Document emBe5wXZRY9wYKjLyRvanLZYe

Am. Ind. Hyg. Assoc. J 43<8):605-621 (1982) * The history of development and validation testing of passive dosimeters is reviewed. Theoretical considerations including possible limiting factors or interferences, are pr sented. Laboratory and fi Id validation tests art \ critically reviewed and results are pr sented for comparative purposes. Evaluation of available data indicates tha passive dosimetry, with some exceptions, is an acceptabl method for monitoring gasses and vapors. Mos importantly, passive systems appear to be as reliable as the now accepted active sampling syst ms. Passive dosimetry -- state of the art review VERNON E. ROSE and JIMMY L PERKINS School of Public Health, University of Alabama in Birmingham, Birmingham. AL 35294 introduction Recognition, evaluation and control are the cornerstones of personal passive dosimeter: personal, because it can be won the application of that mixture ol science and art known as by the worker in close proximity to the breathing /one industrial hygiene. These three tasks. however. are no longer passive, because there is no pump to move the air over , the eminent domain of the industrial hygienist. In the past collector, which equates to fewer calibration and maintc decade, a proliferation of training in the recognition of nance problems. Some quarrel with the term dosimeter, wit workplace hazards has been made widely available to purists preferring to call them collectors, monitors, o workers and management alike. At the other end of the samplers. While many of the devices are collectors an* spectrum has been the training of individuals highly special require the application of subsequent analytical procedures ised in the control of specific hazards, especially those others provide for a more direct measurement ofexposur invoh ing noise and toxic air contaminants. These develop- dose." Their basic appeal, however, is simplicity ol use I ments are welcomed because they contribute significantly to Theoretically, elaborate calibration procedures are unneces the ultimate goal of protecting the health of workers by sary. and all that is needed is a fairly reliable timepiece t prosiding safer and more healthful workplaces. measure exposure duration. There is some recognition tha At the same time, professional industrial hygienists rec temperature and humidity may affect the observations ognize that often the critical step in the process is not recog nition of toxicity, but evaluation of hazard which leads to therefore, most manufacturers advise the user to report thes environmental conditions to the analytical laboratory pro the subsequent development of the most effective means of cessing the dosimeter. control where warranted. This key step of evaluation is the Rather than v iewing passive dosimeters as another wav t unique domain of the industrial hygienist, often supple replace the industrial hygienist, industrial hygienists mus mented by other members of the occupational health and recognize and appreciate the potential of the dosimeter' in safety team. Where evaluation requires the determination of helping to achieve the hygienists' goals That potential is w orker exposure to airborne toxic substances, the industrial significant in that personal dosimeters, if properly used, hygienist has seen a rev olution in the development of sophis offer the opportunity to revolutionize the evaluation step ticated techniques and equipment. The parallels with detector tubes, as well as with noise and The "organ-grinder" impinger sampler is a relic, having been replaced by constant flow, eight-hour battery-operated pumps, light enough to be carried by the worker. The liquid bubbler and impinger have been replaced by the charcoal and chemical substrate sampling tube. The laboratory has come to the field in the form of the portable gas chromato graph and infrared monitor, albeit with a price rise directly proportional to the sophistication of the equipment. Even the once lowly, direct-reading detector tube has become legitimate with the establishment of government programs to certify accuracy and precision. ionizing radiation dosimeters, are obvious. Indeed. th-` parallel with radiation dosimeters, especially film badges, i striking. The opportunity to significantly expand the mea surement of worker exposure to many toxic material' ca provide a quantum leap in our ability to provide sale an healthful workplaces. With any sampling device, however there also must bethe understanding that use ol such dev ice is only one part of the evaluation step. The concepts r proper selection of workers at risk: the understanding e limitations, interferences and similar factors, and v.ltimatel the proper interpretation of the results ate still key ingre dients in the evaluation step. 1 he possibility ol "lake nog; But while mosl evaluation techniques were reaching the live"decisions leading to erroneous assumptions ol salety ) point where the industrial hygiene staff required thcaddition or "false positive" conclusions leading to unwarrante of someone with a Ph D. in electrical engineering, a new expenditures of resources lor conii ok still exists iegardle" device has appeared which has the key of simplicity -- ihe of the measurement device used. Copyright 19B2 Am*t>cjn )n0USt'l*l NvpHTfW* Ass.OfUHH'"' American Industrial Hy^ene Association JOURNAL (43) 8'82 3M 001767 fc( With the rapid proliferation of passive dosimeters in the past several years, it is appropriate that industrial hygienists evaluate the "state-of-the-art"and. as professionals, become involved with the proper application of these monitor ing devices. theories of operation In that passive dosimeters by definition do not use an air moving device to transport contaminated air to a collector, natural forces are relied upon to ensure that a representative amount of contaminant is "seen" by the detector. To dale, one of two principles has been applied in the design of dosimeters. The first, and most widely used, is the principle of diffusion of contaminant molecules through a stagnant gas (air) layer. The second principle involves the absorption in and subsequent permeation of contaminant molecules through a membrane. Diffusional monitors rely on the movement of contami nant molecules across a concentration gradient which for steady-state conditions, can be defined by Tick's First Law of Diffusion:'11 dc w = -da d--\ d> where: W = mass transfer rale, ng sec. D = diffusion coefficient, cm* sec, A = cross sectional area of diffusion path, cm*', and dc dx = the instantaneous rate of change in concentra tion over diffusion path, (ng cm4)cm '. Considering the change in concentration (Cj -- O.) over the total diffusion path length (Xi -- Xu = --IT. equation (1) becomes: W = D -pCT - O.) (2) where: l. = length of the diffusion (static) path. cm. Ci -- ambient concentration of contaminant, ng cm\ and Cn = concentration of contaminant at collecting sur face. ng cm'1. If an effective collection medium is employed, the contam inant concentration at the surface of the collect or (Cn) can be assumed to be zero, and mult iply ing both sides of equation (2) by time, yields: M = DytCTt (3) where: M = total mass transferred, ng, and t = time that the badge is exposed to the contami nated air. sec. It is also interesting to notethat the units of the product of Dand A. divided by l.arecm3 sec. which are the same units associated with active air-moving devices such as personal sampling pumps. Rearranging equation (3) as follow's: it becomes apparent that live tactors affect the measurement of the ambient air concent rat ion of a substance (Ci). Two ol the factors (L. and A) arc physical parameters associated with the construction of the dosimeter, one (M) is provided by measuring the total mass of contaminant collected by the sampler, another is the duration (t) the sampler was exposed to the contaminated atmosphere, and the final factor (D) is an individual properly of each vapor or gas. It also iknown11' that the diffusion coefficient is directly propoitional to the absolute temperature (T) of the vapor, raised to three-halves power and inversely proportional to the atmo spheric pressure (P). Dosimeters that rely on the principle of permeation through a membrane are especially useful w here the contam inant of concern is usually found mixed w ith other interfer ing vapors or gases or when a liquid collecting medium is employed. The goal then becomes to identify a membrane material that is highly permeable to the contaminant of interest and impermeable to most other components in the atmosphere, and or the collecting media. The determination of ambient concentrations of a con taminant using a permeation dev ice can be determined from the formula: C = wk i (6) where: C = concentration of contaminant, ppm. w = mass of contaminant collected, ng. k = permeation constant, ppm-hours pg. and t = exposure time, hours. The permeation constant (k) is determined experimentally and is a function of the specific membrane material and contaminant of interest.`*' sources of measurement error The most obvious sources of error for both types of passive dosimeters are apparent from equations (4) and (6). Com mon to both badges are determinations of the mass of con taminant collected and the time of exposure of the dosimeter to the contaminated atmosphere. For the diffusional moni tor, accurate knowledge of the physical parameters asso ciated with badge construction (length and cross-sectional area) and the diffusion coefficient of the contaminant are important. There are at least nine prediction methods for calculating the diffusion coefficient, and in one study com paring observed and expected values for more than 100 compounds it was not uncommon to have less than 50 percent of the calculated results within 5 percent of the observed.131 Montalvo hasdescribed a procedure for limiting errors associated with computed diffusion coefficients.At 3H 001768 Am Ind Hyf Assoc J (43) August issr leasi one manufacturer, the 3M Company, makes available its procedures for determining sampling rates (DA. L) for its badges.15' Its approach has been to experimentally determine the sampling rate for five or six compounds in a chemical family to establish the relationship between the diffusion coefficient and the measured sampling rates. Sampling rates for other compounds are determined from the diffusion coefficients calculated by the Hirschfcldcr equation and the empirical relationships dev eloped from the test compounds. The rationale for the selection of the Hirschfcldcr equation is not given, but in the study of the nine diffusion coefficient formulas, the author concluded that for higher molecular weight compounds the Hirschfcldcr. Biard and Spatz equa tions were in closest agreement with determined values.'3' Hot the permeation monitor, accurate determination of the permeation coefficient for each monitor is necessary for obtaining accurate results. Factors influencing permeation include: thickness and uniformity of the membrane, affinity of the membrane for the analy te. swelling or shrinkage of the membrane, and possible etching by corrosive chemicals. The problems associated with accurate determinations of the mass of the contaminant collected arc similar to those, involved with other collection devices such as charcoal or silica gel tubes, or to those in which the collection of the contaminant involves a chemical reaction with the collection medium. Using known amounts or concentrations of con taminants to determine collection and or desorption effi ciencies is as critical a step for passive dosimeters as it is for other methods of collection. Saturation of the sorbent as well as the subsequent accuracy of analytical techniques arc also part of the total error associated with the measurement. Another common concern in all types of environmental measurements is the potential for interferences, either posi tive or negative, from other contaminants in the sampled air. As the evaluation of passive dosimeters has matured, increased attention is being paid to possible interferences in multi-contaminant exposure situations, in both the labora tory and field. In evaluating such interferences it should be recognized that there are several potential sites for such interferences to appear, e.g.. effects on adsorption or absorption efficiency of the sampling medium, chemical reactions of two or more contaminants prior to analysis, and the multitude of interferences associated with analysis of complex mixtures of gases and or vapors. These problems also arc found in the more classical sampling and analyti cal methods. Accurate measurement of the time the sampling device is exposed, is essential to most industrial hygiene sampling procedures. For both short-term and full-shift exposure measurements, errors less than one percent, i.e. 9 seconds in 15 minutes and 4.8 minutes over 8 hours, are not unreason able goals. For the diffusion coefficient and possibly the permeation constant, it would appear that three factors have the greatest eflect on variability. These factors are the two already identi fied. temperature and pressure, and. less readily apparent, the velocity of the air external to the badges. Ajupfirar Industrial HvP'Pnp Association JOURNAL <43) 8 8? Considering temperature and pressure, and referring to equation (5) it can be shown that a temperature rise from 5 to 35 C would give a 16 percent increase in the diffusion coefficient, while a rise in barometric pressure from 710 to 810 mm Hg would cause a 14 percent decrease.'1' However, at the same time, the changes in temperature and pressure also are affecting t he concentration (mass v olumc: actually. density is the proper term but most authors use concentra tion) of the contaminant in that concentration is inversely proportional to the temperature and directly proportional to the pressure. Asa result, the total mass(M)collected by the dosimeter is only slightly affected by temperature (M 11J) and is independent of the pressure."1 Consequently. w hilc at ambient temperatures, the diffusion coefficient will increase about 0.5 percent per C, and the total mass collected hy the sampler will increase less than 0.2 percent perC. Therefore, a temperature change from 25 to 30 C. if uncorrected, will introduce a measurement error of less than one percent, while a change from 5 to 35 C. if uncorrectcd. would introduce an error of about five percent. The final source of error to consider is the velocity of the air external to the dosimeter: often this is referred to as face velocity. In an early assessment of face v elocity effects, espe cially the lack thereof. Tompkins and Goldsmith point out that the important consideration is to contain all resistance to contaminant transport within the stagnant air layer inside the device."' As Jonas el at. subsequently noted, the face velocity directly affects the concentration gradient Ci--C in equation (2). and C\ can no longer be assumed to be the ambient concentration when the air external to the badge is stagnant.'6' With zero or low face velocities, the length (1 I of the diffusion pathway is effectively extended, and there is a decrease in the measured ambient concentration. In Tompkins'and Goldsmith's work with the GASBAlXiE'*. they determined experimentally that as long as face veloci ties were greater than 7.5 cm sec (15 fpm) there was "no significant effect on dosimeter response:" however, experi mental results supporting this conclusion were not pre sented."' High face velocities may also affect the concentra tion gradient. Commercially available diffusion devices rely on either a large ratio of diffusion path length to diffu sion tube diameter or a wind screen to limit errors from this condition. One of the most comprehensiv e tests to document sources of error has been conducted under contract for the National Institute for Occupational Safety and Health, and although concluded, it is not yet available as a public report.'" The study involved evaluation of the GASBAPGf and 3M Organic Vapor Monitor'" (the DuPont badge not being available at the time the study was initiated) via challenge with several organic vapors. The factors investigated were precision, effects of storage, maximum and minimum levels of quantification, face velocity effect', effects of temperature and humidity, off-gassing (related to storage), exposure to mixtures, problems associated with applicable analytical methods, and adsorption of the contaminant by the badge itself with subsequent leaching to the sensing surface. The possibility of adsorption by the badge body, thus giving 3H 001769 tr higher results if the contaminant is subsequently released to the active medium, is of special concern in using passive dosimeters to measure very low ambient concentrations such as might be found in air pollution studies. Such interest and concern are evidenced by research on the subject being sponsored by the U.S. Environmental Protection Agency (EPA).181 Because of the lower concentrations involved with air pollution studies as opposed to workplace environments, the EPA also is concerned with the background or postmanufacture contamination levels associated with the sens ing medium. Initially. the focus concerns organics and acti vated charcoal. In summary, although numerous factors may affect the final calculation of concentration, only face velocity and the determination of the diflusion coefficient are unique sources of error for passive collectors. Therefore, if face velocities arc sufficient to prevent "starvation"(probably greater than 7.5 cm sec) and if diffusion coefficients have been accurately calculated orexpcrimentally determined, passive dosimeters should give results comparable to those obtained with tradiional active sampling systems. statistical considerations In evaluating any new monitoring method, extensive labora'orv and field testing is necessary. Interpretation of the results of these tests requires the application of appropriate statistical techniques. The use of statistical techniques which are meaningful and easily understood is important; conse quently. a discussion of the techniques used to evaluate passive dosimeters is appropriate. There are numerous statistical tests which can be applied :o both field and laboratory validation data. The main dif ference between the two situations is the degree of certainty of the "true" concentration of the monitored environment. In the field, the true value is usually an estimate based on the results of a standard sampling and analytical method. In the laboratory. experimental "known" concent rations are evolved and are then used for comparison with the concentrations estimated from sampling and analytical methods. What is often not stated is that a certain amount of error also exists in the determination of laboratory-evolved "known" concentrations. These errors are often difficult to estimate. The "known" concentration is often calculated by weighing a syringe before and after an injection period (mass balance) or simply by injecting or allowing to diffuse a measured volume. It is assumed that the aliquot delivered vas vaporised or diffused into a test chamber of known size. Possible sources of error include adsorption to or leaks from the test chamber, absorption and adsorption to articles placed in the chamber, degradation of the analyte by air oxidation or hydrolysis at high relative humidities, and error in measuring the injected contaminant. A backup monitor ing system may be used to ensure close proximity to the `known "concent rat ion. For example, an infra-red (I R)anay/cr or gas chromatograph may be used as a check on a known" concentration. In other instances an IR analyzer or a direct reading instrument may be the only method for determining "know n" concentrations. In this case the error in the instrument can be calculated or estimated. Charcoal tubes and critical ori fices also have been used to measure "known" concentra tions. If error in both the "known" concentration and the estimated concentration are considered, statistical tests used to validate the experimental method become considerably more complicated; hence, the error in measuring the "known" concentration is usually assumed to be small and unimpor tant.191 Methods described above fordeterminint the "known" concentration vary in their accuracy, a fact which should be considered when evaluating validation data for any sam pling and analytical method. When one is validating a method in the laboratory, there are two main considerations: the variation of the samples or data points about their mean, and the deviation of the sam ple mean from the true mean or "known"concentration. The first consideration often is called precision and is probably the most important and reliable measure as it does not depend on the error in determining the "known" concentra tion. Precision is estimated by determining the coefficient of variation (CV) or relative standard deviation of the data set as follows: CV = ~X 100 (7) where: s = Standard deviation of sample data set. and X = Mean of sample data set. Where samples are taken at several concentrations, it is necessary to determine a pooled coefficient of variation which involves determination of number of levels tested.110' It should be noted that determining the number of concen trations (or more appropriately, the number of statistical levels) is not always straightforward. For example, if 10 samples are taken at each of three concentrations, and if within each concentration five of the samples are collected over four hours while the other five are collected over eight hours, are there three levels or six levels? The important point to consider is whether the differentiation of a level is based upon an anticipated difference in the sample mean. Certainly If the investigator designs the experiment with different time levels, there is an anticipated effect of time on the mean. Unfortunately, when experiments are so designed, statistical analyses at the various levels usually are not performed. The second statistical consideration, the difference between the sample mean and the"known"value, is sometimes called accuracy, but the term bias is more appropriate. It is defined as: b = X ~ Xo X 100 Xo (8) where: X = mean of sample data set. and Xo = "known" value at level tested. If more than one level is sampled, it is necessary to determine the pooled bias of the data set.1111 >08 Am Ind Hyp Assoc J (43) August. 198? 3M 001770 The bias for a given set of data can sometimes be cor rected. If the average bias (either the average of several samples at one level or the pooled bias for several levels) is ^ large, one should note if the components of the bias value (either the individual samples or the levels) are consistently negative or positive. If the bias is large and varies consis tently in one direction, the precision nevertheless may be quite small. In this case a physical or chemical variable may be consistently affecting the method (a systematic as opposed to random error), causing the experimental values to con stantly fall short or long of the "known"concentration. This form of bias should be corrected. In addition to these statistical tests, others have been used to assess the validity of passive monitoring systems. Overall system accuracy"" has been defined as (2 X CV) + absolute bias, expressed as a percent. Others"2' have used the percent age of the "known" concentration accounted for by the sample mean two standard deviations as well as the term systematic error03'1" which is equivalent to overall system accuracy. Relative standard deviation has also been used, and is defined as the equivalent of CV."S> Additionally, some authors report only raw data while others report means without standard deviations or sample sizes, and various other combinations. While all of these statistical determina tions have utility, it seems important for comparative pur poses to consistently use those determinations which give the most information in the simplest form. Certainly, bias and precision meet these criteria. Discussion of one other point seems necessary. NIOSH<10> ) has proposed as a guideline for their own internal purposes that sampling and analytical methods meet a minimum requirement of 25 percent accuracy: that is, the absolute total error of the method should be less than 25 percent in at least 95 percent of the sample population (assuming a nor mal distribution). MOSH derives the maximum precision value for an unbiased method given the accuracy criteria staled. This value (12.8 percent) is the maximum precision value acceptable for an unbiased method. Although the 25 percent accuracy criterion has been criticized by some authors,"6'it wasadopted for NlOSH'sown internal useand is not meant as public policy. However, OSHA adopted the same criterion for the benzene standard, w ithout a complete derivation or explanation. Consequently, this criterion has been criticized and alternatives have been proposed."7' A second important point is that bias is also considered in the 25 percent criterion according to a somewhat complex statistical relationship,"01 but the overall system accuracy as defined earlier"1' is a fair approximation if the method has a true bias. i.e.. its mean is statistically different from the "know n concentration." A final point is that as the number of samples at a level increases, the standard deviation and CV should decrease. These considerations are important w hen evaluating passive monitor validation data, especially in those cases where manufacturers have stated that they have met the 25 percent accuracy criterion, ) For field comparisons of conventional and passive moni tors. different statistical tests are necessary. If passive moni tor values, for example, arc plotted against charcoal tube results (Y vs, X) and more than one concentration is sampled, one would expect an increase in X to cause an increase in Y. If the increase is linear, and the sample values lie on the regression line, the correlation coefficient (r) would have a value of one. If a change in X brings about an equal change in Y. then the slope would also have a value of one. If the individual values for the two devices are indeed equiva lent. the line should intercept the origin. Each of these rela tionships is expected within reason. The difference of the slop from one. the correlation coefficient from one. and the intercept from zero can and should be tested. In order to perform the regression analysis described above we must assume that X (active sampling data) is not subject to error. Of course, we know and can calculate under laboratory conditions the error of active sampling systems. There are at least three reasons why this error is olten overlooked. First, it is assumed that consideration of the error in X would only cause small differences in regression analysis results. Second, in addition to laboratory demon strated error, a range of errors introduced by varying envi ronmental conditions must be considered. While difficult to assess, these errors may have a profound effect on X. and indeed the error in the X measurement may be as great or greater than that for Y. Finally, if the error in X is to be considered the statistical tests are complex. Such tests hav e been discussed for biological problems;"6' however, the the ories apparently have not been applied to sampling and analytical methods even though their appropriateness has been recognized."9' applications From a historical vantage, one of the earliest reports of a "passive" monitor for evaluation of airborne contaminants was patented by Gordon and Lowe in 1927.1201 Their gas detector for carbon monoxide involved an "easily frangible vessel containing a solution of salts including palladium chloride, and a covering for said vessel of a light colored absorbent material." The principle involved breaking the vessel (a small ampule) and noting the subsequent color change of the solution as it reacted w ith the carbon monox ide on the light colored absorbent material. This type of semiquantitative device was certainly a forerunner of those that are available today, though it was undoubtedly affected by air velocity as a stagnant air layer was not employed. An extension of Gordon and Lowe's concepts in the late I960's provided the basis for Plantz et at. to develop a personal dosimeter for measuring hy drazine, unsymmetrical dimethylhvdrazine and monomethylhydrazine.'"" The reac tion of these compounds with a "colorimetric substance" (bindone) produced a purple color, the intensity of which was dependent on both concentration and duration of expo sure. Color standards then were used to estimate the concen tration as a function of the time the badge w as exposed to the contaminated air; consequently, the method was only semiquantitative. In considering sources of error the authors noied that the purple color would also be produced by all volatile bases that were tested, including ammonia, aliphatic amines, aniline and cigarette smoke. American Industrial Hygiene Association JOURNAL (43i 8 8? 3M 001771 609 Of interest in this review, however, are quantitative devices based on the principle of either gas or vapor diffu sion or permeation through a stagnant air layer. The first such device to be reported in the literature was described by Palmesand Gunnison in 1973.'22'Their device employed the principle of gas diffusion to determine airborne concentra tions of sulfur dioxide and will receive further consideration subsequently. To gain the best overview of the various appli cations of these concepts, it is probably best to proceed by considering first the inorganic and then organic gases and sapors. inorganic gases and vapors ammonia In 1978. Mazur et at. described the use of the Abcor GASBADGE to sample employee exposure to ammonia (this device currently is not marketed).121' The investigators replaced thecharcoal pad normally found in the GASBADGE with an acid impregnated absorption pad. Of three acids tested, phosphoric was most successful in providing the best approximation of theoretical concentrations. The> deter mined. however. that x olatile amines, specifically cyclohexylamine, could produce high readings, as high as 185 percent of the synthetic atmosphere. This led them to replace the glass fiberdraft shield on the front of the GASBADGE with a "charcoal impregnated glass fiber filter which had been pretreated with alcoholic KOH containing 0.1 percent sur factant." The charcoal served to adsorb amines as they dif fused into the dosimeter, while the KOH (aided by the wetting agent) eliminated irreversible ammonia adsorption by the charcoal, which would ha\e caused underestimation of the ambient concentration. Additional laboratory exper iments demonstrated that storage time of up to 47 days, prior to analysis, did not appear to ads ersely affect the results. More recently. DuPont has developed a commercially available system for the measurement of several airborne contaminants including ammonia. In 1981. Kring et at. described DuPont's PRO-TEKTM system for ammonia, nitro gen dioxide and sulfur dioxide sampling analysis using a col orimetric readout instrument.124' The ammonia badge relies on molecular diffusion of ammonia and subsequent chemical reaction with a solution of 0.3N boric acid and 0.03N sodium potassium tartarate (s/c). After exposure, the reagent pack is removed from the badge holder and analysis is initiated bypressing reagent "blisters" w hich are ad jacent to theabsorbing solution. This action causes the release of a modified Nessler's reagent and the subsequent development of a colored solution. Forammonia. maximum color intensity is developed at 425 nanometers. The absorbance of the sample is then compared against a standard curve based on Beer's Law. After determination of the precision of the analytical method and verification of the linear range of the color chemistry. laboratory testing was conducted to establish the operational range as well as precision and accuracy of the overall method (see Table I). The minumum and maximum limits of the sampling range were found to be 50 and 500 ppm-hours. respectively. For an eight-hour time weighted average, these values correspond to one-lourth and two and one-half limes the current ACGIH Threshold Limit Value of 25 parts per million.11''1 In considering sources of error, environmental effects including temperature (10 to 40 C). relative humidity (10 to 80 percent), pressure(730 to 790 mm Hg), and face velocity (2.5 to 125 cm sec) were included. Of the environmental factors evaluated, temperature and the concentration of the contaminant were identified as being responsible for 98 percent of the data variation. For ammo nia, a temperature correction factor of 0.6 percent per degree centigrade was suggested. Also investigated was the storage stability of both unexposed and exposed badges. Results indicated that refrigerated storage is necessary to extend the shelf life of unexposed badges. Once the badge is exposed to ammonia and before the reagents arc mixed, the badges can be stored for one (room temperature) to three (refrigerated) weeks without losing any absorbed contaminant. Once the reagents are mixed and color formation is started, the badge should be read within 90 minutes. Additional testing results conducted by DuPont arc shown in Table I.1261 carbon monoxide Shor and Anders, of the 3M Company, have described the 3M "direct-read diffusional monitor" for evaluation of exposures to carbon monoxide.*2,1 The principle involves the reaction of the carbon monoxide and an unreported reagent(s) to give a visible color change from pink to tan. Theoretically, "if any pink color is observable at the end of the exposure period, then the exposure was less than the one time-weighted-average of 400 ppm-hours." They also state that noting the time to the endpoint allows for calculation of the average concentration of carbon monoxide during the exposure period. The authors present summary results of laboratory evaluations using an infrared radiation device to establish "known"concentrations (see Table 1). chlorine Hardy et al. have described a personal chlorine monitor (REAL. Inc.) which employs the principle of permeation of the gaseous contaminant through a silicone membrane and into 10 mL of a fluorescein-bromide solution.*2?1 Colorimet ric techniques then can be applied to determine the chlorine concentration. The authors report a detection limit of 0.013 ppm chlorine for an eight-hour exposure, with a "working range" of 0.1 to 2.0 ppm. They also suggest that for shorter time periods, concentrations of up to five ppm can be deter mined. Other observations included effects of temperature, humidity, absorbent concentration. pH and response time, with all laboratory results presented graphically. Moleculon Research Corporation has recently introduced a chlorine monitoring device which relies on plastic film impregnated with liquid reagents.*291 Exposure of the badge to chlorine gas gives a visible, "blue-purple," color change. Optical transmission measurements, and comparison w ith a standard curve, can then provide quantitative exposures in ppm-hours. The manufacturer's summary results indicating effects of temperature, humidity , w ind v elocity and concen tration are reported as presenting an error at the 95 percen confidence level, which is "less than 15 percent " 3M 0017726)0 Am Ind Hyi Assoc J (43> Aufusi 19S? Chemical TABLE I Inorganic Gases and Vapors Laboratory Results D aimeter* Bias'* Precision' Range" (ppm) Reference Notes Ammonia Nitrogen Dioxide Sulfur Dioxide Hydrogen Sulfide Mercury CO DP DP GB ' GB DP DP MDA GB DP DP GB 3M 3M 3M 0.5 0.5 -3.2 -3.2 -0.9 -4 9 -1 -0.8 05 17 03 7.4 6.9 9.3 ' 21.7 7.5 8.7 41 13 8 58 7.5 15 3 9 2 47 20-50 20-48 ' 6-62 4-11 4-11 4-11 6-9 4 6-5 3 4-11 4-11 1 6-2.2 0.03-0 3mg/m' 0.05-0.2mg/mi 50-1830 24 26 23 1 38 24 36 1 38 24 1 32 33 27 T.C Hl r.rt H1 H1 K >:<; G 'DP-DuPont Pro-Tek Colorimetric System Badges. 3M=3M Company Monitor. MDA=MDA Scientific. GB=Abcor GASBADGE HSee text, equation (8) ' See text, equation (7) '`Some values are rounded to nearest whole number ''Bias consistently negative 'Results derived from Table III of McCammon ef a/1*"'' ''Results calculated from data provided in reference "This product is not currently marketed 'Bias could not be calculated from data given hydrogen sulfide In 1977. Tompkins and Goldsmith described the develop ment of the GASBADGE personal samplcr.(,, Although later work with this device focused on the collection of organics on an activated charcoal substrate, initial studies researched sampling of both organic and inorganic com pounds. (The GASBADGE for organic vapors is now mar keted by National Mine Safety Company, and those for inorganic sapors are not currently marketed.) Applications inv olving inorganic gases relied on collective elements of an "appropriate substrate impregnated with a chemical medium specific for the contaminant of interest." Based on 60 obser vations. the authors' statistical summary is presented in Table I. Eighty percent of the measurements were within 25 percent of the "true" value. Challenge concentrations were established in a "well-mixed" environmental test chamber and were measured with an "independent wetchemistry sampling train." Hardy and associates have described a permeation device (REAL. Inc.) which relies on the permeation of H^S through a dimethyl silicone membrancand subsequent reaction with a solution of 0.2N sodium hydroxide and EDTA.<30' The colored product (methylene blue) is measured spectrophotometrically and compared with a calibration curve to determine ppm-hours. A knowledge of exposure time then allows for the determination of average exposure over the measurement period. The authors note that a critical step in the development of such a device is the experimental deter mination of the permeation constant (see Equation 6). w hich involves calibration of each monitor by exposure to known concentrations of the contaminant. The results of this labo ratory research demonstrated a detection limit of 0.01 ppm for an eight-hour exposure, with a w orking range ol 0.1 to 20 ppm and a linear response up to 200 ppm. The working range corresponds to one one-hundredth to two times the current eight-hour TLV of 10 ppm.1251 Evaluations ol envi ronmental effects indicated that neither temperature, over the range of --3 to39C. nor humidity, from 0 to 99 percent relative, caused any significant variations in response of the device. Further research demonstrated a good response to high concentrations in less than one minute, adequate sam ple stability up to 10 days if EDTA is used in the absorbing solution, and negative and positive interferences, respec tively. from chlorine gas and nitrogen dioxide. Precision and bias were not reported. Anotherapproachforthedetermination of gaseous hydro gen sulfide has been reported by Gracdel and Franev.'311 Their research involved using a semiquantitativc method without a stagnant air layer and relying on the discoloration of lead-stabilized polyvinyl chloride (PN C). The technique involves the diffusion of gas in a poly mcr and. at high HjS levels, the detection rather than the measurement of toxic levels of H2S. Screening applications for low level exposures also are discussed. American Industrial Hygiene Association JOURNAL (43) 8 8? 3M 001773 eii mercury In 1977. McCammon and Woodfin of NIOSH reported the results of a laboratory evaluation of 3M's mercury vapor monitor/32' The monitor's operating principle involves molecular diffusion and deposition of mercury vapor on a gold substrate. The resulting change in electrical conductivity across the gold foil is related to the amount of mercury absorbed by the foil. Three other sampling methods, all of which involved the active movement of air. also were inves tigated and include the LASL tandum sampling tube, the hopcalite tube, and the iodine impregnated charcoal tube. For the passive monitor, precision and accuracy, the effects of face \elocity and temperature, and potential interferences were investigated. Concentrations of mercury sapor in an exposure chamber were monitored w ith an ultraviolet mer cury sapor meter, which in turn was calibrated by measure ments using the LASL method. To determine precision of the monitors. 12 dcsiccs ssere exposed to a test atmosphere. Results from three measurements of the test atmosphere using the LASL method gave an "expected" concentration of 0.056 milligrams of mercury per cubic meter of air (mg m3). ssith a standard desiation (SD) of 0.002 mg m3 and a coefficient of variation (CV) of 0.037. Precision and bias calculated for the data given in Table 111 of McCammon et al.i32) is summarised in Table I. A least squares regression analysis of the combined precision and accuracy results for the passive dosimeters versus the "known" concentrations gave a Y intercept of --0.004 mg m3 and a slope of 1.003. Tests for the effects of face velocities from 25 to 125 cm sec (50 to 250 fpm) did not appear to have any adverse effect on performance, while variation of temperature experimentally confirmed its effect on both the diffusion constant and the concentration of the contaminant. The potential effects of changes in relative humidity were not discussed by the authors. The other area of investigation in this studyinvolved potential interferences of chlorine, sulfur dioxide, and hydrogen sulfide. Concomitant and sequential exposure to mercury and chlorine, the latter at high levels (5.8 ppm), produced a negative bias, but a similar effect was observed with the UV meter and the LASL system. The authors suggested that the mercury and chlorine may have been reacting to form mercuric chloride which was not being measured by any of the systems. The interference effects of sulfur dioxide and hydrogen sulfide, although less than chlo rine. also were confirmed. In 1980. McCammon and his NIOSH and OSHA co workers conducted further tests of the four previously descri bed mercury vapor sa mpling a nd a na lyt ica I met hods/33' The design of this laboratory experiment involved inter comparison of four methods, and the findings suggested that the variability of the iodine charcoal tube method is signifi cantly different from that of the other three methods. It was observed that the other three methods, including the 3M passive dosimeter, exhibited good precision over the concen tration range of 0.05 to 0.2 mg m3 as shown in Table 1. Recently SKC. Inc., introduced a gas monitoring badge (produced by GM D. Inc.) w hich uses the principle of molec ular diffusion (without a stagnant air layer) to collect mer cury vapor/34' The sampling media is referred to as "Hydrar Sorbent." and "extensive" but unpublished field and labora tory testing are cited to show that "chlorine, moisture, etc., do not interfere" with measurements. HYDRAR is devel oped from a "manganese dioxide catalyst material similar to 'Hopcalite.' '' Quantitative determination is made by chemical desorption of the mercury and analysis with atomic absorption. nitrogen dioxide In 1976. Palmese/o/. reported the results of theirevaluation of a personal sampler for nitrogen dioxide (NOj)/33' This work was an extension of their earlier pioneering efforts in developing a personal sampler, employing the principle of gas diffusion, for sulfur dioxide/221 In their design, the sam pling device was a 1,3 cm (0.5 inches) acrylic tube. 7.1 cm (2.8 inches) long. At the "closed "end of the diffusion path (tube) were placed three stainless steel grids coated with triethano lamine (TEA). TEA was selected because: I) it captures NO^ efficiently. 2) it provides a stable sampling surface, and 3) it yields a chemical complex with NOa that is very stable over time. Subsequent analysis y ielded a colored complex u hose absorption was measured at 540 nanometers. Results were then compared against a standard curve which obey ed Beer's Law. The experimental evaluation of the NOa sampler involved chamber measurements compared with "known" values determined by the volume of NO2 introduced to the chamber or the weight loss of NO2 from a permeation tube. Although neither individual nor summary results were pre sented. graphical comparison of the data indicated a close agreement between the passive sampler results and theoreti cal concentrations. The effects of wind velocity and direction as well as stability overtime also were considered. In deter mining w ind effects, the uptake of water vapor, rather than NOa. was measured. The results indicated that there was an increase in average uptake with increased velocity and that the 45 degree incident angle gave the highest uptake (135 percent at 258 cm sec). Across all angles of exposure (0 to 180 degrees) the average uptake increased from 2 to 14 percent as the w ind velocity increased from 50 to 258 cm sec (100 to 516 fpm). Stability studies indicated that the badges could be used for months both after preparation and before exposure as well as after exposure and before analy sis. In the previously described studies with the GASBADGE. Tompkinsand Goldsmith also monitored for nitrogen diox ide.11' Their summary results for 82 observations showed accuracies and precisions as summarized in Table I. Eightyfive percent of the observ ations were w ithin 25 percent of the "true" value. A commercial model of the Palmes passive sampler has been marketed by MDA Scientific. Inc., and additional laboratory testing demonstrated a linear collection effi ciency for any given dose, i.e., concentration X time (see Table I)/3' The sampler exhibited a consistently negative bias as compared to concentration determinations using a continuous monitor and the NIOSH wet chemical method. As noted earlier, the DuPont PRO-TEK system includes a badge for nitrogen dioxide. Laboratory test results of this device are shown in Table l/3,1 12 Am Ind Hyf Assoc J (43) August 19S? 3M 001774 sulfur dioxide As noted previously, pioneering work in 1973 on the design of a sampling device which relics solely on diffusion of gaseous contaminants through a stagnant air layer is attrib uted to Palmes and Gunnison.122' Their initial studies involved experimental work on different tube lengths. The collecting medium was a complex of mercuric chloride and the final analysis involved colorimetric determination. The studies demonstrated that, except for very short diffusion paths (tube lengths), the diffusion monitor satisfactorily duplicated the results obtained by both wet chemical and conductrimetric measurements. Although this study did not go into the ramifications of environmental effects and inter ferences. it should be recognized as an important step in developing a new industrial hygiene technology. The last of the three inorganic gases looked at by Tompkins and Goldsmith in their studies of the GAS BADGE was sulfur dioxide.'1' Summary data for 23 observations on this gas arc included in Table I. One hundred percent of the observations were within 25 percent of the "true" values. The results from DuPont's tests on their sulfur dioxide badge also are shown in Table l/38' organic gases and vapors Most passive dosimeters designed to sample for organic gases and vapors use activated charcoal as the adsorbing medium. As we know from its extensive use in active sys tems. activated charcoal has an affinity for a wide range of organic compounds. Consequently, the discussion on the applications of passive dosimeters for the measurement of organic gases and vapors first will focus on devices using activated charcoal and then will turn to specific organics which rely on other collecting media. Chemical TABLE II Organic Gases and Vapors Laboratory Results Dosimeter* Bias1' Precisionr Range11 (ppm) Reference Notes Carbon Tetrachloride Toluene Formaldehyde Benzene Ethylene Oxide Halothane Enflurane Acrylonitrile Hexane Vinyl Chloride Methyl Chloroform Trichloroethyiene DPA DPA NMS DP DPB NMS NMS 3M 3M 3M NMS NMS R 3M DPA NMS 3M DPA NMS 0.4 0.3 -1.7 1.5 3.3 -4 1 18 -1.4 3 -2.8 0.03 -1.2 -0.2 -5.9 -10.6 2.4 -3.9 -6.9 -0.5 4.4 5.25 1.7 63 4.7 1.7 17 3.2 7.4 4.8 8.7 2 3.7 4.7 4.5 2 7.6 7.7 1.9 3-18 57 228 12-47 0.2-4.2 3-24 08-5.4 13-13.5 300 0.5-20 0.5-20 0.7-19 10-37 1.5-14 160 840 160-840 15-65 20-200 20-200 15-67 10 45 13 FJ 55 46 13 FJ 1 EGJ 53 4 44 4 44 4 39 13 FJ 2 HJ 48 1 J 48 fc IJ 12 FJ 48 U 48 1J 12 FJ A3M = 3M Company Organic Vapor Monitor. DPAand DPB = DuPont PRO-TEK G-AA and G*BB Organic Vapor Badges. NMS = National Mine Safety GASBAOGE. R = Real, Inc. MINIMONITOR, DP = DuPont PRO-TEK System Colorimetric Badges See text, equation |8) cSee text, equation (7) "Some values are rounded to nearest whole number KSmaii sample sire FKnowns were calculated using charcoal tubes with critical orifices. This could affect the bias measure `'Preliminary results "Permeation dosimeter 'Bias consistently negative JResults calculated from data provided in reference American Industrial Hygiene Association JOURNAL (43l S SP 3M 001775 613 activated charcoal devices As of March, 1982. there were four manufacturers of passive dosimeters which rely on diffusion and subsequent adsorp tion on to activated charcoal: National Mine Service Com pany (GASBADGE). 3M Company (Organic Vapor Moni tor), DuPont Company (PRO-TEK, G-AA and G-BB Organic Vapor Air Monitoring Badge), and the Mine Safety Appliance Company (Vaporgard Badge). In 1977, Tompkins and Goldsmith described the first commercial passive dosimeter for monitoring organic vapors.111 The GASBADGE relied on molecular diffusion of the vapor into the badge and subsequent adsorption onto activated charcoal. The authors developed the theoretical principles of the badge's operation, and discussed sensitiv ity to temperature and pressure, face velocity effects, and response time. Preliminary results showing the badges response to benzene, ethyl acetate, methyl ethyl ketone, and 'tyrene also were presented and were described as "very tiicouraging" (see Table II). In the same year. Silverstein reported results of laboratory , id field tcstmgoftheGASBADGE for acrylonitrile.'51'The suits of the exposure of 33 badges to known concentraons in the laboratory are presented in Table 11. The results idicate the acceptability of the GASBADGE for measureents of acrylonitrile over the range of 0.75 to 19 ppm. The , me of exposure in the laboratory was not given, however, eld measurements did cover periods of up to seven hours, '.though temperature and relative humidity ranges were ported, data analysis to determine the effects of these iriables was not presented. Silverstein also looked at csorption efficiencies and determined that the best results >1 percent) were obtained with four mL of two percent etone in carbon disulfide. In 1978, Bamberger etal. conducted a series of laboratory sts toevaluate the GASBADGE.'1-''Theirapproach involved te generation of known concentrations of solvents and tbsequent evaluation with charcoal tubes (active sampling) nd the passive dosimeter. To evaluate the applicability of he dosimeter over a w ide range of compounds, the investiation included seven different organic compounds each jpresentative of a different functional group. Included in this study were: benzene (aromatic), n-butanol (alcohol), n-butyl acetate (ester), isooctane (alkane), methyl chloro form (halogenated alkane), methyl isobutyl ketone (ketone), nd trichloroethylene (halogenated alkene). The diffusion ^efficient (D) used was that supplied by the badge manufacirer, except in the case of isooctane which was reported as aving an unknown coefficient. Computations involving tnis compound relied on the coefficient lor n-octane. A v ariety of experiments was conducted to look at a wide range of questions. Their findings corroborated dosimeter con cerns similar to those of active systems usingcharcoal tubes, minimum and maximum loadings are important, postample contamination and loss can occur if the exposed bsorbent is not adequately sealed, percent recovery for lixtures is consistent with percent recoveries for single compounds, and differences in charcoal lots can give differ* nt results. Other tests confirmed the need for some air movement across the badge face and the lack of ellect ol temperature changes over a small range (11 '('). The results of the simultaneous sampling with the badges and the char coal tubes indicated that the badges had a consistent nega tive bias. The authors suggested that, because the results were so reproducible, corrections for adsorption desorption efficiencies less than 100 percent can be accomplished just as is done for charcoal tube data. In 1979, Hirayamaand Ikeda evaluated the application of the GASBADGE for monitoring exposures to mixed sol vents.<l0' Their research involved different preparations of activated carbon "felt" in place of the supplied collection medium and exposure to mixtures of n-hexane. ethyl acetate and toluene. Summary (graphical) data indicated that the amounts of contaminant absorbed by the dosimeter were proportional to both the vapor concentrations and time of exposure. Halliday and Anderson reported on the use ol ihe GASBADGE in monitoring lor halothane.'111 Six observa tions indicated a range of measurements from minus nine to plus ten percent of the test atmospheres. I'nfortunately. the authors did not report their procedure for determining the concentration of halothane in the test atmospheres. In 1981. Evans and Horstman reported evaluations of desorption efficiencies of charcoal tubes and the GASBADGE for styrene.l4'i> For liquid dosing they found the dosimeter to be similar to the tube, while for vapor dosing the badge was superior. The authors suggested that the differences in results may. have been related to the use of coconut shell carbon in the tubes and petroleum derived carbon in the badge. They did not explain why this difference would affect one method of dosing and not the other. In 1980. Anders and Mullins of the 3M Company pre sented results comparing the 3 M passive monitor with char coal tubes in sampling for mixtures of organic compounds.'u> The laboratory tests included a binary mixture of toluene and methyl ethyl ketone; a tertiary mixture of benzene, toluene and xylene: and complex mixtures of unleaded and leaded gasoline containing various alcohols. Although the investigators cited "excellent" precision and accuracy for the diffusional monitor, sample sizes were small, and the com plicated study design and lack of raw data preclude the determination of precision and bias statistics. Mazur and his co-workers'141 conducted side-by-side laboratory and field tests with charcoal tubes and 3M organic vapor monitors to measure concentrations of halo thane (2-bromo-2-chloro-l.l.l-trifluoroethane) and enflurane (2-chloro-l.1.2-trifluoroethyl difluoromethyI ether). The results of the laboratory studies are presented in Table II and support the authors' conclusions that the dosimeters are a reliable method for the collection of enflurune and halothane. In 1980, Lautenberger el at. described DuPont's passive monitor for organic vapors.'111 Each charcoal strip in the PRO-TEK G-AA Organic Vapor Badge contains approxi mately 300 mg of coconut-based activated charcoal impreg nated in an inert polymer. A dual sampling rate of approxi- i Am ln<J Hyg Assoc J (A3) August 1982 3M 001776 mutely 50 or 100 ml min is determined by iht- removal ol' oik- or both o!'the dosimeter's protective covers. One aspect ol their research involved experimental determination ol the diltusion coefficient ol' several gases and vapors. They reported that values calculated by I ugg'" vvere within 10 percent of their experimentally determined diffusion coeffi cient values. Preliminary experimental results were used to discuss lace velocity effects, ranee and sensjtiv ity. maximum and minimum sampling times, vapor retention, storage sta bility, desorption efficiency, and overall badge efficiency, [he overall accuracy determinations vvere limited to four observations at each of two concentrations of carbon tetrachloride (see Table 111. However, the presentation of raw data, as well as an explanation of the statistical tests applied, is most useful. This same detail of information is also lound in DuPont's validation reports lor toluene and ben/ene (see Table III.'1' I n the ben/ene report. DuPont also describes its PRO-TE K (i-BB badge. This badge has a backup section ol charcoal, which serves the same purpose as the second section in a charcoal tube. /... to aid in determining it the sampler has been overloaded. 1 he 3M Company also marketsan Organic Vapor Monitor with a backup section.'1'1 In studies of the measurement of waste anesthetic gases with passive dosimeters. Jonas cl at. evaluated the GASBADGE, 3M Organic Vapor Monitor and DuPont Pro-Tek in measuringennurane. '" Unfortunately, the badges were not identified in the presentation ol the resultsalthough interpretation of the reported samplergeomeiry would indi cate that A was the DuPont badge. B was the 3M badge.and C vvas the GASBADGE. The results of their laboratorystudies indicated that badge B had the lowest coefficients of variation (CV was not calculated as described in this text)as compared to concentrations determined by infrared analy sis. Badge A had a low CV (9 percent) at 5 ppm and a much higher value (CV=34 percent) at 20 ppm. The C badge had consistently high CV's ranging from 23 to 30 percent. It should also be noted that the charcoal tube CV's ranged from II to 27 percent, that desorption efficiencies for the badges ranged from 0.81 to 1.17. and that IR analyses of tank concentrations were constantly lower than expected. If badge B was the 3 M device, the results of Jonas ei al. support those reported by Ma/ureru/.'141 Further testing of badges A and C seems necessary, however, to confirm their seemingly low precisions. Mazur and his coworkers conducted additional tests comparing the 3M and DuPont badges against charcoal tubes.'*8' Methyl chloroform and trichloroethy lene, two sol vents widely used in vapor degreasing operations, were sampled. In the laboratory phase of the study, the badges and charcoal tubes were exposed to chamber concentrations over the range of 160 to 840 ppm of methyl chloroform and from 20 to 200 ppm of trichloroethylene. Exposure times varied from two to six hours for methyl chloroform and trom four to six hours for trichloroethylene. The laboratorywork indicated that the percent recoveries of the various doses (concentration X time) were in good agreement except for one exposure of the 3M badge which involved a live hour An^r-cjn indu'.tnjl Hygiene Association JOURNAL (43\ 8 82 exposure at 700 ppm. The authors noted that this exposure ol 3500 ppm-hours exceeded the upper exposure limit pro vided by the manufacturer. The overall mean recovery value for each type of sampler was used to correct all subsequent held data. In addition to the recovery measurements, the laboratory phase ol this study also involved determination of storage stability. The authors found no significant losses of methy 1 chloroform or trichloroethylene from either badge following storage of exposed badges for up to three weeks. In 1978. West and Reis/ner reported on the field tests of the MI.NTMONITORTM (REAL. Inc.) permeation personal monitor for vinyl chloride.''' This monitor was a modified version of one previously described by Nelms et aThe collecting medium was activated charcoal, but rather than relying on molecular diffusion, the badge design involved a polymeric membrane and the permeation of vinyl chloride through the membrane and adsorption onto the charcoal. Initial laboratory calibration vvas used to determine the permeation constant ol the device. 1 aboratory results indi cated good accuracies as summarized in Table II. During the same period that Tompkins and Goldsmith'1' were describing the GASBADGE. Bailey and HollingdaleSmith of Great Britain were presenting their ideas for a personal passive sampler for organic gases and vapors. "1. Their design involved the use of either one of two types of membrane and subsequent adsorption onto activated char coal. They found two membranes to be satisfactory: one of thin silicone rubber w hich acted a' a permeation barrier, and the second a porous polypropylene film which allowed lor molecular diffusion of the gas and vapor. They conducted laboratory tests using carbon tetrachloride, styrene and dichlorodifluormethane. Their test results do mix some ter minology. <*... permeation rates for both the permeation dev ice and the diffusional dev ice. but did prov ide an early demonstration of the feasibility of such a dev ice for monitor ing certain organics. The device, the Porton Diltusion Sampler, seems to see its greatest use in Great Britain. acrylonitrile One of the newest applications of passive dosimetry inv olves the use of a porous polymer (Porapak N) as the collecting surface with subsequent thermal desorption and gas chro matographic analysis. Benson and Boyce have described such a device and its utility in sampling for aery lonitrile.'111 Laboratory testing for acrylonitrile inv olved comparison of the dosimeter values with concentrations measured on a gas chromatograph. Initial experimentation indicated that the dosimeter can be used for aery lonitrile concentrations in the range of 4 ppm. but at concentrations of 2 ppm a 40 percent error is reported. aniline In addition to activated charcoal, another widely used adsorbent medium is silica gel. To study the utility of this material. Campbell and Konzen constructed passive dosim eters from glass culture tubes (1.05 cm inside diameter) with 40 60 mesh silica gel as the collecting surface.11"' Laboratory testing involved exposure of the dosimeter to aniline, with exposure concentrations determined by gas chromato graphic analysis of ethanol gas scrubbers Three different 3H 001777 615 size (length) dosimeters were evaluated, with the best results obtained with the intermediate length tube(L = 3.0cm: A; L = 0.3 cm). The authors present raw data and clearly described their statistical techniques. ethylene oxide Mullins and Anders have recently described the 3M diffusional monitor for sampling ethylene oxide in air.*53' In this badge the collecting surface is described as a "chemically impregnated charcoal surface, (where) a reaction occurs producing a stable compound with a vapor pressure sub stantially lower than the parent compound." The authors present statistically summarized data describing the linearity and capacity of the monitor, the recovery of absorbed ethyl ene oxide, environmental effects, sample stability, and the effects of potential interferences. Precision and bias are presented in Table II. formaldehyde Rodriguez et at. have described another 3M diffusional monitor for sampling formaldehyde.*54' In this diffusional monitor, the collecting surface is an "impregnated sorbent" which can then be desorbed in situ with water and the concentration of formaldehyde determined colorimetrically. Laboratory evaluation first involved determination of recov ery coefficients, which at eight ppm-hours (19.5 micro grams) were found to be 1.00 0.04 over six tests. The next step involved determination of the dosimeter's "sampling rate" (DA1 L) by exposing the dosimeters to "known" con centrations of formaldehyde as generated by a permeation tube. The effect of relative humidity on the sampling rate also was investigated, and evaluation of the data did not indicate any statistically significant differences between the rates at 50 and 85 percent relative humidity. The study protocol then involved simultaneous exposures of impingers (modified chromotropic acid method) and dosimeters. The authors concluded that "the measured values by both methods lie within 25 percent of the expected response" and that "less variation is observed in the monitors than in the impingers." However, neither precision nor bias were reported. The authors also investigated effects of storage and determined that at elevated temperatures (38 DC) losses up to 11 percent occurred after one week, however no signif icant loss was seen for samples stored at 23 C. The authors briefly discussed the potential for a negative interference from phenol and described the use of modified calibration curves to address this problem. DuPont's PRO-TEK series of colorimetric Air Monitor ing Badges, includes a badge for formaldehyde. The collec tion principle involves a chromotropic acid-sulfuric acid reaction. Laboratory evaluation(42samples)of thedeviceat seven exposure levels revealed results as shown in Table II.*55, Additional studies also were conducted on tempera ture and storageeffects. The raw data and statistical analysis procedures arc presented. Kriesel*56' has described a new passive dosimeter for form aldehyde which is a modified version of the Palmes tube. At the present time experimental data concerning this device are not available. phosgene Matherne er al. have recently described the GMD. Inc. "passive dosimeter" which pros ides a semiquantitative mea surement of phosgene exposure.*57' The badge involves direct contact between the contaminated air and a chemi cally impregnated tape and therefore does not rely on a stagnant air layer. The treated paper stain intensity is reported to be logarithmically proportional to the phosgene dose overa range of 2 to 100 ppm-minutes. For quantitative measurements the badges can be read colorimetrically. other methods Hill and Fraser have described the use of commercial detec tor tubes modified to act as passive dosimeters.,5'" In then research, common length-of-stain detector tubes were modi fied by cutting off the conical end of the tube and remov mg some of the indicator column material. This leaves an or it ice with a cross-sectional area equal to that of the inside ol the tube and a path length determined by the distance from the end of the tube, to the beginning of the indicator material. One would expect, however, that as the sorbent material becomes exposed, i.e., the length of stain increases, the diffusion path length will also increase, thereby changing the sampling rate. Their evaluation of these devices involved separate laboratory exposures to toluene, ethanol and tsopropanol. The results of their work, although presented only in graphical summary, demonstrate the potential for the use of modified commercial detector tubes as passiv e dosimeters. field validation Relatively few studies have been published in which passive dosimeters have been compared side by side with charcoal tubes or other conventional sampling methods under actual field conditions. For inorganic compounds only two stud ies. involving nitrogen dioxide*59' and chlorine.1'' have been identified. For organic compounds, eight stud ies*2'9 39 4449 51-60 61) have involved field comparisons, with the number of compounds per study ranging from one to 22. In three of these studies, statistical analyses of data were not presented and cannot be performed because of small sample size or insufficient presentation of data. Jones ei al.(b9) conducted a field evaluation for NO_> tn a salt mine, which contained diesel equipment as the NO_> source. At each of 16 different fixed area locations, two Palmes dosimeters and two TEA tubes with pumps were used to sample the atmosphere. The active sampling (pump) method gave a coefficient of variation (CV) of 8.7 percent while for passive tubes the CV was 5.8 percent. Regression analy sis (w here the active system was the X variable) of their data gives a correlation coefficient of 0.69. a slope ol 0.59. and an intercept of 2.05 ppm for data over the range of 3.7 to 5.5 ppm as sampled by the active method. This indicates that the dosimeter gave consistently higher readings which is reflected in the means for the two methods. 4.51 ppm lot passive and 4.14 ppm for active sampling. The authors did not report wind velocities, but low velocities, as would be expected with area samples, should have caused passise 616 Am Ind Hy Assoc J (431 Ajcu1* 3M 001778 values to be low as compared to active values; this was not the case. On the other hand, high face velocities could have led to the observed positive (passive versus active) bias. Hardy ei at.'m reported the raw data results from a field evaluation of a permeation chlorine monitor (REAL. Inc). Thirteen comparisons were made in which the results from a battery operated pump and an impinger sampler were com pared with the measurements from either two or three per meation samplers. To further evaluate their results, the investigators performed a regression analysis using their permeation sample means for each comparison as the dependent variable. The results are quite good with a corre lation coefficient of 0.95. a regression slope of 0.85, and an intercept of 0.15 over a range of 0.05 to I. I ppm as detected by the impinger. It should be noted that with five impinger samples of less than 0.1 ppm. the corresponding permeation devices detected considerably higher concentrations (0.16 to 0.4 ppm). Silverstein139' reported field results for acrylonitrile moni toring using 18 paired samples of GASBADGE passive monitors and active systems (charcoal tubes and pumps) ov er a range of 0.8 to 5.8 ppm as determined by the active method. The differences in results using the active system as a reference ranged from --0.7 to 1.5 ppm. The difference in means. 2.18 for the passive versus 2.75 for the active system, was 25 percent. Further data were not presented. West and Rciszner reported five sets of field results for vinyl chloride sampled w ith permeation dosimeters (REAL, Inc.) and charcoal tubes.u> Further interpretation of their results is presented in Table III. In each case data for the active system are the X values. Four of the correlation coefficient values are very near one. reflecting good correla tion; however, the slopes show quite a large degree of varia bility (0.69 to 1.51). indicating that badge values may fall either well below or above charcoal tube values. In those cases w here the slopes were less than 1.0. very high humidi ties (67 to 91 percent) were reported by the authors. This factor may have interfered with permeation, although the authors reported that humidity had no effect in laboratory validations: hence the variation in slope remains unex plained. The authors noted that the overall field results showed that the badges had a slight positive bias. Hickey and Bishop exposed 78 pairs of side-by-side char coal tubes and 5M Organic Vapor Monitors to complex mixtures of organic chemicals in tire manufacturing operations.'9' Generally the sampling period ranged from three to TABLE III Regression Analysis of Vinyl Chloride Field Data121 N Range (ppm) r Slope Y-Intercept 7 002-1 0 99 1 19 8 0.08-1 8 0 99 0.69 12 002-6.9 1.0 1.31 39 0 05-1.8 0.82 0.81 24 1 48-16 7 0 96 1.08 0 03 005 0.01 0.11 0 43 five hours, and most observations consisted of one monit< and the time weighted average concentration front tw sequentially exposed charcoal tubes. Sixty-four of the se were personal samples, while the remaining 14 were are samples. The samples were collected in two separate plan (30 sample pairs in one plant and 48 in the other). Of the 2 organics potentially available for analysis. 10 were detecte over a sufficiently wide range of concentrations to allow fc appropriate statistical analysis by linear regression. Th results are interesting in that in the first plant. 9 of the I organics measured by the dosimeters showed higher vape concentrations as compared to the charcoal tubes, while i the second plant only three substances had a regression slop greater than one. The combined data for both plants did m indicate that the passive system was consistently hiase when compared to the active system. The authors did poir out that generally the Y-intercepts (Y=passive dosimeu data) were slightly negative, a finding w hich may indicate lack of sensitivity on the part of the dosimeters at lo concentrations. For the remaining 12 compounds, paire t-tests revealed no significant difference between the cha coal tube and passive monitor means at the 95 percei confidence level. The use of t-tests to analyze such data hi been questioned since the means of the two methods may t very similar but the components of paired values can I considerably different.,19) This condition can only be reveale through regression analyses. In 1980. Mazur el at.'44' reported limited field data ft halothane and enflurane measurements using both 3f Organic Vapor Monitors (OVM) and an active systei (charcoal tubes and pumps). For halothane three paire samples were reported. The mean concentration for tf active system was 2.01 ppm w hile 1.9 ppm was reported ft the OVM, a difference (relative to the active system) of fiv percent. Only one data pair was reported for enflurane: 0.4 ppm for the OVM and 0.52 ppm for the active systen Obviously, more data are needed to draw conclusior regarding a comparison of the two methods for these agent. A second study by Mazur el at'4*' reported field cornpar isons of passive dosimeters and active systems (pumps and charcoal tubes) in sampling for trichloroethy lene (TCE) anc methylchloroform (MC). Both DuPont PRO-TEK and 3M Organic Vapor Monitors were used for the passive systems Personal samples included exposure of one each of all thrt monitors. Area sample results involved three av erage value one was the average of three charcoal tubes, and the oiht two. the average of two of each type of dosimeter. For M( 11 personal and 7 area data points collected over time per ods of I to 5 hours at 15 to 21 C and 35 to 40 percent relativ humidity were reported. For TCE. 22 personal and 7 are data points collected over periods of a bout I to 6 hours at I to 24 C and 30 percent relative humidity were reported, regression analysis in which the charcoal tubes were tf independent variable was reported by the authors. In each < the following data sets the presented values involve TC personal and stationary sampling followed by MC person, and stationary sampling. For the DuPont badge, recressic American Industrial Hygiene Association JOURNAL (43) S/82 3M 001779 6 slopes of 1.0, 0.99. 0.99. and 0.98. and correlation coeffi cients of 0.98.0.98,0.94. and 0.94 were obtained. For the 3M badge, regression slopes of 1.08. 1.06, 1.07. and 0.90. and correlation coefficients of 0.98, 0.98. 0.98. and 0.90 were determined. These values appear to be quite good: however, the authors did not report if they tested the statistical signifi cance of these values. They also did not report average of face velocities associated with stationary samples. Evans ei al.{W of Great Britain reported field validation data for the Porton diffusion de\ ice while measuring methyl ethyl ketone. In this case the conventional sampler was a pump and a cassette fitted with a charcoal cloth similar to that used in the Porton device. A regression analysis per formed with their data showed good correlation (0.9) and good slope (0.9): however, the intercept value (4.69) indi cated that, at low concentrations, the "home-made" de\ ice ga\e low'er values than those determined with the conven tional monitor. Concentrations reported for the conven tional device ranged from 11 to 189 ppm. Benson and Boyce*S11 field tested the Monsanto Poropak .s' de\ ice in Great Britain. Conventional samplers consisted of pumps and Poropak N polymer tubes. Sixty-fi\e pairs of samples were obtained, and the range of acrylonitrile mea sured by the tubes was 0.1.1 to 21.65 ppm. Regression analys is of their data indicates only fair correlation (0.63). a low dope (0.46). and a negative intercept (--2.06). These values appear to result from the apparent inability of the passive lev ice to accurately detect concentrations less than 0.5 ppm. Also, comparisons between values over the lower half of .oncentrations sampled showed considerable scatter. The final field study to be discussed suggests perhaps the most serious discrepancies resulting from use of charcoal passive dosimeters.*611 This study was performed by NIOSH personnel in conjunction with industry-wide studies of the iry-cleaning, screen printing, and boat manufacturing indusries. and also included one viscose rayon and one cello phane plant. Carbon disulfide, perchloroethylene, toluene, mcthylisobutyl ketone (MIBK). styrene, and acetone were .amplcd using the 3M OVM. the GASBADGE. and active ystems with charcoal tubes. The presentation of the study design is not clear, but it appears that area samples involved all three devices while personal samples involved charcoal tubes and only one of either passive device. In that this study nvolves six compounds in 64 plants, the volume of data is juite large. In addition to regression analysis, paired t-tests md Wilcox signed rank tests w ere performed by the authors o determine equivalence of data sets. As noted earlier he use of t-tests for determination of equivalence has been questioned.'191 Table IV shows the primary results of this study. As can be seen, the range of correlation coefficients (r) for most com pounds was quite large. Although 12 plants were sur veyed for toluene and MIBK, the data .ere grouped ogether. and therefore ranges of the correlation coefficients ould not be determined. For carbon disulfide, one plant as sur\ eyed w ith the OVM and GASBADGE, and one was urveyed with the GASBADGE only. For the ranges of r .ported in Table IV. the upper values are quite acceptable. with the except ion of carbon disulfide using the GAS BADGE. How;e\er. the correlation coefficient for carbon disulfide using the OVM was 0.95. The correlation data can be summed up as being extremely variable. Table IV also reveals that concentration had an effect on the correlation coefficient for three of the compounds, though this was true for both monitors only when measuring acetone concentra tions. Regression slopes were as variable as the correlation coefficients. The authors tested the slopes to see if they were significantly different from zero, and for acetone and carbon disulfide a difference could not be demonstrated for several of their data sets. This indicated that there was no relation ship between the results obtained w ith the active system and those obtained with the passive dosimeter. For other com pounds. it would have been useful to test the difference ofihc slope from one. which if not significantly different would indicate agreement of the two methods. In tests of equivalence of data sets, the authors noted that for all plant data combined, only toluene showed equality, and this for the charcoal lube-GASBADGE (CT-GB) com parison. However, when results from individual plants are used, the comparison outcomes are quite variable. For perchloroethylene. equality was reported for one of three CT-BG comparisons and for one of two CT-3M sets. For styrene, two of six CT-GB and no CT-3M comparisons showed equality. Foracetone, three of five CT-3M compari sons and one of six CT-GB data sets showed equality. In addition. GB-OVM comparisons showed equality in 6 of 17 comparisons. It is obvious that repeatability was not demon strated in this study. Whether the problem involves the dosimeters, investigative or laboratory techniques, and or env ironmental conditions cannot be determined. In the only other field study of more than one plant, Hickey and Bishop*91 also reported some problems with the consistency of observations. These limited results clearly demonstrate the need for additional field studies of passive dosimeters as compared with standard monitoring techniques. discussion Passive dosimetry (monitoring) is a rapidly developing tech nology as witnessed by the proliferation of devices and applications since Palmes and Gunnison introduced their concepts just under ten years ago.*221 The latest entry into the field comes from the MSA Company and involves an adap tation of their length-of-stain direct reading tubes for inor ganic gases*621 which incorporates the application of molecu lar diffusion and a chemically impregnated paper as the sampling medium. Although research results are not availa ble. as a first approximation one might assume that these devices have precisions and biases similar to those of con ventional detector tubes. For any new technology to be accepted and used by prac ticing professionals, the development of a body of knowl edge demonstrating efficacy is necessary. With environmen tal monitoring techniques, the determination of the efficacy usually starts in the laboratory and culminates in the field. In the case of passive monitors, a body of know ledge based on laboratory testing is rapidly being developed. Of the vari- 3M 001780s Am Inti Hy$ Assoc J (43) August 19T TABLE IV Major Results of a Field Study for Organic Vapors'*1' Concentration Substance Comparison Overall r Range of r Dependency Perchloroethylene Styrene Acetone Toluene MI8K CS. CT-GB CT 3M CT-GB CT-3M CT-GB CT-3M CT GB CT-3M CT-GB CT-3M CT-GB CT-3M 0 62 0.86 0 82 0 76 0 38 045 0 80 0 91 088 0 79 0 30 0 95 0 62-0.99 0.84-0.94 0.65-0.97 0.48-0.86 0 36-0 86 0.25-0.83 NA NA NA NA 0.03 0 38 NA Yes Yes Yes Yes able* that have been studied, threeappearto uniquely affect a diffusion monitors accuracy in measuring airborne concen trations of gases or vapors. The most important factor appears to be determination of the contaminants" diffusion coefficient (or the sampling rate when the dosimeter's geometry is also considered), the wind velocity at the dosimeter face, and the relative humidity of the sampled air. As discussed earliei, there are also a variety of potential sources of error, such as interfering contaminants, sorbent capacity a nd problems associated with analytical determina tions. which are common to both passive and active mea surement techniques. I a bora lory determination of sampling rates (DA L.Jfora specific monitor and a specific contaminant are important and are being provided by several dosimeter manufacturers for an ever increasing number of compounds. Once an appropriate sampling rate has been determined, corrections lor field use. specifically for temperature variations, can be made. The main problem would involve situations where the env ironmcntal temperature fluctuated widely (more than 25 C) and went unnoticed, a very unlikely condition. The research on effects of lace velocities demonstrate that few problems should be encountered where dosimeters are worn by workers as personal monitoring dev ices. Their use as area monitors should be carefully evaluated to ensure that stagnant atmospheres (velocities less than 7.5 cm sec)arc not involved. High wind vclocities(at least what would normally be encountered in the workplace) or wind direction do not appear to have adverse effects on dosime ters with wind screens. Of greatest concern as a result of reviewing the literature on laboratory testing of passive dosimeters is not the results but rather the thoroughness of their presentation. As dem onstrated in Tables 1 and II. where statistical analyses are presented by researchers, or where sufficient data are pre sented to allow the reader todetermine biasand precision, the results are very encouraging ITnlortunatcly the presentation of experimental design, as well as sufficient data and or statistical analvses, ate often lacking. This is true for some Ampncdf' mdustiid' Hyp^ne Ampliation JOURNAL (43 P E7 indiv idual rcsearchersas well as for several manufacturers o the devices, especially those for inorganic compounds. If on< recommendation regarding laboratory testing is made, i would be that those researchers involved in the evaluation o passive dosimeters in the laboratory take the time to repor the conditions of their experiments, especially equipmen used and procedures for determining "known" concentra tions. and as much detail about their results as possible. I summarized data are presented, the author should presen the known concentration at each level, where levels an determined by concentration and time, the number of obser vat ions made with passive dosimeters, and the average valui and standard deviation of the results. Statistical analyses again at each level tested, should involve determination o the coefficient ol variation (precision) and the bias a described in equations (7) and (X). respectively. Once iht evaluations are made at the various test levels, the delerrm nation of a pooled precision and bias is appropriate. Ir addition to these measurements, researchers may alsi choose to present an overall system accuracy . To develop ; better understanding of appropriate statistical technique and their application to passive dosimetry, a review o Lautcnbcrgcr el a/, is recommended.11'' For most active monitoring systems used in industria hygiene the random sampling error is usually associatct with the pump and is traditionally set at 5 percent.1"" It many cases, especially for the measurement of organii vapors, the analytical procedures and consequently thei associated errors are equivalent lor both passive and activi systems. Nevertheless, both systems have random error consequently, one should not expect perfect agreement o the results of comparisons obtained under field test condi tions. Another factor complicating the evaluation ol fielc results is the greatly increased possibility for the introduc tion of operator, or systematic, errors. Since active system: require mechanical pumps, the potential for operator erroi would seem to be greater than for passive systems. Overall, it is apparent that existing field observations comparing passive dosimeters with standard monitoring methods are highly varied. While some studies demonstrate good correlation and slope.'0others show only good correlation.'"" or are extremely varied for both categories.'*'1 Collectively, these references neither support nor refute the use of passive dosimeters. Certainly environmental factor: affect active systems as well as passive systems. In theory, z case can be made that environmental factors (wind anc humidity) affect passive systems to the greatest extent, w hilt temperature and pressure variations most greatly a I fee: active systems. On the other hand one can also state that poor experimental quality control may affect such factors acontamination, time measurement error, and analytics error. Of course, chemical interferences may aflect both systems. As with laboratory experimentation, recommendation: regarding the field testing of passive dosimeters involve a plea lor better reporting ol both lield conditions and result' of analysis. First, for both personal and area monitoring, the estimation and or measurement of face velocity is impor- 3H 001781 619 tanl. Of equal importance is the reporting of airborne con taminants other than the one(s) of interest and environmental variables including temperature, pressure, and relative humidity along with information as to their variation over the period of observation. Again, if raw data cannot be presented, the reported results for each level tested (X value as determined by the standard method) should include the number of observations made with passive dosimeters, and their associated mean and coefficient of variation. Statistical evaluations also should include a regression analysis of the data as outlined earlier. Undoubtedly, additional research is needed on the effect of not considering the error associated with the supposedly independent (X) variable. In summary, passive dosimeters show great promise as an important tool. The results presented in Tables I and II indicate that the precisions of the dosimeters are essentially equivalent to conventional techniques and in many cases the additional five percent error associated with mechanical lumps makes passive dosimeter systems e\en more attracive. This, coupled with their ease of use, lack of required naintcnance. acceptance by workers due to light weight, and mnecessary calibration make passive dosimeters extremely Advantageous. Certainly they will not replace conventional ncthods. as these have their place, especially for area samding. The continued and growing use of passive dosimeters, owever, should generate additional data documenting their . eliability and eliminating doubts about their usefulness. >cknowledgement , he assistance of Dr. H. Kenneth Dillon. Head. Industrial fvgiene Chemistry Section of Southern Research Institute, n critically review ing this paper is gratefully acknowledged. eferences 1. Tompkins, F.C. and R.L. Goldsmith: A New Personal Dosimeter for Monitoring of Industrial Pollutants. Am. Ind. Hyg. Assoc. J. 35:371-377 (1977), 2. West, P.W. and K.D. Reiszner: Field Tests of a PermeationType Personal Monitor for Vinyl Chloride. Am. Ind. Hyg. Assoc. J. 39:645-650(1978). 3. Lugg. G.A.: Diffusion Coefficients of Some Organic and Other Vapors in Air. Anal. Chem 401072-1077(1968). 4. Montalvo, J.G.: Total Elemental Content Passive Personal Monitors. Am. Ind. Hyg Assoc. J. 40:1046-1054 (1979). 5. 3M Company: Organic Vapor Monitor Sampling Rate Vali dation Protocol- St, Paul, MN. 6. Jonas, L.C.. C.E. Billings, and C. Lilis: Laboratory Perfor mance of Passive Personal Samplers for Waste Anesthetic Gas (Enflurane) Concentrations. Am. Ind. Hyg. Assoc. J. 43:104*111 (1981). 7 Woebkenberg, M.L.: Current NIOSH Research on Passive Monitors In Proceedings of the Symposium on the Develop ment and Usage of Personal Monitors for Exposure and Health Effect Studies, pp. 27-33 Evironmental Protection Agency. EPA - 600/9-79-032 (1979). 8. Environmental Protection Agency: Laboratory Evaluation ot Commercially Available Passive Organic Personal Moni tors. Contract Number 68-02-2686. J. Hickey. J.L.S. and C.C. Bishop: Field Comparison of Char coal Tubes and Passive Vapor Monitors with Mixed Organic Vapors Am Ind Hyg Assoc J 42.264-267(1981). 10. U.S. Department of Health, Education, and Welfare: Doc umentation ol NIOSH Validation Tests. NIOSH 77-185. Cincinnati (April, 1977). 11. Lautenberger. W.J., E.V. Kring, J.A. Morello: A New Per sonal Badge Monitor for Organic Vapors. Am. Ind. Hyg. Assoc. J. 41:737-747 (1980). 12. Bamberger, R.L.,G.G. Esposito. B.W. Jacobs.G.E. Podolak and J.F. Mazur: A New Personal Sampler for Organic Vapors. Am. Ind. Hyg. Assoc. J. 39:701-708 (1978). 13. National Mine Safety Company: GASBADGE Product Bulletin Chlorinated Solvents Performance Data. Oakdale, PA (1979). 14. National Mine Safety Company: GASBADGE Product Bulletin Aliphatic and Aromatic Performance Data. Oakdale. PA (1979). 15. Anonymous: Guide for Use of Terms in Reporting Data in Analytical Chemistry, Anal Chem. 52.221 (1980). 16. Shotwell. H.P., J.C. Caporossi. R.W. McCollom and J.F. Mellor: A Validation Procedure for Air Sampling Analysis Systems. Am. Ind Hyg. Assoc. J. 40 737-742 (1979). 17. D'Agostino. R.B. and J.L. Gillespie: Comments on the OSHA Accuracy of Measurement Requirement for Monitor ing Employee Exposure to Benzene. Am. Ind. Hyg. Assoc J. 39 510-513(1978) 18. Carlson. F.D., E. Sobel and G.S. Watson: Linear Relation ships Between Variables Affected by Errors. Biometrics 22:252-267(1966). 19. Tuggle, R.M.: Incorrect Use of 1-Tests, Am. Ind. Hyg. Assoc. J 42:325-326(1981). 20. Gordon, C.S. and J.T. Lowe: Carbon Monoxide Detector. U.S. Patent 1,644,014(1927). 21. Plantz, C.A., P.W. McConnaughey and C.C. Jenca: Colori metric Personal Dosimeter for Hydrazine Fuel Handlers. Am. Ind. Hyg. Assoc. J. 29:162-164 (1968). 22. Palmes, E.D. and A.F. Gunnison: Personal Monitoring Devices lor Gaseous Contaminants. Am. Ind. Hyg. Assoc. J. 34:78-81 (1973). 23. Mazur. J.F.. R.L. Bamberger and G.E. Podolak: Develop ment and Evaluation of an Ammonia Dosimeter. Am. Ind. Hyg. Assoc J. 39:749-753 (1978). 24. Kring, E.V..W.J. Lautenberger, W.B. Baker, J.J. Douglas: A New Passive Colorimetric Air Monitoring Badge System for Ammonia. Sulfur Dioxide, and Nitrogen Dioxide. Am. Ind. Hyg. Assoc. J. 42 373-381 (1981). 25. ACGIH: Threshold Limit Values for Chemical Substances and Physical Agents in the Workroom Environment with IntendedChanges for 1981. American Conference of Govern mental Industrial Hygienists. Cincinnati (1981). 26. DuPont: PRO-TEK Colorimetric Air Monitoring Badge Sys tem Laboratory Validation Report. Ammonia Badge. Type C-10. E.l. DuPont de Nemours and Company. Wilmington. DE (1981). 27. Shor, R.M. and L.W. Anders: Direct Read Carbon Monoxide Monitor Performance Under Possible Use Conditions. Pres entation at American Industrial Hygiene Conference, Portland, OR (May 25-29, 1981). 28. Hardy, J.K., P.K. Dasgupta, K.D. Reiszner, and P.W. West: A Personal Chlorine Monitor Utilizing Permeation Sampling. Env. Sci. Tech. 73:1090-1093 (1979). 29. Moleculon Research Corp.: PROPLASTIC Chlorine Vapor Badge Information, Cambridge. MA. 30. Hardy, J.K., D.T. Strecker, C.P. Savariarand P.W. West: A Method for the Personal Monitoring of Hydrogen Sulfide Using Personal Sampling Am. Ind. Hyg Assoc. J. 42:283- 286(1981). 31. Graedel, T.E. and J.P. Franey: Gaseous Hydrigen Sulfide Determination by Discoloration of Lead-Stabilized PVC. Am. Ind Hyg. Assoc. J. 47:947-953(1979). 0 Am Ind Hyg Assoc J (A3) August 1982 3M 0017S2 32. McCammon. C.S. and J.W. Woodfin: An Evaluation of Passive Monitor for Mercury Vapor. Am. Ind. Hyg. Assoc. J. 38 378-386(1977). 33. McCammon. C.S., S.L. Edwards, R.D. Hull. W.J. Woodfin: A Comparison of Four Personal Sampling Methods for the Determination of Mercury Vapor. Am. Ind. Hyg. Assoc. J. 47:528-531 (1980). 34. Cohen, H.J., R.K.Zahray, A.C. Misiaszek and H.J. M uranko: A New Passive Dosimeter tor Mercury. Presentation at American Industrial Hygiene Conference. Portland, 0R(May 25-29. 1981). 35. ' Palmes, E.D.. A.F. Gunnison, J. DiMattioandC.Tomciyk: Personal Sampler for Nitrogen Dioxide. Am. Ind. Hyg. Assoc. J 37 570-577(1976) 36. McMahon. R., T. Klinger. B. Ferberand G. Schnakenberg: New Technology for Persona! Sampling of NOi and NO\ in the Workplace Presentation at American Chemical Society Exposition Symposium. Las Vegas. NV(August 25-28.1980). 37. DuPont: PRO-TEK Colorimetric Air Monitoring Badge Sys tem Laboratory Validation Report, Nitrogen Dioxide Badge. TypeC-30. E I DuPont de Nemours and Company. Wilmington, DE (1981) 38. DuPont; PRO-TEK Colorimetric Air Monitoring Badge Sys tem Laboratory Validation Report. Sulfur Dioxide Badge. TypeC-20 El DuPont de Nemours and Company, Wilmington. DE (1981), 39. Silverstein. L.G.: Validation of Abcor GASBADGE for Acrylo nitrile and Improved Desorption Efficiency Am Ind. Hyg. Assoc J. 38.412-413 (1977) 40. Hirayama, T. and M. Ikeda: Applicability of Activated Car bon Felt to the Dosimetry of Solvent Vapor Mixture. Am. Ind. Hyg Assoc. J. 40 1091-1095 (1979) 41. Halliday, M.M. and J. Anderson: Determination of Halothane in Operating Theatre Air by Using a Passive Organic Vapor Dosimeter. The Analyst 105 289-292 (1980). 42 Evans. P.R. and S.W. Horstman; Desorption Efficiency Determination Methods for Styrene Using Charcoal Tubes and Passive Monitors Am Ind. Hyg. Assoc. J. 42.471476 (1981) 43. Anders, L.W. and H.E. Mullins: Comparison of Diffusions! Organic Vapor Monitors with Charcoal Tubes for Sampling Laboratory Challenges to Contaminant Mixtures. Presenta tion at American Industrial Hygiene Conference, Portland. OR (May 25-29. 1981) 44. Mazur, J.F., G.E- Podolak. G.G. Esposito. D.S. Rinehart and R.E. Glenn: Evaluation of a Passive Dosimeter for Col lection of 2-Bromo-2-Chloro-1.1,1-Trifluoroethane and 2Chloro-1.1,2-Trifluoroethyl Difluoromethyl Ether. Am. Ind. Hyg. Assoc J. 41 317-321 (1980) 45 DuPont: PRO-TEK Organic Vapor Air Monitoring Badges Laboratory Validation Protocol for Diffusion-Type Air Moni toring Badges with Solid Solvents. E.l. DuPont de Nemours and Company. Wilmington. DE (1981). 46. DuPont: Laboratory and Field Validation Report for PRO-TEK G-BB Diffusion-Type Badges for Monitoring Benzene Vapors. E.l. DuPont de Nemours and Company. Wilmington. DE(1981). 47. 3M Company: #3520 Organic Vapor Monitor with Backup Section. St. Paul. MN. 48. Mazur. J.F.. D.S. Rinehart, G.G. Esposito and G.E. Podolak: Evaluation of Passive Dosimeters for Assessing Vapor Degreaser Emissions Am. Ind. Hyg Assoc. J. 42.752- 756(1981). 49. Nelms, L.H., K.D. Reiszner and P.W. West: Personal Vinyl Chloride Monitoring Device with Permeation Technique for Sampling. Ana! Chem. 49 994-998(1977). 50. Bailey. A. and P.A. Hollingdale-Smith: A Personal Diffu sion Sampler for Evaluating Time Weighted Exposure to Organic Gases and Vapors. Ann. Occup. Hyg. 20 345356(1977). 51. Benson. G-B. and G.E. Boyce: A Thermally-Desorbable Passive Dosimeter for Personal Monitoring of Acrylonitrile. Ann. Occup Hyg 24:55-75 (1981). 52. Campbell, J.E. and R.B. Konzen: The Development of a Passive Dosimeter for Aniline Vapors. Am. Ind. Hyg. Assoc J. 47:180-184(1980) 53. Mullins. H.E. and L.W. Anders: A New Innovative Diffu- sionalMonitor for Sampling Ethylene Oxide in Air Presenta tion at American Industrial Hygiene Conference, Portland. OR (May 25-29. 1981). 54. Rodriguez. S.T.. P.B. Olson and V.R. Lund: Colorimetric Analysis of Formaldehyde Collected on a Diffusional Monitor. Presentation at American Industrial Hygiene Confer ence, Portland, OR (May 25-29, 1981). 55. DuPont: PRO-TEK Colorimetric Air Monitoring Badge Sys tem Laboratory Validation Report. Formaldehyde Badge. Ser ies II. Type C-60. E.l. DuPont de Nemours and Company, Wilmington, DE (1981). 56. Kriesel. R.S.: Formaldehyde Vapor Detection -- New Sam pling Technology. Presentational American Industrial Hygiene Conference. Portland, Or (May 25-29, 1981). 57. Matherne. R.N., P.L. Lubs and E.J. Kerfoot: The Develop ment of a Passive Dosimeter for Immediate Assessment of Phosgene Exposures. Am Ind. Hyg. Assoc. J. 42 681684(1981) 58. Hill, R.H. and D.A. Fraser: Passive Dosimetry Using DetectorTubes. Am. Ind. Hyg. Assoc. J. 47:721-729(1980) 59. Jones, W-, E.D. Palmes, C.Tomczykand M. Millson: Field Comparison of Two Methods for Determination of NO. Con centrations in Air. Am. Ind. Hyg. Assoc. J. 40 437438(1979). 60. Evans, M.. M. Molyneux. T. Sharp. A. Bailey and P. Hollingdale-Smith: The Practical Application of the Porton Diffusion Sampler for the Measurement of Time Weighted Average Exposure to Volatile Organic Substances in Air. Ann. Occup. Hyg. 20:357-363(1977). 61. Zaebst, D.D., M.F. Boenigerand J.R. Burg: Field Compari son of Two Passive Organic Vapor Sampling Devices to the Charcoal Tube. Presentation at American Industrial Hygiene Conference. Houston, TX (May 18-29, 19801 62. McKee, E.S., P.W. McConnaughey and I.M. Pritts: Colori metric Personal Dosimeters for Some Inorganic Contami nants. Mine Safety Appliances Co.. Pittsburgh, PA. 29 December. 1981; Revised 8 February. 1982 American Industrial Hygiene Association JOURNAL (A3) 8 82 3M 001783 621 3M FORMALDEHYDE MONITOR Summary Analytical Method The NIOSH P&CAM 125 Chromotropic Acid Method has been modified to reduce analytical interferences and give an expanded linear response of absorbance as a function of formaldehyde solution concentration. The procedure uses 1 milliliter of 1% chromotropic acid and 5 milliliters of sulfuric acid for sample aliquot volumes ranging from 0.1 to 2.5 milliliters. For the modified method, using a 2.0 milliliter aliquot, the formaldehyde analytical detection limit is 0.5 ug/ml. The precision as measured by the coefficient of variance for the analytical method only is 0.023 and for the overall analytical method is 0.030 Sampling Parameters The formaldehyde recovery from the 3M Formaldehyde Monitor is 1.01 as determined from samples spiked with a range of weights equivalent to sampling 0.1 to 4 ppm for 8 hours. For the 3M Formaldehyde Monitor, the formaldehyde sampling rate is 65.93.7 cc/min. The concentration sensitivity is dependent on the length of the sampling period. For a 4 hour sampling period, a concentration of 0.2 ppm can be quantified, for an 8 hour sampling period, 0.1 ppm and for a 24 hour sampling period, 0.03 ppm. Sampling performance is maintained with air movement as measured by a face velocity as low as 15 linear feet per minute. The 3M Formaldehyde Monitor has a capacity of 850 micrograms (174 ppm-hr). With this capacity, a concentration of 22 ppm can be sampled for 8 hours. The 3M Formaldehyde Monitor can sample accurately formaldehyde in ambient atmosphere with relative humidity ranging from 20 to 90%. From the sampling performance of product stored at 70F and 100F, the shelf life has been document to six months and can be extrapolated to be in excess of one year. The sample stability has been demonstrated to be at least 90 days when stored at temperatures as high as 100F. Analytical & Sampling Interferences The modified chromotropic acid method has eliminated the ethanol interference for samples collected by either the bisulfite impinger or the 3M Formaldehyde Monitor. With ethanol concentrations as high as 200 ppm and 4 hour sampling periods both sampling techniques measured a 1.21 ppm formaldehyde challenge with accuracy greater than 5%. In the presence of phenol at concentrations as high as 10 ppm, the 3M Formaldehyde Monitor measured a 0.99 ppm formaldehyde challenge within +4%. The bisulfite impinger meaured the formaldehyde concentration with a minimum of 3M 001784 interference at phenol concentrations as high as 1 ppm. For the bisulfite impingers at higher phenol concentrations, the phenol interference decreased the accuracy even when using 10% chromotropic acid concentration in the analysis procedure. If high phenol concentrations are encountered with the bisulfite impinger method, the solution concentration of phenol can be quantified with a GC analytical technique and the formaldehyde absorbance from the chromotropic acid method can be corrected for phenol interference. Sampling Performance In the laboratory, the 3M Formaldehyde Monitors have sampled known challenges at various concentrations as low as 0.059 ppm with sampling periods ranging as long as 46 hours. The 3M Formaldehyde Monitors measured the concentrations with excellent precision and an accuracy of 10% or greater. In the field, the formaldehyde concentration levels were measured with both the 3M Formaldehyde Monitor and the standard bisulfite impinger. These evaluations were conducted in a variety of ambient conditions, ranging from a coating/sizing of sandpaper operation to a mobile home. There was excellent correlation between the two techniques as well as extremely good precision when replicated samples were collected. 3M 001785 Table of Contents Introduction i Analytical Method Chromotropic AcidProcedure Analytical Detection Limits Analytical Precision 3 21 25 Sampling Parameters Formaldehyde Recovery Sampling Rate Validation Concentration Sensitivity Face Velocity Capacity Relative Humidity Product Shelf Life Sample Stability Monitor Blank 29 31 35 37 43 45 47 49 51 Analytical & Sampling Interferences Sampling Formaldehyde in Presence of Ethanol Sampling Formaldehyde in Presence of Phenol Evaluation of Analytical Phenol Interference Analytical Method For Monitor Solutions Analytical Method For Impinger Solutions Evaluation of Formaldehyde/Phenol Sampling Analytical Interference Summary of Phenol Interference Errors 53 55 57 73 83 89 Sampling Performance Laboratory Evaluation Field Evaluation 91 99 3M 001786 List of Figures No. Title 1 Absorbance Dependence on Chromotropic Acid Concentration Page 2 .2 Absorbance as a Function of Formaldehyde Standard Solution Concentration 8 3 Absorbance of All Aliquot Volumes as a Function of Formaldehyde Concentration in Final Solution 4 Absorbance Deoendence on Sequence of Sample Addition 16 5 Detection Limit of Analytical Procedure 20 6 Detection Limit of Overall Analytical Procedure 22 7 3M Formaldehyde Monitor Analytical Precision 26 8 Formaldehyde Recovery From 3M Formaldehyde Monitor 28 9 Sampling Rate Validation for 3M Formaldehyde Monitor 30 10 3M Formaldehyde Monitor Concentration Sensitivity 34 11 3M Formaldehyde Monitor Sampling Rate Dependence on Face Velocity 36 12 Formaldehyde Capacity on 3M Formaldehyde Monitor 42 13 3M Formaldehyde Monitor Performance as a Function of Relative Humidity 44 14 3M Formaldehyde Monitor Product ShelfLife 46 15 3M Formaldehyde Monitor Sample Stability 48 16 Sampling Formaldehyde in the Presence of Ethanol 52 17 Sampling Formaldehyde in the Presence of Phenol Solution 54 18 Absorbance as a Function of Formaldehyde Solution Concentration 0.1% Chromotropic Acid 59 19 Absorbance as a Function of Formaldehyde Solution Concentration 1% Chromotropic Acid 61 20 Absorbance as a Function of Formaldehyde Solution Concentration 4% Chromotropic Acid 63 21 Absorbance as a Function of Formaldehyde Solution Concentration 10% Chromotropic Acid 65 3M 001787 List of Figures Continued 22 Phenol Interference as a Function of Chromotropic Acid Concentration 23 Analytical Error in Presence of Phenol as a Function of Chromotropic Acid Concentration 24 Absorbance as a Function of Formaldehyde Solution Concentration 4% Chromotropic Acid 25 Absorbance as a Function of Formaldehyde Solution Concentration 10% Chromotropic Acid 26 Phenol Interference as a Function of Chromotropic Acid Concentration 27 Analytical Error in Presence of Phenol as a Function of Chromotropic Acid Concentration 28 Measured Formaldehyde Concentration Uncorrected for Presence of Phenol Collected When Sampling Known Laboratory Challenge 29 Measured Formaldehyde Concentration Corrected for Presence of Phenol 30 3M Formaldehyde Monitor Performance When Sampling Known Challenges (0.100 ppm) 31 3M Formaldehyde Monitor Performance When Sampling Known Challenges (0.392 ppm) 66 70 75 77 80 84 86 92 94 3M 001788 List of Tables No. Title 1 Absorbance Measured on Formaldehyde Standard Solutions With & Without External Heating (0.135 Chromotropic Acid) Page 4 2 Absorbance Measured on Formaldehyde Standard Solutions With & Without External Heating (1.0% Chromotropic Acid) 4 3 Absorbance as a Function of Formaldehyde Concentration For Various Aliquot Volumes 5 4 Summary of Applicable Aliquot Volumes For Sample Analysis 14 5 Absorbance Dependence on Sequence of Sample Addition 17 6 Absorbance Measured on Standard Formaldehyde Solution For Determination of Analytical Detection Limit 21 7 Absorbance Measured on Desorbed Solution From Monitors Spiked With Known Formaldehyde Weights 23 8 Precision of Analytical Method 25 9 Precision of Overall Analytical Method 27 10 Precision of Overall Method Including Sampling a Known Challenge 27 11 Summary of Recovery From Formaldehyde Monitor Spiked With Known Weights 29 12 Sampling Rate Validation 32 13 Summary of Sampling Rate Validation 33 14 Summary of Concentration Sensitivity 35 15 Sunmary of Face Velocity Dependence of the 3M Formaldehyde Monitor 37 16 Summary of Face Velocity Dependence onAirFlow Rate 39 17 Sampling Data Summarizing Face Velocity Dependence 40 18 Summary of Formaldehyde Capacity Data 43 19 Data Summary When Sampling Known Laboratory Challenges With Relative Humidities Ranging From20 - 90% 45 20 Summary of Product Shelf Life Data 47 21 Summary of Sample Stability 49 3M 001789 List of Table Continued 22 Summary of Blank Levels on 3M Formaldehyde Monitor 23 Data Summarizing Ethanol Interferences 24 Data Summarizing Performance of 3M Formaldehyde Monitor and Bisulfite Impinger When Sampling in the Presence of Phenol 25 Absorbance Measured as a Function of Formaldehyde Concentration for Monitor Solution Using 0.135 Chromotropic Acid 26 Absorbance Measured as a Function of Formaldehyde Concentration for Monitor Solution Using l%.\ Chromotropic Acid 27 Absorbance as a Function of Formaldehyde Concentration for Monitor SolutionUsing 4%Chromotropic Acid 28 Absorbance Measured as a Function of Formaldehyde Concentration for Monitor Solution Using 10351 Chromotropic Acid 29 Summary of Analytical Interference Caused by Presence of Phenol in DesorbedSolutions 30 Summary of Phenol Correction Factor for Phenol Presence in Desorbed Solutions From 3M Formaldehyde Monitors 31 Absorbance Measured as a Function of Formaldehyde Concentration for Impinger Solution Using 4% Chromotropic Acid 32 Absorbance Measured as a Function of Formaldehyde Concentration for Impinger Solution Using 10% Chromotropic Acid 33 Summary of Analytical Interference Caused by Presence of Phenol in Impinger Solutions 34 Summary of Phenol Correction Factor for Phenol Presence in Impinger Solutions 35 Data Summarizing Performance of 3M Formaldehyde Monitor Bisulfite Impingers and Silica Gel Tubes When Sampling Formaldehyde in the Presence of Phenol 36 Summary of Sampling Formaldehyde in Presence ofPhenol 37 Summary of Analytical Errors Resulting From Phenol Interference 51 50 55 58 60 62 64 67 68 74 76 79 81 82 85 88 3M 001790 List of Table Continued 38 Performance of 3M Formaldehyde Monitor When Sampling Known Laboratory Challenges for Extended Sampling Periods (0.059 ppm) 39 Performance of 3M Formaldehyde Monitor When Sampling Known Laboratory Challenges for Extended Sampling Periods (0.100 ppm) 40 Performance of 3M Formaldehyde Monitor When Sampling Known Laboratory Challenges for Extended Sampling Periods (0.392 ppm) 41 Performance of 3M Formaldehyde Monitor When Sampling Known Laboratory Challenges for Extended Sampling . Periods and Compared to BisulfiteImpingers(0.392 ppm) 42 Burlington Industries Field Test 43 Chemical Plant Field Test 44 Field Sampling Formaldehyde in Presence of Phenol and Cellosolve 45 Sampling Formaldehyde in a Mobile Homefor 24 Hour Sampling Period 91 93 95 97 100 101 102 104 3M 001791