Document e524DzjyMxqwoONNoX0GGGmGq

Hyg. Assoc. J. 47(1)27-36(19863 ation of an IS0 Draft Proposal for Sampling and Analysis of Chlorinated Hydrocarbon Solvent Vapors in Workplace Atmospheres T. SPEE TNO Research Institute lor Environmental Hygiene. P.O. Box 214,2600 AE Delft, The Netherlands is0draft proposal for sampling and analysisof chlorinated hydrocarbon solvent vapors in workplaceatmospheres (XSO/TC IJ6/Sc N41)L l s been evaiuated by memmof a round-robin test. For this purpose, the repeatability 8nd reproducibilityfor the determination obUmcMoromethane, trichloromethurc, 1,1,1 -trichloroethane, triebloroetbenc8 d tetnchlorottbene weredetermined at three concentnQrtcrrb. Furthermore, parameters sucb as breakthrough and losson storage were determined. Smpleswere taken by 11participnts from rf.csPratries. After statistical evaluation of the results, it appeared that improvementswere necessaryfor the determination of tetrachlorome- and trichloromethane. For tbe other three compounds, repeatabilitj and reproducibility were sufficient. Possible error sources and polbilities for improvements are discussed. LBpawtlon chlorinated hydrocarbons are widely applied as rdvrxlQ &greasing in the metal industry, industrial textile nd solvents in the paint industry are just some ery of the neurotoxic character and suspected ty of several of the aliphatic chlorinated hydros resulted in a steady lowering of the acceptable corraptrrrtion limits of these compounds in the workplace 8teunphet-e in a number of countries. In order IO increase rdSability of measurement of chlorinated hydro-carbon *M vapors, the 1SO has drafted a proposal for sampling and anatysis of chlorinated hydrocarbon solvent vapors in ~ ~ k p tatcmcospheres.(2) An international round-robin test Wts Qtlplnhed by the TNO Research Institute for EnvironHygiene for the evaluation of this draft proposal. W t s of the round-robin test are described in this Wpel. Summary of the IS0 i' The proposal is compounds: - dichloromethane - trichloromethane - tetrachloromethane - 1, I-dichloroethane - 1.2-dichloroethane - I.l-dichloroethene - 1.2-d1chloroethene - I, I , I -trichloroethane - 1,2,2-trichloroethane - trichloroethene - I. 1,2,2-tetrachloroethane - tetrachloroethene - 1,2-dichloropropane - chlorobenzene - 1.2-dichlorobenzene. suitable for the following _--- sampling time sampling time 4b sompl in# timr sompling time ___) t i m e ( h o u r s ) -&nCentration of chlorinated hydrocarbons vs time (recorded with MIRAN@1A gas analyzer). Date. 30 May 1983. J 147) January 1986 CopyriQhl iD1)6 American fndustrial Hygiene A~s0~1I11011 < 2 1 W- < U 3 4 1. Pr8SSUfl f 8 @ U h t W 2 Critical orificr 3. Humidifirr 4. RH mrasurrmont 5. Syringe and motor drive 6. Vigreur mixing column X Manifold Figure 2-Preparation of known concentrations of chlorinated hydrocarbon solvent vapors in air. The method described is a pumped charcoal tube-solvent desorption gas chromatographic method. A known volume of air is drawn through a glass or metal tube by means of a sampling pump; I g of activated coconut charcoal appears to be necessary for the adsorption of the chlorinated hydrocarbon vapors in the concentration range of interest. The organic vapors collected are desorbed by a suitable solvent and analyzed by means of gas chromatography with flame ionization detection. The peak heights or peak areas are -_-compared to those of standard solutions in carbon disul- phide. Breakthrough is determined in each sample by means o f a back-up section in the tube. Desorption efficiencieshave to be determined for each lot of tubes by preparation of spiked samples on charcoal. The proposal is fully described in reference.'*' General Procedure of the Experiment 1 In the IS0 draft proposal 15 compounds are listed to which the proposal is meant to be applicable. As it was not possible to test all 15 compounds in one run, 5 of them were selected for the test, namely. tetrachloromethane, trichloromethane, I , I , I-trichloroethane, trichloroethene, and tetrachloroethene. All participants took samples simultaneously in duplicate from a common sampling manifold in which a test atmosphere with a constant concentration of a known fixed mixture of the chlorinated hydrocarbons to be tested was maintained. The test atmosphere was generated dynamically, the flow available being a t least fivefold in excess relative to the test gas consumption due to sampling. Three concentration levels for each mixture were established on subsequent days. The relative humidity of the test atmosphere was 50f2%. Each participant took eight sam- ples (four duplicates) of each concentration level -four in the morning and four in the afternoon. The sampling time of each sample was 90 min at a flow rate of about 0.I L/min, resulting in a total volume of about 9 L per sample. The samples were analyzed at the laboratories of the partici- determined by the participants themselves. The results were evaluated statistically according IS0 International Standard 5725.'" Description of the Test Gas Atmosphere Generator Test gas mixtures with known concentrations o nated hydrocarbons of interest were generated b te~hnique.'~A' diagram of the test gas generat an absolute filter. The purified air contained less ppm of water vapor. and the concentratlons of t vidual chlorinated hydrocarbons were below 0. f Zero test gas was humidified by passinganadju istered. On each of the test days, the relative hum1 & 2%. Gravimetrically prepared mixtures of hydrocarbons were injected directly into th air stream (12-18 L/min) by a motor drive In order to diminish concentration fluctuations stream of zero gas. Fresh mixtures of the liquid components were each day. The chemicals were of reagent grade q syringes used had a volume of 100 m L for the h 21 Am Ind Myg Asme. 1.(47) 113.0 667.0 22.5 132.0 864.0 m.0 179.0 1056.0 286.0 1685.0 117 655 22 130 908 5086 185 1032 296 1586 5.4 3.8 7.4 3.2 5.2 4.2 5.0 4.6 4.0 4.8 Duplicate measurements of the flow rate (beginning and end of each test day) were within 0.2%. The repeatability of the determination of the quantity of the liquid metered had been determined before performance of the test and appeared to be 0.002%at the lowest concen1 tration level, 0.00696 at the middle level and 0.03% at the highest level (95%' confidence interval, mean of ten determinations). The total error estimate expressed as the root sum of squares of all errors -assuming these are independent is I % at the 95% confidence level. Constancyand Homogeneityof the Composition of tbt Test Cas Atmosphere Slight fluctuations in the composition of the test gaGiGiZphere may occur by fluctuations in the gas stream or in the injection system, or by irregularities during the evaporation ofthe injected liquids (the description of the test gas atmosphere generator). The constancy of the gas concentration is determined by means ofinfrared spectrometry and has been found to be better than 5% for all concentration ranges during the sampling periods (Figure 2). The volume of the test installation is 2 L. The gas flow being 20 L/ min and the sampling time 90 min, the discrepancy in the composition of the samples which are taken simultaneously due to fluctuations of the gas mixture composition will be at most 0.005%. s s s S,W s S W W S s 8 i .i The homogeneity of the gas mixture and the reproduclbllity of the determination have been investigated by taking five samples from different positions of the test installation. The results are summarized in Table 1. The reproducibility falls within the limits calculated in the Results Section. Performance of the Test The parricipating laboratories are listed below in alphabetic order. *Akzo Corponre Research Arnhem Nederland *AIIgctneine Verslcherungsanslalt Wien Ostrrreich .&rufsgenossenahaftliches Sankt Augustin Bundesrepublik lnsutut fur Arbeltssicherheit De utschla nd *Centnlab soest Nederland *Directoraat-Generaal van Voorburg Nederland de Arbeid *DSM-Research and Patents *Health and Safety Executive Geleen London Nederland England *IMG--TNO OINRS *NIOSH .Shell Nederland Chemie B.V. Delft Vandoeuvre Cincinnati Rotterdam Nederland France U S.A. Nederland in all other Tables the participants are represented by a random number. The participants used their own sampling and calibration equipment. Analyses were performed on different gas chromatographic (GC) columns. The equipment used and the dates of analysis are summarized in Tables I1 an4 111. Sampling was carried out according to Paragraph 7 ofthe proposal."' The sampling rate was 100 mL per minute and the saITipttRg time amounted to 100 minutes. Each participant took eight samples per day (four duplicate samples). in this way 88 samples were collected frqm each concentration range. The sample tubes were tragshrted and stored at a temperature of-20C or less. G a s c h r o m a t o p p h k q preparation of the calibration cu the desorption efficiency were u Paragraph 8 of the proposal." Results Results of Analysis Graphic representations of the results of analysis samples are presented in Figures 3 to 7. About 10% results appeared to be missing, mostly because the ch tubes had not been loaded with any material at all. Aa reason for unsuccessful samples was breakage of the coal tubes or vials during workup of the samples. 3 Desorptlon Efffclency (DE) Desorption efficiencies reported by the participaaj presented in Table 1V. Desorption efficiencies were mined according to Paragraph 8.3 of the propo alternative method described in Annex C was not 1 Breakthrough i Preliminary breakthrough experiments were car' before performance of the round-robin test with t tubes containing 100 + 50 mg of coconut charcoal ad taining 800 + 200 mg of coconut charcoal in the f r a l back-up sections, respectively. The specifications , tubes met those Prescribed in Paragraph 5.2-1- , Proposa1- The experiments were performed with the highest ct j t i o n s of analytes to be used in the round-robin 1 r lathe humidity used in all experiments was 50 f4 SO-proposal, a method for the determination ofth Participant number TABLE 111 Round-Robin Test for Chlorinated Hydrocarbons Analvsis Eauiement and Dates ot Analvsis Gas chromatograph manufacturer tw Chromatography column stat. phar kngth 1.0. -~ Detector Tima h ( m n sampling and aMfySiS (woks) '1 Packard Becker GC433 2 Varian 3 Packard 4 H.P. Car@Erba S Perkin Elmer 6 Perkin Elmer 7 H.P. 8 H.P 9 PerkinElmer 10 PerkinElmer 11 lntenmat 3700 SE-30 429 Cpsll5 CB 5880 SE-54 2900 OVlOl Sigma 3 6 C R i l 5 Sigma 1 209c SE-30 + 0.5% polypurgent 5840 A SE-30 5730A GP2096SP21W 0.1% carbowax F17 OVlOl F22 OVlM IGClPOFL SE-30 *= not specified '= two separationmethods have been u s d by this participant Mahod 1: SE-30.50 m. 0.25 mm ID Metnod 2: M055 MuItidimensoMI Switching -rn prscolumn: SE-30 6m W" main-column: SE-30 35 m x 0.25 mm 40.0 m 50.0 m 25.0 m 25.0 m 25.0 m 2.0 m 50.0 m =.Om 3.0 m 50.0 m 4.5 m 0.28 mm 0.32 mm 0.32 mrn 0.32 mm 0.32 mm W inch 0.7 mm 1.2 mm 2.0 mm 0.25 mm %inch FID FID FID FID MPECO FID FID FID ECD FID FID FtO A 2 1-2 3 <1 <1 <1 58 <1 <I 2-4 . .SWnsbcd Evaluation of the R d t s m Tbc rcsplts were evaluated according to IS0 5725. The -ropriate. For tbk rlLt. following tests are prescribed in this Standard.'" 0 Cochran's maximum variance test for the determination of repeatability.'5' In this test, the highest variance found is divided by the mean of the remaining variances and the was chosen: difkrat u a r rexultingvalue is compared to a pumerical criterion, dejxnd- t and back-up s c u b m s o f ~ m k ~ * O D thenumber of replicatesand the number of participants. Dixon's outliers test for the determination of reproduci- bility. In the Dixon test, the highest and the lowest result of *in Tabk V. These res- ph.r pllp tbc analyses found by the participants are compared to the remaining results. This means that, contrary to Cochran's Usf Dixon's test is a two-sided one. Repeatability and reproducibility are defined as follow^.'^' The repeatability r is the value below which the absolute difference between two single test results obtained with the same method on identical test material. under the same mrcemtrotion rb@/Im3) l2 I - 0 i z 3 L s 6 7 o 9 ion 340.- -320 300- no - 260 1 f True rolue 256 m @ / d 210 200- -18. I I I 160 - 1 140, 1 L I 123~~6789ion --- portacapont number Fqure 4-Round-rabin test for chlorinated hydrocarbons Resutts of analysls of trlchloramethane Mean values and 95% confidence intervals. 31 4' 100 ttiiii I T A 1 1,1II, s 6 7 o smn I___Ic_ porticipont n u m b # Figure 5-Round-robin test for chlorinated hydrocarhns Results of analysis of trichloroethane. Mean values and 95% confidence intervals. -L participart numbw F Q Pa-~ ~ La for chlorinated hydrocar f?esmits a'rpiyss d-oethene. Mean values an mdcteRceirerrars conditions (same operator. same apparatus. same labora- tory, and a short interval of time). may be expected to lie with a specified probability In the absence of other imii0- tions, the probability is 95%. The reproducibiiity R is tbt value below which the absolute difference betwren twos&gle test results obtained with the same method on ideaticzl test material, under different conditions (different opcrr- tors, different apparatus, different laboratories and: or da- ferent time), may be expected to lie with a specifiedproboW ity; in the absence of other indications, the probobilitv k 95%. Cochran's test and Dixon's test are explained Rckr- ence 3. Briefly summarized, the procedure of the s t - ms - tw A4 uation is as follows: 1. The results are inspected for peculiarities s u d as ~ r s - tematically high or low values within oat hbartg missing values, etc. 2. The variances per "cell"(= group of results pcr hkn, wq d a n calculated in the normal w -A=-5. T k gm d are then compared wi a k z 9etar+4COC&n's test. In this test, MS&~RR Y by the sum of all varia -c =stmax P 2 s', i=I TL d & d e n s t a table of critica QI ik -bet of laboratories d s q q pc'ancentration."' If th s k Ilf LCRP in the mbk, the result grrar-lnrhcorsidcredasan outlier. If t &rL 5?k Icrd, tbe rcsdt is considered am&r-Q~&~~l~tJirnimEeSdtr.agglingv -E- bi& nmmber of st CI kmu outlying laboratory. lf m32 11 IW,~SSOC1.(47) I test must k m p t c d with the remaining am move arc found y a t r w m e mtbjecW to Dixon'hrtat. Fop %heman dues ofthe remaining cells arc ian~edia increasing order, the rtsult L ...&.Dixon's criterion depends on the bf H (Hking the remaining number ofcells clce of Cochrur's test). In our case, Dix- IScatculatedaccording to the formulas: d 3- z(: @rbetest criterion, 2is the mean value of each rcWyper level and Hthe laboratory with the high- mn value. The greatest of the two values of Qll is @against a table of critical val~es`o~n' 1% and on hdfor outliers and stragglers I *-I1- A.-L 2 3 4 5 67 9 Dn True value 29.5 m g k d I True voluc 226 mg/m3 23L5 True value 1259 m o l d 'T i- T A J2 3 4 5 4 i 14 n 8 19 porttocpoat numbrr or chbrinated hyd sI&6intervals. January. 1986 5, r 4 R a m alcuhted fro0 tbc rcmainiag 7 L r m r r f a y c l ; p b l e din Rtfmnce 3. n cocLp.-'r--rpZrrpr.indrpaarntieciqpadntn. uAmnbuemrboefrroefpcliocnatceesnptrear- t i m e & 1ltc0BDply witb thii requirement. ln order t o Ip.Lc.lLctadr d a b k for inclusion in the calculation of d q~ducibiiitya, number of values were cf - e r l ~ s k - p k c o n c e n t r a t i o n c e l i s randomly. A f t a e t F - - 1 4the Cochran and Dixon tests were coIIp.Icd @This operation was performed with one, nn,13phzt *tiom per concentration cell. lncom- p k w d d i swhich had not been rejected at these n;rles mmic dfor the calculation of repeatability and Tbr lglsd zh:statistical evaluation are presented in Tabk W (tk fdl Itatistical evaluation is available on Ir Ta&k VI1 an overall picture of the outliers is l k nkk sboa that 27% of the results are ehmi- ouad Icy C-i test- Eliminations are concentrated in al- if. and 8. Furthermore, two results are sejccsd b~W ' r test.. Further investigation has shown -that tLarrbrsnammost likely are caused by a n error in t& -qp d rk standanis. Tk SJWw- of tbc llaults on carbon tetrachloride - mp10 8091r. Fortbeother c o m p o 6 , x i a - tiOahILtgeprrrl meanvary from -9 to +5%, indicating tLu 10 LaE i m l d by the method of sampling and -rf*- o tbe Roposrl Corrrrr OJtk tbr partidpants of the round-robin test arc e--tL s ICR sensitiveenough for determination of -eat in the concentration range ofinterest; ---for-* drimurrsof sdwnts, capillary columns are mar ILUpacked ones. It is suggested that a list dnrrbie c&mnas be included in the Standard; --ibr pgplei compounds with entirely different -a A mar distinct description of compounds .sLicL- m Iw laarmd simultaneously is necessary. afj9tS;19dlfd ofa simple method for purifica- tiadC5ekawtkooaantratrons ofcontaminantsare -hLais=ggestcd; ai& ~zxqnatsstates that a more systematic studj ofrbc- of the different sources of errors erhorJI Lr If the inaccurac) of the volume -1 -- gecors to be a major error source. alter- sbould be considered; ,ubrtanwsare not in accordance TkhaELptjLgLI WC(Ditbechrcoaltubes were not loaded 33 nrrmb.r CCI, 1 2 3 98 4 100 5 6 7 96 9 100 9 10 11 98 'not specified TABLE IV 0.rorpSlon Eff)ci.ncios o..orpaon.mcicncy CHCh C & C G CHCtCCh CChCcls 93 94 93 92 99 98 99 97 102 95 100 100 100 100 99 99 97 99 99 96 nolspccifffl p e r compourd % 4 36 100 95 with any test material is remarkable. The origin of this phenomenon is not clear. Most common sampling failures can be discounted in this experiment : OLPakage between charcoal tube and pump. This possi- bility is excluded by the fact that the sampling flow was determined with the charcoal tube in line. Pumpfarlure. Working of the pump could be ascertained by the fact that some pumps had a stroke counter and on others running of the pump was clearly visible by movement of the belt. Low ahorption efficiency. No breakthrough was observed. oLms-z#uringtransit or storage. In this case an equal loss of anaiyte from all tubes would be expected. Failure of the sampling system. This possibility can virtually be excluded. Any small leakage of the sampling systern will cause a considerable lowering of the concentration of the material downstream of the leakage due to a venturi effect. This was found neither in the samples nor on the monitor. Furthermore, it is very unlikely that all sampling air will be extracted from the leakage. *Poor hading of the tubes. Although the three participants who were confronted with unloaded sample tubes used the same make and lot of charcoal tubes, six other participants using the same tubes did not suffer from this phenomenon. Consequently, the sampling failures certainly cannot be assigned to the charcoal tubes as such. The strong differences in adsorption efficiencies of chlorinated hydrocarbons found in another experiment in which the test atmosphere was generated in an independent way, however,@' lead to the conclusion that sampling of chlorinated hydrocarbons is less straight forward than it seems to be. From Table V11 it can be concluded that the number of outliers differsstrongly per participant. Four laboratoriesare responsible for 85% of the outliers. From the reply form three major causes of error can be assigned: ])In a few cases the results of one component deviate throughout the test. This can be assigned to failures in the preparation of the standards. 2)ln some cases the amount of material found on the t was in reasonable agreement with that on the other t but the volume of air sampled differed strongly fro other volumes. A failure in volume determination due to lack of constancy of the pump, may be theca ...- 4the failure. 3)ln concentraticoe -c F;.,--11J S l J "1 -1. kc" Ln' . yr"n" - sn' nnpntf too low on one tube. Poor loading of the sample tu be the cause of the failure: in these cases; however, sing-le laboratory reported a breakthrough. Apart from theseI ,,vJJ.vfaLiluire causes, a critica parison of the work.-up and analysis procedure of the ent laboratories is I t i p in the proposal. 'Concerning the quality of the method, the f o b eidelines for the reproducibility are up to 25%: good 25-50%: acceptable more than 50%: unacce'nY'"-"h"lP ' Comparison of these criteria with the results in Ta suggests that the method is not suitable for the det tion of tetrachloromethane and trichloromethane. I ments are necessary. For the other three compou reproducibility is sufficient for practical use. For tric TABLE V trichloromethane 56.0 <O.l <0.1 tetrachloromethane 34.0 l , l , l - t ~ c h , o r ~ ~ a ~59.0 <<o0..l1 <0.1 <o.l 13.0 3.0 5,0 tric hloroethene c 0.1 <0.1 <0.1 < 0.1 c of the round-robin test. 34 ABI. M.My& Arsa: J. 1471 R r % a d - - , N c w b k . N Y 10018. -3. IS0 Precrsron of Test Methods Determinatron of n f t b ~aSiitutcsare obtainable .* * I, I , Rue de - :._-_I--A ability and Reproducibilrty by lnterlaboratory Tests (IS05725). Geneva, 1981 pp. 1-42. -4. IS0 Part I: Calibration Methods; Part IV. Conti tion Methods. Gas Analysis Preparation 0 Gas Mixtures. Dynamic VolumetrtcMethods (IS 'iIf Geneva. 1982. pp. 11-28 (Part I); pp. 1-6 (Part IV). -5. SO: Statistics Vocabulary and Symbols, 1st e.! References 4 I . nAw---iw---mwm +b-,u-4fm--m-rwr-iwmiru .mu-1 *-a. -I---- s-1 I.rr(..d.i-l UUI~IIIIIIS~IU~ IIIYYI...-. Y.uyw-i--...r-- 3534). Geneva. 1977. p. 28. 6. Van der Wal, J.F.: Unpublished mSUltS. ists: Documentationof the Threshold Limit Values*. 4th ed. 7. MacFarn, E.F., R.J. Lishhlrr and J.H. Parkor. Cincinnati, Ohio: ACGIH. 1981. -- ---2. 1SO: Workplace A..t-m...mnIm- h.-nms Determina.tio-n o.f Concentrations of Chlcirinated Hydrocarbon Solvent Vapours ~nAir Judging Acceptability of Analytical Methods. 42~358-365(1970). 8. Mi-, 0. and E.F. Cmrlord: Crrterion for Jud by A.H.Brown (ISO/TC 146/SC2/WG4 N41). Delft, The Acceptability of Analytical Methods.Anel. Chem./9 Netherlands, 1983.pp. 1-16. (1977). 5 November 1984: Reviml 19 June 1985 36