Document 9JrY7N3RD5yvD1XVngExXVnaR

M&6-3397 000017 P-2 M M - i<7 E. I. du Pont de Nemours end Company, Inc. Polymer Products Department BLOOD Code Bo. B U65.50OO Original 3 Apr 01 Page -1 of 22 PeterminatioB of Perfluorooctanoic Acid Gas Chromatograohic Method I. Scone and Amplications Phis is a method for the determination of perfluorooctanoic acid (Co) and its salts in blood. Approximately 1 g of sample is required for measurement of concentrations down to 0.01 ug/g. ~ The method has been evaluated for analysis of human and rat -whole blood, but preliminary experiments indicate that it could also be used for senna. A similar procedure can be used for the analysis of aqueous solutions (as from air impingers), and a modified method used for other solutions and solid samples. II. Principle (See Bote 1.) The water is removed from sample aliquots by freeze-drying (lyophilization) to permit derivatization. Addition of methanolic HC1 to the dried residue, along with periluorodecanoic acid (C10) internal standard, converts the acids, which are not amenable to GC analysis, to more volatile methyl esters. These are extracted from the reaction mixture j-nto hexane solution for GC analysis. An electron capture detector (BCD} is used because of its sensitivity and selectivity, with either a packed or a capillary column. _+ & B C 1- - CTF15C0 m + '-a3H Ct?15COCCH3 + MOE M+ = BHU+ , Ba+ , E+ III. Interferences As in any GC analysis, conroounds with the seme retention time as the compound of interest will not be distinguished from it. These can include impurities present in the reagents or other components in the samples themselves. ' Interfering peaks from solvents, derivatizing reagent, etc., ere not generally observed in this laboratory with the reagents specified below (See Bote 2), but each new batch should be checked and a blank included in every analysis. A small peek at the Cg ester position is found in reagent blanks containing perfluoroaecanoic acid internal standard, probably from Cg present as an impurity in that material. This will give an intercept slightly greater than zero in the calibration plot, but will be a significant contribution to the Cfl peak area cniy at the lowest concentrations (about 1/3 of the total for'0.01 yg/g Co, when about 1 ug/g C1Q is added). Company Sanitized. Does not contain T SC A CBI 000018 p.3 Page 2 o f 22 A very small peak at that position has also been found in some "bleak blood" sasples, as shown in the example in Figure 2a. Whether this represents a trace of Cg (about 0.001 - 0.005 Wg/g) or some other component of the blood would be difficult to determine at this low concentration and with the limited number of samples examined to date. IV. Sensitivity. Precision, and Accuracy The method has been used over the concentration range 0.01-150 Wg/g Cg. Above 10-20 yg/g, however, interferences with the C^q Internal standard are generally encountered so that some modification oust be made for analyses at higher concentrations (See Rote 15). Due to the relatively narrow linear range of the BCD, the calibration curve must always cover the regiou of interest for quantification. From the data available at this time, precision is estimated to be - 10i relative. Kecoveiy of spikes is quantitative within this uncertainty when calculated relative to the C^q internal standard, although absolute recovery is affected by the presence of dried blood solids (See Rote 3)._ Accuracy will also be affected by the choice of Cg standard (particularly when the packed colunn analysis is used), which should correspond to the fluorosurfactant composition of the samples as nearly as possible (See Rote U) . V. Apparatus Instruments and equipment used in this laboratory are specified here; equivalent apparatus can be substituted. 1 , Gas Chromatograph and Supplies - For packed column analysis. (See|Appendix 1 for capillary column analysis.) Hewlett-Packard 5830 GC with EP 18803 Electron Capture Detector, equipped for on-column injection with glass-packed columns 10 ft x 2 m m id glass columns (HP configuration 5 with 3CD adapter), packed with 10S OV-210 on 70/80 mesh Chromasorb W.AW.DMCS. and conditioned at 200C Hamilton 701R 10 yL syringe Company Sanitized. Does not contain TSC A CBI 000019 P-4 I 2. Iyouhilizer Code Ho. B b65.5000 Original 3 Apr 8l Page 3 of 22 < Labconco Bo. 75352 bendi-top freeze dryer (12-port, dry ice cooled) Labconco Bo. 75b06 (150 mL) or Bo. 75bo8 (300 mL) Fast-Freeze flasks, with Ho. 75U76 stainless steel adapters Vacuum, mump with auxiliary cold trap McLeod gage or electronic vacuum gage reading in the 5-0.005 torr range 3- Thermostated Beaction Block Pierce H o . 18800 Beacti-Therm Heating Module, with Heacti-Block to hold 2-dram vials and thermometer (Bloch C, Ho. 1880b, has 12 holes, but they must be enlarged to about 19 mm to accommodate the 2-dram vials. Bloch B, Ho. 18802, can he used as is, but will hold only 9 vials and provides poorer thermal contact.) Lab-line Multi-Blot Heaters can be used with the same blocks, and will hold 2-6 in a single unit. b. Vials and Septum Cans . Wheaton Ho. 22bB8b or Pierce No. 13023 2-dram (about 7 mL) screw cap septum vials (borosilicate glass) Pierce Ho. 12713 T3FL0N-- silicone septa and Ho. 13216 open-top screw caps . 5 . Pinets and Dispensers Gilson P200 Pipetman variable volume micropipet for quantitative delivery of 20-200 PL aliquots 1 , 2 , and 5 mL volumetric pipets for measuring saurole and reagent aliquots. (Brinkman Dispensette bottle-top dispensers are a great convenience for repetitive delivery of solvents.)6*8910 6. Disposable hypodermic syringes, 1-3 mL, if blood samples are to be takes from closed "Vacutainers". 7- Volumetric flasks, 10, 25, ana 50 mL 8. Branson 3-220 ultrasonic cleaner (See Note 5.) 9 . Analytical balance 10. Common laboratory equipment (including a refrigerator for storage of samples) . Company Sanitized. Does not contain TSC A CBI ooozo P-5 Page U of 22 VI. Reagents 1. Cg standards: Perfluoro-n-octanoic Acid (PCR Research Chemicals), PC-IU3 asmwnium perfluorooctanoate (3K), or other material in use In the area to he monitored (See Rote h.) 2. Perfluorodecanoic acid (PCR Research Chemicals) 3. Methanol (Fisher BPLC (A-U52) or Fisher Certified ACS (A-ll2)) (See Rote 2 .) **- Derivatization reagent, 3% HC1 in methanol, prepared from Applied Science Ho. 18053 Instaat Msthanolic HC1 Kit, but substituting Fisher HPLC or ACS methanol for the Lipqpure methanol provided (See Hote 2). The reagent should be stored in the refrigerator, and can be kept for about 1 month. 5. Hexane, Phillips Seeetrograde or Applied Science U p o p u r e (See Hote 2 .) 6 . Veter, distilled or deionized 7 . "Blank" Blood, to spike for standards. Heparinized "Vacutainer" blood collection tubes (3-D Ho. 6^80) have generally been used for this, as well as for samples. Vacutainers containing liquid EDTA anticoagulant can also be used. NOTE: All blood samples must be stored under refrigeration. 8. Perfluorodecanoic acid (C.-) standard solution in methanol (See Hote 6 .) 10 Prepare a stock solution and sequential dilutions as follows: Solution X ! Preparation l& q ], Ug/mL Weigh (to 0 .1 mg) about 25 mg perfluorodecanoic acid Into a vial. Add 10 mL methanol and shake to dissolve. (Solvent volume can be adjusted as needed to correspond to the actual sample wt.) 2500 [F] Ug/mL 1755 II Dilute 2 mL I to 50 mL 100 in methanol. 70.2 III Dilute 2 mL II to 10 mL in methanol 20.0 lh.0 Company Sanitized. Does not contain TSC A CBI 000021 Code Ho. 3 U65.5000 Original 3 Apr 8l Page 5 of 22 ' 9. Perfluorooctanoic acid (Cg) standard .solution in water (See Note 6 .) Ps^iluorooctanoic acxd is only sparing ly soluble in water, and dissolves vary slowly. At least'1-2 hr ahking is required at this concentration (about 2 mg/mL), and It is best to prepare the stock solution one day in advance of the dilutions. Acid con centration can be checked by titration. Concentrations will vary, but solutions are needed from which at least four spiked blood standards can be prepared to cower the range of interest. For the example shown in the chromatograms ana calculations below (0.01-0.9 yg/ml), a stock solution and sequential dilutions would be prepared as follows: Solution Preuaration [Cg], Pg/mL [P], Pg/mL I ` Weigh (to 0 .1 mg) about 2200 151k k5 mg perfluorooctanoic (for k U m g acid into a vial. Add sanle) 20 mL distilled water and shake to dissolve. (Solvent volume can be adjusted as needed.) II Dilute 3 ml I to 50 mL 132 in HgO. 90.6 III Dilute 5 mL II to 25 mL in H2O. 26.k 18.2 IV Dilute 2 mL II to 50 mL in HgO. V Dilute 3 mL IV to 25 mL in 0. VII. Oneratine Conditions VIII. Safety and Health Precautions 5.26 0.63k 3.63 0.k36 Human blood should be considered potentially infectious, end the Laboratory Guidelines (See Reference 1 ) for handling, storage, end disposal of samples should be followed. The 3M Com)any has found in preliminary studies that Cg caused birth defects when fed to rats in a laboratory experiment. Female emoloyees of childbearing capability should not engage in any portion of this procedure which offers significant chance for exposure. Although no health problems are known for workers exposed to'cg, it has been found in blood samples and may be very slowly eliminated frcm the body (See Reference 2 ). Standard laboratory safety practices for handling toxic, embryotoxic, etc., materials end for corrosives should be used in working with these perfluorinated acids and their solutions. Company Sanitized. Dees not contain TSC A CBI 000022 P-7 Page 6 of 22 Y H I . Safety and Health Precautions (Coat'd) The electron capture detector contains a radioactive source (^3Ni), and the manufacturer's instructions for safe operation must be observed. Care should be taken, to avoid contact with the low-temperature baths used for freeze-drying. IX. Procedure A. Preparation of Blood Samples and Spiked Standards for Analysis (See Note T.) All blood samples should be well shaken, before taking aliquots to insure & uniform distribution of cells and plasma. 1. Spiked standards For each standard to be prepared, and for one or more blanks, pipet a 1 mL aliquot of "blank" blood into a 2-dram vial. Spike each with Cg standard solution to give the range of concentrations required, as shown in the following example: Suike 33 pi 26.U jig/mli C8 in water 110 uL 5-28 yg/mL Cg in water 69 " hi " 110 UL 0.631* Ug/mL Cg in water 16 " Blank (no spike) [Cg], Ug/mL blood 0.871 0.581 0.36b 0.216 0.0697 0.010 0 [F], Ug/mL bkx>d 0.600 o.uoo 0.251 0.ll*9 0 .0U80 0.0070 0 Mix gently, keeping the solution in the bottom of the vial, and sonicate for about 2 min for additional mixing (See Hote 5) Store in the refrigerator until needed (See Hote 8). 2. Samples If the sample volume is sufficient, prepare in duplicate. For each samnle, pipet a 1 sL aliquot into a tared 2--dram vial and weigh to'0.1 mg. (Aliquots can also he taken with a disposable syringe, if it is felt preferable not to open the containers). Company SanHfced. Does nol contain TSC A CBI OOOi p.8 B. Freeze-Drying _ Code Bo. 3 U65.5000 Original 3 Apr 8l Page 7 of 22 1. A schematic diagram of the freeze-drying apparatus is shown in Figure 1. Bach of the 12 ports has a separate waive so that flasks can be attached or removed while the system is under vacuus. The apparatus should be pumped down to an acceptable vacuum (See Bote 9) and the cold traps filled before the drying flasks are attached. 2. Freeze the samples before placing them under vacuum to prevent bumping. To reduce the chance of contamination or samole loss acl1 vial with a light filter beforehand.; a small piece ' 0 "iinvipe held In place with. & rubber band works well. Freeze the solution in each (See Bote 10) and keep the vials at dry ice temperature until all of them are ready. 3. -Place up to four vials in each crying flask and attach to the drying chamber. Use pressure should drop below 0.5 torr within a few minutes if no leaks are present, and the ssarnies should remain frozen as sublimation takes place. Allow them to cry for about U hr (See Bote 11). C. Derivat ization. 1. To the dried residue in each vial, add 2 mi aiethaaolic HC1 and 50 pi. Cj_o standard solution (about 20 pg/ml in methanol). Prepare a reagent blank similarly, using an empty vial. Close with a septum screw cap and shake well. Sonicate for about 2 min (See Bote 5) and shake again. 2. Thermostat for 1 hr at 65C, then allow the vials to cool to room temperature and stand overnight (See Bote 12). 3. Add 2 mi each of hexane and distilled water and shake for about 2 min. ~ k. After the phases separate, the upper hexane layer can be sampled directly for GC analysis (See Bote 13). The solutions Iere stable in this form for at least several days, although hexane will gradually evaporate after the senturn has been pierced. ' D. GC Analysis - Packed Column. See Appendix I for capillary column analysis. 1. GC Conditions - instrument and column as shown in V, Apoaratus. Temperatures: Injection port Detector Coi nmr. oven 200C 325C 100C, isothermal Company Sanitized. Does not contain TSC A CBt 000024 P-9 Page 8 of 22 Carrier gas: 90% argon/lO* methane (or nitrogen if recommended for the instrument) flow about 30 mL/min Injection Volumei 2 yL Sensitivity: Attenuation as needed to keep peak height measurable, 25-2-3 on HP 5830. Bun time: 11-25 min, depending on sensitivity (See Note l U .) After Installing the column and establishing the conditions above, check the ECE base frequency and noise level according to the manufacturer's instructions, to insure that they are stable and within acceptable limits. 2. Bun each solution twice (or run duplicates once each), bracketing samples with standards of similar concentration in the second series of injections. Representative chromatograms are shown in Figure 2 for blanks, standards and samples. Measure either peak height or peak area. As noted above, the calibration curve must cover the concentra tion range of the samples. If some samples are found to fall above the range of prepared standards, the hexene solution can be diluted with that from a blank (e.g., 100 yL + b00 yL for 5 X dilution) to reduce the Cg concentration without changing that of the internal standard. This type of dilution can also be used if peaks interfering with the internal standard are observed at high Ca concentrations in either sample or standard preparations (See Hote 15) X. Calculations (See Note l6 .) 1 . Normalised C0 Peak Values A tabulation of peak heights for analysis of a series of standards (prepared as above) and samples as shown in Figure 3a- To correct for any variations in injection volume, reagent volumes, etc., the raw Cg peak heights (or areas) are first normalized relative to the internal standard average: _ . _ c10 (average) Co (corrected) = Cg x Ll0 Calculation of the C^g averege also provides a measure of the orecision in sample preparation and GC analysis. *-o m P anySnHbed. O o M nolcwrtatnTSCACB, 000025 p. 10 Code Ho. B U65.5000 Original 3 Apr 81 Page 9 of 22 -- 2. Calibration Plot A calibration curve obtained by plotting corrected Cg peak height vs concentration of tbe spiked standards is shown in Figure 3b. In this case, straight lines can be fitted to sets of points at higher (0.61-0.15 yg/mL) end lower (0.15-0 yg/mL) concentrations, and the slopes rid intercepts calculated by linear least squares analysis. In some plots no really linear region may be found, and a smooth curve should be drawn through the data points. 3 . Calculation of Sample Concentrations' Using the corrected Cg peak heights (or areas) for the samples, corresponding values of yg F/sample can be calculated from the T.Tfl equations (as in Figure 3) or can be read directly from the calibration curve. Divide each value by the sample weight (in grams) to obtain g F/g blood, and average the results for duplicaLa runs or duplicate preparations. Report as Ug F/g blood (See Note 6). XI. Calibration (See Under X, Calculations.) XII. Botes 1. Initial method development was based on a procedure for determination of perfluorooctanoic acid in serum or plasma recently published by 3M (See Reference 3). In that method, Cg is extracted from acidified serum into hexane/ether, the solution concentrated by evaporation, and diazomethane added for derivatisation. With the lyophilization procedure described here, sample handling, time, and material required ere reduced, and the use of diazomethene is eliminated. However, concentration of Cg from a large volume of sample as in an extraction, is not possible. A sample preparation procedure similar to that in tbe 3M method has been used for aqueous solutions in 'this laboratory and can be followed by the same methanolic HC1 derivatisation and GC analysis as lyophilization. The extraction has net been tested with whole blood in either method, however. GC analysis was initially done with a packed column (as in the 3M method but with a different stationary phase). A capillary column has also been used recently for better resolution of the branched Cg Isomers, and modifications for that analysis are given in Appendix I. I Company Sanitized. Does not contain TSC A CBI 000026 p. 11 Page 10 of 22 2. The solvents specified in Section VI vere found preferable to several tried for this analysis, but others may also be satisfactory. Sven with the same reagents, the appearance of the blank may depend on the condition of the detector. Kith the 2CD used here, contamination results in a significant decrease is. selectivity and increased reagent interference. 3. Recovery was determined by spiking blood, vater, and methanol solutions at equivalent concentrations, obtaining the response factor for Cg relative to C,g in GC analysis of the vater and methanol spikes, end using this factor to calculate the Cg concentration in the blood spikes. Average recovery (Cg found/added) was 100 i 55 for 18 samples of human blood spiked with 0 .0lh-U6 Ug T/aL (FC-lb-3 or perfluoro-n-octanoic acid) end for T seamles of rat blood spiked with 0.007-67 Kg F/mL. Absolute Cg and Cj_q peak heights vere about 15-205 lover in the presence of dried solids from 1 g blood, however, so that a corresponding error would result from external standard calibration against aqueous solutions. The question of whether recovery of spikes provides an accurate measure of recovery of the same material from samples should also be considered, particularly with a complicated substrate such as blood. Agreement between the results of this analysis and those of Wickbold Torch analysis for organic fluorine in a series of test samples seems to indicate that the spikes are & reasonable model for samples. k. The FC-ll3 samples examined to date all appear to be a mixture of . about 70-755 straight-chain material with several breached isomers present at about 1-105 (determined by F-19 HMH, capillary GC, and GC/MS). In contrast, commercial perfluoro-n-octanoic acid (?CR) contains about 955 straight chain Cg with about Uj (CF^jC^C^^UCO*- and traces of the other isomers. As shown in the chromatograms in Figures 2c and 2d, in the packed column GC analys i s , only one of these is really separated from the main peak (small peak at 3*7 min) while others are visible as a shoulder following i t . Chromatograms of sass^les and standards obtained using the capillary column analysis are shown in Figure h (Appendix I); under those conditions, at least partial resolution of four Isomers from the main peak can be obtained. Since ECD response appears to be higher for the isomers than for the straight-chain ester (by a factor of 2 or more), the total area/ concentration ratio also becomes higher as their percentage increases. In the packed column analysis, the response factor calculated for Cg relative to C^g is about 30-l05 higher for ?C-lb3 standards than for perfluoro-n-octanoic acid standards, so that it becomes very important to use the one most similar to the samples to be analyzed. Although small variations in isomer distribution can be seen in human blood samples, alT of those examined to date resemble perfluoro-n-octanoic acid much more than FC-1^3, and it has been an appropriate standard Company Sanitized. Does not contain T SC A CBI 000027 p. 12 Code No. 3 U65-5000 ` Original 3 Anr 81 ' Page 11 of 22 m . Notes (Cont'd) k. (Coat'd) for quantification within the uncertainty cf the method. In a series of rat blood samples, however, considerable variation in distribution was observed. As seen from the chromatograms in Figure U , neither standard would have been a good approximation to some samples, and analyses were done with the capillary cclum. 5. Gentle sonication (with the vials immersed about half-way in the water bath of a sonic cleaner) has been used to aid in mixing^ of spikes and In promoting contact of derivatizing reagent with the dried blood solids. Additional experiments are planned to determine whether this step is necessary or can be eliminated. 6. Since results that are to be compared with nonspecific analyses for * total organic fluorine must be expressed as fluorine rather than es Cq , both concentrations are given for the standard solutions. Values for uercent fluorine used In the conversion are: perfluoro- octanoic acid, 68.8255, ammonium perfluorooctanoate, 66.10?, perfluorodecanoic acid, TO.22?. 7. An aliquot volume of 1 mL is generally used for both samples and standards, but it can be reduced to 0.5 mL when concentrations are expected "to toe relatively bigh. Coagulation is generally a protoles vitoh rat blood samples kept for more than a week or two before analysis. When it is no longer a homogeneous liquid, the entire sample must be lyophiiized, mixed veil, sad portions of the dried powder weighed out for _analysis The exact ratio of dry to liquid weight must be determined for each saanle, but is about 0.2, so that 0.2 g powder will be comparable to 1 mL liquid. This procedure greatly increases the time required for sample handling and the probability for error in the weigit, and it should be avoided if possible. 8. Soiies nrenared for method development have been allowed to stand at least overnight before analysis, but a period of a few hours aDcears bo give oquivaletlx, results. O. For thf freeze-drying auolication, pressures less than 0.5 torr ' seem to be satisfactory, although 0.025-0.05 torr may be achieved. Readings will also depend on the type of gage used, since the McLeod gage does not read the pressure of water vapor as electronic gages do. Company Sanitized. Dees not contain TSC A C M 000028 p. 13 Page 32 of 22 10:' A small dry ice/acetone bath or the center cold veil of the drying chamber cen he used for rapid freezing of the solutions in the hotton of the vials. As noted in Section VIII, care must he taken to avoid contact with the cold hath. 11. A U~hr drying time has been adopted for preparation of blood samples by comparison vith 1 mL vater samples, vhere drying can be monitored by disappearance of the ice. Experiments vith spiked blood samples indicate that longer drying times (8-15 hr) can also be used without loss of Cg. 12. Derivatizotion mixtures from aqueous samples can be partitioned vith hexane /water immediately after cooling. In the presence of dried blood solids, however, additional time for equilibration of the C8 end Cio esters seems to be necessary. After standing overnight, peak areas for both Cg and Cio are lover than in reaction mixtures prepared from aqueous or methanolic standards, but the relative response factors are equivalent (See Note 3). Preliminary experi ments indicate that it might be passible to eliminate the overnight equilibration by spiking vith aqueous C10 before lyqphilizstion, rather than vith methanolic C10 after. 13. Although a clear hexane layer is obtained after standing for about l/U-1/2 hr, the solution has a relatively dark rust color and - undoubtedly contains nonvolatile or high-boiling impurities. In the packed column analysis, these ere injected onto the column head; and at the end of the day, the oven should he programmed to 180C and held for several hours to reduce huild-up. No deterioration of performance was observed in a column used for several weeks of analyses, but it may prove preferable to adopt flash vaporization vith a glass injection port liner which can be changed daily, as in the capillary column analysis. lU. As seen in the chromatograms in Figure 2, some small, broad peaks can appear late in the run. These appear to come from the Cno standard and might be eliminated in purified material; others seem to be characteristic of older blood samples. At the high sensitivity used for low Cg concentrations, these can interfere vith the following run and must he allowed to elute Defers the ne^t injection. 15. Several minor components are present in both the perfluoro-n-octescic acid end F 0-1^3 standards which elute near the internal standard under these GC conditions and can interfere with quantification. Similar peeks ere also often observed in samples at higher concentrations. With the Ct_q concentration generally added (about 1 Ug/g), this seems to occur above about 20 ppm Cg in the capillary column analysis end about 10 ug/g vith the packed column. In some cases, an obvious shoulder can be observed on the Ct_q peak, varying in site vith the Cg concentration. In others, an unresolved peek may he indicated by an unusual change in Ct_q peak size, but is only seen if a sample is prepared without added internal standard. Depending on the type of analysis to be done, various techniques can be used for quantification under these conditions. 000029 Code Ho. B 1*65-5000 Original 3 Apr 81 Page 13 of 22 (a) Standards can be prepared only up to about 10 pg/g, and ^ aay samples found to be above this range diluted and run again, as described in IX, Procedure, Part D. (b) If only relatively high concentrations are of interest, the Cjq concentration can be increased and/or the sample size decreased so that the interference becomes negligible. (c) With the increased resolution of the capillary column, components of the perfluoro-n--octanoic acid standard can be resolved iron the Ctq and the problem eliminated for standards. Samples must still be examined for interference, however. (d) The calibration curve can be used for an external standard calculation, with uncorrected Cg peak areas or heights. While good reproducibility is possible with care in measuring the 'injection and reagent volumes, it is obviously more critical here then in an internal standard calculation and must be checked regularly at each point. 15. Prom the calibration plot in Figure 3, it can be seen that the BCD response is not linear with concentration over the ent-'re range. Linear plots may be obtained in some cases, but they still have concentration-dependent response factors. The calculations described here thus make use of a calibration curve, in which the internal standard is used to normal ise raw peek values before plotting. For an BCD which shows a more linear response and constant relative response factors, calculations could be done by a general internal standard procedure instead. (A calibration plot of this sort has been observed with a Yarian 3T00 GC under similar conditions). XIII. References 1. "Experimental Station Rules and Procedures for Human and Other Primate Blood and Blood Products", 3/80, Safety Office, Experimental Statical. 2. F. A. libel, S. D. Sorenson, end D. E. Roach, "Health Status of Plant Workers Exposed to Fluorochemicals - A Preliminary Report" Am. Ind. Eygl Assoc. J. fel. 58U (1980). J* Beiisle and D. F. Eagen, "A Method for Determination of Perfluoro- octaaoic Acid in Blood and other Biological Samples", Anal. 3iochem. 101, 369 (1980). ` Origin: Experimental Station Prepared by: S. S. Stafford Approved by: Approved by Laboratory Methods Committee, 9 Nov 8l L. J. Papa 2S-567 Company Sanitized. Does not contain TSCA CB I 000030 VIQUIIE .1 - APPARATUS FOR PHBEZE-PnyiNG SAMPLES (a) Pryinw Chamber and Vacmun Line Dry Ice/Iaopropanol Cold Well, (b) Vacuum Fort. Drying Flask, and Samples Crying Chamber Code Ho. B U65.5000 Original 3 Apr 8l Page lU of 22 Company Sanitized. Does not containJBCACBI o o Ol CO b* Code Ho. B U65.5000 Originel 3 Apr Si Page 15 or 22 FIGURE 2 - CHROMATOGRAMS KIR Cs DETERMnwTTOiir, pararen rnT.nMW wat.y s i S GC conditions as on p. 8. Attenuation as noted for 0-6 min, 2^ after 6 minutes. j Chart speed 1 cm/min. Retention time (minutes T indicated by each ueak. (a) Blank blood, no Cq spike or C1Q standard added (9A/80) (b) Standard, blood spiked vite 0.01 tig/mL perfluoro-n-octanoic a d d (0.007 UfC F/mL) (9A/80) C0 ester "5 attn 23 Jr ^ Cl (e) Standard, blood spiked vith 0.9 pg/mL perflucro-n-oetanoie acid (0.6 p k F/mL) (9/10/80) (d) Semole E2251I*-2S-1 (9/10/80) Company Sanitized. Does not contain T SC A CBI 000032 FTOURE 3 - OAI.OULATIOH OP COWIBOTED PEAK HEIOIITa AND CALIBRATION PIOT (u) PtiMk height values Tor runs of seven sUimlurd and five sample solutions (duplicate ruim were made, and shown un Ilia calibration plot, bub only one is included In the table). (h) Calibration plot, corrected Cg peak height vs. concentration (expressed as fluorine) Code Ho. B It65.5000 Original 3 Apr 8l Page 16 of 22 Company Sanitized. Does not contain T SC A CBI Peak Hal|lit, m at att27 Run Solution [Cq], wgF/tuaplc Ho. flu Cg - Raw C10 Corrected - - 79 a * ClT iiUcil SlandrdE o.6ov 8 93 0.399 0.250 0.11*9 6 71.1 10 1*9.9 12 3b.l 0.Ot80 23 12.6 0.0070 1* 3.80 BlanK t c10 3 1.93 7 78.1 77.5 79.9 78.2 81.T B0 102 71.9 50.9 33.7 12.T 3.67 . i.W Avg. (11 valuaa) - 79 3 Calculated* Staples 0.1*91 0.1*09 9 BT.8 7 72.5 0.330 3 1*8.3 0.191 11 36.3 0.136 20 38.1 80.8 7*.5 82 7 0 .8 75.5 85.8 73.5 1*6.5 1*0.7 39.9 ^Sample concentrations calculated from linear least oquures fit, 0.6-0.15 ppin standards (0 points); li * 11.98 + 150.3 [Fj co 0.15-0 ppm standards (5 points) ; II " 2.13 + 2lli.l| [F] o o CO CO Code Ho. B U65.5000 Original 3 Apr 81 Page IT of 22 APPETTOIX I Modifications for Can't11 ary GC Analysis As discussed in the method, a capillary column he substituted for the original packed column in analyses where separation of the Cg isomers is necessary for quantification, or Where determination of the isomer distribution is of interest. Chromatograms obtained with the column and conditions currently in use are shown in Figure U. A wide-bore WCOT is used to permit large injec tions (l-3 pL, es with the packed column) for detection of low concentrations although the efficiency is probably less than that of a lower capacity column. The major isomer peaks are generally resolved well enough for quantification, however, or at least to indicate whether a significant peak is present (as in Figures Uc and d). Comparison of these chromatograms with those in Figure 2 shows that the analysis time is presently longer than with the packed column; this may be reduced b y further column modifications. Preparation of samples and standards is the same for both analyses. Modifications needed in apparatus, GC conditions, and calculations for the capillary analysis are as follows: 1. Apparatus (7) HP5830 GC with HP18803 Electron Capture Detector, equipped for splitless injection with glass capillary columns. A flame ionization detector should also be available on this or another capillary instrument. Glass injection port liners (inserts) for use with the capillary system in splitless mode. Silanized glass w o o l for packing injection port liners. Glass WCOT column, 50 m x 0.5 mm, 0V--210 (high load). (The column currently in use is a custom-coated column from Alltech Associates. It can also be specially ordered from Quadrex or Chrompaek, but is not presently available as a stock column.) 2. GC Analysis (|IX, Procedure, Part D) Instrument and column as above; straighten the column ends, recoat them with FFA? in CH2CI2 , and install the column according tc the manufacturer's instructions. Before inserting the glass injection port liner, pack it loosely with about 1 cm silanized glass wool. This should be positioned just below the point of injection, to prevent nonvolatile sample components from reaching the column. Company Sanitized. Does not contain TSC A CBI 000034 p. 19 Page 18 of 22 GC Conditions: Temperatures: Injection port ' Detector 200C 325C Column oven programs (a) 15 min 55C, 8/min to i30c 5 min 130C "' (1>) 10 min 6oC, 2/min to T6C then 6/min to 120QC, 6 min 120C Carrier and make-up gas: 90% argon/10? methane column pressure 3 psig, about 2.J mL/min flow at 100C make-up about 26 mL/min (total flow about 30 mL/min) splitter about 100 mL/min Injection -volume: 2 yL, splitless injection with 60 sec hold before ' opening splitter vent Sensitivity: Attenuation as needed to keep peak height measurable 25-213 on HP5830 Hun time: About 30 min (depending on program) plus cool-down. Temperature program and column flow rate will vary with the condition of the co l u m , instrument configuration, carrier gas, etc., and must be chosen and modified as needed to maintain adequate resolution. Column oven program (a) was used for the chromatograms in Figure U; program (b) and similar ones have also been used. After installing the column and establishing the conditions above, check the ECD base frequency and noise level according to the manufacturer's instructions to insure that they are stable and within acceptable limits. Base frequency should also be checked before each day's operation. Maintenance of a clean system is especially imoortant for temperature- programmed ECD analyses at high sensitivity. A clean injection nort line- and new septum (low-bleed) should be installed at the end of each day, and the column ovpu temperature raised to 130C (or the m r i m m program temperature} overnight and when the instrument is not in u s e . 3- Calculations (X) The calculations ere done in two parts, first determining the concentration of the main component, then adding a percentage correction for the secondary isomers. (a) Concentration of uerfluoro-n-octanoate Calculation of normalized CQ peak values, preparation cf the calibration curve, and use of it and the sample weights to calculate ug/g for each Company Sanitized. Does not contain TSCA CBI (j O0 0 3 3 Code Bo. B U65.5000 Originel 3 Apr 81 Page 19 of 22 sample ere done as in the original method. Since only the area of the main peak is used .-however, the concentrations of the standards must first he corrected to correspond to it rather than to the total Cg concentration. For the perfluoro-n-octanoic acid standard used is this laboratory, the concentration of straight-chain acid is taken to be 95? of the total (See Bote t above). For some sarnies, such es the one shown in Figure Ud, the isomer peak Just ahead of the peak may not be large enough to be resolved but still makes a significant contribution to the total. Peak height rather than area can be used to minimize the error in quantification of the main peek. (b) Correction for secondary canroonents . (l) Relative response factors As discussed in Bote U above, ECD response for the branched isomers is higher than that for the straight-chain ester, so that relative response factors must be determined in order to calculate their concentrations. Since FID response appears to be more nearly equal for all o f tbe isomers, the factors can be estimated by comparing peak area percent values for standards analyzed on the same column by the two different detectors. Differences in carrier gas, reagent interference, and relative peak sizes make it difficult to reproduce the chromatography exactly, but conditions should be chosen to make the analyses as similar as possible. Calculate average area percent values by both detectors for the groups of five Cg isomer peaks, then use the ratios to determine ECD response factors relative to the main peak, as in the following example for an FC-1 U3 standard: Peak Number, lU.95-17.22 min in Figure Ub ECD area percent (average of several runs) FID area percent (average of several runs) #1 7-38 #2 #3 (main peak) 1 1 .91* 51.76 #k #5 10.02 18.91 3.55 U.38 72.88 8.U5 10.73 w ' / Ai ) x 1 y W i n / S C D \ Ai /FID 2.93 3.8b 1.00 1.67 2 .U8 Company Sanitized. Does no! contain T SC A CBI 000036 p. 21 Page 20 or 22 For perfluoro-n-octanoic acid^standards (Figure ta) the only measurable isomer peak is 5.' The factor calculated for it generally differs somewhat from that for FC-lU-3 standards, as might be expected from the incomplete resolution and different contributions from neighboring peeks. For samples ranging in type between the two standards, the average factor is usually taken. (2) Percent correction for total concentration For each sample run, use the peak areas and response factors to calculate the concentration of isomer peaks relative to the TMfn peak as in the following example (peak numbering as in the factor calculation above): Figure Ue; Sample 80-68609, Run #21, 12/30/80 Main Component Concentration T.63 lig/g F Peek Ho. K Peak Area (integrator units) i of 1 2.93 2 3.8U 3 1.00 (main peak) h 1 .6J 5 2.73 83 560 TO U60 59 UOOO 32 U30 20 9300 U .80 3 .09* (100) 3-27** 12.91 Total % correction = 2U .1 T.63 x 1 .2U1 = 9-5 ppm F total "Peak #2 is frequently visible as a shoulder, but not resolved sufficiently for quantitative measurement. Depending on the size, the correction may he considered insignificant or can be estimated b y comparison with tbe other isomers. **This value ip probably too high, by comparison with peeks #1 and #3 in the chromatogram in Figure Ue. Since the correction is only a few percent, however, the effect on the total concentration will not be appreciable. These calculations clearly involve many approximations, especially for samples unlike either of the standards in isomer distribution. Determination of the isomer concentrations is difficult in those cases, since the "response factors" for a group of poorly resolved peaks will be influenced by the relative sizes and contributions from each other as well as by detector response. For peaks #2 and gh in particular, the individual values must be considered estimates only. Since the contribution of each isomer to the Company Sanitized. Does not contain TSCA CBI ^00037 p. 22 Code Bo. B W5-5000 Original 3 Apr 8l Page 21 of 22 total concentration is usually small, however, the overall uncertainty will probably still b e less than if the composite peak were used and no correction is made for differences in conposition. In mast of the blood samples examined to date, the composition has appeared similar to that of perfluoro-n-octanoie acid (See Bote h ) , so that the isomer correction is small and calculated for the well-resolved peaks 1 and k. For those samples, the packed column analysis should give equivalent results (within the uncertainty of the method), but the capillary analysis has the advantage of verifying that a significant error has not been introduced by undetected isomer peaks. I Company Sanitized. Does not contain TSCA CBI 000038 /3/B coodl cions as on p.io anerature program: \c 35*C, 8"/in 130*C, ,130 *C . urt speed 1 cm/nia tenuclan 2^ for 20 min, lafter <a) Standard, rat blood spiked with %>12 ppm perfluoro-o-oc Canole acid Loue uw> & **-:j^UUL Original 3 Apr 8l Page 22 of 22 esters e cio ester >15 ppm) TS !is= 5. a ssaa- S 'S **' ^ i t rr Company Sanitized. Does not contain T SC A CBI 000039 BEST COPY AVAILABLE