flow rates to allow specific sampling times.Fig. 4 presents data from a typical experiment run at two different humidities and two different flow rates. The data indicate that the sampler is adversely affected by low humidity, since the capacity is reduced at humidity levels below 10%. Flow rate seems to have little effect on these results, within experimental error. The results of this experiment suggest that the atmospheres used in further generation experiments should be kept at low humidity, in order to test the sampler under conditions that most severely restrict its performance. Sampling and Analysis Evaluation To assess the performance of a method, certain additional experimental parameters should be evaluated through a series of defined experiments. The effect of environmental conditions on sampling efficiency of the sampling medium can be evaluated by a factorial design.8.9.29 The relative humidity, flow rate, temperature and sampling times, determined in the experiment described above to have most severely limited the sampler capacity, should be used in these experimental runs.However, including these experimental factors in the experimental design could permit the examination of their interactions with other factors. At a minimum, the effect of concentration on analyte recovery should be investigated. Three sets of 12 samples should be collected from an atmosphere containing concentra- tions of 0.1 , 1 .O and 2.0 times the exposure limit at the humidity and temperature determined in previous experiments to reduce the sampler capacity the most.If the analyte has a short-term or ceiling exposure limit in addition to an 8 h time-weighted average, an additional 12 samplers should be collected at the short-term exposure limit or ceiling limit for the recommended sampling period at the appropriate flow rate. Potential interfer- ences in the work environment should be included as a factor in the generation experiments to assess their impact on method performance. Concentrations up to twice the exposure limit 50 40 h 8 5 30 e 22 CrJ 20 ?.? m v 3 r 10 0 ---+-- 0.75 (<lo% RH) + 1.5 (<lo% RH) + 0.75 (ca. 80% RH) -.-c- 1.5 (ca. 80% RH) 0 20 40 60 80 100 Formaldehyde collected/pg Fig. 4 formaldehyde at two different flow rates (1 min-1) and humidities.Graph showing the capacity of a reagent-coated sampler forAnalyst, September 1996, Vol. I21 90-’ 1167 -t- Ambient Storage ----k-- - Refrlgeratod Storage i ----_ value for the interference should be included. Other environ- mental factors may be studied, but will require a more comprehensive experimental design. Those factors found to exert an effect on analyte recovery should be investigated further to determine if their impact is predictable in terms of easily observed c0nditions.~9 If these effects are not predictable, the utility of the method will be limited, based on the conditions defined by this experiment. If only concentration is evaluated, the analyte recovery should be the same or at least predictable at all concentrations after correctable biases have been included, such as desorption efficiency.Sample Stability When samples need to be shipped back to a laboratory for analysis, the integrity of the analyte on the sampling media is very important. The preliminary experiment for the determi- nation of analyte stability on the sampling medium should have indicated if additional precautions, such as refrigeration, were required to maintain sample integrity. To assess more fully the stability of the analyte on samplers, samples should be collected from a generated atmosphere and stored under defined condi- tions (i.e., ambient or refrigerated, light or dark). The samples can then be analysed at specified times to track sample stability. A suggested concentration of half the lowest exposure limit should be sampled with 30 samplers for a minimum of half of the MRST.The humidity and temperature of the generator should be at the same level as defined in the sample capacity experiment to reduce sample capacity. The samplers should be divided randomly into one group of 12, one group of six and four groups of three, with the group of 12 analysed as soon after collection as possible (day 0). The group of six should be analysed after 7 d. The four remaining sets of three samples should be analysed after 10, 14,21 and 30 d. The conditions of storage are determined by the nature of the analyte. If there is an indication of analyte instability on the sampling medium, refrigeration of the samplers may be required. However, under normal conditions, storage for the first 7 d should be at room temperature.Samples should be stable for a minimum of 7 d under ambient conditions to simulate shipping to a laboratory for analysis. If the average analysis result of the samplers analysed on day 7 differs from the set analysed on day 0 by more than lo%, the method does not meet the sample stability criterion. Either additional precautions, such as shipment on ice and refrigerator storage, may be required or the method may have to be modified to address this problem. If a plot of recovery versus time indicates that the recovery decreased by more than 10% after the initial 7 d storage period, sample instability is a problem. If samples need to be stored for longer periods, more restrictive storage conditions are required. Remedial action, such as cold storage, may solve this longer term storage problem. After remedial precautions have been made in the method, the sample stability of the method must be re-determined.Fig. 5 represents the long-term storage data for samples of methamidophos collected on OVS sorbent tubes.26 These samples were stable for at least 21 d under ambient storage conditions, but recovery dropped below 90% of the day 0 results for the samples stored for 30 d under ambient conditions. When the experiment was repeated with the samples stored under refrigeration, all results were within 10% of the day 0 results. i,-~? Precision, Bias and Accuracy As mentioned in the Background section, a goal of method evaluation is the determination of the ability of the method to produce accurate results.The information collected in the previous experiments can be combined to assess the bias, precision and accuracy of the method. A more exacting treatment of the statistical calculations involved is described in the literature.l3 To help simplify the calculations, a computer program was written to ease the data analysis for the determination of method accuracy.30 Sampler results from the analytical recovery experiment, the factorial design experiment at 0.1, 1.0 and 2.0 times the exposure limit value, the sampler stability experiment (at half the exposure limit), and the environmental factors experiment are used in the calculations of method presision. The calculations for the estimated method precision, SrT, have been described previously.2-4~13 Before calculating this estimate of method precision from the four sets of samplers, the homogeneity of the precision over the range of concentrations studied should be checked using a test, such as Bartlett’s t e ~ t .~ ~ * l 3 If the precision is not constant over concentrations, the sample set collected at 0.1 times the exposure limit should be removed and Bartlett’s test re- calculated. Homogeneity of the method precision is an anssumption required for statistical pooling of the data to obtain SrT. Bias is assumed to be homogeneous over the evaluation range. This assumption should be tested by estimating the bias at each concentration and testing these for homogeneity using the procedures described in the literature. 13 Method bias should be less than 10%.A test for this is also described in the literature. 3 For a rudimentary estimate of method accuracy, the bias and precision estimates can be used with the graph presented in Fig. 6 to estimate ac~uracy.3~ The bias and precision estimates are plotted on the abscissa and ordinate of the graph, respectively. The intersection of these points on the parabolic grid in the graph can be used to estimate the accuracy of the method. Typically, the 2.5% and 97.5% confidence limits are calculated around the bias and precision estimates.13 These values are used with Fig. 6 to provide an estimate of the confidence interval estimate of the accuracy of the method. If both upper and lower confidence limits on the accuracy are less than 25%, the method meets the accuracy criterion.If both limits exceed 25%, the method definitely does not meet the accuracy criterion. If the upper limit exceeds 25% and the lower limit does not, the result is inconclusive and further study is required to see if the method will meet the accuracy criterion. If the results for four concentrations do not satisfy or fail the 25% accuracy criterion, then the set of samples collected at 0.1 tima the exposure limit should be excluded from the data set. The SrT and the bias should be re-calculated on this reduced data set before performing the accuracy analysis described above. For the 12 samplers collected at the ceiling limit, the accuracy analysis described above should be repeated using only the data collected at the ceiling limit.E i 1 f 30 6o O j I I 4 \ 0 5 10 15 20 25 30 Storage timeld Fig. 5 Graph showing the long-term stability for samples of the pesticide methamidophos, over a period of 30 d. Results from samples stored under both ambient and refngerated conditions are represented.261168 Analyst, September 1996, Vol. 121 Field Evaluation Although field evaluation is not required in method evaluation, it does provide a further test of the method, since conditions that exist in the field are difficult to reproduce in the laboratory. Also, unknown variables may affect the sampling results when field samples are taken. This type of evaluation is recommended to study further the performance of the method in terms of field precision, bias and interferences and the general utility of the method.5,8,9 A field evaluation of a method also allows the developer of the method to determine its ruggedness. Although this may be a subjective judgement, first-hand experience with the method in the field may suggest changes to the sampler or method that might make the method more easily used in the field and less subject to variability. Both the collection of area samples and personal samples should be included in the field evaluation of the method. Area samples should provide an estimate of field precision and bias. Personal samples may confirm these values and provide a means to assess the utility of the method. A statistical study design should be prepared, based on the variability of the method and the statistical power required to observe differences between the independent method and the method under evaluation.3 I If this type of study is not feasible, a minimum of 20 pairs of samples of the method under study and an independent method should be used for personal sampling.Placement of the samplers on the workers should be random to prevent the biasing of results due to the 'handedness' of the worker. Workers sampled should be in areas where both low and high concentrations of the analyte may be present. As a minimum, sets of six area samplers paired with independent methods should be placed in areas of low, intermediate and high analyte concentration. If the atmosphere sampled is not homogeneous, precautions may have to be taken to ensure that all samplers are exposed to the same concentra- tions. This can be done by using field exposure chambers, such as those described in the literature.32J3 Field precision and bias of the area sampler results of the method under study should compare with laboratory evaluation results, if precautions have been taken to ensure that all samplers have been exposed to the same homogeneous atmosphere.Differences in precision and bias can be investi- gated using either a paired t- test'y or ANOVA.l8 Sources of variation should be studied and corrections carried out where necessary. Evaluation of personal sampler results should be done cautiously, since observable differences may be due to work practices or other situations that are beyond the control of the method. Documentation As a final step in the development and evaluation of a sampling and analytical method, a final report should be prepared to document the research.The report should describe what was determined about the method. If the results of the statistical analysis of the data indicate that there is not 95% confidence that the accuracy of the method is less than or equal to +25%, the report should state this fdct and show the estimates of bias, precision and accuracy. In some instances, the method may actually have an accuracy of less than 25%, but a larger sample size must be used to prove this statistically (see Appendix 1 of ref. 13). The final report may be either a technical report or a failure report. The technical report (acceptable method development) documents the successful development of the analytical method. This report may be prepared in a format appropriate for submission to a peer-reviewed journal for publication.The failure report (no acceptable method developed) documents the research performed on an attempted method development for an analyte or analytes. The report should describe the failure of the method and also other areas of the method research that were successful. Recommendations to solve the failure of the method may be included. If an acceptable method is developed, a sampling and analytical method prepared in appropriate format should be prepared. The format of the resulting analytical method should 0.7 0.0 1.1 1 .B 0.t5 0.10 0.a 0.00 b.? 0.8 I -0 1 .r l i Bias, B Fig. 6 which yield the value of accuracy indicated on the. curve. Nomogram relating accuracy in percentage units as a function of the bias ( B ) and the precision (&).Each parabolic curve is the locus of all pointsAnalyst, September- 1996, Vol. 121 1169 provide clear instructions for the method. Sampling, sample work-up and analysis procedures should be described clearly. The necessary equipment and supplies for the method should be listed clearly in the method. A summary of the evaluation of the method should be included, in addition to a discussion of method applicability and lists of interferences and related references. Conclusion In order to foster the development of universally applicable industrial hygiene sampling and analytical methods, standar- dized procedures are needed for the evaluation of these methods. Based on the procedures that have been used and documented previously, an approach for method evaluation has been formulated, based on analytical feasibility and statistical validity.The details of these experimental procedures have been described in this paper, along with a brief discussion of the statistical treatment of the data. The end result of this process is a sampling and analytical method that has a defined level of accuracy. The authors acknowledge the assistance of Martha Seymour, NIOSH, and Pam Iraneta, Waters Chromatography, for the formaldehyde analytical recovery and capacity data. Mention of company names or products does not constitute endorsement by the Centers for Disease Control and Prevention. References Occ~iipational Safety and Health Act of 1970, Title 84 U.S.Code, Section 6 . Anderson, C., Gunderson, E., and Coulson, D., in Chemical HazardJ in the Wurkplace, ed. Choudhary, G., American Chemical Society, Washington, DC, 198 1, pp. 3-1 9. Busch, K., and Taylor, D., in Chemical Hazards in the Workplace, ed. Choudhary, G., American Chemical Society. Washington, DC, 198 1, Gunderson, E., Anderson, C., Smith, R., and Doemeny, L.. Develop- ment and Validation of Methods for Sampling and Analysis of Workplace Toxic Substances, DHHS/NIOSH Publication No. 80- 133, National Institute for Occupational Safety and Health, Cincinnati, OH, 1980. Health and Safety Executive, Protocol for AsseJsing the Perfornwnce of a Pumped Sampler,for Gases and Vapour-s, MDHS 54, HSE, London, 1986. Shotwell, H. P., Copperas, J. C. D., McCollum, R.W., and Mellor, J. F., Am. Znd. Hyg. Assoc. J., 1979, 40, 737. Elia, V. J., Fisher, R. P., Jain, A. K., and Rovell-Rixx, D. C., Workplace Measurement Methods Field Evaluation Protocd., Inter- nal Report, National Council of the Paper Industry for Air and Stream Improvement, Corvallis, OR, 1982. Workplace Armosphrre.~~eneraI Requirenwnts for the Perfor- mance of Procedures.for the Measurement of Chemical Agents, CEN EN 482, European Committee for European Standardization, Brus- sels, 1994. pp. 503-5 17. 9 10 11 12 13 14 1s 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 Pumped Sorbent Tubes for the Determination of Gases and Vu~~ours-Requirenzents and Test Method, CEN prEN 1076, Euro- pean Committee for European Standardization, Brussels, 1995. Occupational Safety and Health Administration, Method Evaluation Guidelines of the Organic Methods Evaluation Brunch, Internal Report, OSHA, Salt Lake City, UT, 1989.Occupational Safety and Health Administration, Evaluation Guide- lines of the Inorganic Methods Evaluation Branch, Internal Report, OSHA, Salt Lake City, UT, 1990. Melcher, R., Langer, R., and Kagel, R., Am. Ind. Hq’g. Assoc. .I., 1978, 39, 349. Kennedy, E. R., Fischbach, T. J., Song, R., Eller, P. E., and Shulman, S. A., Guidelines for Air Sampling and Analytical Method Develop- ment and Evaluation, DHHS (NIOSH) Publication No. 95- 1 17, National Institute for Occupational Safety and Health, Cincinnati, OH, 1995. Streicher, R. P., Kennedy, E. R., and Lorberau, C., Analyst, 1994, 119, 89. Saalwaechter, A., McCammon, C., Jr., Roper, C., and Carlberg, K., Am. Znd. Hyg. Assoc. .I., 1977, 38, 476. Foerst, D. L., and Teass, A. W., in Analytical Technology in Occupational Health Chemistry, ed. Dollberg, D. D., and Verstuyft A. W., American Chemical Society, Washington, DC, 1980, pp. 169- 184. Posner, J., and Okenfuss, J., Am. Znd. Hyg. Assoc. .I., 1981, 42, 643, Box, G. E. P., Hunter, W. G., and Hunter, J. S., Statistics for Experimenters, Wiley, New York, 1978, pp. 170-174. Box. G. E. P., Hunter, W. G., and Hunter, J. S., Statistics for Experimenters, Wilcy, New York, 1978, pp. 49-53. Woodfin, W., Gas and Vapor Generating Systems f u r Ldm-atories, DHHSNIOSH Publication No. 84- 1 13, National Institute for Occu- pational Safety and Health, Cincinnati, OH, 1984. Nelson, G. O., Controlled Test Atmospheres, Ann Arbor Science Publishers, Ann Arbor, MI, 197 1. Nelson, G. 0.. Gas Mixtures: Preparution and Control, Lewis, Ann Arbor, MI, 1992. Cassinelli, M. E., Appl. Occ~up. Environ. Hyg., 1991, 6, 215. Generation of Aerosols and Facilities f o r Exposure Experiments, ed. Willeke, K., Ann Arbor Science Publishers, Ann Arbor, MI, 1980. Hinds, W. C.,Aerosol Technology, Wiley, New York, 1982, pp. 379- 39s. Kennedy, E. R., Abell, M. T., Reynolds, J., and Wickman, D., Am. Ind. Hyg. Assoc. J . , 1994, 55, 1 172. Jonas, L. A., and Rehrmann. J. A., Carbon, 1973, 11, 59. Vahdat, N., Swearengen, P. M., Johnson, J. S., Priante, S., Mathews, K., and Neidhardt, A., Am. Znd. Hyg. Assoc. J., 1995, 56, 32. Box, G. E. P., Hunter, W. G., and Hunter, J. S., Statistics for Experimenters, Wiley, New York, 1978, pp. 291-350. Abell, M. T., and Kennedy, E. R., Am. Znd. Hyg. Assoc. J . , submitted for publication. Fischbach, T., Song, R., and Shulman, S., Am. Ind. Hyg. Assoc. J . , 1996, 57,440. Cassinelli, M. E., Hull, R. D., and Cuendet, P. A., Anz. Znd. Hyg. Assoc. J., 1985, 45, 599. Kennedy, E. R., Smith, D. L., and Geraci, C. L., Jr., in Formal- dehyde-Analytical Chemistry and Toxicology, ed. Turoski, V., American Cheniical Society, Washington, DC, 1985, pp. 151-159. Paper 6/01 01 I C Received Fehruury 12, I996 Accepted April 10, 1996
ISSN:0003-2654
DOI:10.1039/AN9962101163
出版商:RSC
年代:1996
数据来源: RSC
|
15. |
What is the best sorbent for pumped sampling–thermal desorption of volatile organic compounds? Experience with the EC sorbents project |
|
Analyst,
Volume 121,
Issue 9,
1996,
Page 1171-1175
R. H. Brown,
Preview
|
PDF (632KB)
|
|
摘要:
Analyst, September 1996, Vol. 121 ( I 171-1175) 1171 What is the Best Sorbent for Pumped Sampl i ng-Thermal Desorpt ion of Volatile Organic Compounds? Experience With the EC Sorbents Project* R. H. Brown Health and Sufety Laboratory, Health and Safety Executive, Broad Lane. ShefSield, UK S3 7HQ A project supported under the EC Measurements and Testing Programme and involving 12 participants has evaluated potentially useful sorbents for the collection and measurement of airborne volatile organic compounds (VOCs) in the workplace. The standard method uses Tenax as sorbing agent, but this is not suitable for very volatile or very polar compounds because of its low capacity for these compounds. Consensus on the best sorbents has been reached on the basis of an evaluation of sorbent performance, using about 20 test compounds with a wide range of volatilities and polarities.The sorbent which satisfied the agreed acceptance criteria for the largest number of compounds was Chromosorb 106. It is therefore recommended for use when sampling the more volatile and polar organic vapours encountered in workplace atmospheres. However, a number of other sorbents also satisfied the acceptance criteria for the majority of compounds and are suitable for a more limited (higher molecular mass) range of organic vapours. These sorbents included Carbotrap, Tenax GR, Tenax TA and Carbopack B. Carboxen 569 was found suitable for sampling the most volatile compounds (ethylene oxide, methanol and propane). Keywords: Volatile organic compounds; pumped sampling; thermal desorption Introduction In connection with EC Directives, it is necessary to develop methods for the measurement of hazardous substances in workplace air in accordance with the general performance requirements of EN 482.1 However, acceptable methods are not generally available for the more volatile and more polar organic vapours.A standard method for the measurement of volatile organic compounds in workplace or environmental air involves the sampling of the polluted air on a suitable sorbing agent, followed by recovery of the sample by thermal desorption. The standard method uses Tenax as sorbing agent, but this is not suitable for very volatile or very polar compounds because of its low capacity for these compounds. A project involving 12 participants has evaluated potentially useful alternatives to Tenax, using test compounds representing very volatile and very polar compounds.Consensus on the best sorbents has been reached on the basis of an evaluation of sorbent performance, which includes sorbent sampling capac- ity, analyte recovery, combined precision and accuracy in compliance with EN 482 and EN 1076,2 recovery after storage * Presented at AIRMON '96, Salen, Sweden, February 5-8, 1996. and background levels of analyte (which determine detection limits). Feasibility Study Potential sorbents were short-listed to five on the basis of preliminary tests on sampling capacity and desorption recovery of a limited list of test compounds. Sample Tubes Sorbents were packed into stainless-steel tubes, 90 mm x 6.3 mm od X S mm id and containing approximately 60 mm of sorbent of 40-60 mesh where available (200-SO0 mg of sorbent, depending on the sorbent density).Tubes were conditioned as recommended by the manufacturer and sealed hand-tight with Swagelok or equivalent fittings with PTFE inserts. Selection of Sorbents Manufacturers' products were selected on the basis of their suitability for thermal desorption (as opposed to solvent extraction) and their availability at the start of the project (May 1993). These sorbents were Anasorb 747, Anasorb CMS, Carbograph I, Carbopack B HT, Carbosieve S, Carbosieve S 11, Carbosieve S 111, Carbotrap, Carboxen 563, Carboxen 564, Carboxen 569, Chromosorb 106, Spherocarb, Tenax TA, Tenax GR, Thermotrap TA. Inevitably, some sorbents were missed, or not considered significantly different from those tested, and some new sorbents have appeared since.Selection of Organic Vapours For the feasibility study, four representative vapours were used, one from each of four main classes of organic solvent: hexane (aliphatic hydrocarbon), 1,2-dichloroethane (chlorinated hydro- carbon), methyl ethyl ketone (ketone) and ethyl acetate (ester). The last three compounds have also been shown to exhibit significant thermal instability on desorption from carbon-type sorbents.3 Sampling Capacity The sampling capacity of the sorbents for the four test compounds was determined as a breakthrough volume, i.e., the volume at which 1% of the applied vapour concentration appears in the effluent from the sample tube when the sorbent is packed into a sample tube and a test atmosphere is pulled through the tube with a sampling pump. This volume was determined from chromatographic retention volumes and sorbent tube theoretical plates essentially by the method of Brown and Purnell.4 Sorbents were selected on the basis of a sampling capacity of at least 1 1 per sorbent tube.I172 Analyst, September 1996, Vol.121 Desorption Recovery Desorption recovery was determined by desorbing tubes spiked with a known amount of test compound as described in MDHS 72, paragraphs 48 and 59.5 The known amount was injected as a volume of vapour in air equivalent to about 20 pg of each compound. Sorbents were selected on the basis of a desorption recovery of at least 95% for all test vapours. Conclusions Carbograph I, Carbopack B, Carbotrap B, Chromosorb 106 and both types of Tenax passed the sampling capacity test. The remaining sorbents either had too low a capacity or an excessively high capacity, which would mean an excessive desorption temperature for convenient recovery, or showed signs of decomposition.These seven sorbents were subjected to the desorption recovery test. All except Carbograph passed the test; Carbo- graph passed the test only on reducing the desorption tem- perature to 140 "C, which would be impractical for many of the compounds on the full analyte list. Main Phase Tests Introduction Nineteen test vapours representing very volatile or very polar organic compounds that commonly occur in workplace air were selected and are listed in Table 1.Two compounds (hexane and methyl ethyl ketone) were also used in the feasibility study. The five sorbents which were short-listed in the feasibility study were evaluated independently by the participants (includ- ing the coordinator) against the full range of test compounds. Each participant evaluated one sorbent, both hexane and a- pinene and approximately half of the remaining compounds (list I or 2), so that there was some duplication and hence confirmation of data. For each sorbent/vapour combination, the following were determined; breakthrough volume (sampling capacity); de- sorption recovery (thermal stability); analytical precision; sampling errors (from standard atmosphere and pump); storage stability; and detection limitbackground levels. Breakthrough Volume (Test 1) Three different procedures were used.A screening method employed two sorbent tubes in tandem. An aliquot of gas standard containing approximately 10 pg of each test compound (list 1 or 2) was injected into the first tube in a flow of clean air at 50 ml min-1. After 20 min ( I 1 total flow), the flow was discontinued and the two tubes separated and analysed. Breakthrough is defined as more than 5% of the analyte in the second tube relative to the total recovered. A chromatographic method employed the same procedure as in the feasibility study. The third method used direct determination of the break- through capacity from a standard atmosphere generated at a concentration of approximately 1 mg m--3 and 80% relative humidity (RH) at 20 "C passed continuously through the sample tube.The effluent was monitored with a gas chromatograph or mass spectrometric detector and the volume determined at which the effluent concentration reached 5% of the inlet concentration. Desorption Recovery (Test 2) Desorption recovery was determined as in the feasibility study except that two loading levels were used, I .0 and 0.1 pg. The test gases were loaded either from gas dilutions or as liquid solutions. The tubes were also loaded with water equivalent to 100 ml of air at 80% RH to simulate high humidity. The initial acceptance criterion was a desorption recovery of at least 95% for all test vapours and both loading levels. Analytical Precision and Sampling Errors (Test 3) The sampling errors were determined by estimating the overall uncertainty (OU), as defined in EN 482,' for one test compound only, hexane.It was assumed that the sampling errors due to the pump were independent of the analyte, so that the overall error for other compounds could be determined by a comparison of analytical errors only. The OU for hexane was determined by using a pump in conforrni ty with the CEN pre-standard CEN/TC 137/WG2/N90, and using at least six replicate sample tubes, drawing a 1 1 sample of a standard test atmosphere of hexane at a concentra- tion of 1 mg m-3 and 50% RH at room temperature (20 "C) through the tubes. The acceptance criterion was a OU of better than 30%. Storage (Test 4) For the storage tests, sample tubes were loaded with a known amount of test compound as described in the desorption recovery tests and desorbed and analysed either immediately (within 24 h) or after 2 weeks of storage at room temperature.Only one spike loading, i.e., 1.0 pg, was used. The acceptance criterion was that the mean results from the stored tubes should not differ from those of the unstored tubes by more than 10%. Background Levels and Detection Limit (Test 5) Conditioned but unspiked tubes were analysed at maximum sensitivity, i.e., a small splitting ratio, and three of the largest chromatographic peaks were selected. The acceptability crite- rion was that the main peak should be < 30 ng. Results Table 1 gives the results of the sampling capacity determina- tions (Test 1) expressed as a breakthrough volume (BTV). R indicates that the sorbent-analyte combination failed the screening test.A normal entry indicates the sampling capacity (breakthrough volume) in 1 8-1, where this capacity is greater than 1 per tube. An italic entry indicates a capacity of less than 1 1 per tube and starred entries borderline cases. The borderline cases were also evaluated by the direct method, which in the main confirmed the indirect chromato- graphic method. However, there were some differences be- tween the behaviour of porous polymer sorbents (Tenax TA and Tenax GR) and the carbonaceous sorbents (Carbopack and Carbotrap). In the case of porous polymer sorbents, the direct measurement of breakthrough was always greater than the indirect measurement, although the ratio varied between 1.25 and 22. For carbonaceous sorbents, the first measurement was sometimes greater (up to four times) and sometimes less (up to nine times), probably reflecting a greater influence of humidity.The cases where the direct volume was greater were all for volumes less than I 1 per tube; cases where the direct volume was less were mostly for volumes greater than 1 0 1 per tube so that, overall, the results did not affect the conclusions reached on the basis of the indirect determinations. The overall conclusion from the sampling capacity studies was therefore a ranking of the five short-listed sorbents, measured as the number of analytes passing the test (out of 19; a half score represents a bordlerline case) as in Table 2. The results of Test 2 were broadly in the range 90-1 10% recovery at both loading levels.However, some laboratoriesCompound Propane Pentane Hexane Benzene Dichloromethane Trichloroethane Methanol Elhano1 Butanol Methyl acetate Methoxyethanol Methyl ethyl ketone Acetonitrile Butyl acetate a-Pinene Decane Ethylene oxide Propylene oxide Hexanal List 1 2 I , 2 1 2 1 2 1 2 2 1 1 2 1 1, 2 1 2 I 2 IAbOrdtOry No. 7 I 1 3 Feas.t 8 2 5 Feas.+ 6 1 4 Feas: 9 12 Feas.+ Feas? 10 15.6 22.8 R R 17.2 10.4 R 6000 R 3.4* Tenax TA R Tenax GR 1 10 10.5 8 37 1.8 1.5 0.2 R 17 3.8 10 6 1 1 5000 6.8 21 22 9.14 R R 15.4 17 6.95 1.37 4.2 1203 R Table 1 Chromatographic determination of sampling capacity [breakthrough volume (BTV)/I 8-11 R = rejected in screening test; * = checked in direct method: itulic = rejected (BTV < 1 1 per tube); n.d. = not determined. + Feas.= results from the feasibility study. 107 R R 13 R R 9.3 14 R R g* 28 R 4.3" n.d. R 8.1 13 n.d n.d. 52 386 0.29 19 77 95 71 5.3 75 22 28 689 Chromosorb 106 Carbopack B 39 21 1 82 11.5 1.3 225 23.4 43 5.5" 0.08 626 110 22 0. I 12 15 192 0. I R 1343 34-430 66 1087 139 390 0.7 < 0.1 27 0.97 I .3 17.2 4 24 6123 16 X loz 30 x 103 R 104 X lo3 2.5 x 104 1.37 < 0.1 3.4 7.5 0.4 2800 n.d. Carbotrap 25.6 586 650 119 0.29 6.23 1.18 49 0.05 150 > I x 10'0 1 X 10'" 5419 - 0.76 7661174 Analyst, September 1996, Vol. 121 experienced difficult in loading at the lower (0.1 pg) level. Also, there was some inconsistency between pairs of laboratories undertaking the same tests. Sorbents were rejected, therefore, only in cases where there was a consistent recovery outside the range 90-1 10%.The results of Test 3 are presented in Table 3. The overall uncertainty (bias plus twice the standard deviation) meets the CEN criterion of better than 30% in all cases, although the porous polymer results are consistently lower than the carbona- ceous sorbents results. In addition, the analytical errors were determined from the desorption recovery tests (at the 1.0 pg level) and are presented for Chromosorb 106 in Table 4. These results indicate that the precisions for compounds other than hexane differ by only a few per cent from those for hexane, so that in all cases, the overall uncertainty will remain well under 30%. The results of Test 4 were broadly in the range 90-110% recovery after storage for 2 weeks. However, there was some inconsistency between pairs of laboratories undertaking the same tests, suggesting that some laboratories had failed to take adequate precautions to seal the tubes.Sorbents were rejected, therefore, only in cases where there was a consistent recovery outside the range 90-1 10%. The results of Test 5 indicated that Tenax TA, Carbopack and Carbotrap met the criterion easily, but inconsistent results were obtained for the other two sorbents, indicating that the criterion could be met by proper conditioning. Preliminary Conclusions On the basis of Tests 2-5, the ranking in Table 2 was modified, indicating a final acceptance 'score,' with Chromosorb 106 as the most promising sorbent. No sorbent, however, was fully satisfactory for all the analytes tested, and Chromosorb 106 was significantly poorer than Tenax TA or Carbograph or Carbo- pack in its blank level, which may render it impractical for very low-level monitoring.Table 2 Performance of sorbents on the basis of main phase tests No. of analytes for which all tests Sorbent satisfactory Tenax GR 9 Chromosorb 106 16 Carbopack B 6 Tenax TA a Carbotrap 1 1 Table 3 Relative overall uncertainty (ROU) Sorbent ROU (%) Tenax TA 6.5 Tenax GR 4.2, 11 Chromosorb 106 8.6, 14 Carbotrap 15.3, 14.9 Carbopack B 20. I Table 4 Precision of analysis (RSD) and storage of test compounds on Chromosorb 106 and Carboxen 569 Storage RSD (%) recovery (5%) Chromo- Carboxen Chromo- Carboxen Organic compound sorb 106 569 sorb 106 569 Propane Pentene Hexane Benzene Dichloromethane 1,1 ,l-trichloroethane Methanol Ethanol Butanol Methyl acetate Methox yethanol Methyl ethyl ketone Ace tonitrile Butyl acetate a-Pinene Decane Ethylene oxide Propylene oxide Hexanal * Not determined.1.8 1.7 2.1; 3.6 2.9 1.9 2.4 5.9 1.3 I .8 5.7 2.2 4.1 3.4 4.2; 2.5 4.2 3.6 3.5 1.7 nd* 115 112 104 100 100 101 96 101 113 121 103 112 104 104 104 103 98 64 nd Additional Tests On the basis of the preliminary conclusions, therefore, some additional tests were undertaken on the Chromosorb 106 blank level and on a suitable sorbent for ethylene oxide, methanol and propane, for which none of the five short-listed sorbents had been suitable. Chromosorb 106 Blank Level Additional tests were undertaken with better defined conditions of conditioning and both before and after storage at room temperature for 2 weeks.The results still vary slightly between laboratories, but they suggested that the blank level was around 20-1 00 ng total VOCs when desorbed at 190 "C and about 1000 ng total VOCs when desorbed at 250 "C. Other Sorbents Additional tests were conducted on Carboxen 569, broadly following the main phase test protocols. The chromatographic determination of sampling capacity gave results of ethylene oxide 140, propane 7.2 and methanol 4.0 1 g-I. The last value was confirmed by direct measurement (4.5 1 g-1). A 300 mg tube therefore has sufficient capacity for all three compounds. Precision and storage data are also given in Table 3. Overall Conclusions The sorbent which satisfied the above acceptance criteria for the largest number of compounds was Chromosorb 106.It is therefore recommended for use when sampling the more volatile and polar organic vapours encountered in workplace atmospheres. However, a number of other sorbents also satisfied the above acceptance criteria for the majority of compounds and are suitable for a more limited (higher molecular mass) range of organic vapours. These sorbents included Carbotrap, Tenax GR, Tenax TA and Carbopack B. Some of these sorbents have a lower blank level than Chromosorb 106 and may be more suitable for trace level air monitoring. Carboxen 569 was found suitable for sampling the most volatile compounds (ethylene oxide, methanol and propane). The support of the EC Measurements and Testing Programme under contract MAT 1 -CT92-0038 is acknowledged. The author also acknowledges the support of the other participants in the EC study: Professor F.Bruner, Universiti degli Studi, Urbino, Italy; Dr. Danilo Cottica, LaboratorioAnalyst, September 1996, Vol. 121 1175 Igiene Industriale, Pavis, Italy; Dr. P. Ciccioli and Dr. G. Bertoni, Instituto sull’hquinamento, Monterotondo, Italy; E. Goelen, VITO, Mol, Belgium; Dr. J. Kristensson, University of Stockholm, Stockholm, Sweden; Dr. M. J. Quintana, INSHT, Baracaldo, Spain; Dr. Rudolf Knutti and A. S. Mogl, Abteilung Arbeitsmedizin und Arbeitshygiene, Zurich Switzerland; Dr. M.-L. Mattinen, Dr. J. Tuominen and E. Sippola, VTT, Espoo, Finland; Dr. K. J. Saunders, Keris, Basingstoke, UK; H. Schlitt and Dr. H. Knoppel, EC Commission/Joint Research Centre, Ispra, Varese, Italy; Dr.A. Venemat, AKZO-ARLA, Arnhem, The Netherlands and Dr. Peder Wolkoff, National Institute of Occupational Health, Copenhagen, Denmark. Full details of the results of the EC study are reported in the Final Report,6 which is available from the author. This report also contains a Standard Operating Procedure, Workplace Atmospheres-Determination of Concentrations of Volatile Organic Compounds in uir-Pumped Sorbent TuhelThermal DesorptionlCapillary Gas Chromatographic Method, in IS0 format. It is hoped that this will form the basis of an I S 0 method for the determination of VOCs in indoor air, ambient air and workplace air. It is with regret that we report the death of Professor Venema during the duration of the project. Note: the selection of particular sorbents or a particular manufacturer’s products does not imply a preference by any member of the consortium or by the EC. Other equivalent sorbents may be used. References Workplace Atmospheres4eneral Requirements for the Perjor- mance of Procedures for the Measurement of Chemical Agents, EN 482: 1994, ComitC EuropeCn de Normalisation, Brussels, 1994. Workplace Atmospheres-Requirements and Test Methods for Pumped Sorbent Tubes for the Determinution of Gases and Vapours, EN 1076: 1995, ComitC EuropeCn de Normalisation, Brussels, 1995. Wright, M. D., M. Phil. Thesis, University of Surrey, 1990. Brown, R. H., and Pumell, C. J., J . Chromatogr., 1979, 178, 79. Volatile Organic Compounds in Air, Methods f o r the Determination of Hazardous Substances, MDHS 72, Health and Safety Executive, London, 1992. Brown, R. H., Study of Sorbing Agents for the Sampling of Volatile Compounds from Air, EC contract MAT1 -CT92-0038, Final Report, 1995. Paper 6102398C Received April 9, 1996 Accepted May 23, 1996
ISSN:0003-2654
DOI:10.1039/AN9962101171
出版商:RSC
年代:1996
数据来源: RSC
|
16. |
Validation of a diffusive sampler for the determination of acetaldehyde in air |
|
Analyst,
Volume 121,
Issue 9,
1996,
Page 1177-1181
Roger Lindahl,
Preview
|
PDF (721KB)
|
|
摘要:
Analyst, September- 1996, Vol. 121 ( I 1 77-1 181) I177 Validation of a Diffusive Sampler for the Determination of Acetaldehyde in Air* Roger Lindahl, Jan-Olof Levin and Maud Martensson Natioiiul Institute foi- Wor-king L fe, POB 7654, S-90713 Umed, Sweden A diffusive sampler for the determination of acetaldehyde in air was evaluated. Acetaldehyde was trapped on a filter impregnated with 2,4-dinitrophenylhydrazine and the hydrazone formed was analysed using HPLC with UV detection. As the capacity of the standard sampler was too low, a modification was made. The capacity of the modified sampler was sufficient for 8 h sampling at 90 mg m--3 and the method was sensitive enough for the measurement of less than 4.5 mg m-3 in a 15 min sampling period. The sampling rate of 17.1 ml min- 1 (RSD = 6.7%) was not influenced by the sampling time, concentration or relative humidity.The sampler can be used for both personal sampling and area sampling. Keywords: Acetaldehyde; difliisive aii- sampling; passive sanrplei-; ocwptiional hygiene Introduction Acetaldehyde is mainly used in the chemical industry for the production of acetic acid. It is also used as a hardening agent in photography and in the food, drug, leather, plastic and paint indush-ies. Acetaldehyde is reporled to be an animal teratogen, mutagen and carcinogen. The vapour is irritating to the eyes, nose and throat and inhalation or ingestion may cause headache, drowsiness, dizziness, respiratory problems and damage to the kidney or lungs. I In the European Union (EU), current occupational exposure limits range from 45 to 180 mg m-?.The exposure limit is either a ceiling value (normally 1.5 min) or an 8 h time-weighted average. The current value from the American Conference of Governmental lndustrial Hygienists (ACGIH) is 45 mg m-3 as a ceiling value.2 Pumped sampling of acetaldehyde can be performed with gas dispersion bottles and different reagent solutions. 2-(Hydroxy- methy1)piperidine (2-HMP) coated on XAD-2 and 2,4-dini- trophenylhydrazine (DNPH) coated on XAD-2 or silica are examples of solvent-free pumped sampling methods. 1 In occupational hygiene, diffusive sampling has been con- sidered as an efficient alternative to pumped sampling for both personal monitoring and area ~arnpling.-7,~ A diffusive sampler, the GMD Model 570 sampler, has been developed by us and is specially designed for use with a reagent-coated filter. The sampler has been validated for measurements of several aldehydes and amines, using filters impregnated with DNPH and 1 -naphthyl isothiocyanate, respectively.”-” The above diffusive sampler has been used for measurements of low levels of acetaldehyde in indoor air.lO The conclusion of that work was that the sampler was suitable for measuring acetaldehyde in indoor air, but further validation was required.To our knowledge, no diffusive sampling method for measuring acetaldehyde at occupational hygiene levels has been reported. This paper describes the validation of the commercially available GMD Model 570 sampler for the measurement of ~ - _ _ ~ ~ ~ Piesented at AIRMON ‘96, Salcn.Sweden, February 5-8, 1996. acetaldehyde in air. The sampler has also been modified for improved performance. Experimental Chemicals Chemicals were of analytical-reagent grade unless indicated otherwise. Solvents used for HPLC analysis were methanol (Merck, Darmstadt, Germany) and water [purified with a Milli- RQ system (Millipore, Milford, MA, USA)]. For coating filters, phosphoric acid (Merck), glycerol (May & Baker, Dagenham, Essex, UK), ethanol (99.99%), acetonitrile (HPLC Grade S; Rathburn Chemicals, Walkerburn, UK) and 2,4-dinitrophenyl- hydrazine (Fluka, Buchs, Switzerland) (recrystallized twice from 4 mol I-’ HCI) were used. For dynamic generation, acetaldehyde (puriss. p.a.; Fluka) was used. Acetaldehyde 2,4-diniti-ophenylhydrazone was prepared from DNPII, acet- aldehyde and concentrated HC1 and recrystallized twice from ethanol.Diffusive Sampler The diffusive sampler used in this project, the GMD Model 570 (GMD Systems, Hendersonville, PA, USA), shown in Fig. 1, is especially designed for the sampling of reactive compounds using a reagent-coated filter. The housing, measuring 60 X 30 X 5 mm, is made of polypropylene. The impregnated filter, 20 X 45 mm, is placed beneath a 2.9 mm thick screen of the same size. Within an area of 20 X 20 mm, the screen has 112 holes with a diameter of 1.0 mm. The filter part beneath the holes is used for sampling (sampling filter) and the other half is used to quantitate the filter blank (control filter). The tape is marked into the two sections by a small ridge on the back of the screen plate. A sliding cover is used to seal the holes when the sampler is not in use.For aldehydes a DNPH-coated filter is used. Filters for Diffusive Sampling The first investigation showed that the capacity of the filters supplied with the GMD standard sampler for formaldehyde was Sliding cover L I 7 - - - * - . - . * - ------..I- - - - . - * - - - * Screen Coated filter / / Fig. 1 Diffusive sampler for reactive compounds.1178 Analyst, September 1996, Vol. 121 too low. New filters for the modified method were coated in the following way: a solution for coating the filters was made from 300 mg of recrystallized DNPH, 0.5 ml of concentrated phosphoric acid, 1.5 ml of 20% glycerol in ethanol and 9 ml of acetonitrile. Glass-fibre filters (2 X 2 cm) were cut from round filters (Type AE, 0.3 pm pore size, diameter 25 mm, SKC, Eighty Four, PA, USA).These were then dipped into the coating solution and allowed to dry on a glass surface. One filter was placed under the sampling part of the sampler screen and another under the control part. The amount of DNPH on the standard Whatman-type filter supplied with the sampler is approximately 0.7 mg per 2 X 2 cm filter. The coating procedure above results in a coating of approximately 3.5 mg of DNPH per 2 X 2 cm filter. Reference Method As a reference method, pumped sampling with Waters Sep-Pak DNPH-silica (Part No. 37500) (Waters-Millipore, Milford, MA, USA) was used. The sampling flow rate was about 20 ml min-1 for all tests. For testing the reference method, 10 PI aliquots of acetaldehyde in methanol were injected into a glass tube connected in front of the cartridge.An air flow of 200 ml min-1 through the glass tube allowed the acetaldehyde to evaporate and reach the cartridge as vapour. The air was controlled with respect to relative humidity and flow. The apparatus has been described earlier." Generation of Acetaldehyde An acetaldehyde-air mixture was generated with the equipment outlined in Fig. 2. The acetaldehyde was injected with a syringe pump (Carnegie Medicine, Stockholm, Sweden) into a glass nebulizer (J. E. Meinhard, Santa Ana, CA, USA). The syringe was chilled to prevent the acetaldehyde from boiling. The syringe pump flow varied from 0.23 to 9.2 pl min-1 and the air flow though the nebulizer was 0.9 N1 min-1.The aerosol from the nebulizer was mixed with air ( 5 N1 min-1) and evaporated in an evaporation chamber with an internal volume of about 0.5 1 (Fig. 2). The air mixture was then further diluted and transported to an exposure chamber that has been described earlier.'2 The air flow in the exposure chamber was 40 1 min-I. The air was controlled with respect to relative humidity and temperature. The wind velocity in the exposure chamber was 0.3 m s-1 for all experiments. The concentration in the exposure chamber was calculated from the amount of acetaldehyde delivered and the total air flow. The concentration was confirmed with the reference method. The experimentally determined value was mostly within +lo% of the calculated value. In all the experiments the calculated value was taken as the true value of the delivered concentration.Sample Analysis The acetaldehyde 2,4-dinitrophenylhydrazone was eluted from the filter by shaking for 1 min with 3.0 or 10 ml of acetonitrile. An aliquot of 10 pl was injected into the liquid chromatograph. An HPLC system consisting of two Waters M-6000 A pumps, Syringe pump Evaporation chamber Exposure chamber \ \ ! tf Air t Humidified air a Waters M-710 B autosampler and a Shimadzu (Kyoto, Japan) adsorbance detector was used. The system was controlled from a computer with a Waters Millenium data system that was also used for evaluating the chromatograms. The column was a Spherisorb ODS 2 (150 X 4.6 mm id) (Phase Separations, Queensferry, Clwyd, UK). The mobile phase was 65% methanol in water at a flow rate of 1 .O ml min--l.The hydrazone was detected at 365 nm. The acetaldehyde hydrazone can exist as two different isomers (cis or trans form) and both have to be considered in the analysis. Fig. 3 shows a chromatogram with the two isomers. The analytical detection limit was 0.1 ng of hydrazone per sample. The method detection limit was, owing to blanks, 0.1 pg of acetaldehyde per sample, corresponding to 0.4 mg m-3 for a 15 min sample. Sampling Exercises A facility for the generation and distribution of dynamic standard atmospheres of aliphatic aldehydes in air has been developed by the Flemish Institute for Technological Research (VITO) in Mol, Belgium.13 The unit offers the ability to charge air simultaneously with water vapour and formaldehyde, acrolein, glutaraldehyde and acetaldehyde at occupational hygiene levels.Acetaldehyde was generated using a capillary dosage technique. The test gas flows through a glass manifold with sampling ports for active sampling and exposure chambers for diffusive samplers. The design is such that up to 30 participants can sample simultaneously from the same aldehyde calibration mixture. Performance tests have shown that the levels of aldehydes generated in the system remain stable to within f2% and that the aldehyde concentrations are known with a total uncertainty of 3%. The aldehyde sampling exercises were performed on two occasions during 1995.14 Results and Discussion Diffusive Sampling The theory of diffusive sampling is described by Fick's first law: mlt = DA(C - Co)/L where rn = mass collected on the sorbent (ng), t = sampling time (s), D = diffusion coefficient (cm2 s-I), A = cross- sectional area of the opening of the monitor (cm2), C = external concentration (ng cm-3 = mg m-3), Co = concentration of the analyte above the surface of the sorbent (ng cm--? = mg m-3) and L = length of the diffusive zone of the monitor (cm).With an ideal adsorbent or a chemosorbent with no reverse reaction, the assumption can be made that Co = 0, which reduces the abo e relationship to mlt = DACIL = SC I 1 DNPH Acetaldehyde- J 0 2 4 6 8 10 Time / min Fig. 2 Generation of acetaldehyde. Fig. 3 sampler exposed to 4.5 mg m--3 of acetaldehyde for 15 min. HPLC trace showing the analytical part of a filter from a diffusiveAnalyst, September 1996, Vol.121 1179 where S = sampling rate, DA/L (cm3 s-l). The diffusion coefficient ( D ) is a physical parameter associated with the analyte and is independent of the sampler construction, whereas A and L are parameters associated with the sampler construction and are independent of the analyte. The sampling rate can be theoretically calculated from the diffusion coefficient and the geometry of the sampler, but these theoretical values often differ from the measured values. An experimentally determined sampling rate is calculated after analysis of diffusive samplers exposed in atmospheres of known analyte concentrations, The most important protocol describing how to test a diffusive sampling method has been adopted by the European Committee for Standardisation (CEN). 15 This protocol describes how tests should be performed to examine effects on sampling rate from parameters such as sampling time, concentration, relative humidity, temperature, storage and wind velocity.These tests must be performed under laboratory conditions with accurate control of the above-mentioned parameters. The GMD sampler has been validated for sampling of aldehydes6~8~9 and amines. 16317 For the analytes formaldehyde, glutaraldehyde, butylamine, dimethylamine, diethylamine and allylamine, the mean deviation of the experimentally deter- mined sampling rates from the theoretically calculated values (diffusion coefficients calculated according to Hirschfelder et ~ 1 . ~ 8 ) was -16% with an RSD of 4.5%. With the use of this deviation as a correction factor, the 'theoretical' sampling rate for acetaldehyde was calculated to be 19.4 ml min-l.The diffusion coefficient 0.1230 cm2 s-l for acetaldehyde was calculated according to Hirschfelder et al. 18 Reference Method As a reference method to verify the concentrations in the exposure chamber, pumped sampling with a Waters Sep-Pak DNPH-silica cartridge was used. To validate the method for measuring acetaldehyde, a number of recovery experiments were performed. The results in Table 1 show that the recoveries were in the range 96-101%. Experiments with the Standard Diffusive Sampler In the experiments with the GMD sampler with the standard filter used for formaldehyde, the diffusive sampler was validated according to the protocol adopted by the Commit6 Europe& de Normalisation (CEN).15 The effects of sampling time, concentration, relative humidity, zero exposure and storage were investigated. The effects of wind velocity and sampler orientation were studied in connection with validation of the sampler for formaldehyde sampling.6 In that study, the wind velocity at the sampler face was varied between 0.05 and 1.0 m s-'. The sampling rate was constant within the wind velocity range studied. Perpendicular orientation of the sampler resulted in a slight increase in sampling rate at high wind velocity. At wind velocities below 0.02 m s-1 the sampler performed well, allowing static area sampling of indoor air.6.8 As wind velocity and sampler orientation are parameters associated with the sampler and not the analyte, these effects were not studied further.Table 1 Recovery of acetaldehyde from Waters Sep-Pak DNPH-silica cartridge with flow rate 200 ml min-I, sampling time 15 min and relative humidity 50% Amount/pg Recovery (%) RSD (%)* 18 98 1 81 101 3 176 96 2 * No. of determinations n = 3. The highest acetaldehyde concentration in the test was 90 mg m--3. The results at this concentration showed a drastic decrease in sampling rate with increasing sampling time (Fig. 4). The storage test at - 18 "C followed the declining line and the zero exposure test was even lower (Fig. 4). In the zero exposure test, the samplers were exposed to 91 mg m-3 of acetaldehyde for 30 min and then exposed to clean air for 7 h. The lowest level tested was 4.5 mg m--3 or one tenth of the lowest EU exposure limit.From the tests with different sampling times and relative humidities at this concentration, a similar but not so drastic decrease in sampling rate with increasing time was observed. The sampling rate of 16 ml min-1 for 15 min exposure decreased to 7 ml min-1 for 7 h exposure time. l4 T Acetaldehyde 90 mg m3 0 4 0 50 100 150 200 250 300 350 0 - 400 450 Sampling timehin Fig. 4 Effect of sampling time on sampling rate for acetaldehyde with standard diffusive sampler. 0, Tests with different sampling times and relative humidity; B, zero exposure; and 0, storage for 14 d at -18 "C. Table 2 Sampling rate of the modified diffusive sampler at various acetaldehyde concentrations, sampling times and relative humidities (RH) with face velocity 0.3 m s-1 Concentration/ Time/ RH Sampling rate/ RSD mg m--3 min (%) ml min-1 (%I n 4.5 4.5 4.5 4.5 180 180 52 90 90 90 90 90 90 90 15 15 480 480 15 15 248 360 3 60 480 480 60 60 60 20 16.8 80 18.0 20 17.0 80 17.1 20 17.6 80 16.3 50 19.2 20 17.4 80 17.1 20 16.2 80 15.6 50 18.6 50 15.4 50 15.1 4.4 2.0 3.1 2.9 2.9 1.3 4.0 6.1 3.9 7.4 4.1 3.0 3.8 2.8 6 6 6 6 6 6 6 6 6 6 6 6* 57 6* * Zero exposure for 7 h with opened sampler after 60 min acet- aldehyde exposure.Storage test: immediate analysis. * Storage test: analysis after 14 d at room temperature. Table 3 Multiple regression analysis of the influence of sampling time, concentration and relative humidity on sampling rate of the modified acetaldehyde sampler (the parameters are normalised to a - 1 to +1 scale) Parameter estimate/ Variable mi min-1 Standard error Intercept (uptake rate) 17.1 0.34 Time -0.41 0.37 NS* Concentration -0.21 0.22 NS Relative humidity -0.09 0.33 NS * NS = not significant (5% significance level).1180 Analyst, September 1996, Vol.121 Table 4 Results from the aldehyde sampling exercises Exercise I* Exercise 2* Sampling Mean/ Bias RSD Sampling Mean/ Bias RSD Sampler time/min mg m-3 (%) (%) n time/min mgm-3 (%) (5%) n Sep-Pak 60-80 54.7 -4.3 3.8 8+ 60 44.2 -9.6 0.8 6 GMD, standard filter 90 18.7 -67.3 10.5 6 60 11.0 -77.5 12.6 6 GMD, glass-fibre filter 90 59.3 3.8 2.5 6 60 47.5 -2.9 2.4 6 Reference value* 57.1 48.9 * Relative humidity = 80% in exercise 1 and 39% in exercise 2. Two consecutive sampling periods. + The reference value is calculated from the amount of acetaldehyde delivered and the air flow.To investigate whether the decrease in sampling rate came from a reverse reaction effect, i.e., formation of acetaldehyde from acetaldehyde hydrazone followed by acetaldehyde back- diffusion from the sampler, some additional experiments were cauied out. The samplers were exposed to 4.5 mg m--3 of acetaldehyde for 20 min and then exposed to clean air for 7 h. During the zero exposure the samplers were either closed or opened. The results from the opened samplers were about 30% lower than those from the closed samplers or 11.7 compared with 16.3 ml min- * . This shows that there was back-diffusion of acetaldehyde from the diffusive sampler at these levels. Validation of the Modified Sampler As the studies on the standard DNPH-coated filter indicated that this filter did not have a capacity large enough for the sampling of acetaldehyde, a modification of the sampling filter was investigated.With the use of the same sampler body but with glass-fibre filters and a more concentrated coating solution, a higher capacity was achieved. Details of the coating of the filters are described in the Experimental section. The modified diffusive sampler was studied according to the CEN protocol.15 The effects of sampling time, concentration and relative humidity were investigated by a three-factor factorial design. The individual results from each experiment are given in Table 2. The mean sampling rate was determined to be 17.1 ml min-’ with an RSD of 6.7% (n = 66). Statistical analysis was performed with the use of multiple regression and the results are given in Table 3.As can be seen, there was no influence of the parameters tested on the sampling rate. From the zero exposure test no reverse reaction could be seen (Table 2). Storage tests showed no decrease in sampling rate after 2 weeks of storage at room temperature, as Table 2 shows. The deviation of the experimentally determined sampling rate from the ‘theoretical’ value of 19.4 ml min-1, as calculated above, is - 12%. Sampling Exercises In order to study the performance of the diffusive sampler for acetaldehyde under conditions of multiple exposure, the sampler was included in two sampling exercises at the facility of VIT0.14 The results are given in Table 4, and include the pumped Sep-Pak method, the GMD sampler with the standard filter and the modified GMD sampler with the coated glass fibre-filter.The acetaldehyde measurements were acceptable for the Sep- Pak pumped sampling method with biases from -4.3 to -9.6%, as can be seen from Table 4. As expected from the validation results, the GMD standard sampler results were too low, with negative biases of more than 60%. The GMD glass-fibre filter method performed excellently, with biases from -2.9 to +3.8% and with RSDs of less than 3%. Conclusions The GMD Model 570 diffusive sampler has been validated for measurement of airborne acetaldehyde in the range 5- 180 mg m--3. The standard GMD impregnated filter tape was not suitable for sampling acetaldehyde as the capacity was too low and there was also a back-diffusion effect.These effects resulted in a decreased sampling rate with increasing sampling time and concentration of acetaldehyde. A modification of the coated filter with the use of a glass- fibre filter and larger amounts of reagent resulted in a diffusive sampling method with increased capacity. The determined sampling rate for acetaldehyde was 17.1 ml min-I with an RSD of 6.7%, corresponding to a relative overall uncertainty19 of 13.4%. The capacity was high enough to permit 8 h sampling at 180 mg m-3, which corresponds to twice the highest EU occupational exposure limit. No decomposition could be seen in the storage tests. The sampler can be used for short-time sampling (15 min) down to at least 4.5 mg m-3. The sampler has been shown previously to perform well at extremely low wind velocities, which makes the method suitable for both static and personal monitoring.This project was financed by the European Commission, Community Bureau of Reference (BCR) contract No. 5563/1/9/397/92/9-BCR-S( 10). The authors are indebted to Dr. S. Vandendriessche of the EC Standards, Measurements and Testing Programme for his valuable assistance throughout the project. They also thank Mr. E. Goelen of the Flemish Institute for Technological Research (VITO) in Mol, Belgium, for the opportunity to participate in the aldehyde sampling exercises. References 1 2 3 4 5 6 7 8 9 10 US Department of Labor, Occupational Safety and Health Admini- stration, OSHA Analytical Laboratory, Salt Lake City, UT, OSHA Analytical Methods Manual, American Conference of Governmental Hygienists (ACGIH), Cincinnati, OH, 1988, Method 68. American Conference of Governmental Industrial Hygienists, 1995-1996 Threshold Limit Values for Chemical Substance3 and Physical Agents, ACGIH, Cincinnati, OH, 1995.Diffusive Sampling. An Alternative Approach to Workplace Air Monitoring, ed. Berlin, A., Brown, R. H., and Saunders, K. J., Royal Society of Chemistry, London, 1987. Clean Air at Work-New Trends in Assessment and Measurement for the 1990s, ed. Brown, R. H., Curtis, M., Saunders, K. J., and Vandendriessche, S., Royal Society of Chemistry, Cambridge, 1992. Levin, J.-O., Anderson, K., Lindhal, R., and Nilsson, C.-A., Anal. Chern., 1985,57, 1032. Levin, J.-O., Lindahl, R., and Anderson, K., Environ. Technol. Lett., 1988, 9, 1423. Noble, J. S., Strang, C. R., and Michael, P. R., Am. Ind. Hyg Assoc. J . , 1993, 54, 723. Levin, J.-O., and Lindahl, R., Analyst, 1994, 119, 79. Lindahl, K., and Levin, J.-O., J . Chromatogr. A, 1995, 710, 175. Brown, V. M., Crump, D. R., Gardiner, D., and Gavin, M., Environ. Sci. Technol., 1994, 15, 679.Analyst, September 1996, VoL. I21 1181 11 Anderson, K., Levin, J.-O., Lindahl, R., and Nilsson, C. A., Chemosphere, 1984, 13, 437. 12 Levin, J.-O., Lindahl, R., and Anderson, K., Environ. Sci. Technol., 1986,20, 1273. 13 Goelen, E., Lambrechts, M., Geyskens, F., and Rymen, T., Znt. J . Environ. Anal. Chem., 1992, 47, 217. 14 Geyskens, F., Lambrechts, M., and Goelen, E., Fadit_)! for the Generation and Distribution of Aldehyde Atmospheres in Air at Oci upational Hygiene Levels, Flemish Institute for Technological Research, Mol, Belgium, 1995. European Committee for Standardization (CEN), Workplace Atmos- pheres4iffusive SumplerJ for the Determination of Gases and Vapors--Requiren7enfs and Test Methods, EN 838 : 1995, CEN, Brussels, 1995. 15 16 17 18 19 Levin, J.-O., Lindahl, R., Anderson, K., and Hallgren, C., Chemo- sphere, 1989, 18, 2121. Lindahl, R., Levin, J.-O., and Anderson, K., J . Chromatogr., 1993, 643, 35. Hirschfelder, J. O., Bird, R. B., and Spotz, E. L., Truns. Am. Soc. Mech. Eng., 1949, 71, 921. European Committee for Standardization (CEN), General Require- ments for the Performance of Procedures for Workplace Measure- ments, EN 482 : 1994, CEN, Brussels, 1994. Paper 6102031 C Received March 23, I996 Accepted May 13,1996
ISSN:0003-2654
DOI:10.1039/AN9962101177
出版商:RSC
年代:1996
数据来源: RSC
|
17. |
Elemental carbon-based method for occupational monitoring of particulate diesel exhaust: methodology and exposure issues |
|
Analyst,
Volume 121,
Issue 9,
1996,
Page 1183-1190
M. Eileen Birch,
Preview
|
PDF (1699KB)
|
|
摘要:
Analyst, September 1996, Vol. 121 (1183-1 190) 1183 Elemental Carbon-based Method for Occupational Monitoring of Particulate Diesel Exhaust: Methodology and Exposure Issues* M. Eileen Bircha and Robert A. Caryb uUS Department of Health and Human Services, Public Health Service, Centers for Disease Control and Prevention, National Institute for Occupational Safety and Health, Division of Physical Sciences and Engineering, 4676 Columbia Parkway, Cincinnati, OH 45226, USA hSunset Laboratory, 201 7 19th Avenue, Forest Grove, OR 971 16, USA Diesel exhaust has been classified a probable human carcinogen, and the National Institute for Occupational Safety and Health (NIOSH) has recommended that employers reduce workers’ exposures. Because diesel exhaust is a chemically complex mixture containing thousands of compounds, some measure of exposure must be selected.Previously used methods involving gravimetry or analysis of the soluble organic fraction of diesel soot lack adequate sensitivity and selectivity for low-level determination of particulate diesel exhaust; a new analytical approach was therefore needed. In this paper, results of investigation of a thermal-optical technique for the analysis of the carbonaceous fraction of particulate diesel exhaust are discussed. With this technique, speciation of organic and elemental carbon is accomplished through temperature and atmosphere control and by an optical feature that corrects for pyrolytically generated carbon, or ‘char,’ which is formed during the analysis of some materials. The thermal-optical method was selected because the instrument has desirable design features not present in other carbon analysers.Although various carbon types are determined by the method, elemental carbon is the superior marker of diesel particulate matter because elemental carbon constitutes a large fraction of the particulate mass, it can be quantified at low levels and its only significant source in most workplaces is the diesel engine. Exposure-related issues and sampling methods for particulate diesel exhaust also are discussed. Keywords: Diesel exhaust; diesel particulate; soot; carbon; carbonaceous aerosol Introduction The widespread use of diesel equipment has generated concern about occupational exposures to diesel engine exhaust, which has been classified as a probable human Carcinogen.’ The National Institute for Occupational Safety and Health (NIOSH) considers diesel exhaust a potential occupational carcinogen and has recommended that employers reduce workers’ expo- sures.2 This recommendation was based on results of five independent animal studies in which rats exposed to unfiltered diesel exhaust showed an increased incidence of benign and malignant lung tumours.1 An increased incidence of lung tumours was observed in one study of mice exposed to filtered diesel exhaust, but the total incidence of lung tumours in this particular study was comparable to that for historical con- trols.’ * Presented at AIRMON ’96, Sllen, Sweden, February 5-8, 1996. Various estimates of unit cancer risk [defined as the risk (lifetime) per unit of lifetime exposure (in pg m-3)] of exposure to diesel exhaust have been calculated. In a draft quantitative risk assessment conducted by State of California Office of Environmental Health Hazard Assessment (OEHHA), carcino- genicity data from one animal bioassay3 and one human study4 were used to predict risks of cancer in humans exposed to ambient levels of particulate diesel exhaust.The calculated risk range based on the animal (rat) data was 3 X 10-5-3 X 10-4 per pg m-3. Different models of carcinogenicity and different measures of exposure were used in the calculation of the range of estimates. Based on results of the human study by Garshick et al.4 and two different measures of cumulative exposure, lifetime unit risks (95% upper confidence level) of 3.4 X and 2.3 X 10-3 were calculated.Of these two estimates, the rounded value of 3 X 10-4 was proposed as the lifetime unit risk of exposure and this value reportedly is ‘consistent with the current evidence. ’ s Particulate diesel exhaust, like particulate air pollution in general, also is of concern with respect to non-cancer health effects. The US EPA has proposed an inhalation Reference Concentration (RfC) of 5 pg m-3 for the non-cancer health effects of diesel exhaust,G and the OEHHA has proposed to adopt this value for the chronic inhalation reference exposure level (REL) in California.5 The RfC for a substance is an estimate of a daily exposure of humans, including sensitive subgroups, that is ‘likely to be without appreciable risk of deleterious effects during a lifetime of exposure.’s A com- prehensive review of the potential health effects of exposure to diesel exhaust has recently been published? Because diesel exhaust is a highly complex mixture contain- ing thousands of compounds, 1 7 2 some measure(s) of exposure must be selected.As tumour induction in animals is associated with exposure to unfiltered diesel exhaust, a measure of exposure to the particulate fraction of the exhaust was sought. Previously, specific soot-borne organic compounds have been targeted; however, this approach is limited in that selected compounds usually are present only at low (often non- detectable) levels.8 Further, although considerable research effort has been devoted to chemical characterization of the solvent-extractable fraction of diesel soot, a unique marker for diesel exhaust has not been found.Even if a unique marker(s) could be identified, the exhaust composition is highly vari- able,g.lo so any single cbmpound or compound class probably would not reflect exposure to the diesel aerosol mass concentra- tion. A review of analytical methods for chemical character- ization of the organic fraction of particulate diesel exhaust has been published previously.11 Gravimetric methods for diesel particulate matter have been employed for exposure monitoring. One approach12 involves gravimetric determination of respirable combustible dust1184 Analyst, Septenihrr 1996, Vol. 121 (RCD). In this method, combustible material in a respirable aerosol sample is determined as the difference in filter mass before and after combustion at 500 "C.Similarly, a low- temperature ashing method has been applied to estimate the diesel fraction of respirable dust. ' 3 Other gravimetric methods have involved direct determina- tion of either 'fine particulate material' 14 or sub-micrometre aerosol. Impactors with optimized cut-points (sub-micrometre) have been developed for application in coal mines to minimize interference from coal dust. These devices effect separa- tions of particles according to their aerodynamic diameter through inertial impaction. Because diesel exhaust particles are essentially sub-micrometre, 9-26 their separation from larger, mechanically generated dust is possible. Although size-selective sampling can improve the selectivity of gravimetric analysis, sub-micrometre aerosol (e.g., mists, cigarette smoke and fuels) can arise from non-diesel sources.Thus, even if very small masses could be determined precisely, the. presence of other sub-micrometre aerosols could make the accuracy of such measurements highly questionable, especially when determining low levels (e.g., < 200 pg m-3) of particulate diesel exhaust. Aside from the potential interference problem, a small portion of diesel aerosol may be excluded in some cases if a sub-micrometre size fraction is collected. The choice of a 0.8 pm cut-point was based on size distribution data collected in coal mines, where vehicles are normally equipped with water scrubbers that tend to collect larger (i.e., > 1 pm) particles. In other workplaces, larger diesel particles could be present and these will be excluded if a impactor with a sub-micrometre cut- point is employed. Because previously used methods for measuring occupa- tional exposures to particulate diesel exhaust lack adequate sensitivity and selectivity, a new approach was needed.As diesel exhaust particles are composed primarily of carbon, carbon measurement is a logical approach but the many non- diesel sources of organic carbon make a total carbon measure- ment interference prone. For this reason, use of an elemental carbon marker was proposed.27 Of the exhaust components that have been determined, elemental carbon is the superior measure of exposure to particulate diesel exhaust because elemental carbon constitutes a large portion of the particulate mass, it can be quantified at low (e.g., environmental) levels and its only significant source in many workplaces is the diesel engine.Selection of an elemental carbon marker also was based on previous work by Fowler,"' who evaluated various analytes as indices of 'over-all diesel exposure.' Included in the evaluation were C02, CO, SO2, NO, NO2, total and fine particulate material, volatilizable carbon (organic) and elemental carbon. Of these constituents, elemental carbon was reportedly the most reliable measure of 'diesel exhaust as an entity' (i.e., it reflected exposures to the largest number of exhaust components examined). In addition to being a specific and sensitive measure of particulate diesel exhaust. use of an elemental carbon marker was proposed in view of preliminary results of an animal study28JY involving groups of rats exposed to unfiltered diesel exhaust or carbon black containing very little organic content.In this study, tumour incidence was similar in the two groups, which reportedly indicates that fine particulate carbon itself may play a primary role in the formation of malignant tumours in rats exposed to high levels of diesel particulate.30 The relevance of this finding with respect to humans exposed to much lower levels of diesel particulate is uncertain. A full report on the study ha(; recently been published.30 Various methods, most of which involve thermal speciation, have been used for determination of organic and elemental carbon (OC and EC, respectively) in carbonaceous aerosols.Thermally evolved OC and EC are oxidized to C02, which is then quantified either with a non-dispersive infrared detector or electrochemically. Alternatively, the C02 can be reduced to CH4, which is then quantified with a flame ionization detector. Although various techniques have been used for the analysis of carbonaceous aerosols, different methods generally give similar results for total carbon. In an interlaboratory comparison of methods for the analysis of carbonaceous aerosols,?' good agreement between total carbon results of 11 laboratories (pooled RSD = 9.0%) was obtained for a wide variety of 'reference' samples (e.,?. , organic aerosol, soot, wood smoke and diesel exhaust). In contrast, large disagreement was seen in the EC and OC results reported by 10 laboratories (pooled RSDs for OC and EC were 25.8% and 52.3%, respectively). Unlike the case with total carbon, there are no reference standards for speciation of different carbon types in complex carbonaceous aerosols.For this reason, methods that speciate EC and OC are considered operational32 in the sense that the method itself defines the analyte. Results of laboratories performing such analyses can be compared on a relative basis, but it is not possible to assess the accuracy of results. Although a standard reference material (SRM 1650) for diesel particulate matter is available from NIST, this standard applies only to specific soot- borne organic compounds. In this paper, an EC-based approach for monitoring occupa- tional exposure to particulate diesel exhaust is described.A thermal-optical analysis technique was investigated for this purpose. This particular technique was investigated because the instrumentation has desirable design and performance features (e.g., flame ionization detector, pyrolysis correction, minimal undesired oxidation of EC, no carbonate interference) that afford it greater sensitivity and selectivity than other methods for OC and EC determination. Selectivity is especially im- portant when analysing complex carbonaceous aerosols, as is evident from the results of the methods comparison study.? In the study, four laboratories took measures to minimize or correct for pyrolytically generated EC, but only the thermal- optical method's EC results were lower than those reported by laboratories that did not correct for pyrolysis.In addition to pyrolysis, the higher EC results reported by these laboratories may also be due to the presence of carbonate-source carbon (CC), which was found in wood smoke and automotive exhaust samples. A discussion of the thermal-optical instrument's design, operation, and performance is provided in this paper. Exposure-related issues and results of investigation of various sampling strategies are also discussed. Experimental Reagents and Materials Standard solutions (Supelco, Bellefonte, PA, USA) of PAHs were 296% pure as determined by capillary GC coupled with flame ionization detection (FID). All chemicals (Aldrich, Milwaukee, WI, USA) used for the preparation of aqueous solutions were of analytical-reagent grade or better.Coarsely ground (-20 mesh) coal samples were obtained from the Penn State Coal Sample Bank at the Energy and Fuels Research Center, Pennsylvania State University. Coal samples analysed for carbon were sieved to 50 mesh prior to analysis. Pallflex 2500QAT-UP quartz-fibre filters (Pallflex, Putnam, CT, USA) were used for all samples (air and spiked filter punches). To remove possible carbon contamination, quartz-fibre filters were pre-cleaned in either a muffle furnace or a low-temperature asher. Both commercial and prototype samplers were used for collection of diesel aerosol. Details on the sampling devices employed are given in a following section (see Sampling). Thermal-Optical Inslrumentation A schematic diagram of the thermal-optical instrument is shown in Fig.1. The instrument is a modified version of aAnulyst, Septenther 1996, Vol. 121 1185 previously described designT3 In thermal-optical analysis, speciation of organic and elemental carbon is accomplished through temperature and atmosphere control. An optical feature corrects for pyrolytically generated EC, or ‘char,’ which is formed during the analysis of some materials (r.g., cigarette smoke and pollen). He-Ne laser light passed through the filter allows continuous monitoring of filter transmittance. Because of the high temperatures employed during the analysis, quartz- fibre filters are required for sample collection. Normally, a 1.54 cm? rectangular portion (taken with a punch) of the filter deposit is analysed, and organic and elemental carbon are reported as pg C per cm2 of deposit area.Total EC and OC on the filter are calculated by multiplying reported values by the sample deposit area. In this approach, a homogeneous filter deposit is assumed. FID is used for quantification (as CH4) of evolved carbon, and instrument calibration is achieved through injection of a known volume of methane into the sample oven. Additional details on the analytical procedure are described below (see Carbon Speciation). Coal Dust Generation System An in-house dust generation system was used for evaluation of prototype impactors, which were designed for use in under- ground coal mines (see Sampling). The generation system is a modified version of one described previously.34 Instruments were interfaced to the system dust chamber to allow fluidization of coal dust (TSI Model 3400 fluidized bed generator) and measurement of its mass concentration (TEOM Mass Measure- ment Systems, Rupprecht and Patashnick, Albany, NY, USA). A charge neutralizer (X-STATIC charge neutralizer) was located inside the chamber.Coarsely ground (-20 mesh) coal dust was sieved to 50 pm for use in the fluidized bed generator. A stainless-steel cyclone (TSI) operated at 9 1 min-1 was used with the generator for further size classification. The reported cut-point (Ds0 value) for the cyclone at this flow rate was 3.5 Elm. Results and Discussion Carbon Speciation An example of the instrument output, called a ‘thermogram,’ is shown in Fig. 2. The three traces appearing in the therrnogram correspond to temperature, filter transmittance and detector response (FID). The analysis proceeds essentially in two stages.In the first, organic and carbonate carbon (if present) are volatilized from the sample in a pure helium atmosphere as the temperature is stepped to about 820 “C. Evolved carbon is catalytically oxidized to CO2 in a bed of granular MnO? (held at about 900 “C), reduced to CH4 in a nickel-firebrick methanator (at 450 “C) and quantified as CH4 by FID. During the second stage of the analysis, pyrolysis correction and EC measurement are made. The oven temperature is lowered, an oxygen-helium ( 1 + 9) mix is introduced and the oven temperature is then raised to about 860 “C. As oxygen enters the oven, pyrolytically generated ‘EC‘ is oxidized and a concurrent increase in filter transmittance occurs (Fig.2). Correction for the char contribu- tion to EC is accomplished by measuring the amount of char oxidation required to return the filter to its initial transmittance value. The point at which the filter transmittance reaches its initial value (vertical solid line) is defined as the ‘split’ between organic and elemental carbon. Carbon evolved prior to the split is considered ‘organic’ (including carbonate), and carbon volatilized after the split and prior to the peak used for instrument calibration (final peak) is considered ‘elemental’. If desired, the presence of carbonate can be verified through analysis of a second portion (punch) of the filter after its exposure to HCI vapour. In the second analysis, the absence or diminished size of the suspect peak (typically the fourth ‘OC’ peak) is indicative of carbonate in the original sample.Instrument Calibration Because a standard for OC, EC and CC in carbonaceous aerosol samples is not available, OC standards must be used for instrument calibration. In this approach, accurate determination of carbon in samples of unknown composition requires an instrument response that is independent of the compound and matrix type. To investigate whether the efficiency of the thermal-optical method’s catalytic oxidation-reduction process is independent of sample type, a range of analyte types were selected for analysis. Standard solutions containing PAHs and aqueous solutions of various water-soluble organic compounds were analysed. An aliquot of solution was applied to a 1 .S cm2 filter (quartz) punch.Prior to application of the standard solution, the punch was pre-cleaned in the analyser to remove possible carbon contamination. Results for four of the six PAHs examined are plotted in Fig. 3. Because two of the compounds (naphthalene and Fig. 2 Example thermogrdm for sample containing rock dust (carbonate source) and diesel exhaust. The three tnices correspond to temperature, filter transmittance and detector (FID) response. Peaks correspond to organic (OC), carbonate (CC), pyrolytic (PC) and elemental (EC) carbon. Thc final peak is a methane calibration peak Fig. 1 Schematic diagram of thermal-optical instrumentation. Gas stream selected by four-port switching valve (V 1). Pure He used during first stage of analysis; 02-He ( 1 + 9) mixture used during second stage (see text for details).1186 Analyst, September 1996, Vol.121 acenaphthylene) were too volatile to be determined reliably, results for these compounds were not reported. Response of each of the other PAHs examined appeared linear (r2 2 0.996, 3 5 n s 13), but benzo[u]pyrene and benzo[k]fluoranthene responses were higher than predicted. High results for these two compounds are attributed to the presence of residual solvent (acetone), which did not completely volatilize from the filter punch. No discernible difference between the responses of dimethylbenz[a]anthracene and fluoranthene was seen. Better agreement between actual and reported values was obtained with the aqueous standard solutions, where interfer- ence of residual organic solvent was not a problem.Results of linear regression of the data (i.e., pg C reported \ws-sus pg actual) for each aqueous standard and all standards (aqueous) combined are listed in Table 1. Based on the confidence intervals of the slopes, no compound dependence was noted with the compounds examined. The response of individual compounds was linear (r2 2 0.999, 4 s n I 19) and the slopes of-the regression lines were all near unity, indicating that the efficiency of the method's oxidation-reduction process (i.e. , conversion of analyte carbon to CH4) was near 100%. In addition to the OC standards, other carbonaceous materials were analysed by the thermal-optical method and by two other independent laboratories (Galbraith Laboratories, Knoxville, TN and M-H-W Laboratories, Phoenix, AZ, USA).The materials examined and the analytical results are listed in Table 2. Good interlaboratory agreement was seen between results reported by the three laboratories. The RSD of the mean percentage of carbon found in the samples was <7% for all materials examined. Good reproducibility also was obtained by individual laboratories; the variability in the mean of two analyses was less than 2%. Analytical Range and Limit of Detection The analytical range and LOD of the thermal-optical method were determined with aqueous solutions of OC standards. An upper filter loading of about 50 pg C cm-2 was obtained when 190, 1 / - C A I / FLUOR BaP BkF DMBaA 1 0 20 40 60 80 100 120 140 160 Carbon actual/pg Fig.3 Plot of amount of carbon reported versus actual amount. Results obtained for four PAHs: fluoranthene (FLUOR), benzo[a]pyrene (BaP), benzo [ k ] fluoranthene (BkF) and dimethylbenz[a] anthracene (DMBaA). Table 1 Linear regression results for aqueous solutions of OC standards Slope,* Analyte n r2 CLf Sucrose 7 >0.998 0.98, 0.04 Caffeine 4 >0.999 0.98, 0.10 GI y cine 4 0.999 0.96, 0.09 EDTA 19 >0.999 0.99, 0.01 KHP 9 >0.999 1.01, 0.01 All aqueous 43 0.999 0.99, 0.01 SE of 11 Intercept estimate 0.94 1.06 0.85 0.63 1.45 1.03 0.15 0.35 0.30 0.89 -0.21 0.28 * pg C reported/pg C actual. 95% confidence limits of slope. the carbon was evolved as a narrow peak, which occurred with caffeine. In general, when a given peak in the thermogram represents more than about 50 pg C cm-2 a smaller sample punch should be analyzed to avoid an off-scale response.Unlike the case for individual solutions of OC standards, various types of carbon are usually present in field samples and the different types are evolved over different temperature ranges. In this situation, much higher carbon loadings (e.g., 200-300 pg cm-2) are possible. Although carbon loadings as high as 300 pg cm-2 can be analysed, the split between OC and EC can be affected if the filter is so heavily loaded that changes in transmittance cannot be monitored. The loading at which this occurs depends on the optical properties of the sample; however, in general, EC loadings less than 20 pg cm-2 are recommended. For an 8 h sample collected at 2 1 min-l on a 37 mm filter, this represents a maximum EC concentration of about 180 pg m-3.In practice, pyrolysis correction is unnecessary when EC loadings are this high because the char contribution to the measured EC is only minor (note: an impactor with appropriate design specifications must be used in coal mines). In such cases, the user can simply designate the OC-EC split prior to the EC peak and a maximum EC concentration of about 0.5 mg m-3 can be determined when collecting 8 h samples at 2 1 min-1 on 37 mm filters, or an upper EC loading of about 50 pg cm-2. An option for assignment of the split time (i.e., overriding the transmittance-based value) by the analyst is included in the data analysis software. If measurement of higher EC concentrations is desired, lower sampling rates or shorter sampling times (or both) should be used.In many workplaces, EC concentrations are well below this upper limit,-75 a notable exception being the mining industry. Requirements for applica- tion of the thermal-optical method in the mining industry will appear e1sewhe1-e.~~ The LOD of the thermal-optical method was calculated following a standard operating procedure.37 Accordingly, low- level calibration standards covering a range from less than the expected LOD to no greater than 10 times that expected were prepared and analysed. A 0.05 mol 1-1 volumetric standard solution (Aldrich) of the disodium salt of EDTA was diluted with distilled, de-ionized water and aliquots of the solution were applied directly to 1.54 cmz quartz-fibre filter punches, which is the punch size normally taken from the filter sample for analysis.As with all standards, the punches were pre-cleaned in the analyser prior to application of the standard solution to ensure that they were carbon free. Loadings on the punches covered a range from 0.23 to 2.82 pg of carbon and five levels were examined. The three lowest carbon levels were analysed in duplicate and the two highest were analysed once, giving a total of eight analyses. Table 2 Analysis of carbonaceous materials: interlaboratory comparison C reported (96) Sample Lignite SubB' hvAa" Ivb* Anthracite SRM 1649 SRM 1650 Humic acid Gal brai th 51.51 60.17' 61.04 75.46 77.98 17.55 77.19 41.84 M-H-W 55.40 63.34 62.049 77.0 1 80.64 18.95 77.01 44.44 Thermal- optical 53.27 57.12j. 70.40 75.72 79.76'' 17.60 74.60 41 .05 Mean C (%) RSD(%) 53.39 2.98 60.21 4.22 64.49 6.51 76.06 0.89 79.46 1.39 18.03 3.60 76.27 1.55 42.44 3.41 * ASTM coal rank; subbituminous (subB) and bituminous (hvAB, lvb) coals.Mean, n = 2, RSD = 0.26%. Mean, n = 3, RSD = 1.90%. 8 Mean, n = 2, RSD = 0.14. 1 Mean, 17 = 2, RSD = 6.97%; second sample analysed 6 months after first analysis.Analyst, September 1996, Vol. 121 1187 Results of linear regression of the low-level calibration data (i.e., pg C reported versus pg C actual) were used to calculate the analytical LOD according to LOD = 3 s,,/m, where sy is the standard error of the regression and m is the slope of the regression line. An LOD of about 0.23 pg of carbon (or about 0.15 pg C per cm2 of filter) was calculated (intercept = 0.02, SE of y estimate = 0.08, r2 = 0.998, m = 1.05 f 0.02). In terms of concentration, if air samples were collected for 8 h at 2 1 min-1 on 37 mm filters, an LOD of about 2 pg C per m3 of air could be expected. Hence the method is capable of EC determination at background (i.e., environmental) levels, which are typically about 2-4 pg m-3 in air.35 Interferences A variety of carbonaceous aerosols are present in the work- place; however, most of these aerosols are primarily OC and do not pose an interference in the determination of EC.Unlike other thermal methods, char generated during thermal-optical analysis of some materials (e.g., cigarette smoke, spores, pollen and plant debris) does not interfere because the method optically corrects for carbonization. The potential contribution of EC by cigarette smoke was examined.Filter samples of cigarette smoke were collected for analysis. The mean EC found was only 0.86% (0, - = 0.48%, y1 = 6) of the total carbon content. This result is in good agreement with that obtained previously in an industrial hygiene study by other NIOSH investigat~rs,~S where less than 2% of the carbon found in cigarette smoke was elemental. The mean EC fraction reported in this paper may be slightly lower than that found previously because the current temperature pro- gramme differs slightly from that used previously. It is important to recognize that the temperature programme em- ployed in the analysis was optimized for specificity against cigarette smoke, and other temperature programmes could produce different results.For example, when a different temperature programme was used for the analysis of cigarette smoke, the evolution of high-temperature carbon and the concomitant increase in filter transmittance happened too quickly and the OC-EC split occurred earlier than it did with the optimized programme. Consequently, 20-30% of the carbon in cigarette smoke was designated elemental when a less than optimum temperature programme was used. Depending on the thermal programme employed, EC fractions in this range could potentially be found with other carbon analysis methods. Various inorganic materials may be present in some work- places. Coloured inorganics (e.g., iron oxide) ordinarily do not present a problem with respect to the OC-EC split because their absorbance is usually negligible (relative to EC) and is constant during the analysis.Although metals that form black oxides could affect the OC-EC split, this situation can be remedied through manual assignment of the split. Non-absorbing in- organics (e.g., ammonium sulfate or nitrate) that scatter light and vaporize during the analysis also do not affect the determination of EC unless they are present at high loadings (e.g., > 100 pg cm-2) and the amount of EC is low (e.g., background levels). Other airborne materials that could be encountered include wood smoke, welding fumes, automotive (gasoline) exhaust and soils. Because wood smoke is largely (195%) OC, its EC contribution should be relatively minor in most situations. Welding fumes consist mostly of organic and metallic species (e.g., manganese, iron), which ordinarily do not interfere in the analysis.Gasoline-powered vehicles emit much less soot than do diesel vehicles,39 so their contribution to the measured EC should be negligible. Finally, because soils are composed largely of inorganic matter, they also do not present an interference (soils high in humic acid may be an exception if present at high levels). Although interferences in most workplaces are not expected, it is not possible to predict all possible scenarios. Obviously, the method should not be applied for monitoring particulate diesel exhaust when other major sources of EC ( e . g . , carbon black) are present. In general, the best means of determining whether the EC contribution from a particular airborne material could be significant is through analysis of the bulk material.If the dust is mechanically generated, as is coal dust, a size-selective sampling approach can be used to minimize its collection (see below). If the material is a combustion aerosol that cannot be excluded on the basis of size ( e . g . , cigarette smoke), its EC contribution will depend on the concentration and EC content of the aerosol. Sampling General industry When interference from other dusts is not a problem, which is often the case in general (i.e., non-mining) industry, size- selective samplers are not required. In this situation, any of a variety of samplers can be expected to give equivalent EC results because sub-micrometre diesel particles will be collected with the same efficiency (near 100%).To confirm this assumption, a sampler comparison study was conducted at an express-mail facility that uses diesel-powered trucks for package transport. Included in the comparison were open-face 35 and 25 mm cassettes, the Model 298 personal cascade impactor (Graseby/Andersen, Atlanta, GA, USA), the IOM inhalable dust samplel-40 and four different prototype impactors designed by researchers at the University of Minnesota and the US Bureau of Mines18 (BOM), the Mine Safety and Health Admini~trationl~ (MSHA) and NIOSH.'S Two each of eight sampler types were used and all samplers were operated at 2 1 min-l. The prototype impactors were modified to accom- modate quartz-fibre filters, which are necessary for thermal- optical analysis.The impactors were designed for use in coal mines and are not required in general industry, but they were included in the evaluation so results obtained with these devices could be compared with those found using commercially available samplers. Because diesel particles are largely sub- micrometre, significant differences in the EC results found with the different sampler types were not expected. Results of thermal-optical analysis of diesel exhaust samples collected with the eight different sampler types are given in Table 3. Because the precision of the mean EC concentration determined with the IOM sampler was poor (RSD = 31%), results obtained with this sampler were not included in the Table 3 Sampler comparison study: thermal-optical analysis of carbon in diesel particulate samples Meant Cut- Meant point*/ EC/ RSD OC/ RSD mgm-3 (%) mgm-3 (%) Sampler pm 25 mm cassettef - 37 mm cassette* - Model 298 cascade IOM sampler - BOM impactor 0.8 University MSHA impactor 0.8 NIOSH impactor 1 impactor 8 0.9 impactorq 0.8 3.95 4.16 3.72 4.34 4.20 4.09 3.96 4.02 0.08 0.02 0.03 0.3 1 0.05 0.04 0.10 0.02 8.99 9.74 6.88 8.38 9.58 9.27 8.44 7.54 0.00 0.04 0.08 0.04 0.00 0.03 0.0 1 0.04 * Approximate 50% cut-point.* n = 2. + open-face configuration. Developed jointly by the Seven stages loaded to give 0.93 pm cut-point. University of Minnesota and BOM.1 I88 Ariulyst, Septeniher. 1996, Vol. 121 statistical comparison of samplers. The RSD of the mean EC concentration found using results for the other seven sampler types was 5.6% (mean EC = 4.01 mg m-3, n = 14).To compare differences between means for different sam- plers, estimates of the between- and within-sampler variances were calculated (0.05 16 and 0.0507, respectively). Based on results of a one-tailed F-test (F6,7 = 1.02), differences between sampler means were not significant [critical value of F =: 3.87 ( P = 0.05)j. Higher intersampler variability was seen in the OC concentration (mean OC = 8.63 mg m-3, RSD = 12.396, ii = 14), which can be expected when filters are used for collection of organic aerosol containing volatile or semi-volatile components. Suniplirig in c*oul nii1it.s Although interferences are not expected in most workplaces, interference from coal dust was expected when sampling for particulate diesel exhaust in coal mines.To address sampling requirements in this workplace, the prototype impactors described above were used to examine the potential contribution of coal-source EC. Two each of the four prototypes and two commercial impactors (Model 298 personal cascade impactor, Graseby/Andersen) were included in the evaluation. An in- house aerosol generation system (see Experimental) was used for generation of coal dust. A brief summary of results relevant to sampling in coal mines is provided below. A complete discussion of the results, as well as issues related to sampling in the mining industry in general. will appear Before evaluating the performance of the impactors, homoge- neity of the dust concentration within the chamber was confirmed in a separate experiment.Eight nylon cyclones each in series with a 37 inm plastic cassette loaded with a pre- weighed PVC filter were used for sample collection. The samplers were placed in a circular arrangement near the base of the chamber, and two personal cascade impactors were operated in the centre ( i . ~ . , within the circle of cyclones). Results of this experiment indicated that the dust concentra- tion was sufficiently homogeneous. The RSD of the mean dust concentration (1.26 mg m-3) found with the cyclone-filter samplers was 4.6% (11 = 8). Results for the cascade impactors were higher than those seen with the cyclones; however, a direct comparison with the cyclone results was not considered appropriate because the cyclones were operated at 2 1 min-1. Ordinarily, cyclones are operated at 1.7 1 min-I when collecting a respirable dust fraction, but this flow rate was not used because the prototype impactors being investigated were designed for operation at 2 I min-' (see Table 3 for cut-points). In addition, a direct comparison was not considered appropriate because the respirable dust concentration determined with cascade impactors must be estimated by applying weighting factors to different stages.In view of these variables, the respirable dust level ( I .41 mg nip3, RSD = 1 1 .O%) calculated with the cascade impactor data was in fairly good agreement with that found with the cyclones. After verifying that the dust distribution within the chamber was homogeneous, the performance of the prototype impactors was evaluated at two coal dust concentrations.As with the gravimetric run described above, the samplers (cyclone plus impactor) were placed in a circular arrangement and cascade impactors were positioned in the centre. For the purpose of randomization, prototype impactors of a given type were diametrically opposed. Again, all samples were collected at 2 1 inin-'. At a respirable dust concentration of 1.50 mg m--3 (total dust = 2.86 nig m-?), the performances of the MSHA, BOM and University of Minnesota samplers were comparable; however, the total carbon concentration found with the impactor designed by NIOSH researchers was over five times higher than the mean concentration (approximately 54 pg m--3) found with the other three prototype impactors. When the respirable dust level was about 5 mg m--3 (total dust 12 mg m--3), only the BOM and University impactors, which are four-nozzle designs with approximately 0.8 ym cut-points, appeared effective in excluding larger-sized dust.The mean carbon concentration found with these two impactors was about 156 yg m-3 and the corresponding mean EC level was about 30 pg m--3. The carbon loading found on the final filter of the MSHA impactor was five times the mean loading found with the BOM and University impactors and the loading on the NIOSH impactor was nine times higher. In summary, the results of this investigation indicate that the thermal-optical method could be applied in coal mines with only a relatively minor contribution of coal-source EC provided that an impactor with appropriate design specifications is used.Even when the respirable dust concentration was as high as S mg mp7, which is 2.5 times the compliance (US) level, the EC concentration was only about 30 pg m-3. These results are consistent with those obtained in field studies, where only low levels (515 pg m--3) of EC were found in electric-powered (i.e., non-diesel) underground coal mines when impactors with sub- micrometre cut-points were used.27?-36 Although this back- ground is higher than typical environmental levels (2-4 pg m-3), it is relatively minor when compared with the EC concentrations found in diesel-based mines, which generally are much higher than those seen in non-mining occupations. For example, EC concentrations found in preliminary surveys41 of diesel-based mines typically ranged from about 1 SO to 500 pg m-3 and sometimes exceeded 800 pg m--3.Exposure Issues Uncertainties in human risk Although various methods for detecting potential human carcinogens have been developed, elucidation of a causal relationship between exposure to a suspect agent and human cancer ultimately requires epidemiological evidence.42 Often, results of epidemiological studies are equivocal unless strong relationships exist and they are not specific when exposures involve a wide variety of agents.42 Current epidemiological evidence suggests that long-term employment in occupations where significant exposure to diesel exhaust occurs is asso- ciated with a 'modest' increase in risk (relative risk 1.2-1 .5) for lung ~ a n c e r ; ~ 3 however, because the apparent increase in risk is small, it is sensitive to misclassification of subjects (e.g., smoker versus non-smoker).4-3 Animal studies provide evidence that unfiltered diesel exhaust is a pulmonary carcinogen in rats when high concen- trations are inhaled chr0nically;~3 evidence for exposed mice and Syrian hamsters is questionable and negative, respectively.Confounding the assessment of human carcinogenic risk using rat data is the validity of the extrapolation, which has been questioned.7~30 Future research is necessary to establish whether the mechanisms responsible for malignancies in rats are operative in humans. Exposure criteria: current status As discussed in the previous section, there are many uncer- tainties and complex issues involved in the quantitative assessment of human cancer risk associated with exposure to diesel exhaust.At present, health-based exposure criteria have not been established, but the American Conference of Govern- mental Industrial Hygienists (ACGIH) has proposed a TLV of 150 yg of sub-micrometre particulate per m-3 air (see 1995-96 Notice of Intended Changes list). Standards based on technical feasibility have been set in Germany and British Columbia. British Columbia limitsAnal-yst, September 1996, Vol. 121 118'3 respirable combustible dust (RCD) concentrations44 in mines (coal mines excepted because of coal dust interference) to 1.5 mg m--l. Based on an estimated 33% contribution to the RCD by non-diesel sources ( ~ . g . , oil mist), this standard corresponds to an average diesel particulate level of about 1 mg m-3.R e p ~ r t e d l y , ~ ~ reduction of the standard to 0.75 mg m-7 has been recommended. Germany has established carbon-based exposure criteria for diesel particulate matter.4h-47 Technische Richtkoazentrationen (TRK) values, which apply to carcino- gens/mutagens and consider technical and economic aspects, were set at 200 pg of carbon per m-3 of air in non-mining industries and 600 pg m-3 in non-coal mines and other underground worksites. A carbon analysis method involving combustion and coulometric determination of total carbon (as C02) initially was employed,48 but this method was later modified to permit speciation of OC and EC.49 Although an EC exposure standard for diesel particulate has not yet been established in Germany, a standard is expected to be set in 1996. Reportedly,s" EC standards of 0.3 mg m--3 for underground worksites (coal mines excepted) and 0.1 mg m--7 for surface workplaces have been proposed.An additional provision has been proposed in the case of surface worksites where the OC fraction of the total carbon exceeds 50%. In this case, a total carbon standard of 0.15 mg will apply. Conclusion Because animal studies link the carcinogenic effects of exposure to diesel engine exhaust emissions to the particulate fraction of the exhaust, a sampling and analytical method for carbonaceous aerosols was investigated. Of the various methods that have been used for carbon analysis, a thermal-optical technique that speciates different carbon types in a filter sample was examined because this approach offers maximum selectiv- ity relative to other methods.Although different carbon types are quantified by the method, EC is the most specific measure of particulate diesel exhaust. In addition to its selectivity ad- vantage, the thermal-optical method is capable of determination of EC at environmental background levels. The method is both practical and inexpensive (currently about $30 per analysis), making it well suited for routine exposure monitoring and evaluation of control technology for diesel particulate matter. In this study, a variety of samplers were used for the collection of diesel aerosol at an express mail facility and the EC results for different sampler types were not statistically different. Because open-face cassettes are inexpensive, easy to use, require no modification, give homogeneous deposits (typically less than 2% variability across 37 mm filter) and are readily available, these samplers are recommended unless the sampling environment dictates otherwise.In some instances (e.g., when other dust is present at levels that could overload the filter). collection of a respirable dust fraction should be considered. In coal mines, an impactor with a sub-micrometre cut-point is necessary to minimize collection of coal-source EC. Additional details concerning sampling requirements for this particular workplace will appear elsewhere.36 Although an EC-based method (5040) for particulate diesel exhaust has been submitted for inclusion in the NIOSH Manual of Analytical Methods (NMAM) and industrial hygienists both within and outside the Institute have been using the method for occupational monitoring, exposure criteria for particulate diesel exhaust have not yet been established in the USA.A TLV of 150 pg m--3 air has been proposed, but there is much uncertainty in its basis and the value applies to sub-micrometre particulate, not EC. Many interferences can be expected when determining diesel particulate matter by gravimetric means. Interferences also are expected if total carbon is used as a measure of the diesel particulate concentration. As an alternative, the proposed standard for diesel particulate could be expressed in term\ of EC, but to do so requires a knowledge of the EC fraction of the particulate. Because this fraction is variable, an approximation would have to be made.Based on field and laboratory studies, an estimated EC fraction of 50% appears reasonable. A more detailed discussion of this estimate will appear elsewhere.35 Although an estimated value contributes added uncertainty in the standard development process, this contribution is minor relative to the uncertainties in the assumptions made in arriving at the proposed TLV. Environmental background levels of EC can cerve a\ a reference point for determining the extent to which workplace EC concentrations are elevated, but the riskr asrociated with these elevated levels are uncertain. Typically, background levels of EC are about 2 4 pg m-3 air. For comparison. an overall mean EC exposure of 14 pg m was seen in the trucking industry.38 In the mining indu\try, EC concentration\ often exceeded 100 times environmental levels and typically range from about 150 to 500 pg ni-?.In other work places, EC concentrations below 100 pg rn-? were more common. A review of occupational exposure data for EC will appear elsewhere.35 Until health-based exposure criteria are established, refer- ence to epidemiological studies as an indicator of potential health effects may be a prudent approach when considering whether elevated levels of diesel particulate warrant remedial action (e.g., exhaust filters, increased ventilation). To date, the bulk of epidemiological evidence indicates that the increased risk of lung cancer for workers exposed to diesel exhau\t i\ comparable to that for environmental tobacco smoke.In a widely cited retrospective cohort study of diesel-exposed railroad workers,4 a relative lung cancer risk of 1.4 (9S% CI = 1.11, 1.89) was found. Geometric mean exposures to respirable particulate (after correction for cigarette smoke) ranged from 17 pg in- for clerks to 134 pg rn-' for locomotive shop workers. In general, the average levels of diesel exhaust found in most occupational settings (mining industry excepted) are estimated7 to be less than 100 pg m-?. Overall, these findings suggest that occupational exposure to relatively low concentrations (e.g., 100 pg m- ?) of diesel exhaust may pose an elevated risk for lung cancer. At present, actual (i.e., measured) exposure data are lacking and the accuracy o f past human exposure estimates is uncertain.An ongoing epidemiological study [by NIOSH and the National Cancer Institute (NCI)] of miners exposed to diesel exhaust may help resolve some of the uncertainties involved in assessment of the potential carcino- genic risk for humans. Additional research is necessary to addresr the potential adverse health effects of fine particulate matter, especially carbonaceous material bearing trace levels of adsorbed carcino- gens, as does diesel particulate. Such particles have a high affinity for organic compounds and have been shown to increase the long-term retention of known carcinogen\. Al- though it appears that tumour induction in animals may be caused primarily by excessive lung burdens of particulate carbon, it has not been determined whether the mechanisms responsible for tumorigenesis in rats are operative in humans.Regardless of whether potential adverse health effects in humans originate largely from the particles themselves, their adsorbed genotoxins, or from a combination of the two, monitoring and control of the particulate component are necessary if effects exist that are particle-related. Aside from its potential carcinogenicity, it is important to recognize that other principal components of diesel exhaust, including carbon monoxide and oxides of nitrogen and sulfur, also are associated with adverse health effects. Exposure criteria already exist for these gas-phase species, and adoption of an EC standard would not obviate the need to monitor their concentra- tions as well.1190 Analyst, September 1996, Vol.121 Mention of company name or product does not constitute endorsement by the Centers for Disease Control and Prevention. The views expressed in this paper are those of the authors and do not necessarily reflect the views or policies of the National Institute for Occupational Safety and Health. References 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 IARC Monographs on the Evaluation of Carcinogenic Risks to Humans, Vol. 46, Diesel and Gasoline Exhausts and Some Nitroar- enes. International Agency for Research on Cancer, Lyon, 1989. NIOSH Current Intelligence Bulletin No. 50: Carcinogenic Effects of Exposure to Diesel Exhaust, DHHS(NI0SH) Publication No. 88- 1 16, National Institute for Occupational Safety and Health, Cincinnati, OH, 1988.Mauderly, J. L.. Jones, R. K., Griffith, W. C., Henderson, R. F., and McClellan, R. O., Fundam. Appl. Toxicol., 1987, 9, 208. Garshick, E., Schenker, M. B., Mufioz, A., Segal, M., Smith, T. J., Woskie, S. R., Hammond, S. K., and Speizer, F. E., Am Rev. Respir. Dzs., 1988, 137, 820. California Environmental Protection Agency, Health Risk Assess- ment for Diesel Exhaust (Draft), Office of Environmental Health Hazard Assessment, Sacramento, CA, 1994. US Environmental Protection Agency, Health Assessment Document for Diesel Emission, EPA/600/8-90/057A (External Review Draft, 1993), Office of Health and Environmental Assessment, Environ- mental Criteria Assessment Office, Research Triangle Park, NC, 1993. Health Effects Institute, Diesel Exhaust: a Critical Analysis of EmissionJ, Exposure, and Health Effects, Special Report of the Institute’s Diesel Working Group, Health Effects Institute (HEI), Cambridge, MA, 1995.Blade, L. M., and Savery, H., Health Hazard Evaluation Reports, Surveys of Consolidated Freightways, Inc., Peru, IL (HETA No. 88-077-1969) and Pocono Summit, PA (HETA 87-232-1948), National Institute for Occupational Safety and Health, Cincinnati, OH, 1989. Schuetzle, D., Environ. Health Perspect., 1983, 47, 65. Diesel Particulate Emissions: Measurement Techniques, Fuel Effects and Control Technology, PT-42, ed. Johnson, J. H., Baines, T. M., and Clerc, J. C., Society of Automotive Engineers (SAE), Warren- dale, PA, 1992. Levsen, K., Fresenius’ Z. Anal. Chem., 1988, 331, 467. Gangal, M. K., and Dainty, E. D., in Proceedings of the 6th U.S.Mine Ventilation Symposium, ed. Bhaskar, R., Society for Mining, Metallurgy, and Exploration, Littleton, CO, 1993, pp. 83-89. Cornwell, J., and Knutti, E., Health Hazard Evaluation Report, Survey of ASARCO Troy Unit Mine, Troy, MT (HETA No. 88-104-2207), National Institute for Occupational Safety and Health, Cincinnati, OH, 1992. Fowler, D. P., Industrial HygienelEnvironmental Sampling Program to Develop Qualitative and Quantitative Data for Diesel Exhaust Emission Exposure, Coordinating Research Council, Atlanta, GA, Final Report on CRC-APRAC Project No. CAPM-24-78, 1985. McCawley, M., and Cocalis, J., Ann. Am. Conf. Gov. Ind. Hyg., 1986, 14, 271. Burkhart, J. E., McCawley, M. A., and Wheeler, R. W., Am. Ind. Hyg. Assoc. J., 1987, 48(2), 122.Haney, R. A., Diesel Particulate Exposures in Underground Mines AIME 90-40, Society for Mining, Metallurgy, and Exploration, Littleton, CO, 1990. McCartney, T. C., and Cantrell, B. K., in Diesels in Underground Mines: Measurement and Control of Particulate Emissions (lnforma- tion Circular 9324), Proceedings ofthe Bureau ofMines Information and Technology Transfer Seminar, Minneapolis, MN, September 29-30, 1992, US Bureau of Mines, Washington, DC, pp. 24-30. Dolan, D. F., Kittelson, D. B., and Whitby, K. T., Am. Snc. Mech. Eng. Pap., 75-WAIAPC-5, 1975. Vuk, C. T., Jones, M. A., and Johnson, J. H., SAE (Soc. Automot. Eng.) Tech. Pap., 760131, 1976. Groblicki, W. H., and Begeman, C. A., SAE (SOC. Automot. Eng.) Tech. Pap., 790421, 1979. Lipkea, W. H., Johnson, J.H., Vuk, C. T., SAE (Soc. Automot. Eng.) Tech. Pap., 780108, 1979. 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 Amann, C. A., and Siegla, D. C., Aerosol Sci. Technol., 1982, 1, 73. Kittelson, D. B., Kadue, P. A., Scherrer, H. C. , and Lovrien, R. E., Characterization of Diesel Particles in the Atmosphere, CRC AP- 1 Project Group Final Report, Coordinating Research Council, Atlanta, GA, 1988. Baumgard, K. A., and Johnson, J. H., SAE (SOC. Automot. Eng.) Tech. Pap., 920566, 1992. Johnson, J. H., Bagley, S. T., Gratz, L. D., and Leddy, D. G., SAE (Soc. Automot. Eng.) Tech. Pap., 940233, 1994. Birch, M. E., presented at the 203rd Meeting of the American Chemical Society, San Francisco, CA, April 5-10, 1992. Mauderly, J.L., Wolf, R. K., Bond, J. A., Harkema, J. R., Henderson, R. F., and McClellan, R. O., Am. Rev. Resp. Dis., 1988, 137(4), 404. Mauderly, J. L., teleconference communication, 1992. Mauderly, J. L., Snipes, M. B., Barr, E. B., Belinsky, S. A., Bond, J. A., Brooks, A. L., Chang, L.-Y., Cheng, Y. S., Gillett, N. A., Griffith, W. C., Henderson, R. F., Mitchell, C. E., Nikula, K. J., and Thomassen, D. G., in Pulnionary Toxicity of Inhaled Diesel Exhaust and Carbon Black in Chronically Exposed Rats, Part I , Neoplastic and Nonneoplastic Lung Lesions, Research Report No. 68, Health Effects Institute, Cambridge, MA, 1994. Countess, R. J., Aerosol Sci. Technol., 1990, 12(1), 114. Cadle, S. H., and Groblicki, J., paper presented at the Symposium on Particulate Carbon: Atmospheric Life Cycle, Warren, MI, October, 1980, General Motors Report GMR-3452, ENV No. 86, General Motors, Warren, MI, 1980. Johnson, R. I., Jitendra, J. S., Cary, R. A., and Huntzicker, J. J., ACS Symp. Series, 1981, No. 167223. Carsey, T. P., Am. Ind. Hyg. Assoc. J., 1987, 48(8), 710. Birch, M. E., and Cary, R. A., Aerosol Sci. Technol., in the press. Birch, M. E., and Stanevich, R., in preparation. Eller, P. M., in Guidelines for Air Sampling and Analytical Method Development and Evaluation, NIOSH Technical Report, DHHS (NIOSH) Publication No. 95- 1 17, National Institute for Occupational Safety and Health, Cincinnati, OH, 1995, pp. 65-68. Zaebst, D. D., Clapp, D. E., Blade, L. M., Marlow, D. A., Steenland, K., Hornung, R. W., Scheutzle, D., and Butler, J., Am. Znd. Hyg. Assoc. J., 1991, 52(12), 529. Zweidinger, R. B., in Toxicological Effects of Emissions from Diesel Engines, ed. Lewtas, J., Elsevier, Amsterdam, 1982, pp. 83-96. Mark, D., and Vincent, J. H., Ann. Occup. Hyg., 1986, 30, 89. Stanevich, R., and Birch, M. E., paper presented at the American Industrial Hygiene Conference and Exposition, Salt Lake City, UT, May 20-24, 1991. Cohen, S. M., and Ellwein, L. B., Chem. Res. Toxicol., 1992, 5 , 742. Mauderly, J. L., in Environmental Toxicantsquman Exposures and Their Health Effects, ed. Lippmann, M., Van Nostrand Reinhold, New York, 1992, ch. 5. Dainty, D., and Gangal, M., Mining Research Laboratories (MRL) Report 91 -015, Energy Mines and Resources Canada, Canadian Center for Mineral and Energy Technology (CANMET), Ottawa, Ontario, 1991. Watts, W. F., in Diesel Exhaust: a Critical Analysis of Emissions, Exposure, and Health Effects, Special Report of the Institute’s Diesel Working Group, Health Effects Institute (HEI), Cambridge, MA, Lehmann, E., Bundesanstalt fur Arbeitsschutz, Dortmund, personal communication, 1993. TRK-Wert fur Dieselmotoremissionen (ZH 1/120.44), Bunde- sarbeitsblatt 6/1992, pp. 53-55; TRGS 910: Federal Ministry of Labour and Social Affairs. Lehmann, E., Rentel, K.-H., Allescher, W., and Hohmann, R., in Measurement of Workplace Exposure to Diesel Exhaust, Schriftenr. Bundesanst. Arbeitsschutz, Gefaehrliche Arbeitsst. GA 33, 1989 (in German). Auffarth, J., Bundesanstalt fur Arbeitsschutz, Dortmund, letter to E. R. Kennedy, NIOSH, Cincinnati, OH, November 13, 1991 [see also 2nd (1991) edition of Lehmann et al.]. Dahmann, D., personal communication, 1996. 1995, pp. 109-123. Paper 6/01 5Y6D Received March 6, 1996 Accepted June 27, I996
ISSN:0003-2654
DOI:10.1039/AN9962101183
出版商:RSC
年代:1996
数据来源: RSC
|
18. |
Comparative study of an inhalable and a total dust sampler for personal sampling of dust and polycyclic aromatic hydrocarbons in the gas and particulate phase |
|
Analyst,
Volume 121,
Issue 9,
1996,
Page 1191-1196
Hilde Notø,
Preview
|
PDF (1232KB)
|
|
摘要:
Analyst, September 1996, Vol. 121 ( I 191-1196) 1191 Comparative Study of an lnhalable and a Total Dust Sampler for Personal Sampling of Dust and Polycyclic Aromatic Hydrocarbons in the Gas and Particulate Phase* Hilde Not@, Kristin Halgard, Hanne Line Daae, Ragne K. Bentsen and Wijnand Eduard National Institute of O~*cupational Health, PO Box 81 49 Dep, N-0033 Oslo, No m a y For many years the closed-face plastic cassette has been widely used for monitoring 'total' dust in working atmospheres. In the late 1980s a different personal sampler was developed at the Institute of Occupational Medicine in Edinburgh. This so-called IOM sampler was designed to meet the criteria for inhalable dust (EN 481). In this work, a comparison of the closed-face 25 mm Gelman total dust sampler and the TOM inhalable sampler was made for exposure measurements of dust and PAHs among workers in an electrode paste plant.The two samplers were modified in order to permit sampling of both particulate and gas-phase PAHs. Three groups with different tasks were examined. The particle size distribution was determined using personal inhalable dust spectrometers and the GRIMM particle counter. The results showed that higher concentrations of dust and particulate PAHs were measured with the TOM inhalable sampler than the Gelman total dust sampler and the difference between the samplers was significant. This is in agreement with previously reported studies. In some samples collected with the IOM sampler very large particles were observed. There was no significant difference between volatile PAHs measured with the two samplers.For semi-volatile PAHs a significant difference between the samplers was observed, which was ascribed to evaporation loss from the filters during storage. Keywords: Personal samplers; dust; aerosols; PAHs; workplace monitoring Introduction For many years personal air sampling using different types of samplers has been used to monitor the exposure of workers to aerosols for risk assessment purposes. These samplers were assumed to collect all airborne particles, so-called 'total' dust. However, many of them have shown lower collection efficien- cies than relevant for health-related aerosol sampling.' The collection efficiency of these total dust samplers decreases with increase in aerodynamic diameter, especially for particles > 10 ym.2.3 A different personal sampler was designed at the Institute of Occupational Medicine (IOMj in Edinburgh, UK,4 in order to meet the criteria for inhalable dust recently adopted by the European Committee for Standardization (CEN).S It has been tested in wind tunnel experiments and has been shown to approximate the inhalable convention more closely than other devices. It is therefore referred to as an inhalable sampler.At the * Presented at AIRMON '96. Salen, Sweden, February 5-8, 1996. National Institute of Occupational Health (NIOH) in Oslo, Norway, the traditional samplers for collecting aerosols in workplace air have been 37 and 25 mm closed-face filter cassettes. In recent years, cassettes made of conducting graphite-filled polypropylene were used to minimize possible effects from static electric charge on the sampling efficiency.Only a few field studies with the purpose of comparing inhalable samplers with total dust samplers have been reported.6 These studies showed that higher concentrations of dust were measured with the inhalable sampler. Similar results have been found at our institute.7 This work was conducted in an electrode paste plant to study the difference between the IOM inhalable sampler and the 25 mm Gelman total dust sampler for sampling of dust and PAH compounds. A wide range of PAH compounds from volatiles to airborne particulates have been found in work environments. Several of the particulate PAH compounds are classified as being carcinogenic to humans, but they are usually present at low concentrations in workplace air.Pyrene, a non-carcino- genic semi-volatile PAH compound, is often found at higher concentrations, and its metabolite 1 -hydroxypyrene can be detected in the urine of PAH-exposed workers.8 1-Hydroxy- pyrene has therefore been used as a biomarker for exposure to PAHs,9--" but the relationship is not very strong. As the measurement of volatile, semi-volatile and particulate PAH compounds includes sampling of both the particulate and the gas phase, both samplers were adapted for the simultaneous collection of these phases.12.13 The particle size distribution was measured because the performance of the aerosol samplers also will depend on the size of the sampled aeros01.l~ Experimental Sampling Equipment Personal dust samplers A 25 mm black graphite-filled polystyrene closed-face Gelman total dust sampler (Gelman Sciences, Ann Arbor, MI, USA) and an IOM inhalable sampler made of aluminium (No.701, Rotheroe and Mitchell, Aylesbury, UK) were used (Fig. 1j. Both samplers were equipped with an acrylic copolymer membrane filter with pore size 0.8 ym (Versapore 800, Gelman Sciences). The empty spaces behind the filters were completely filled with XAD-2 (SKC, Blandford Forum, UK) to avoid the formation of air channels through the adsorbents. The mass of the adsorbent was approximately 0.4 and 1.0 g in the Gelman total dust sampler and the IOM inhalable sampler, respectively. Stationary measurements using tubes (SKC Cat. No. 226-30-06) with two seperated layers of XAD-2, a main layer of 0.4 g and a backup layer of 0.2 g, showed a breakthrough of less than 1% measured in a very high exposed area.The samplingwas performed with Casella AFC 123 personal pumps (Casella, London, UK) operated at 2 I min-I. Pasticle size c.harac'tesi=atiOn The particle size distribution was determined using personal inhalable dust spectrometers (PIDS) developed at the IOM. I s The clean stages and top of the PIDS were sprayed with silicone oil and left overnight for evaporation of solvents. Special covers were used to avoid spray in and around the impactor holes. The stages were weighed. mounted and monitored for leakage. A flow rate of 2 I niin was maintained with personal pumps (PS 101. NIOH) for 24 h. A portable GRIMM particle counter (GRIMM Labortechnik, Ainring, Germany) equipped with a radial sampling head for air sampling at 1.25 m s-' and a 256 kb storage card was also used.Collection of Samples Dust aixl PAHJ The two personal samplers were mounted side by side near the breathing zone of the worker. The inlet of the Gelman total dust sampler faced downwards whereas that of the IOM inhalable sampler faced outwards. The flow rate was 2 1 min-1 and the sampling period was 7-8 h. After sampling, all the filters and samplers were brought back to the laboratory in a horizontal position. The samples were collected in a plant producing electrodes for use in the metal production industry. The production process can be divided into different steps such as grinding of calcined anthracite in a ball-mill, mixing of electrode paste from calcined ground anthracite and coal tar (elevated temperature), filling of moulds with electrode paste and transport of moulds.Operators with three different tasks were examined: Fig. 1 Schematic diagram\ of (upper) the IOM inhalable sampler and (lower) the Gelman total dust sampler equipped with adsorbent behind the filter. Mixers spent most of their time sitting in a control room supervising the mixing and grinding processes. Every 15th minute they left the room to take samples of the paste to check the plasticity. Twice a shift they also took samples from the ball mill. Mould,filless were responsible for filling of warm electrode paste into the moulds. Their activity and exposure to dust and PAHs were strongly dependent on the size and shape of the moulds.Tsuck dsivcss transported the moulds to and from the filling point. They also took the cooled electrodes out of the moulds. Pai.ticle sizr nieusurewients Like the TOM samplers, the PTDS are constructed to collect the inhalable aerosol fraction. The PlDS were mounted in the same way as the inhalable sampler and with the same flow rate of 2 1 min-1. As low exposure was expected for two of the tasks, each PIDS was carried for three subsequent shifts. The GRIMM personal particle counter was used for real-time logging of the particle size distributions for periods of 5-30 min. Short-time exposure within each task was measured. Analytical Procedures Gru linietric analysis For practical reasons, the adsorbent from the IOM inhalable samplers was transferred to screw-capped glass vials and wrapped in aluminium foil immediately after sampling.The inner cassettes with filters were stored in sealed 37 mm Millipore cassettes. The filters and adsorbent were stored together in the capped Gelman samplers. All samples were stored at 4 "C until analysis. The Gelman filters and the 10M inner cassettes with filters were conditioned at room tempera- ture and low humidity overnight and then the following night at room temperature with a relative humidity of 40-5096 before weighing. Pcrsticle size charactesization After exposure, the stages in the PIDS were acclimatised at room temperature overnight before weighing. Fine dust was estimated by the combined mass on stages 6, 7 and 8 and the backup filter, medium-sized dust by stages 4 and 5 , and coarse dust by stages 1, 2 and 3 and the entry.These fractions had particle skes <4.8, 4.8-10.6 and > 10.6 pm, respectively.IS The same size fractions were computed from the GRIMM particle counter with the software provided and interpolation within size channels. PAH deterininatioir Chmicals and instrumentation. Dichloromethane, cyclohex- ane, N,N-dimethylformamide (DMF) and sodium sulfate were obtained from Merck (Darmstadt, Germany). The PAH refer- ence compounds used are listed in Table 1. Samples were analysed with an HP 5890 Series I1 gas chromatograph (Hewlett-Packard, Wilmington, DE, USA) with a flame ioniza- tion detector (FID) and a CP-SiI 8 CB fused-silica column (Chrompack, Raritan, NJ, USA, Cat. No. 745 1, 25 m >( 0.25 mm id, df = 0.25 pm).Injections were made in the splitless mode with a 2 min splitless time. Helium was used as the carrier gas at a flow rate of 1 ml min I . The injector and FID temperatures were 300 "C. The oven temperature was pro- grammed as follows: held at 35 "C for 2 min, increawd from 35 to 150 "C at 6 "C min-I and then from 150 to 310 "C at 10 "C min-1 and held at 3 10 "C for 15 min. For identification of PAH compounds, an HP 5890 Series I1 gas chromatograph connected to an HP 5971 mass-selectiveAnalyst, September 1996, Vol. 121 1193 detector (MSD) was used. The gas chromatograph was equipped with an HP-5 column (Hewlett-Packard, Part No. 190915-433, 30 m X 0.25 mm id, d f = 0.25 pm). The injector and MS interface temperatures were 300 "C.The temperature programme was as described above. The multiplier voltage was 70 eV. The analyses were performed in the scan mode and the mass range was 40-350 u. Saniple preparation and unalysis. 3,6-Dime th ylphenanthrene was added as an internal standard to the samples before preparation. The filters were extracted with cyclohexane in an ultrasonic bath. PAHs and polar compounds were extracted from the cyclohexane solutions into DMF containing 3% of water. The DMF phase was diluted with an equal volume of water and extracted with cyclohexane. The cyclohexane phase was dried with anhydrous sodium sulfate before concentration of the sample at 50 "C under a stream of nitrogen.l6 The adsorbent was desorbed with 2.0 ml of dichloromethane for 30 min. To avoid loss of the volatiles, the desorbed solution was analysed directly without concentration.The desorption efficiencies in dichloromethane for most of the compounds in Table 1 have been studied previously.17 The IOM inner cassettes and the Gelman total dust samplers were washed inside with 2 ml of cyclohexane and the solution was treated in an ultrasonic bath for 15 min. All samples were analysed by GC-FID. The PAH compounds were identified comparing the GC-FID retention times with the retention times of the standards. Total ion chromatograms from GC-MS analysis of selected samples were also used for identification. An internal standard method was used for the quantification of PAH compounds. Statistical Methods The results often had a skewed distributions and were compared using non-parametric methods.Results from pairs of samples collected with the IOM inhalable sampler and the Gelman total dust sampler were compared using the Wilcoxon rank sum test. Associations between the samplers were shown by scatter plots. In addition, the ratio of the inhalable sampler to the total dust sampler was computed for each pair of samples. The ratios were compared with 1 using the Wilcoxon signed rank test and the Table 1 PAH compounds used as standards and in the quantification of total PAHs No. Compound No. Compound 1 2 3 4 5 6 7 8 9 10 11 12 13 14 1s 16 17 18 19 20 21 22 23 Naphthalene 24 2-Methylnaphthalene 25 1 -Methylnaphthalene 26 Acenaphthy lene 28 Acenaphthene 29 Biphenyl 27 Dibenzofuran 30 Fluorene 31 9-Methylfluorene 32 9,lO-Dihydroanthracene 33 2-Methylfluorene 34 1 -Methylfuorene 35 Dibenzothiophene 36 Phenanthrene 37 Anthracene 38 Acridine 39 Carbazole 41 3-Methylphenanthrene 42 Benzo mquinol ine 40 2-Methylphenanthrene 43 2-Methylanthracene 44 4,s-Methylenphenanthrene 45 1 -Methylphenanthrene 9-Meth y lanthracene Fluoranthene Pyrene Benzo [ a] fluorene Benzo[h] fluorene 1 -Methylpyrene Benz[u]anthracene Chrysene Benzo[h]fluoranthene Benzo[k]fluoranthene Benzo[ jlfluoranthene B enzo [ el pyrene Benzolalpyrene Pery lene Indenor 1,2,3-rd]pyrene Dibenz[u,h] anthracene Benzo[ghi] perylene Anthanthrene Coronene Dibenzo[a,e]pyrene Di benzo[a ,il pyrene Dibenzo[a,h]pyrene Wilcoxon rank sum test was used for comparisons of ratios between different groups. Results and Discussion Dust In total, 27 pairs of samples were collected, 11 from mixing, 10 from mould filling and six from truck driving.The results are shown in Figs. 2 and 3. The concentration of airborne dust was higher for the inhalable sampler than the total dust sampler in 26 out of 27 measurements and the difference was significant. Significant differences between the samplers were also found within each task. The ratios of the inhalable sampler to the total dust sampler had a median of 4.4 (range 0.92-64), p < 0.001. The ratios were I .32 (range 0.924.6), p < 0.00 I , among the mixers, 5.7 (range 1.2-64), p < 0.001, among the mould fillers and 6.2 (range 2.1-6.2), p < 0.05, among the truck drivers. Information about particle size was considered important because of its expected influence on sampling efficiency.The Fig. 2 Dust concentrations found with the IOM inhalable sampler and the Gelman total dust sampler among electrode paste plant workers. Results from all measurements together and divided by task are shown. Significance levels of the difference between the samplers by Wilcoxon matched pair test are shown: ***p < 0.001, ** p < 0.01, * p < 0.05 and NS = not significant. All measurements n = 27. mixing n = 11, mould filling n = 10, truck driving IE = 6. The box extents indicate the 25th and 75th percentiles, the capped bars indicate the 10th and 90th percentiles and the symbols mark all data outside the 10th and 90th percentiles. The solid line marks the median and the dotted line displays the arithmetic mean. ._ - 10 "- 0 7 E 81 0 0 77-- T-- 0 2 4 6 8 1 0 ' Gelman/mg m-3 2 Fig.3 Relationship between dust exposure measured with the IOM inhalable sampler and the Gelman total dust sampler. 0, = mixing; 0, = mould filling; and A, = truck driving. The 1 : 1 line (.-.) shows identical samplers.1194 Analyst, September 1996, Vol. 121 results from the two different types of particle size measure- ments are summarized in Table 2. A total of seven measure- ments were made with PIDS, one from mixing, two from mould filling and four from truck driving. Both methods showed that the highest exposed task (mixers) also was exposed to the largest particles and that the mould fillers were exposed to the finest aerosol. The observed particle size could not explain the great difference between the samplers for the mould fillers and truck drivers. Other factors were therefore considered.Very coarse particles collected with the inhalable sampler were observed on some filters from mould fillers and truck drivers. These workers were exposed to low dust levels. In short periods some workers may have been facing strong directional winds. Under such conditions, the IOM inhalable sampler is known to oversample large particles whereas the Gelman total dust sampler actually undersamples compared with the criteria for inhalable dust sampling. The extrathoracic fraction measured with the PIDS was much lacger than with the GRIMM particle counter. This may be due to the different sampling efficiencies of the inlets. The inlet characteristics of the GRIMM particle counter have not been documented.It may also be possible that the GRIMM particle counter measurements did not cover periods with exposure to the largest particles as the duration of these measurements was relatively short. Continuous logging of particle size distribution gave information about relative differences within and between tasks. Short-period high exposure with a special particle size distribution was observed. PAH Results Particulate PAHs The total amounts of particulate PAH compounds found with the samplers are shown in Figs. 4 and 5. The IOM inhalable sampler was designed to measure all particles that entered the inlet and not only those collected on the filter. The PAHs found on the inner cassette wall (wash solution) were therefore combined with the PAHs on the filter.Particulate PAHs for the Gelman total dust sampler consisted of PAHs from the filter only, whereas only trace amounts of PAH compounds were found on its walls. The concentration of particulate PAHs was significantly higher for the inhalable sampler than to the total dust sampler calculated from all measurements. Significant differences between the samplers were also found within the mould fillers and truck drivers. The ratios of the inhalable sampler to the total dust sampler had a median of 2.15 (range 0.30-13), p < 0.001. The ratios were 1.35 (range 0.30-4.5), p = 0.06, among the mixers, 2.2 (range 1.2-13), p < 0.001, among the mould fillers and 2.9 (range 1.7-5.3), p < 0.001, among the truck drivers. Table 2 Particle size distributions of workers with different tasks estimated with PIDS over the whole work shift and short-term measurement with the GRIMM particle counter Size fraction (%) Sampler Task PIDS Mixing, n = 1 Mould filling, n = 2 Truck driving, n = 4 Mixing, normal exposure Mould filling, in mist Mould filling, normal Truck driving GRIMM Mixing, high exposure < 4.8 ELm 17 5-12 16-2 1 23 67 75 57 62 4.8- 10.6 ELm 15 60-80 32-48 30 14 24 34 19 > 10.6 w 68 8-35 3 1-52 47 19 1 9 19 Volatile PAHs The proportion of all PAH compounds found in the adsorbent was 93% (s 2%) of the total amount (PAHs on filter + adsorbent) for the inhalable sampler and 97% (s 6%) for the total dust sampler.The differences between the two samplers, for all measurements and divided by task, were small and never significant (Figs.6 and 7). The ratios of the inhalable sampler to the total dust sampler had a median of 1.2 1 (range 0.36-1.75), p = 0.09. The ratios were 1.08 (range 0.36-1.47), not significant (NS), among the mixers, 1.20 (range 0.44-1.75), NS, among the mould fillers and 1.44 (range 0.88-1.58), p < 0.05, among the truck drivers. This was expected? as the collection efficiency of PAHs in gas phase is not likely to depend on the geometry of the sampling inlet. Semi-volatile PAHs The proportion of PAH compounds found in gas phase was computed for a subset of the samples from all tasks. The results showed that PAH compounds with low boiling-points were found only in the gas phase whereas compounds with high Fig. 4 Concentrations of particulate PAHs found with the 10M inhalable sampler and the Gelman total dust sampler among electrode paste plant workers.Results from all measurements together and divided by task are shown. All measurements n = 26, mixing n = 10, mould filling n = 10 and truck driving n = 6. For further descriptions, see Fig. 2. 120 I ;::I 0 90 5 GeIman/pg m-3 Fig. 5 Relationship between particulate PAH exposure measured with the IOM inhalable sampler and the Gelman total dust sampler. 0. mixing; 0, mould filling; and A, truck driving. The 1 : I line (-..) shows identical samplers.Analyst, September 1996, Vol. 121 1195 boiling-points were found solely in the particulate phase. PAHs with intermediate boiling-points were found in both phases (Fig. 8). This is well known,'* but the present results also Fig.6 Concentrations of PAH compounds found on the adsorbent in the IOM inhalable sampler and the Gelman total dust sampler among electrode paste plant workers. Results from all measurements together and divided by task are shown. All measurements n = 26, mixing n = 10, mould filling n = 10 and truck driving n = 6. For further descriptions, see Fig. 2. 2500 2000 PI 1500 0, =L \ 2 1000 G 500 0 0 0 0 0 0 0 0 Q o n A o 0 & "0" Fig. 7 Relationship between PAH compounds found in the adsorbents of the IOM inhalable sampler and the Gelman total dust sampler. 0, mixing; 0, mould filling; and A, truck driving. The 1 : 1 line (...) shows identical samplers. 8o i a 401 h 601 m A Gelman 20 4 0 I I I , I I I I ~ ~ ~ I T 1 6 7 8 1 1 13 14 15 19 23 25 26 27 29 31 32 35 36 37 38 39 40 42 45 Compound Fig.8 Partitioning of PAH compounds between filter and adsorbent in the IOM inhalable sampler and the Gelman total dust sampler. PAH compounds are arranged by increasing boiling point. Arithmetic mean and standard error from 16 pairs of measurements from all tasks are shown. Differences between the samplers are significant for the semi-volatile PAH com- pounds. showed significant differences between the samplers for several semi-volatile PAH compounds. The Gelman total dust sampler showed higher concentrations of the semi-volatile compounds found on the adsorbent than the IOM inhalable sampler. One explanation could be the different method of construction. Another cause could be differences in sample handling. For practical reasons, the adsorbent in the inhalable sampler was separated from the filter immediately after exposure. In the Gelman total dust sampler, the filter and adsorbent were stored together until analysis.Semi-volatile compounds for this sampler may therefore have diffused from the filter to the adsorbent. As the time for such diffusion to take place was limited to a few minutes for the samples collected with the IOM inhalable sampler, semi-volatiles may have been lost from the filter during storage in the Millipore cassettes. Pyrene The difference between the samplers for the semi-volatile PAH compound pyrene was further studied. Pyrene is generally found both as a gas and on particles in working athmospheres.' Better estimates of 'real' workplace exposure to particulate and gas-phase pyrene may eventually improve the relationship between PAH exposure and urinary 1 -hydroxypyrene excretion, provided that other exposure routes such as ingestion and absorption through the skin and exposure outside the workplace are not important. The proportion of pyrene in the gas phase compared with the particulate pyrene was found to be 75% (s 12%) for the total dust sampler and 50% (s 24%) for the inhalable sampler (Fig.8). Although a larger amount of particulate pyrene was found with the inhalable sampler than the total dust sampler, the total dust sampler showed higher values for the pyrene in the gas phase (Fig. 9). These results indicate that the measurement of pyrene and other semi-volatile PAH compounds in air without the use of an adsorbent to include the gas phase is incomplete, and that special care has to be taken to avoid loss from the filter after sampling of semi- volatile PAH compounds.One solution would be to store filters with the adsorbent for sufficient time to allow the 'mobile' pyrene to migrate to the adsorbent. Further studies are needed to confirm these results and to optimize sample handling. Fig. 9 Concentrations of pyrene found on the filter and adsorbent in the IOM inhalable sampler and the Gelman total dust sampler among electrode paste plant workers. Results from all measurements are shown. n = 26 for the IOM and Gelman samplers for both filter and adsorbent. For further descriptions, see Fig. 2.1196 Analyst, September 1996, Vol. 121 Conclusions The IOM inhalable sampler measured higher concentrations of dust and particulate PAHs than the Gelman total dust sampler for electrode paste plant workers.For volatile PAH compounds the two samplers gave similar results. Large particles could be the cause of unexpected high mass ratios between the samplers. It is likely that semi-volatile PAH compounds will be lost from the particulate phase if filters are stored separated from the adsorbent before analysis. This work was supported by the Confederation of Norwegian Industries (NHO), National Fund for Occupational Health, grant number 0728. We thank T. Aaen, B. Faaness and all the workers at Elkem Carbon, Fiskaa, Norway, who participated in this study. We also greatly appreciate the technical assistance of 0. Synnes and T.Nilsen and useful advice and encouragement from S. Hetland, Y. Thomassen and S. 0vrebg at the National Institute of Occupational Health, Oslo, Norway. References I 2 3 4 5 Tsai, P. J., and Vincent, J. H., J . Aerosol Sci., 1993, 27, 919. Buchan, R. M., Soderholm, S. C., and Tillery, M. I., Am. Znd. Hyg. Assoc. J., 1986, 47, 825. Vincent, J. H., Aerosol Sumpling Science and Pructice, Wiley, New York, 1989. Mark, D., and Vincent, J. H., Ann. Occup. Hyg., 1986, 30, 89. European Committee for Standardization (CEN) Workplace Atmos- pheres-Size Fraction Definitions for Measurement of Airborne Particles. Report EN 48 1, CEN, Brussels, 1993. Ref. EN 48 1 : 1993 E, 1993. 6 7 8 9 10 11 12 13 14 15 16 17 18 Vincent, J. H., Sci. Total Environ., 1995, 163, 3. Hetland, S., and Thomassen, Y., Pure Appl. Chem., 1993, 65, 2417. Hansen, A. M., Poulsen, 0. M., and Christensen, J. M., Znt. Arch. Occup. Environ, Heulth, 1991, 63, 247. Tjoe Ny, E., Heederik, D., Kromhout, H., and Jongeneelen, F., Am. Znd. Hyg. Assoc. J., 1993, 54, 277. Buchet, J. P., Gennart, J. P., Mercado-Calderon, F., Delavignette, J. P., Cupers, L., and Lauwreys, R., Br. J. Znd. Med., 1992, 49, 761. Jongeneelen, F. J., van Leeuwen, F. L., Oosterink, S., Anzion, R. B. M., van der Loop, F., Bos, R. P., and van Veen, H. G., Br. J . Znd. Med., 1990, 47, 454. Brandt, H. C. A., De Groot, P. C., Molyneux, M. K. B., and Tindle, P. E., Ann. Occup. Hyg., 1985, 29, 27. Anderson, K., Levin, J. O., and Nilseon, C. A,, Chemosphere, 1983, 12, 197. Vinzents, P. S., Thomassen, Y., and Hetland, S., Ann. Occup. Hyg., 1995, 39, 795. Gibson, H., Vincent, J. H., and Mark, D., Ann. Occup. Hyg., 1987,31, 463. Bjorseth, A., Bjgrseth, O., and Fjeldstad, P. E., Scand. J . Work Environ. Health, 1978, 4, 212. Daae, H. L., and Not@, H., 44. Nordiska Arbetsmiljiimiitet (44th Nordic Meeting in Work Environment), NAdendal, Finland, 1995, Abstract (in Norwegian). Cautreels, W., andvan Cauwenberghe, K., Atmos. Environ., 1978,12, 1133. Paper 6102220K Received March 29, 1996 Accepted June 17, 1996
ISSN:0003-2654
DOI:10.1039/AN9962101191
出版商:RSC
年代:1996
数据来源: RSC
|
19. |
Measurement methods and strategies for non-infectious microbial components in bioaerosols at the workplace |
|
Analyst,
Volume 121,
Issue 9,
1996,
Page 1197-1201
Wijnand Eduard,
Preview
|
PDF (812KB)
|
|
摘要:
Analyst, September 1996, Vol. 121 (1197-1201) 1197 Measurement Methods and Strategies for Non-infectious Microbial Components in Bioaerosols at the Workplace* Wijnand Eduard National Institute of Occupational Health, P.O. Box 8149 Dep., N-0033 Oslo, Norway Exposure to micro-organisms can be measured by different methods. Traditionally, viable methods and light microscopy have been used for detection of micro-organisms. Most viable methods measure micro-organisms that are able to grow in culture, and these methods are also common for the identification of micro-organisms. More recently, non-viable methods have been developed for the measurement of bioaerosol components originating from micro-organisms that are based on microscopic techniques, bioassays, immunoassays and chemical methods.These methods are important for the assessment of exposure to bioaerosols in work environments as non-infectious micro-organisms and microbial components may cause allergic and toxic reactions independent of viability. It is not clear to what extent micro-organisms should be identified because exposure-response data are limited and many different micro-organisms and microbial components may cause similar health effects. Viable methods have also been used in indoor environments for the detection of specific organisms as markers of indoor growth of micro-organisms. At present, the validity of measurement methods can only be assessed by comparative laboratory and field studies because standard materials of microbial bioaerosol components are not available. Systematic errors may occur especially when results obtained by different methods are compared.Differences between laboratories that use the same methods may also occur as quality assurance schemes of analytical methods for bioaerosol components do not exist. Measurement methods may also have poor precision, especially the viable methods. It therefore seems difficult to meet the criteria for accuracy of measurement methods of workplace exposure that have recently been adopted by the CEN. Risk assessment is limited by the lack of generally accepted reference values or guidelines for microbial bioaerosol components. The cost of measurements of exposure to microbial bioaerosol components may be high owing to expensive analyses and highly variable exposure levels.The use of qualitative indicators of microbial growth, recording of health effects, specific immunoglobulin G antibody levels to prevalent species in serum of exposed workers and stratified sampling may help to reduce the costs of exposure assessment. An example of a combined strategy for assessment of health risks from handling mouldy timber is shown. Keywords: Bioaerosol components; exposure; work environment; measurement methods; risk assessment; strategy; micro-organisms * Presented at AIRMON ’96, Salen, Sweden, February 5-8, 1996. Introduction The first volumetric measurements of micro-organisms in air were published by Pasteur’ in 1861 in the study on the doctrine of the spontaneous generation of life. He filtered urban air of Paris through a depth filter of gun cotton, dissolved the filter in a mixture of ethanol and diethyl ether and found several thousand ‘corpuscules organis&’ per litre of air by counting with a light microscope.About a century later, Gregory and Lacey2 found that spores from actinomycetes could be liberated in large numbers from mouldy hay, also using a light microscope. Previously, many different fungi had been cultured from mouldy hays from farms of ‘farmers lung’ patients in search of the cause of this form of allergic alveolitis, but exposure challenge of patients with extracts from these micro- organisms failed to provoke attacks of the disease. Inhalation of spores from actinomycetes, especially Saccharopolyspora rec- tivirgula, (previously called Micropolyspora faeni and Faeni rectivirgula) was subsequently shown to be the most important cause of ‘farmer’s lung’ disease3 in the UK.These studies illustrate the important role that microscopic techniques have played in aerobiology. Most microscopic methods detect micro-organisms independently of their ability to grow and belong to the so-called non-viable methods. Cells with metabolic activity can also be counted with the fluores- cence microscope after fluorescent ~taining,~ but this method does not seem to have been used for exposure measurements in the work environment. In most studies, micro-organisms have been detected by culture techniques. In the 1980s, other non- viable methods were developed based on microscopic tech- niques, bioassays, immunoassays and chemical methods.5Jj These methods are important developments because allergic and toxic reactions from exposure to bioaerosols probably do not depend on the viability of the micro-organisms as exposure challenge of patients with allergic alveolitis and asthma with antigens and allergens extracted from micro-organisms may cause such reactions.7.8 Risk assessment in working populations exposed to non- infectious bioaerosols such as workers in sawmills, waste water treatment plants, cotton factories, tobacco factories and farmers is hampered by the lack of internationally recognized occupa- tional exposure limits.The establishment of such limits requires more data on exposure-response relationships of bioaerosol components in working populations. The aim of this paper is to give an overview of measurement methods for microbial bioaerosol components and some of their properties which may serve as an aid for selecting methods in future studies and to make suggestions on strategies for risk assessment.Bioaerosol Components and Properties The micro-organisms are a large group of organisms that include bacteria, fungi, viruses, amoebae and algae. In the work1198 Analyst, September 1996, Vol. 121 environment, bacteria and fungi have been studied extensively compared with the other taxonomic groups. The bacteria can be divided into the Gram-negative and Gram-positive bacteria, depending on their ability to retain the Gram stain. This property coincides with differences in the structure of the cell wall. An important sub-group of the Gram-positive bacteria is the actinomycetes, which resemble the fungi both morpho- logically and by production of large numbers of airborne spores.It is not clear if non-infectious micro-organisms should be identified at the species level as many different micro- organisms and microbial components may cause similar health effects.9 Results from a few animal studies suggest, however, that there may be differences between exposure-response relationships of different species. 10,11 Further, several fungal species produce mycotoxins and Gram-negative bacteria con- tain endotoxins with species-dependent toxicity. The role of the viability of non-infectious micro-organisms is also unclear as no studies seem to have been published on the influence of viability on health effects from exposure to non-infectious micro- organisms.It may be possible that viable micro-organisms produce stronger effects from the production of allergens or toxins after deposition in the lung. Further studies are needed to clarify these points. Components of microbial cells, also called primary meta- bolites, may serve as markers of micro-organisms in bioassays, immunoassays and chemical and molecular biological methods. Material from dead and disintegrated micro-organisms may also be assessed by these methods, which may not be possible by microscopic and viable methods. Some primary metabolites have inflammatory properties, e.g., endotoxins that are present in the cell walls of Gram-negative bacteria and glucans that can be found in the cell walls of many fungi.12 Secondary metabolites are compounds that may be excreted by micro- organisms into the environment, e.g., mycotoxins and enzymes, and may be found in particles from colonized materials as well as in the micro-organism itself.Mycotoxins can have high toxicity, but their contribution to health effects from bioaerosol exposure needs further clarification. 1 3 3 1 4 tion of the collected micro-organisms may also be possible if characteristic morphological details can be observed. The greater resolution of SEM allows the observation of finer surface details than with light microscopy, which is an advantage for the small-sized spores from the fungal genera Aspergillus and Penicillium and the actinomycetes. Micro- organisms that have been stained with a fluorochrome can be counted with the fluorescence microscope.The fluorescent stain facilitates the recognition of micro-organisms in the presence of other particles and in complex aggregates. Bacterial cells can also be counted, which is an advantage compared with the other microscopic methods as bacterial cells often lack characteristic morphological features. The Limulus amoebocyte lysate assay (LAL) for measure- ment of endotoxins is the most commonly used bioassay in bioaerosol analysis. Several modifications of this method have been developed in order to increase sensitivity and specificity. The filter material, the composition of the extraction fluid, storage conditions and the presence of other bioaerosol components, including glucans, have been shown to influence the results.16J7 At present there is no general agreement on a standard procedure. The LAL-based methods assess the bio- logical activity of the toxins rather than the amount.As endotoxins may have different toxicities, the results are compared with the activity of standard endotoxin, usually from Escherichia coli, and expressed as endotoxin units (em.) or as E. coli endotoxin equivalents in micrograms. A bioassay for glucans based on the LAL method has also been described.18 Immunoassays have been described for antigens and aller- gens, e.g., ELISA and the radio allergo-sorbent test (RAST). These methods are suitable for the measurement of specific antigens and allergens. The specificity of these methods may be a disadvantage in studies of non-allergic reactions when the bioaerosol contains components from many different species.Chemical methods include GC-MS of chemical markers such as 3-hydroxy-fatty acids for endotoxins, muramic acid for bacteria and ergosterol for fungi.6 Mycotoxins can be measured by chromatographic technique^'^ but exposure to airborne Measurement Methods Bioaerosols may contain different components originating from micro-organisms: single cells and spores, aggregates of micro- organisms, aggregates with other particles, fragments and primary and secondary metabolites. Micro-organisms may also have different biological properties such as species and viability. These components and properties can be assessed by a variety of viable and non-viable methods (Table 1). Measurement methods for bioaerosols have been thoroughly described in recent publicationss-1s and are briefly summarized below. Several viable methods are based on impaction on nutrient agar plates, e.g.. the slit sampler, which is a single-stage impactor, and the multi-stage Andersen sampler.The agar plates are cultured directly and viable micro-organisms that grow into colonies are counted. Impingers can be used for the collection of micro-organisms in a fluid such as the AGI-30 all- glass impinger. Sub-samples of the fluid sample can be cultured on nutrient plates, after dilution, so that larger numbers of micro-organisms can be sampled than with impactors. Different groups of micro-organisms may also be detected by culture of sub-samples on various media and under different conditions. Filter samples may be re-suspended and the suspension cultured in the same way.Most non-viable methods are based on filter sampling. Spores from fungi and actinomycetes may be recognized by their morphology with light microscopy and SEM. Classifica- Table 1 Microbial bioaerosol components and properties that can be measured by different methods Method Light microscopy Scanning electron microscopy (SEM) Fluorescence microscopy Limulus amoebocyte lysate assay (LAL) Enzyme-linked immuno- sorbent assay (ELISA) Radio allergo-sorbent test (RAST) Gas chromatography-mass spectrometry (GC-MS) Chromatogaphy Culture Bioaerosol component/ biological property Micro-organisms, especially fungal Some potential for classification Micro-organisms, especially spores from fungi and actinomycetes Some potential for classification Micro-organisms, spores and cells, also in complex aggregates with other particles Endotoxins, glucans spores Antigens, allergens Markers at species/genus level Chemical markers mainly at group My cot oxins Viable micro-organisms Identification by morphology of level colonies and vegetative structures and growth on selective mediaAnalyst, September 1996, Vol.I21 1199 mycotoxin has mainly been evaluated by identification of toxicogenic species using viable methods. Specific DNA markers can be detected with the polymerase chain reaction (PCR) technique. This technique is very sensitive and may be used for the assessment of low levels of pathogenic micro- organisms and genetically altered micro-organisms. Measure- ments of chemical and molecular biological markers in bioaerosols in the work environment are still in their infancy, but may have potential for the assessment of exposure to microbial components from broad groups and at the species level, respectively.The identification of micro-organisms is traditionally per- formed by means of the morphological characters of colonies, cells, mycelium, fruiting bodies and spores and by growth on selective media. Non-viable methods may also be used, especially when viable organisms cannot be cultured on nutrient agar plates, e.g., microscopy of spores with specific morphol- ogy, fluorescence microscopy of micro-organisms stained with specific antibodies conjugated with a fluorescent marker, ELISA of specific antigens and PCR of specific DNA markers.Measurement Errors The accuracy of sampling and analysis by viable and micro- scopic methods has been reviewed recently.20 At present, systematic errors are difficult to evaluate because standard materials are lacking and validity can only be studied by comparative measurements with other methods in homoge- neous test atmospheres. In most studies the Andersen sampler and the AGI-30 impinger have been used as references and these methods generally show the highest yield among the viable methods. Measurement methods for micro-organisms generally do not comply with the criteria for inhalable dust sampling that were recently adopted by the Comit6 Europken de Normalisation (CEN).21 However, it should be possible to meet these criteria for methods that use filter sampling.Viability loss from sampling stress is an important source of sampling error in viable methods. Losses may be large for vegetative bacterial cells, especially when collected on filters, whereas spores are more robust. Delays from transportation of samples to the laboratory before analysis by culture may lead to decreased or increased colony counts depending on the type of micro- organism.22 Transportation by mail of spore samples on filters in aerosol monitors showed only small losses when the samples were analysed by SEM.23 Losses of up to 70% during preparation of samples for fluorescence microscopy have been reported recently.24 The presence of aggregates of micro- organisms may also give rise to errors. The number of viable micro-organisms in an aggregate size is usually neglected in analysis by culture as one aggregate may form just one colony.However, bioaerosols that are collected or redispersed in a liquid may show higher colony counts compared with directly cultured samples from disruption of aggregates. Micro-organ- isms in aggregates can be counted with microscopic techniques unless the aggregates are so large and complex that the number of micro-organisms is impossible to count. A procedure for the estimation of the number of micro-organisms in such aggregates has been described recently.24 Comparisons of studies may be substantially biased if differences between measurement methods have not been considered. For example, results from dilution plated samples and directly cultured samples, and the bias of using viable methods as a surrogate for non-viable methods and vice versa may be very large as results from viable methods may vary from <0.1% to up to 100% of the levels found with non-viable methods.Quality assurance schemes for bioaerosol analysis in the workplace are lacking so even results from different laboratories using the same method may differ. The precision of viable methods has been studied both in the laboratory and in field studies. Field studies are more valid for accuracy assessment as the composition of bioaerosols in work environments is generally more heterogeneous than in labo- ratory tests, where mainly monodisperse aerosols of mono- cultures have been used. The spatial distribution of aerosols in field studies may not be as good as in laboratory test atmospheres but may be improved if a field exposure chamber can be used.2”26 This inhomogeneity is not a component of the precision of the method and may partly explain the higher RSDs of 9-51% that have been found.Only a few studies have been published on the precision of nonviable microscopic meth- 0ds.27~28 These studies show that a major source of random error is the Poisson variability of the counting process. In addition, the aggregate size (micro-organisms per aggregate) has a large influence on the precision; 200-1 000 micro-organisms need to be counted in field samples to achieve a counting precision of 10% whereas a count of 100 is sufficient if aggregates are absent.28 It will be difficult for bioaerosol measurement methods to meet the performance criteria that have been adopted recently by the CEN,2g where an over-all uncertainty for measurements near the occupational exposure limit of less than 30% is required.However, measurement methods with RSDs of 30% contribute only a small error compared with the total random error of an exposure estimate as the variability of the exposure levels in the work environment is often high30931 and exposure to bioaerosols does not seem to be an exception (Fig. 1). Recommendations Which bioaerosol component should be measured depends on the expected health risk. As viability is not essential for non- infectious health effects such as inhalation fever, mucous membrane irritation, chronic bronchitis, asthma and hay fever, non-viable methods can be used for the measurement of cells, spores, endotoxins, glucans, chemical markers, antigens and allergens.Specific responses to certain species of micro- organisms or allergens, e.g., specific asthma and hay fever require assessment of specific organisms, but for other health effects it is not clear to what level micro-organisms or microbial bioaerosol components should be identified. Allergic alveolitis day Fig. 1 Variability of mould spore exposure levels of sawmill workers working in the same department; 8 h TWA concentrations from stationary measurements on 40 consecutive work days are shown. The geometric mean exposure is 106 spores m--3 and the geometric standard deviation is 3.3.1200 Analyst, September 1996, Vol. 121 and reassessment Medical examinations Follow-up of workers , and exposure may also be a specific response in patients.However, many different micro-organisms have been recognized as causes of this disease9 and species that are involved seem to be present in high concentrations and dominate the microbial flora. It is therefore possible that the combined exposure to all micro- organisms is more important for the development of allergic alveolitis than exposure to specific micro-organisms. For infectious diseases specific micro-organisms are of primary interest and viability is essential. Specific DNA markers and antigens may also be measured as surrogates if viable methods do not exist or have low sensitivity. For risk assessment of non-infectious occupational diseases, filter sampling is recommended because viability of micro- organisms is not essential, personal sampling is straightforward, there are few limitations on the sampling period and compliance with criteria for inhalable dust sampling2' seems possible.Bioaerosol components should be analysed by non-viable methods but the choice between bioaerosol components and analytical methods is less obvious and should be a subject for further study. Allergenic components can be analysed by immunoassays such as ELISA and RAST, and by microscopic techniques if allergenic micro-organisms can be recognized morphologically. Qualitative information on the composition of the microbial flora could be obtained by viable methods. Identification of species may show the presence of toxicogenic and allergenic species and is also important for the selection of non-viable methods.A special case is the detection of specific micro-organisms by viable methods as markers of indoor microbial growth in buildings.14 Viable methods are poor surrogates for non-viable methods because the relative yield of viable micro-organisms is highly variable, sampling periods are short and may be extremely short in highly contaminated work environments, few methods are suitable for personal sampling and compliance with criteria for inhalable dust sampling is poor. Exposure levels to bioaerosols may show high variability and a large number of measurements may be needed to obtain precise exposure estimates from measurements. The number of measurements may be reduced if sampling can be stratified by more homogeneously exposed tasks or special exposure conditions.Long-term exposure levels may then be computed Occurs mouldy Yes timber open? assessment bv: timber perio- dically ? Is the exposure level high examinations exposure a no no T Ye5 --b I exposure to mould I - (spores not likely 1 Fig. 2 timber.34 Combined strategy for evaluation of health risks from mouldy from measured exposure levels and the duration of these tasks and conditions.32 Other Methods for Exposure Assessment As bioaerosol measurements are expensive, cheap qualitative methods are interesting tools for exposure assessment such as the recording of indicators of microbial growth. Such results may also give useful information for a stratified sampling strategy. Specific immunoglobulin G (IgG) antibodies against moulds and actinomycetes can be found in the serum of regularly exposed workers.These antibody levels show a relationship with exposure levels during the preceding months.33 As the IgG antibody response may differ substantially between individuals, risk assessment is improved if it can be based on groups of workers with similar exposure levels. Intraindividual changes of specific IgG antibody levels indicate changes in exposure level more accurately. Serology may therefore be used for the assessment of exposure to specific micro-organisms but the results may be difficult to interpret if exposure to diverse microbial flora is to be assessed. A combined strategy for the assessment of health risks from exposure to fungal spores from mouldy timber in the wood industry has been published recently in Norway.34 This strategy includes the recording of mouldy timber, recording of febrile and respiratory symptoms, measurement of serum IgG antibody levels to prevalent species and exposure measurements (Fig.2). Conclusions Filter sampling and analysis by non-viable methods are recommended for the measurement of exposure to non- infectious microbial bioaerosol components in the work environment. Relationships between methods for measurement of exposure to bioaerosol components must be taken into account when different studies are compared. Further studies are needed to clarify which microbial components are most relevant for health effects from non- infectious bioaerosol exposure in the work environment.Quality assurance schemes of analytical methods for bio- aerosol components are needed for future compliance testing. References 1 2 3 4 5 6 7 8 9 10 I I 12 13 Pasteur, L., Ann. Sci. Nut. (Zool.), 4me Ser. 1861, 16, 5. Gregory, P. H., and Lacey, M. E., J . Gen. Microbiol , 1963, 30, 75. Pepys, J., Jenkins, P. A., Gregory, P. H., Festenstein, G. N., Lacey, M. E., and Skinner, F. A., Lancet. 1963, 2, 607. Madelin, T. M., and Madelin, M. F., in Bioaerosols Handhook, ed. Cox, C. S., and Wathes, C. M., CRC Press, Boca Raton, FL, 1995, p. 361. Bioaerosols Handbook, ed. Cox, C. S.. and Wathes. C. M., CRC Press, Boca Raton, FL, 1995. Larsson, L.,. Acta Pathol. Microhiol Immunol. Scund., 1994, 102, 161. Parkes, W. R., Occupational Lung Disorders, Butterworths, London, 2nd edn., 1982.Pepys, J., and Jenkins, P. A., Thorax, 1965, 20, 21. Lacey, J., and Dutkiewicz, J., J . Aerosol. Scz., 1994, 25, 1371. Thurston, J. R., Cysewski, S. J., and Richard, J. L., Am. J . Vet. Rrs.. 1979,40, 1443. Fogelmark, B., Lacey, J., and Rylander, R., Int. J . Exp. Patlwl., 1991. 72, 387. Rylander, R., Am. J . Ind. Med., 1994, 25, 19. Flannigan, B., and Miller, J. D., in Health Inzplications ofFungi in Indoor Environments, ed. Samson, R. A,, Flannigan, B., Flannigan, M. E., Verhoeff, A. P., Adan, 0. C. G., and Hoekstra, E. S., Elsevier, Amsterdam, 1994, p. 3.Analyst, September 1996, Vol. 121 1201 14 15 16 17 18 19 20 21 22 23 Healtlz Iniplications of Fungi in Indoor Environments, ed. Samson, R. A., Flannigan, B., Flannigan, M. E., Verhoeff, A.P., Adan, 0. C. G., and Hoekstra. E. S., Elsevier, Amsterdam, 1994, p. 531. Jensen, 1’. A.. Lighthart, B., Mohr, A. J., and Schaffer, B. T., in Atniospheric, Microbial Aerosols. ed. Lighthart, B., and Mohr, A. J., Chapman and Hall, New York, 1994, pp. 226. Walters, M., Milton, D., Larsson, L., and Ford, T., Appl. Environ. Microhiol., 1994, 60, 996. Douwes, J., Versloot, P., Hollander, A., Heederik, D., and Doekes, G.. Appl. Environ. Micwhiol., 1995, 61, 1763. Obayashi, T., Tamura, H., and Taaka, S., Prog. Clin. Biol. Res., 1987, 231, 357. Frisvad, J. C., and Gravesen, S., in Heulth Inzplicutions of Fungi in Indoor Environmenls, ed. Samson, R. A,, Flannigan, B., Flannigan, M. E., Verhoeff, A. P., Adan, 0. C. G., and Hoekstra, E. S., Elsevier Science, Amsterdam, 1994. p. 28 I . Eduard, W., PhD Dissertation, University of Wageningen, 1993. Comitk Europken de Normalisation (CEN), Workplace Atmospheres. Size Fiuction Definition for Measurement of Airborne Purticles, EN 48 1, CEN, Brussels, 1993. Thome, P. S., Lange, J. L., Bloebaum, P., Kullman, G. J., Am. Ind. Hyg. Assoc. J . , 1994, 55, 1072. Eduard, W., Lacey, J., Karlsson, K., Palmgren, U., Strom, G., and Blomquist, G., Am. Ind.Hyg. Assnc. J., 1990, 51, 427. 24 25 26 27 28 29 30 31 32 33 34 Heldal, K., Skogstad, A., and Eduard, W., Ann. Occxp. Hyg., in the press. Seim, H. J., and Dickeson, J. A., Am. Ind. Hvg. Assoc. J . , 1983. 44, 562. Skogstad, A., Eduard, W., and Huser, P. O., Ant. Ind. Hyg. Assrtc.. J . , in the press. Kapyla, M., and Penttinen, A., Grana, 198 1, 20, 13 I . Eduard, W., and Aalen, O., Ann. Occup. Hyg., 1988, 32, 471. ComitC EuropCen de Normalisation (CEN), Workpkfce Atmospheres. General Requirements for the Pet formatice of ProcxJdures fijr Measurement of Chemical Agents, EN 482, CEN, Brussels, 1994. Rappaport, S. M., Anti. Occup. Hyg., 1991, 35, 61. Kromhout, H., Symanski, E., and Rappaport, S. M., Ann. Occup. Hyg., 1993, 37, 253. Nicas, M., and Spear, R. C., Am. Ind. Hj~g. Assoc. J . , 1993, 54, 21 1. Eduard, W., Occup. Hyg., 1995, 1, 247. Arbeidstilsynet, Health Risk,fiom Mouldy Timber, Orientering Best. Nr. 537, Tiden Norsk Forlag, Oslo, 1996 (in Norwegian). Puper 6/01 854H Received March 18, 1996 Accepted May 31, 1996
ISSN:0003-2654
DOI:10.1039/AN9962101197
出版商:RSC
年代:1996
数据来源: RSC
|
20. |
Microbial volatile organic compounds—what substances can be found in sick buildings? |
|
Analyst,
Volume 121,
Issue 9,
1996,
Page 1203-1205
Bengt Wessén,
Preview
|
PDF (423KB)
|
|
摘要:
Analyst, September 1996, Vol. 121 (1203-1205) 1203 Microbial Volatile Organic Compounds-What Substances can be Found in Sick Buildings?* Bengt Wessen and Karl-Olof Schoeps Pegasus Laboratory, P.O. Box 97, 750 03 Uppsala, Sweden There is a relationship between damp buildings and health complaints. Damp conditions in building constructions also favour the growth of micro-organisms. Growth of micro-organisms results in the production of volatile organic compounds, which has been shown to have an impact on Indoor-air monitored via a microbial volatile organic compound (MVOC) analysis. In order to widen the applicability of MVOC analysis, it is necessary to increase this analysis by including more volatiles. By active sampling on Anasorb 747 and selected ion monitoring on a mass spectrometer equipped with a quadropole detector, it is possible to determine these volatiles with sufficient accuracy in indoor air of non-industrial buildings.Keywords: Microbial volatiles; MVOC; mouldy odour; damp buildings; sick buildings Introduction Users of damp buildings often suffer from ‘sick building syndrome’ (SBS) symptoms of different kinds. SBS symptoms occur in buildings with new and old moisture problems. These SBS symptoms can be manifested in many different ways, such as irritated eyes, throat and skin. There is a wide variety of factors, alone or in combination, that can be the cause and/or trigger of SBS. Numerous investigations have shown that micro-organisms may play an important role. Thus, mouldy odour showed high odds ratios for SBS symptoms in a survey from Finnish Day-care Centres,’ and this was also shown in an Australian home study of 40 houses.2 In a survey of 9000 children in North America, there was shown to be a high correlation between home dampness indicators and respiratory health problems.’ In Sweden there have also been extensive reports from field workers that microbial growth in building construction does cause health problems to the users of the building.That volatiles in sick buildings can be of special interest has previously been shown by Norback et al.4 and Joki et ~ l . , ~ who found that volatiles from microbial cultures had an impact on cilia1 cells in respiratory airways. The main activity of micro-organisms is directed to the decomposition of complex organic compounds into simpler compounds, carbon dioxide and microbial biomass.During this process, a wide range of by-products can be produced: some are unique to micro-organisms, some can also be natural compo- nents of building materials and others are the same as those which can originate from outdoor sources, such as exhausts from engines.6 These volatiles are so diverse in composition that they will belong to all volatile classes. In order to distinguish microbial volatiles from other sources, it is of utmost importance to choose chemicals that can only be of microbial origin and to use these as markers of microbial impact. This means that there will always be an underestimate * Presented at AIRMON ’96, Salen, Sweden, February 5-8, 1996. of ‘true’ microbial impact as only a few volatiles will pass as original microbial volatile organic compounds (MVOCs).From an analytical viewpoint, the use of the selected ion monitoring (SIM) principle, offers a great advantage. The MVOC peaks often occur in low concentration and these peaks are therefore often interfered with by other, more commonly found volatile organic compounds (VOCs) that occur in higher concentrations in indoor air. Experimental A range of volatile substances (MVOC class a) were chosen (Table 1) on the basis of the causality of microbial origin or a combined origin with micro-organisms and others when sampled in the field.7.8 In order to determine the reproducibility of the SIM method in MVOC analysis, the following study was performed in a laboratory environment. Sampling For each compound, 10 PI of a solution containing 800 ng (for geosmine 200 ng) of compound per millilitre of dichlorome- thane were applied to a piece of glass-wool inside a glass tube.Samples were collected with portable constant-flow pumps. Air samples were collected in a laboratory environment by sucking air through these glass tubes, to which adsorbent glass tubes containing Anasorb 747 (210 mg of beaded carbon) were connected. The adsorbent tubes had been conditioned and sealed by the manufacturer (SKC, Eighty Four, PA, USA).The capacity of Anasorb 747 is high, it can retain more than 1 mg of VOCs before breakthrough. In non-industrial environments, no breakthrough was found when sampling air volumes of up to 120 1. The air flow was 200 ml min-1 and the sampling period was 240 min.After the sampling period, the tubes were sealed with PTFE caps and kept in a freezer until analysis. Desorption of Adsorbents Each adsorbent was transferred into a 3.5 ml glass bottle and 1 ml of dichloromethane was added. The glass bottles were placed on a shaker for 30 min. After desorption, the solvent was transferred into a 2 ml glass vial with a Pasteur pipette and sealed with a crimp-top seal with rubber liners. Table 1 Selected hydrocarbons produced by micro-organisms (MVOC class a) 3-Methylfuran 3-Methylbutan-1 -01 3 -Methylbutan-2-01 Pentan-2-01 Hexan-2-one Heptan-2-one Octan-3-one Octan-3-01 Oct-I-en-3-01 Oct-2-en- 1-01 Geosmine Dimethyl disulfide Butan- 1-01 2-Methylpropan- 1-011204 Analyst, September 1996, Vol. 121 Chromatography Analysis was performed with an HP 5890 gas chromatograph equipped with a HP 5971 mass-selective detector. The analyt- ical method was performed according to Strom et al.,9 but using a 60 m capillary column.MVOC Analysis For each substance (Table l), a mass spectrum was obtained, together with the GC retention time. SIM ion values within 0.1 u were determined to obtain the highest response. Three ions were selected for each compound. The SIM method increases the detection level for specific compounds by two to three orders of magnitude. The detection limit for each compound is in the range of 0.1-0.5 ng ml-I dichloromethane. Results and Discussion The validity of the analytical system including sampling is presented in Table 2. The results show a notable small deviation from the mean, which indicates that the method is reliable.Some of the compounds show lower recoveries than would be expected and others higher values, the reason for which can only be speculated upon. The low recoveries of the terpenes could be due to a low release from the adsorbent. The higher values for the alcohols could be due to possible background emissions from the laboratory building. Field measurements of MVOC (class a) from problem buildings with SBS complaints often show high concentrations compared with normal atmospheric samples (Table 3). How- ever, measurements of other parameters such as airborne micro- organisms seldom show high concentrations in non-industrial Table 2 Recovery of MVOCs (class a) from Anasorb 747 in a laboratory environment, presented as a percentage of the theoretical value Compound Geosmine Metox y p yrazine Oct- 1-en-3-01 3-Methylfuran Oct-2-en- 1-01 3-Methylbutan- 1-01 3-Methylbutan-2-01 Pentan-2-01 Octan-3-01 Hexan-2-one Heptan-2-one Oc tan- 3 -one Total MVCOs Mean recovery (%)* SE(%) s(%) 50.9 1 61.12 100.58 74.92 121.25 94.85 107.33 116.2 99.34 142.06 176.88 133.56 I I 1.44 1.48 I .52 4.94 2.47 7.56 7.13 2.66 2.22 2.15 4.65 8.08 3.69 2.6 1 4.68 4.80 15.61 7.82 23.90 22.54 8.42 7.01 6.80 14.70 25.54 11.67 8.25 buildings with SBS complaints10 compared with normal atmospheric samples.It is obvious, however, that the range of the MVOCs (class a) analysed in problem buildings varies considerably, and in some cases it can be difficult to distinguish a problem building from a reference building or the outdoor air when comparing the sum of MVOCs or the concentration of specific MVOCs.This means that it is important to study and characterize more relevant volatiles that can be produced by the micro-organisms growing on specific building materials in a specific problem building. It has also been shown that this is the case, as both Wesskn et al." and Sunessonl2 found that micro- organisms produced different VOCs when growing on different materials. A list which includes more MVOCs has been assembled from recent published data (MVOC class a + b)I13l3 and unpublished laboratory growth studies on actinomycetes, moulds and basidiomycetes (Table 4). Data are continuously being collected and for the first 2 months of 1996 it can be seen that the new compounds correlate with the original hydrocarbons (Table 5).In individual cases, however, the inclusion of new MVOCs has been of great use to explain the cause of problem buildings (Fig. 1). In all these seven cases of mouldy buildings, while using the MVOC class a from Table I , it was not possible to correlate microbial impact with the indoor environment as the MVOC concentration was too low, but with the inclusion of the extra substances from Table 4 List of MVOCs (class a + b) including new compounds 3-Methylfuran Pentan-2-01 Octan-3-one Oct-2-en- 1-01 Butan-1 -01 Thujopsene Endoborneol 2-Pentylfuran 3-Methylbutan- 1-01 3-Methylbutan-2-01 Hexan-2-one Heptan-2-one Octan-3-01 Oct- 1 -en-3-01 Geosmine Dimethyl disulfide 2-Methylpropan- 1-01 Ethyl isobutyrate Karveol Terpineol Fenchone Nonan-2-one 4-Methylheptan-3 -one Table 5 Comparison of different MVOC classes from Swedish buildings [MVOC]/pg m--3 Class a + b Class a Class b Outdoor Indoor Outdoor Indoor Outdoor Indoor Mean 0.04 0.52 0.05 0.40 0.09 1.01 SE 0.00 0.07 0.01 0.04 0.01 1.12 s 0.04 1.12 0.07 0.35 0.09 1.83 Minimum 0.00 0.05 0.00 0.04 0.00 0.07 Maximum 0.21 9.03 0.41 1.55 0.55 17.38 No.of cases 76 229 76 229 76 229 * n = 10. Table 3 MVOCs (class a) in Swedish problem buildings during 1996 [MVOC]/pg m--3 Outdoor Indoor Mean 0.04 0.52 SE 0.00 0.07 S 0.04 1.12 Minimum 0.00 0.05 Maximum 0.21 9.03 No. of cases 76 229 1 2 3 4 5 6 7 Cases Fig. 1 Effect of introducing more MVOC substances in the MVOC analysis of seven problem buildings with SBS complaints.Analyst, September 1996, Vol.121 1205 Table 4, it can be seen that there is a relatively high microbial 8 9 VOC impact on the indoor air of these problem buildings. References Jaakkola, J. J. K., Jaakkola, N., and Ruotsalainen, R., J . Exp. Anal. Godish, T. J., Godish, D. R., Hooper, €3. M., and Hooper, M. A., in Indoor Air: an Integrated Approach, ed. Morawaska, L., Bofinger, N. D., and Maroni, M., Elsevier, Oxford, 1995, pp. 21 1-214. Spengler, J., Nakai, S., 0zkaynak, H., and Schwab, M., in Indoor Air: an Integrated Approach, ed. Morawska, L., Bofinger, N. D., and Maroni, M., Elsevier, Oxford, 1995, pp. 189-195. Norbiick, D., Michel, I.. and Widstrom, J., Scand. J . Work EnLiron. Health, 1990, 16, 121. Joki, S., et al., in Indoor-Air '93, Helsinki, Proceedings, Indoor Air C93, Helsinki, 1993, vol.1, pp. 259-263. Bayer, C. W., and Crow, S., in Indoor-Air '93, Helsinki, Proceedings, Indoor Air C93, Helsinki, 1993, vol. 2, pp. 33-38. Miller, J. D., Laflamme, A. M., Sobol, Y., Lafontaine, P., and Greenhalgh, R., Int. Biodeterior. Bull., 1988, 24, 103. Environ. Epid., 1993, 3, Suppl. 1, 129. 10 11 12 13 Borjesson, T., Stollman, U., Adamek, P., and Kaspersson, A., Cereal Chem., 1989,66,300. Strorn, G., West, J., WessCn, B., and Palmgren, U., in Health Implications .f Fungi in Indoor Environments, ed. Samson, R. A., Flannigan, B., Flannigan, M. E.. Verhoeff, A. P., Adan, 0. C. G., and Hoekstra, E. S., 1994, pp. 2.5-279. Strom, G., et al., in Indoor-Air 'YO, ed. Walkingshaw, D. S., Indoor Air C93, Ontario, 1990, vol. 1 , pp. 173-178. WessCn, B., Strom, G., and Schoeps, K.-O., in Indoor Air: an IntegratedApproach, ed. Morawska, LA., Bofinger, N. D., and Maroni, M., Elsevier, Oxford, 1995, pp. 67-70. Sunesson, A.-L., Dissertation, UmeA University, 1995. Larscn, T. O., and Frisvad, J. C., in Health Implications of Fungi in Indoor Environments, ed. Samson, R. A,, Flannigan, B., Flannigan, M. E., Verhoeff, A. P., Adan, 0. C. G., and Hoekstra, E. S., 1994, pp. 251-279. Paper 610201 6J Received March 23, 1996 Accepted June 17, I996
ISSN:0003-2654
DOI:10.1039/AN9962101203
出版商:RSC
年代:1996
数据来源: RSC
|
|