Document qMM7Zw42MXOqb1KyQmpjYyvK

THE MIGRATION OF VINYL CHLORIDE MONOMER FROM PVC PIPE INTO WATER J.D. Btnzer Diamond Shamrock Corporation T.R. Evans Research Center P.O.Box 348 Painesville, Ohio 44077 Introduction Vinyl chloride monomer (VCM) has been shown to be carcinogenic by inhalation.1 There is unpublished date by Maltoni and coworkers which indicate that, at high dietary levels, VCM is also carcino genic by ingestion. Because of this, the Pood and Drug Administra tion moved to revoke the prior sanctioned status of certsin PVC products used in food contact applications. PDA has defined potable water as a fooditem and indicated^ that it would revoke the prior sanction of PVC potable water pipe unless industry can supply data which demonstrates that "...there is little likelihood of VCM becoming a component of food:" In order to satisfy the above, FDA required an experimental study to determine the relationship between residual vinyl chloride monomer (KVCM) in potable water pipe and the amount of vinyl chloride monomer (VCM) that may be present in water in such pipe under static conditions. Laboratory experiments were conducted to relate the residual vai to the amount of vcm extracted by water. A variety of pipe samples were prepared containing a wide range of residual VCM levels. The extraction studies were performed at several different times to determine the effect of the age of the pipe on the amount of VCM extracted. Experiments were carried out under both static and flowing conditions. The samples were analyzed using both headspace and vapor extraction techniques. Experimental Two procedures were developed for the analysis of VCM in water. They are referred to as the vapor extraction method and the head-, space method. Vapor Extraction Method v The analysis is based on sparging of VCM by nitrogen gas out of water onto activated charcoal. The VCM is desorbed from the char coal with carbon disulfide and analysed by gas chromatography. The solvent desorption technique was chosen over thermal elution since solvent desorption allows several analyses of the same sample vs. a "one shot" analysis by the thermal elution technique. Three liters of water are placed in a 51 3-neek flask. A nitrogen bubbler is placed in one neck with the end below the surface of the water. The second neck holds a reflux condenser with an adsorption trap containing 5g of activated charcoal at the condenser outlet. The third neck is closed with a ground glass stopper. Nitrogen is bubbled through the water at a flow of 1.4 1/nun. while the water is stirred with a magnetic stirrer. After 1 hr. the charcoal is removed and prepared for analysis. The charcoal is placed in 4 sealed hypo vial, and the sample chilled in dry ice. 10.0 ml of carbon disulfide are added and'the vial is shaken for 30 minutes. A 2.0 ia 1 aliquot of the carbon disulfide is taken from the sealed hypo vial and injected*into the gas chromatograph. The area counts of the VCM pea:, are recorded. Samples were analyzed on e Hewlett-Packard S830A gas chromato graph equipped with a flame ionization detector. The col van was 3 feet x 1/8 inch stainless steel packed with 80/100 mesh Porapak Q. Instrument Conditions: Colimn temperature 115*C for 2 minutes, program at 30*C/min. to 145*C, hold for 2 minutes; Injection temperature 2S0*C; FID tetqwrature 275*Ci helium flowrate 36 ml/min. Solutions for spiking water were prepared by filling 6 ml HypoVials (Pierce-Chemical) liquid full with methanol and then sealing them with a Hycar septus and crimp cap. The vial was frozen in dry ice for IS min. A known volisse of pure (99+%) vinyl chloride monomer (Matheson) was injected into the chilled methanol. The vial was then shaken for 30 min. and warmed to room temperature. Methanol was used for spiking since it is water soluble. GC standards were prepared in the same manner except that carbon disulfida was used es a solvent. Solutions of distilled water were spiked at levels from 0.001 mg/kg to 0.1 mg/kg VCM. The percent recovery wes determined in each case end is given in Table I. The average recovery wee 94.2%. A calibration curve is given in Figure 1; demonstrating linearity over a hundredfold range in concentration. Based on the 3 t sample, a consistent detection limit of 0.002 mgAg is obtained. Headspace Method The analysis is based upon analyzing VCm in the headspace above a water solution contained in a sealed system using gas chromatography. Samples are prepared by pipetting IS ml of water into a 22 ml Hypo-Vial (Pierce-Chemical) which is then sealed using e Hycar septw and aluminum crimp cap. The samples are placed on a wrist action shaker for 30 minutes. Samples are analyzed by injecting e 1.0 ml volume of air into the headspace; then a 1 ml air sample is withdrawn and injected into the gas chromatograph. The area counts of the VCM peak are recorded. Instrument conditions ere identical to the vapor extraction method. The percentage of VCM above the water as s function of VCM concentration in water was determined for spiked deionized water. The results are given in Teble II. The results indicate the percent of VCM distributed in the headspace is 21.5. A calibration curve constructed from the data is given in Figure 2. The date indicate that good linearity is obtained over a range of .001 to 0.1 mgAg. Extraction Experiments One inch schedule 40 pipe (1M ID) was selected as the pipe size used for all tests because it has the highest surface to volume ratio of any common size PVC pipe. Twenty-five foot test samples were used m order to contain a volume of 3.0 1 of water. This is the minimum volume required for the vapor extraction analytical technique to permit a detection limit of 0.002 mgAg. Shorter lengths of pipe were used for long-term sampling because of storage problems associated with 2S* sections. Nhen shorter lengths were used, only headspace analyses were performed. Deionized water was used for ell studies in order to eliminate any potential matrix effects from contaminants in tap water. 1 ; Several conversations were held with water system personnel to determine typical residence time for water systems. No absolute values were offered, but most felt that 2-3 days was b good average value. The bulk of our data was run for/cither 48 or 72 hrs. Additional data was collected at 24 hr.,/ and, in some case, as Ion as 180 days, to provide as complete a ringe of time intervals as possible. '` The ends of all pipe sections were terminated with 1" steel ceps which were tapped to receive small ball valves. One end of GJ co C 30 T1 Che ball valve was terminated with a 1/4* Swagelok fitting. Pipes were filled by' elevating one end, drawing a vacuta on the pipe and sucking water into the other end through a piece of 1/4* Teflon tubing. Care was taken to displace all the air with water. When the pipe was liquid full, the ball valves were closed, and the pipe section was stored at roosi temperature for the duration of the test. using previously documented procedures.3 The age of the pipe in days at the start of Che extraction period is given in addition to the vcn level in the water. For all test result*, the following two significant conclusions are drawn: For pipe samples exposed under flowing conditions, the ends of the pipe were connected to a leak tight centrifugal puap (ColeParmer "Micro Puap*). Water was circulated through the pipe for the duration of the test interval at a flowrate of 0.74 l/min. At the conclusion of the teat interval, the pipe was pressurized with `4 psi prepurified nitrogen through the valve at one end. The other end was fitted with a scptua. Headspace samples were with drawn using a 20 at glass syringe fitted with a 20 guage needle. Three 15 el sanples were taken for headspace analysis. Each sample was transferred to a Hypo-Vial which was immediately sealed. The remaining water was transferred under nitrogen pressure to the vapor extraction flask through 1/4" Teflon (duPont) tubing. The entire transfer system was cloaad to the atmosphere so that any effluent gases passed through the charcoal trap. Nitrogen for sparging the water was passed through the pipe first, then into the flask to flush out ^ny entrapped pockets of VCM. Blank determinations were run on 3.0 f of water to insure that the water and all reagents were free of VCM. No VCM ( <0.002 mg/kg) was detected In any blank. Result* and Discussion 1. A decrease in the residual VCM content of the pipe reduces the VCM level in water. 2. A decrease in the extraction period reduces the VCM level In water. The most interesting results of these studies are the VCM levels in water for the pipes containing <1.0 and 2.5 mg/kg RVCM. In all cases, the VCM levels in water are at or below 0.002 mg/kg. This is true even for fresh pipe (17 and 19 days old) and long extraction times (150 and 160 days). The low RVCM pipe currently being produced will probably result in no detectable VCM in water using currently accepted analytical methods. A comparison of the results for the static end pumped runs does not show any consistent pattern. For the 2-day data, there was a three fold increase in VCM extracted when the test mode was changed from static to pumped. However, a repeated recycling from static to pumped showed essentially consistent static data and a decrease in the VCM extracted for the pixsped data. The 3-day data indicates similar results for both static and ptwped test nodes, considering all the static versus pumped data, it appears that in the majority of cases,there it a negligible difference between the two modes of testing. Comparison of Analytical Methods: Acknowledgements Accuracy data for the headspace method is shown in Table I and for the vapor extraction method in Table It. Both methods have been found to give comparable results on laboratory samples. A coeg>arison of results is given in Table 111. The author wishes to thank Reuss Griffiths for the RVCM analyses and Gil Oiets. Dennis Cocco and Wes McCoy for their assistance in extruding the pipe and fabricating the test loop*. The precision of the method vet determined at both the high and low ends of the calibration range. For the headspace method, the coefficient of variation was 16.9% at the low end and 7.8% at the high end. For the vapor extraction method, the values were 2.3% and 3.0%. respectively. Although both methods give detection limits of 0.001 mgAg on spiked recovery sanples. there was some day to day variation in sensitivity when the water from pipe aample* was analyzed. Consequently, 0.002 mg/kg was adopted as the detection limit since this value could be consistently obtained on all samples. p. l. viola, a. Biogotti, a. Capueo, Cancer Research. 31. 516-522 (1971) 40 Federal Register 40529, September 3. 1975 Results of attraction Experiments Results of all extraction studies are given in Table TV. Residual vinyl chloride monomer (RVCM) levels were determined TABLE I 3 R. r. Griffiths, `Residual Vinyl Chloride Monomer in pvc Products," HOM-1-JOA-O end HOM-1-30C-0 (Diamond Shamrock Corp. Plastics Div. Test Procedure). Recovery of VCM from Water Using Vapor Extracting Method Spike Level (mg/kg) % Recovery* 0.001 0.005 0.010 0.050 0.100 Average 99 100 105 72 95 94.2 Average of duplicate determinations ii O T 1 TABLE II Recovery of VCM from Spiked Samples using Headspace Method Spike Level (mg/kg) 0.001 0.005 0.010 0.025 0.050 0.075 0.100 Recovered* 24.4 21.4 22.5 21.0 21.9 18.5 21.1 Average 21.5 `Average of duplicate determinations. TABLE III Comparison of Headspace and Vapor Extraction Methods of Analysis for VCM in Water Sample Headspace Vapor Extraction Number(Concentrations in mg/kg) 1 0.002 2 0.006 3 0.006 4 0.003 5 0.009 6 0.036 0.002 0.007 0.006 0.002 0.006 0.032 1 ! 23831003 T 1 ;R e s id u a l VCM i n P ipe (m g/kg) ro CD oCM rC-M CM H GO ^ fn ^ ONO B oo O O rM H O ro o H 00 O CM O CM H rH oo P* CO C"* VD GO Oo vOOo* GOoH cOm NoOldinOo fi^n ro o o o o o 0} >1 <0 o CM O VO CO O <30 CM o V CM CM CO CM OiCiOCNUOkOVOkO 00C\0HO*H . rH i-M o o oo co cn vo cm cm oo voooooorc-MopO *ion<o Hr- o o cc CM O VO m o co <N O V CM CM CM o r- o O H o H o ro O Oo v vV CM CM CM CM OBOHOlOOB OlJlOBOHOH fH r-1 o o o v v vV o KJ o 0 u *0 0 0 u 4J U 8, H H JJ W 4J 0$4 H -P a) a 4J V 04 <Q o e <0 <0 flj 4J P +J 4J <0 4J <G -M jjj CO 04 (0 (0 <0 cu CO 04 CO 04 c <u o 0 *H JJ JC o 4J <C <$ O' U D C P a& CM CM CM CM CM CO CO O UO a co 8 T n*. 1 Xntoyrator kcmpOnM vi. Concentration o vqh in Motor Vapor ptutctlan Mathod Concentration of vot in Motor (ag/ltq a 10*1 Pi*. 2 Integrator ftaaponoo vo. VO* toocontrotlon Noodapoco Hatha* Z3Q 91005 3* T i