Document 8R0ak8qVD05Q1dw5nX75XYDpm

i JOSEPH E KELLER JEROME H HECKMAN CHARLES M MEEHAN WILLIAM H BORGHESANI. JR ROBERT R TIERNAN MALCOLM D MACARTHUR WAYNE V BLACK DAVID L HILL MARTIN W BERCOVICI JOHN S EIDRED CAROLL C HARRIS MICHAEL F MORRONC LARRY s SOLOMON JOHN 5 DUSF.CK LAW OFFICES Keller and Heckman 1150 17TM STREET, N. W. SUITE lOOO C H RI 5TI N E A MEAGHER SHIRLEY S. FUJIMOTO peter l, de l* cruz LAWRENCE p HAL PR I N DEBORAH 5huR TRINKER C, DOUGLAS JARRETT EDWARD L KORWEK ROBERT L FLEShnER JONATHAN P LEVINE SHEILA A MILLAR RUSSELL H FOX LEE M WEINER ANGELENA C LE BLANC WASHINGTON, D. C- 20036 202) 457-1100 November 12, 1982 TELECOPIER ! 202) 296*7682 CABLE ADDRESS KtLMAN writer's direct dial number (202) 457-1110 Stephen Higgins Acting Director Bureau of Alcohol, Tobacco and Firearms Department of the Treasury 1200 Pennsylvania Avenue, N.W. Washington, D.C. 20226 Re: Use of Polyvinyl Chloride (PVC) for Manufacturing Plastic Liquor Bottles Dear Mr. Higgins On November 3, 1982, the Department of Treasury released an Environmental Assessment of polyethylene terephthalate (PET) for use in liquor bottles. Page 13 of the Assessment contains a response to Comments that we filed on behalf of our client, The Society of the Plastics Industry, Inc., that polyvinyl chloride (PVC) also be approved for liquor bottle use. The reply states that ATF does not intend to address the use of PVC for manufacturing liquor bottles until there is clear evidence that FDA sanctions the use of PVC for such an application. We are writing at this time to request that you address a new in quiry to the Commissioner of Food and Drugs regarding the status of PVC for packaging distilled spirits. FDA last commented to ATF regarding this issue in a letter dated January 13, 1981, from Richard J. Ronk, Deputy Director, Bureau of Foods. In that letter FDA indicated that it was reassessing its policies regarding indirect additives largely because of the recent court decision involving acrylo nitrile bottles (Monsanto v. Kennedy, 613 F.2d 947 (D.C. Cir. GENC017192 Stephen Higgins November 12, 1982 Page 2 Keller and Heckman 1979)). On April 2, 1982, FDA announced a significant new policy regarding "constituents" in packaging materials. 47 Fed. Reg. 14464. The notice specifically refers to residual vinyl chloride monomer as the type of "constituent" that would be covered by the policy. The Notice also stated that FDA would implement the policy on a case-by-case basis as administrative needs required. On August 31, 1982 Gerard L. McCowin, Director, Division of Food and Color Additives, Bureau of Foods, wrote to me stating, "we wish to develop proposals, utilizing the constituents policy, to initiate final action on the food-contact use of ...vinyl chloride polymers." In this connection, Mr. McCowin asked for information on residual vinyl chloride levels and markets for PVC to enable FDA to make calculations from an actual data base rather than by using estimates. We have now provided FDA with information (a) showing the industry can provide PVC bottles suitable for packaging distilled spirits with a residual vinyl chloride level that does not exceed 10 parts per billion and (b) demonstrating that migration of vinyl chloride to the contents of such bottles will not exceed the upper limit established by the application of the constituents policy. Thus, in our opinion, FDA should now be in a very sound position to make a clear statement that the use of PVC bottles with an appropriately low residual vinyl chloride level is suitable for packaging distilled spirits. As you may be aware, ATF already has a substantial amount of data in its files regarding the suitability of PVC as a liquor container. A petition requesting that ATF approve polyvinyl chloride (PVC) for liquor bottle use was filed by SPI on January 22, 1980. On June 5, 1981, you requested the submission of new sample bottles for storability testing. The requested samples were hand-delivered to Dr. Charles Midkiff at ATF National Laboratory on August 24, 1981. In light of this background, we would hope that you will have everything necessary to make a suitability determination regarding PVC for use as a liquor bottle once FDA affirms the satisfactory status of this plastic from a health standpoint. GENC017193 Stephen Higgins November 12, 1982 Page 3 Keller and Heckman We would greatly appreciate your prompt attention to this matter. Very, truly yours, U/ /;. /. 7 meronie H. Heckman cc: Dr. Arthur Hull Hayes, Commissioner of Food and Drugs Dr. Sanford Miller, Director, Bureau of Foods Mr. Richard Ronk, Deputy Director, Bureau of Foods GENC017194 DEPARTMENT OF THE TREASURY Bureau of AlcoholtTobacco and Firearms Washington, D.C. 20226 J Nnber:g2_ Dit* Polyethylene Terephthalate Liquor Bottles Proprietors of Distilled Spirits Plants, Importers and Others Concerned: Purpose: The purpose of this circular is to inform you of a forthcoming ATF Ruling concerning the approval of polyethylene terephthalate containers for bottling distilled spirits. The ATF Ruling will read as follows: The Bureau of Alcohol, Tobacco and Firearms has been requested to approve the use of polyethylene terephthalate (PET) as a suitable material for the manufacture of liquor bottles. Section 5301(a) of Title 26, United States Code, provides that the Secretary may regulate the kind of containers designed or intended for use in the sale of distilled spirits. Sections 19.11, 194.11, 250.11 and 251.11 of Title 27, Code of Federal Regulations, provide for liquor bottles to be made of glass or earthenware or of other suitable material approved by the Food and Drug Administration, which has been designed or is intended for use as a container for distilled spirits for sale for beverage purposes and which has been determined by the Director to adequately protect the revenue. The Bureau is aware of the need to recognize advancing technology which provides materials adaptable or developed for packaging distilled spirits. At the same time, the Bureau is aware of its responsibilities for protecting the Federal excise tax revenues, as pre scribed by the Internal Revenue Code (Title 26, U.S.C., Chapter 51); for insuring an orderly marketplace, in accordance with the requirements of the Federal Alcohol Administration Act (Title 27, U.S.C., section 205(e)); and for protecting the environment, pursuant to the National Environmental Policy Act (Title 42, U.S.C., section 4332). GENC017I95 ?iv..*:.-- K" ' fi-m 4) 2 The Bureau recognizes that, in some minor instances, a proof gain of up to two degrees of proof per year (and a corresponding water volume loss) may occur. However, this effect can be minimized by avoiding high storage temperatures, by insuring that the wall thickness of PET containers is uniform, and by insuring market turnover. In any event, ATF has concluded that this characteristic of PET packaging poses no jeopardy to the revenue because the taxable commodity, the alcohol, does not travel through the package wall. The quantity of alcohol does not change between the time of bottling and the point of tax determination. Therefore, it has been determined that the use of PET for liquor bottles provides adequate protection to the excise tax revenue and is a suitable material for this use. The consumer obtains the same amount of alcohol contained in the product at the time of bottling, since only water is lost. Since the same amount of alcohol is contained in the product with only a very minor, if any, increase in proof, the label on the product pro vides adequate information to the consumer regarding alcohol content. In accordance with the requirements imposed by the National Environmental Policy Act, the Bureau prepared an assessment of the projected effect on the environ ment of PET liquor bottles. The conclusion reached on the pertinent issues was that neither container use nor disposal would have a significant environmental impact. Held, polyethylene terephthalate (PET) liquor bottles conforming to the following specifications may be used as containers for distilled spirits: (1) The bottles must be rigid or semi-rigid with molded shape or design which cannot be altered by pressure without damage to the bottle, and the wall thickness of the bottle must be as uniform as possible; (2) The bottle must be manufactured in an approved standard of fill; and > GENC017196 3 (3) The material used to construct the PET bottle must meet the Food and Drug Administration's health and safety specifications for the packaging of alcoholic beverages for consumption as promulgated in the Code of Federal Regulations. Inquiries: Inquiries concerning this circular should refer to its number and be addressed to the Assistant Director, Regulatory Enforcement, Bureau of Alcohol, Tobacco and Firearms, 1200 Pennsylvania Avenue, NW., Washington, DC 20226. Attention: Commodity Classification Branch. GENC017197 PART II-A 27 CFR 19.11: MEANING OF TERMS (Liquor Bottle) (Also 194.11/ 250.11, 251.11) Polyethylene terephthalate (PET) containers are approved for bottling distilled spirits when the containers meet the requirements of the Federal Alcohol Administration Act and the materials used to construct the bottles meet Food and Drug Administration specifications for the packaging of distilled spirits. ATF Rui. 82- ! 2. The Bureau of Alcohol, Tobacco and Firearms has been requested to approve the use of polyethylene terephthalate (PET) as a suitable material for the manufacture of liquor bottles. Section 5301(a) of Title 26, United States Code, provides that the Secretary may regulate the kind of containers designed or intended for use in the sale of distilled spirits. Sections 19.11, 194.11, 250.11 and 251.11 of Title 27, Code of Federal Regulations, provide for liquor bottles to be made of glass or earthenware or of other suitable material approved by the Food and Drug Administration, which has been designed or is intended for use as a container for distilled spirits for sale for beverage purposes and which has been determined by the Director to adequately protect the revenue. G E N C 017198 2 The Bureau is aware of the need to recognize advancing technology which provides materials adaptable or developed for packaging distilled spirits. At the same time, the Bureau is aware of its responsibilities for protecting the Federal excise .tax revenues, as pre scribed by the Internal Revenue Code (Title 26, U.S.C., Chapter 51); for insuring an orderly marketplace, in accordance with the requirements of the Federal Alcohol Administration Act (Title 27, U.S.C., section 205(e)); and for protecting the environment, pursuant to the National Environmental Policy Act (Title 42, U.S.C., section 4332). The Bureau recognizes that, in some minor instances, a proof gain of up to two degrees of proof per year (and a corresponding water volume loss) may occur. However, this effect can be minimized by avoiding high storage temperatures, by insuring that the wall thickness of PET containers is uniform, and by insuring market turnover. In any event, ATF has concluded that this characteristic of PET packaging poses no jeopardy to the revenue because the taxable commodity, the alcohol, does not travel through the package wall. The quantity of alcohol does not change between the time of bottling and the point of tax determination. Therefore, it has been determined that GENC017199 3 the use of PET for liquor bottles provides adequate protection to the excise tax revenue and is a suitable material for this use. The consumer obtains the same amount of alcohol contained in the product at the time of bottling, since only water is lost, since the same amount of alcohol is contained in the product with only a very minor, if any, increase in proof, the label on the product pro vides adequate information to the consumer regarding alcohol content. In accordance with the requirements imposed by the National Environmental Policy Act, the Bureau prepared an assessment of the projected effect on the environ ment of PET liquor bottles. The conclusion reached on the pertinent issues was that neither container use nor disposal would have a significant environmental impact. Held, polyethylene terephthalate (PET) liquor bottles conforming to the following specifications may be used as containers for distilled spirits: (1) The bottles must be rigid or semi-rigid with molded shape or design which cannot be altered by pressure without damage to the bottle, and the wall thickness of the bottle must be as uniform as possible; . f"; F' :::: >" c:::: nr f-,r t....... F-- oo r^. ot--t O O 4 (2) The bottle must be manufactured in an approved standard of fill; and (3) The material used to construct the PET bottle must meet the Food and Drug Administration's health and safety specifications for the packaging of alcoholic beverages for consumption as promulgated in the Code of Federal Regulations. GENCO17201 5 PART II-A 27 CFR 194.11: MEANING OF TERMS (Liquor Bottle) Polyethylene terephthalate (PET) containers are approved for bottling distilled spirits. See ATF Rul. 82-/2, page > GENCO17202 6 PART II-A 27 CFR 250.11: MEANING OF TERMS (Liquor Bottle) Polyethylene terephthalate (PET) containers are approved for bottling distilled spirits. See ATF Rul. 82- 1 , page GENCO17203 7 PART II-A 27 CFR 251.11: MEANING OF TERMS (Liquor Bottle) Polyethylene terephthalate (PET) containers are approved for bottling distilled spirits. See ATF Rul. 82- iZ, page l O<N o O % o JOSEPH E <CLLR JEROME h wccmnan CHARLES M, MEEHAN WILLIAM H, 9QRGHESANI, JR. ROBERT 9. nernan MALCOLM o. MACARTHUR WAYNE V, SLACK OAVIO L. HILL MARTIN W BERCOVIC! JOHN S. ELOREO CAPOLE C, HARRIS MICHAEL ? MOPPONC LARRY S, SOLOMON JOHN S OUACCX law omcis Keller and Heckma.v U30 17** STRIZT, N1. w, SUIT'S iOOO CHRISTINE A MEAGHER Shirley S 'tjjimoyo aETER L. aK^CRUJ LAWRENCE R HALPPIN 0E3ORAH SHufl 7RINKCR C, OOUOLAS jASP ETT EDwaRO l_ XORWEK ROBERT L. TLE3HNER JONATHAN P LEVINE SHEILA A. MILLAR Russell h. pox lee m WC1NCR ANGELINA C. LX 9LANC Washington, d, c. 20030 ______ '202) 457*1100 November 12, 1982 TELECOPIER ' 20?) 296 -?S2 cable aOORCSS "kelman WRITER'S DIRECT OlAL NUM9ER (202) 457-1110 Dr-. Sanford A. Miller Director Bureau of Foods Food and Drug Administration 200 C. Street, S.W. Washington, D.C. 20204 Re: Polyvinyl Chloride for Use as a Container for Alcoholic Beverages Dear Dr. Miller: In a letter to me dated August 31, 1982, Mr. Gerad McCowin, Director of the Division of Food and Color Additives, requested information available to The Society of the Plastics Industry, Inc. (SPI) bearing on residual monomer levels in poly vinyl chloride (PVC) plastics, and potential markets for such products. We understand that this information would be useful to the Agency as it moves to implement the constituents policy with respect to PVC food-contact applications. We have not yet been able to collect any meaningful information regarding present and potential uses of PVC that would provide the Agency with sound data for calculating a consumption factor, i. e., for deter mining the percentage of the diet that might be packaged in PVC plastics. However, we have been informed that with respect to PVC bottles, the industry can provide products with residual monomer levels not exceeding 10 parts per billion (ppb) by weight. We are writing to you at this time solely with respect to PVC bottles because the only application for which PVC is not currently permitted involves bottles, liquor bottles to be GENC017205 Dr. Sanford A. Miller November 12, 1982 Page 2 Keller and Hzcsxan specific. We have received and are enclosing copies of letters from the following resin suppliers and fabricators of bottles stating that they can produce such materials: Aim Packaging, Inc., Brockway Plastics, Continental Group, Ethyl Corporation, Occidental Chemical Corporation; Owens-Illinois, Inc., and The Proctor and Gamble Company. It should be emphasized again that the data we are pro viding here relates only to PVC bottles. Film products fabri cated from PVC or copolymers thereof utilize different techno logical processes so that the standards and criteria applicable to bottles may not be relevant. Furthermore, we think it clear and correct, as stated in the Food and Drug Administration's (FDA) September 3, 1975, Notice of Proposed Rule Making, that PVC products such as films and can enamels are of no regulatory concern; nor are we aware of any data reported since 1975 to change this conclusion. In a letter dated January 23, 1981, addressed to Mr. Wil liam T. Drake, at the Bureau of Alcohol, Tobacco, and Firearms (ATF), Mr. Richard J. Ronk responded to an ATF inquiry regarding the regulatory status of PVC for use as a container for alcoholic beverages. Mr. Ronk stated in relevant part that the Agency was "not likely to publish any final regulation or a new proposal regarding polyvinyl chloride" until the Agency had resolved what its policy should be in dealing with trace quantities of carcinogenic residues in food-contact materials. Mr. Ronk also wrote that the Agency was evaluating data which indicated that if current good manufacturing practices are followed, PVC packag ing material "with encouragingly low residual VCM levels" can be produced. Since FDA has now developed its "constituents policy" (47 Fed. Reg. 14,464 (Apr. 2, 1982)) which it is implementing on a case-by-case basis, and because PVC bottles can now be produced with residual vinyl chloride levels sufficiently low as to comply with the safety requirements embodied in the new constituents policy, we are hereby requesting that FDA immediately re-evaluate the status of PVC for use as packaging GENCO17206 Dr. Sanford A. Miller November 12, 1982 Page 3 K-ET.T..EK ajtd HecKM-AO* material for alcoholic beverages. V As discussed more fully below and in the technical appendices to this letter, vinyl chloride migration from PVC bottles containing a 10 ppb maximum residual vinyl chloride (RVCM) concentration cannot exceed the "very safe dose" level calculated by PDA's risk assessment pro cedures. Consequently, the use of these bottle will pose no health or safety problem with respect to the possible migration of vinyl chloride. It is our hope you will concur in our opinion that there is no longer any basis for concern as to the safety of using PVC bottles for packaging distilled spirits provided the residual monomer level does not exceed a concentration of 10 pob by weight in the bottle walls. It is our understanding that FDA has determined, utiliz ing its conservative risk assessment procedures, that a dietary intake of 0.22 ug of vinyl chloride per day will be "very safe." Inasmuch as the Agency considers the average diet to consist of 3 kg of food (both solid and liquid combined), the safe concen tration of vinyl chloride in food is 0.073 ppb by weight: 0.22 = 0.073 ug/kg 3 In other words, if the entire daily diet, both solid and liquid, were packaged in PVC, a migration level of 0.073 ppb would be permitted and the dietary intake could not exceed 0.22 yg/day. In dealing with indirect additives, FDA utilizes a "con sumption factor" to provide an estimate of how much of an indi rect additive might be present in the total diet based upon measured migration data. Thus, if a packaging material is used to package one-third of the diet, migration could be permitted into the packaged portion of the diet at three times the safe level in the total diet. For example, if the safe level of a substance were found from appropriate toxicological studies to be 1 part per million (ppm), migration into the food that actually contacts the packaging material could be as high as 3 ppm. Stated */ In this connection, please take due note of today's release of a new Bureau of Alcohol, Tobacco, and Firearms ruling on other plastic liquor bottles. ATF Rule 82-12 (copy attached). GENCO17207 Dr. Sanford A. Miller November 12, 1982 Page 4 Keuler actd somewhat differently, if migration into food contacted by a particular packaging material is as high at 3 ppm, the total diet will contain only 1 ppm because only one-third of the diet contacted the packaging material. Depending upon the actual consumption factor, j number not now known with any precision for vinyl chloride-containing packaging materials, the permitted level of vinyl chloride migra tion will be higher than 0.073 ppb. If the consumption factor were 0.1 (i. e. , if 10% of the diet were packaged in PVC-based materials) migration from such materials could be permitted at a level of 0.73 ppb; if the consumption factor were 20%, migra tion could be permitted at a level of 0.365 ppb, and so forth. Although we recognize that it is not feasible to set a generally applicable level of vinyl chloride migration until a reliable consumption factor is determined, it is, nevertheless, alto gether proper to clear the use of PVC at this time for any appli cation where potential migration does not exceed the safe level (0.073 ppb). If FDA affirms the safety of PVC as a container for distilled spirits at this time, the otherwise permitted level of migration from other vinyl chloride-based products based on consumption factor considerations would not be adversely affected as is shown in detail in Appendix A to this letter. In other words, the Agency can deal with the liquor bottle question without considering possible inequities that might affect other uses of vinyl chloride-based products. An Agency decision at this time to permit the use of PVC as a con tainer for alcoholic beverages would not adversely affect any future regulatory migration levels that may be applicable to other PVC uses, provided the level of migration from the presently considered uses does not exceed 0.073 ppb. Appendix B to this letter provides a technical explanation demonstrating that containers with residual monomer levels not exceeding 10 ppb will not leach vinyl chloride into the contents at a level in excess of 0.073 ppo. Stated summarily, the quantity of vinyl chloride available to migrate is so low, and the rate of migration of vinyl chloride from PVC containers made with an RVCM content of 10 ppb or less is so slow that the concentration of vinyl chloride in the contents even after an exaggerated shelf-life exposure at moderately elevated tempera tures will not exceed the safe 0.073 ppb level. , GENC017208 Dr. Sanford A. Miller November 12, 1982 Page 5 Reixer anb Heckxaj^ One final point needs attention. Mr. Ronk1s letter noted that, at the time of writing, FDA could not document the existence of a prior-sanction permitting the use of PVC for packaging alcoholic beverages, despite the statement made by Mr. William F. Randolph, now Deputy Associate Commissioner for Compliance at FDA, at an International Conference in Managua, Nicaragua, in May, 1969, to the effect that FDA recognizes a prior-sanction for PVC for all food-contact uses. Even if this confusion has not yet been clarified, there appears to be no reason why FDA need take any administrative action against the specific use of otherwise satisfactory PVC containers for packag ing distilled spirits until it can deal appropriately with the status of PVC containers for all food-contact applications. It appears to be both unreasonable and discriminatory to proscribe the use of such containers for one particular category of foods (i.e. , alcoholic beverages) when there is no public health or safety problem involved. Rather, we suggest, that if any further action other than a withdrawal of its 1975 rule making proposal is deemed necessary, the prior-sanctioned status of PVC for all food-contact uses be dealt with in some other regulatory action in due course. Summarizing, the industry can provide alcoholic beverage containers with a residual monomer content not exceeding 10 ppb and vinyl chloride will not migrate from these containers to the packaged food at a level higher than 0.073 ppb. Conse quently, favorable FDA action clearing the use of such products is consistent with the constituents policy. Furthermore, action to clear this limited application will not adversely affect the interests of any other portion of the industry since the per mitted vinyl chloride migration levels for other applications will be, if anything, higher as a result of this action than they would be if all packaging materials were subject to the same migration limitation based on an overall consumption factor; We trust the information we are providing here will permit you to conclude that the use of the PVC alcoholic bever age containers will pose no safety concerns provided the residual monomer level in such containers does not exceed 10 ppb and thus enable you to advise the Bureau of Alcohol, Tobacco, and GENC017209 Dr. Sanford A. Miller November 12, 1982 Page 6 KjeiX.gR A2TD HjECKM-ACT Firearms to this effect without further delay. Should you have any questions, however, please let us know so that we can pro vide answers as promptly as possible. Very truly yours Enclosures cc: Mr. Stephen Higgins Dr. Arthur Hull Hayes Mr. Richard J. Ronk APPENDIX A ALLOCATION OF PERMITTED MIGRATION LEVELS UNDER THE "CONSTITUENTS POLICY* Any vinyl chloride-based packaging material will pose no monomer migration problem if the migration level does not exceed the "very safe" concentration of 0.073 parts per billion (ppb). V Although such uses, per se, will pose no public health or safety concern, it may not be immediately apparent that the early allocation to bottles of a portion of the very safe quan tity of vinyl chloride that may be permitted in the diet will not adversely effect the interests of those packaging materials which are not being dealt with at the same time. This, however, is the case. To demonstrate this mathematically, in general terras: Let the consumption factor, i. e. , the fraction of the total diet contacted as a result of all uses of vinyl chloridecontaining packaging materials, be "c." Let "b" be the fractional portion of the diet contacted by bot tles only ("b" must be less than "c") . Let 0.22 Ug be the very safe level of vinyl chloride monomer (VCM) in the total diet and Let 3 kg be the weight of the total diet. (Actual numbers are used here for convenience, but there is no loss in generality. It would be perfectly in order to substi tute symbols for these specific values.) *_/ We understand that the actual FDA document that sets forth the derivation of this figure is not available for public distribution at this time. Nevertheless, we have been informed that the very safe dose has been determined to be 0.22 Ug per day. If the entire diet, both liquid and solid, were packaged in PVC, a migration level of 0.073 ppb would assure that the 0.22 ug level would not be exceeded. GENC017211 2 Then: (1) The safe concentration in the total diet is: HNf-= '073 ppb (2) The weight of diet contacting vinyl chloride-based products is 3c. (3) The safe level of migration, taking the consump tion factor into consideration is: 0.22 = C.073 ppb 3c c (4) If bottles with migration at or below 0.073 ppb are utilized, the weight of diet packaged in bottles is (3 b: kg, and the weight of vinyl chloride contributed to the diet by this use is: (0.073 ppb)(3b) kg = (0.073)(3b) Ug = 0.22b Ug (5) The remaining weight of vinyl chloride that may be contributed by all other uses is: (0.22 -0.22b) ug = (0.22) (1-b) Ug (6) The weight of diet that is packaged in all other vinyl chloride-containing materials is: (3c-3b) kg = 3(c-b) kg Note: (c-b) is less than (1-b) (7) The permitted concentration of vinyl chloride migrating from all other uses is: 0.22 (1-b) 0.073 (1-b) > 0.073 ppb 3 (c-b) ' (c-b) Thus, regardless of the actual numerical value of the very safe dose, restricting a portion of the total consumption factor to migration at the very safe concentration in the total diet must result in an increase in the permitted level of migration for all other uses. GENC017212 APPENDIX B VINYL CHLORIDE LEVELS IN CONTENTS OF BOTTLES The following considerations demonstrate that the vinyl chloride content of substances packaged in polyvinyl chloride (PVC) bottles cannot exceed 0.073 parts per billion (ppb) pro vided the bottle walls contain no more than 10 ppb of residual vinyl chloride (RVCM). Typical wall thicknesses of bottles do not exceed 20 mils (0.020 inches). Consequently, the volume of PVC in 1 square inch of wall surface is 0.020 cubic inches or 0.33 cc. Inasmuch as the density of PVC compounds is somewhat less than 1.4 grams per cc, the weight of PVC in the wall section is somewhat less than 0.46 g (0.33 x 1.4). If the RVCM content is no higher than 10 ppb, the maximum weight of vinyl chloride in that section will be 4.6 x 10-9 g. The Food and Drug Administration (FDA) considers a weight of 10 g of contents to be in contact with each square inch of package surface. Consequently, if all the vinyl chloride were to migrate into the contents, the resulting concentration would be 0.46 nanograms per gram of food or 0.46 ppb. However, it has been early demonstrated that migration of residual monomer from PVC packaging materials initially proceeds in two directions, part entering the contents of a package, the remainder diffusing to the exterior. Thus, the maximum concentra tion in the contents could not exceed 0.23 ppb if all of the available vinyl chloride were to migrate into the contents. This cannot occur because, at equilibrium, there will be a parti tioning between the concentration of vinyl chloride in the con tents and the vinyl chloride in the solid walls of the con tainer so that the 0.23 figure calculated above would be an unapproachable limit. Furthermore, over a very long time period, the direction of migration would reverse because the exterior air would represent an infinite sink; after an initial buildup of vinyl chloride concentration in the contents, the vinyl chloride would then diffuse from the contents through the wall into the , exterior air. Nor is this all. Diffusion experiments conducted by Behrens and co-workers utilizing PVC with relatively high residual monomer contents have demonstrated that over a period of appro ximately a year at 3QC less than 1/3 of the available vinyl chloride will diffuse. Thus, less than 1/3 of the limiting 0.23 ppb will migrate in one year at 30C. The time and tempera ture is quite excessive as simulants for reasonably expected GENC017213 2 shelf-life and storage conditions. Thus, even under these con ditions, less than 0.073 ppb can be anticipated. Copies of relevant publications are attached. Studies conducted by Ethyl Corporation and reported to FDA have demonstrated that the diffusion constant for vinyl chloride migration is, in fact, not constant when the residual monomer concentration drops somewhat below 50 ppb. Thus, the calculations based on Behrens' work are high and may be as much as an order of magnitude too high when applied to the containers of interest here which bear not more than 10 ppb of residual monomer. Copies of the Ethyl reports are attached for ready reference. Although not amenable to calculation, it is known that vinyl chloride tends to accumulate in the headspace over dilute vinyl chloride solutions. Thus, the vinyl chloride content of the solutions as calculated above will be lower by a signifi cant although not readily quantifiable amount because much of the vinyl chloride will be in the headspace and will dissipate when the container is opened. Furthermore, each time the con tainer is used, this same phenomenon will be repeated so that the initial monomer content will decay rapidly. In summary, bottles made with RVCM levels not exceeding 10 ppb will not yield VCM to the contents at levels exceeding 0.073 ppb. GENC017214 Q ^tiuiesfeiTitunion isA^^Srr '- :^r* :^fWiS~r;Tl-.'^S(6iJjs.tf*'13! - v 3 A 2 7 3 ( 1 72S)(--2-?S5^^3 15 )3 D 11/11/32 172-2 -C T " Vi " *T P " ^ C ID a>SS35?!AA VL TEEN 3C?T CLINTON CH 129 1I-1I 2532? 1ST :vq irsQvr u . ;->'ECvv AN, K.ZLLE? AND HECKMAN ?PT DLY K3K, CL? 115 2 17?:J' ST NC?7U,WZ3T SUITE' 1222 VA?-I" 310%' PC 2222 S r'T'p v5 vr^yv*\| PE'-CNBIVE T2 THE INOUIPY VAEE BY 7-E CS?I) SOCIETY OF THE 3LASTIC3 :y:"ST?v LETTER 3T OCTOBER 25 19S2, THIS is to advise you that we ar- -^v? r: Tu-T VE CAM KANUFACTUPZ 3VC BOTTLES 'WHICH 'WILL CONTAIN NOT IC - - ?! 3 3E3 BILLION C" RESIDUAL PVC *CMCMZP. o* * t'cji T r w.a 1201`SP (*S-9t V -TQ 'Ml'IHl'l THIS NO^CNE? LEVEL IS BASED ON ASSURANCES -AC-'ASIMS HAS RECEIVED R30M CJP PVC rGCD SPACE MATERIAL T ^0 t c 6 T" C T'~ET c A - - B T1 17 v OF B0OVIPINS CCV = Q'JND BILLION MONOMER LEVEL. mi* * rt/T] v '=> :t'-uc r v::;v:v iriOJTIVi V? I%- ~7 a p 7 '' P TMTHM pu : - p v 5 -* c c=: c- .40452 lAL THA,7 WOULD INC?: or c3i?at:cns a:y SP A5-59I GENC017215 pia/tic/ October 27, 1982 Jerome H. Heckman <el1er & Heckman 1150 17th Street, N.W. Suite 1000 Washington, O.C. 20036 Jerry: Re: G. R. `-lunger's letter dated 10/25/82 This is to confirm that given the current status of resin manufacturing, Brockway Plastics is capable of manufacturing rigid and semi-rigid PVC food containers with a RVCM of no more than 10 parts per billion. From conversations I have had with Or. Dixler, this should fall within the maximum*allowaole standards being considered by FOA and further market data as to products, etc. is not necessary, Should this not be the case, please contact me. Good 1uck... Robert W. Scnroeder Vice President, Sales/Marketing BCCK'A'AV =LA5TC33 ZrviSJCN = !_ --'' Q 1 ^ NASrL'A, l\i c=. \yV IA .V> P S ^ 1= S C3QS^ E-2CCC GBMC017216 VIA TELECOPIER (202) 296-76S2 ETHYL CORPORATION TLLSZjlMCH A>"D Dsmo.niDiT DjLMJEXXZDTT November 11, 1982 P%CA4C HL^LT to: + o. sox a*t j^atoh ftouac,LA.70n Mr. Jerome H. Heckman Keller and Heckman 1150 17th Street, N.W. Suite 1000 Washington, D.C. 2003d Dear Mr. Heckman: Please be advised that Ethyl Corporation can provide com mercially rigid PVC compound with a residual vinyl chloride monomer content of less than 10 parts per billion. Finished products (e.g., bottles) produced from this compound by processors will also contain less than 10 parts per billion vinyl chLori.de. Very truly yours, AJH:dsp Mr. R. M. Mars ton A. J'. Haefner Director, Plastics Applications Research and t GENC017217 George K London Vice President October 27, 1982 CONTINENTAl~CtJP Packaging Fores! Product! Iniurance Energy Jerome H. Heckman Keller and Heckman 1150 17th Street, N.W. Suite 1000 Washington, D.C. 20036 Dear Jerry: In a letter dated October 25 Monk Munger asked us to advise you whether we can produce bottles of food grade PVC which would have less than 10 parts per billion of vinyl chloride monomer. We are advised by our resin suppliers that they will deliver resin with a residual monomer level of 5 parts per billion or less. We are reasonably certain that there is no way we can increase the residual mono'mer level in our bottle making process. It would seem then that we would be safe in stating that we can produce bottles which consistently have residual vinyl chloride monomer present at 10 parts per billion or less. Very truly yours, GKL : hoc G. R. Munge GENC017218 Occidental Chemical Corporation November 8, 1982 Jerome H. Heckman, Esq. Keller and Heckman 1150 17th Street, N.W. Suite 1000 Washington, DC 20036 Dear Mr. Heckman: Please be advised that Occidental Chemical Corporation can provide rigid PVC film and sheet with a maximum residual vinyl chloride monomer of 10 parts per billion. We can also provide PVC compounds to manufacturers of film and sheet with a certified RVCM of 10ppb. max. which would in turn enable them to provide film and sheet with the same amount of RVCM. We trust this is the information you want, and if we can be of further service to you, please let me know. Sincerely RLR:sic Market Development Manacer c:iy PVC Pesms/PVC Fabricated Products GENC017219 Occidental Chemical Corporation November 3, 1982 Jerome H. Heckman, Esq. Keller and Heckman 1150 17th Street, N.W. Suite 1000 Washington, DC 20036 Dear Mr. Heckman: Thus will serve to confirm our recent conversation that we can supply rigid food-grade PVC compounds for the manufacture of PVC bottles with a residual vinyl chloride monomer content of 10 parts per billion maximum. Therefore, bottles produced from this material would also have a lOppb. max. RVCM. The markets which we envision for food-grade PVC bottles would be: vegetable oils, liquor and peanut butter. Ic is difficult to predict .with any accuracy what percent of a particular food market will convert from glass to plastic or, for that matter, to which plastic, but it is our understanding that assurance of a lOppb. max. RVCM in the bottle could allow a total diet to be packaged in rigid PVC with an adequate safety margin, We do know that the liquor packaging industry does want a choice of glass and several plastics, including rigid PVC, and that early experience with food-grade PVC showed economical and safety benefits to the consumer and also overall energy savings for all of the processes involved in the production, shipment, handling and disposal of the comakers. The choice of which plascic the distiller would use is dependent 1 on many facccrs, but it is clear that they should have a choiceWith the imminent approval of PET,' for example, and an unpredict able delay in the approval cf PVC, we fear that many plastic battle manufacturers and disciliers will be forced to waste a considerable amount cf money for investment in bottle design, tool inc, production molds, etc., only to have to re-mvest the same money, perhaps only months later, to take advantage of PVC'3 benefits once than maoariai is "approved". CXY ^1^___________________________ PVC Resins, PVC Fabricated Products GENC017220 Occidental Chemical November 3, 1382 Page 2 Consequently, if the FDA is going to eventually move favorably on the PVC issue, we think it should be done now to afford the distillers with enough intelligent choices to minimize economic waste. Please let me know if I can be of further service- I enjoyed talking with you, as always. Sincerely, RLR:sic Market Development Manager GENCO17221 TELECOPY 11 -- 1' O'' [Keller & Heel: mar. off-line telecopy (202) 296-7632] TO: Hr. Jerome H. Hcck-man Keller 4 Hechrsan 1 150 '7th St. 7 N . W. Suite 1000 Wasninet-en, D.C. 20016 uear Jerry: oer so t--ne -ocifety of the Plastics Industry office directed a letter to us to inquire as to whether we can produce bottles rratde of polyvinylchloride for food grade applications, and whether such bottles could be certified to contain less than ten (10) parts per billion residual vinyl chloride monomer. We r.ave beer: in touch wish our- ocrr'0..nd suppliers in this connection and rev been assures that tney can provide us with compound that contains less than parts per billion KVQJ. This being the esse, it is very clear that we would have no difficulty in producing, bottles which contain less than ten parts per Pillion vir.-vi chloride man oner. CJ 4> ' % r. f * i t \j n * m j i : .**. - t ^ Group Vice President, 'Twens-II 1 i.nois, Inc. C.ne cc-aj3te iOicOC, GhiC u 7566 Plastics 4 Closures GENCO17222 THE PROCTER & GAMBLE'COMPANY VCSYOAL; 'cCHNlCAl CENTS* November 4, 1982 CINCINNATI OHIO 452 Mr, Jerome H. Heckman Keller and Heckman 1150 17th Street, N.W. Suite 1000 Washington, D.C. 20036 Dear Jerry: Confirming our telephone conversation of yesterday, our experience at Procter & Gamble supports the view chat SPI advise FDA that manufacturers of rigid and semi-rigid PVC food packaging can meet a RVCM limitation of 10 ppb in their finished articles. Our recent experience in checking RVCM in five different lots of PVC bottles from three different bottle manufacturers is offered as evidence for our support. Analyses showed results in the range of 3-7 ppb, with an average of 4.2 ppb RVCM. In all cases, Hooker's PVC 2160L was the basic resin used. If comments from other SPI members do not verify our findings, or if it is decided that the PVC industry will not commit to a 10 ppb RVCM limit for food packaging, I would like to discuss this situation with you promptly. I look forward to the next meeting of our Food, Drug & Cosmetic Packaging Materials Committee on December 8, when we will undoubtedly be discussing this subject further. Sincerely GENC017223 rti fa- '"0 ran Jd DEPARTMENT OF THE TREASURY Bureau of Alcohol,Tobacco and Firearm Washington, D.C. 20226 NumtMf:g2- Dt* Polyethylene Terephthalate Liquor Bottles Proprietors of Distilled Spirits Plants, Importers and Others Concerned: Purpose: The purpose of this circular is to inform you of a forthcoming ATF Kuling concerning the approval of polyethylene terephthalate containers for bottling distilled spirits. The ATF Ruling will read as follows: The Bureau of Alcohol, Tobacco and Firearms has been requested to approve the use of polyethylene terephthalate (PST) as a suitable material for the manufacture of liquor bottles. Section 5301(a) of Title 26, United States Code, provides that the Secretary may regulate the kind of containers designed or intended for use in the sale of distilled spirits. Sections 19.11, 194.11, 250.11 and 251.11 of Title 27, Code of Federal Regulations, provide for liquor bottles to be made of glass or earthenware or of other suitable material approved by the Food and Drug Administration, which has been designed or is intended for use as a container for distilled spirits for sale for beverage purposes and which has been determined by the Director to adequately protect the revenue. The Bureau is aware of the need tc advancing technology wh ich provides m; or developed for packag ing distilled : same time, the 3ureau i s aware o: its for protecting the Fede ral excise tax scribed by the Internal Revenue Code i Chapter 51); for insuri ng an orderly r accordance with the req uirements of t) Administration Act (Tit le 27, D.S.C., and for protecting the environment, pi National Environmental Policy Act (Tit section 4332) . ecogmze trials adaptable .rits. At the sponsibilities `venues, as pre.tle 26, u.s'.c., rketplace, in Federal Alcohol ;ction 205 (e)); suant to the i 42, U.S.C., i; El GENC017224 2 The Bureau recognizes that, in some minor instances, a proof gain of up to two degrees of proof per year (and a corresponding water volume loss) may occur. However, this effect can be minimized by avoiding high storage temperatures, by insuring that the wall thickness of PET containers is uniform, and by insuring market turnover. In any event, ATP has concluded that this characteristic of PET packaging poses no jeopardy to the revenue because the taxable commodity, the alcohol, does not travel through the package wall. The quantity of alcohol does not change between the time of bottling and the point of tax determination. Therefore, it has been determined that the use of PET for liquor bottles provides adequate protection to the excise tax revenue and is a suitable material for this use. The consumer obtains the same amount of alcohol contained in the product at the time of bottling, since only water is lost. Since the same amount of alcohol is contained in the product with only a very minor, if any, increase in proof, the label on the product pro vides adequate information to the consumer regarding alcohol content. In accordance with the requirements imposed by the National Environmental Policy Act, the Bureau prepared an assessment of the projected effect on the environ ment of PET liquor bottles. The conclusion reached on the pertinent issues was that neither container use nor disposal would have a significant environmental impact. Held, polyethylene terephthalate (PET) liquor bottles conforming to the following specifications may be used as containers for distilled spirits: (1) The bottles must be rigid or semi-rigid with molded shape or design which cannot be altered by pressure without damage to the bottle, and the wall thickness of the bottle must be as uniform as possible; (2) The bottle must be manufactured in an approved standard of fill; and '> GENCO17225 3 (3) The materiel used to construct the PET bottle must meet the Food and Drug Administration's health and safety specifications for the packaging of alcoholic beverages for consumption as promulgated in the Code of Federal Regulations. Inquiries: Inquiries concerning this circular should 'referto its number and be addressed to the Assistant Director, Regulatory Enforcement, Bureau of Alcohol, Tobacco and Firearms, 1200 Pennsylvania Avenue, KW., Washington, DC 20226. Attention: Commodity Classification Branch. Acting Director GENC017226 PAST II-A 27 CFR 19.11: MEANING OF TERMS (Liquor Bottle) (Also 194.11, 250.11, 251.11) Polyethylene terephthalate (PET) containers are approved for bottling distilled spirits when the containers meet the requirements of the Federal Alcohol Administration Act and t'he materials used to construct the bottles meet Food and Drug Administration specifications for the packaging of distilled spirits. ATF Rul. 82- I 'Z- The Bureau of Alcohol, Tobacco and Firearms has been requested to approve the use of polyethylene terephthalate (PET) as a suitable material for the manufacture of liquor bottles. Section 5301(a) of Title 26, United States Code, provides that the Secretary may regulate the kind of containers designed or intended for use in the sale of distilled spirits. Sections 19,11, 194.11, 250.11 and 251.11 of Title 27, Code of Federal Regulations, provide for liquor bottles to be made of glass or earthenware or of other suitable material approved by the Food and Drug Administration, which has been designed or is intended for use as a container for distilled spirits for sale for beverage purposes and which has been determined by the Director to adequately protect the revenue. GENCO17227 2 The Bureau is aware of the need to recognize advancing technology which provides materials adaptable or developed for packaging distilled spirits. At the same time, the Bureau is aware of its responsibilities for protecting the Federal excise .tax revenues, as pre scribed by the Internal Revenue Code (Title 26, U.S.C., Chapter 51); for insuring an orderly marketplace, in accordance with the requirements of the Federal Alcohol Administration Act (Title 27, U.S.C., section 205(e)); and for protecting the environment, pursuant to the National Environmental Policy Act (Title 42, D.S.C., section 4332). The Bureau recognizes that, in some minor instances, a proof gain of up to two degrees of proof per year (and a corresponding water volume loss) may occur. However, this effect can be minimized by avoiding high storage temperatures, by insuring that the wail thickness of PET containers is uniform, and by insuring market turnover. In any event, AT? has concluded that this characteristic of PET packaging poses no jeopardy to the revenue because the taxable commodity, the alcohol, does not travel through the package wall. The quantity of alcohol does not change between the time of bottling and the point of tax determination. Therefore, it has been determined that GENC017228 3 the use of PET for liquor bottles provides adequate protection to the excise tax revenue and is a suitable material for this use. The consumer obtains the same amount of alcohol contained in the product at the time of bottling, since only water is lost. Since the same amount of alcohol is contained in the product with only a very minor, if any, increase in proof, the label on the product pro vides adequate information to the consumer regarding alcohol content. In accordance with the requirements imposed by the National Environmental Policy Act, the Bureau prepared an assessment of the projected effect on the environ ment of PET liquor bottles. The conclusion reached on the pertinent issues was that neither container use nor disposal would have a significant environmental impact. Held, polyethylene terephthalate (PET) liquor bottles conforming to the following specifications may be used as containers for distilled spirits: (1) The bottles must be rigid or semi-rigid with molded shape or design which cannot be altered by pressure without damage to the bottle, and the wall thickness of the bottle must be as uniform as possible; itiv.vV P"E~ w GENCO17229 4 (2) The bottle must be manufactured in an approved standard of fill; and (3) The material used to construct the PET bottle must meet the Food and Drug Administration's health and safety specifications for the packaging of alcoholic beverages for consumption as promulgated in the Code of Federal Regulations. Ir/:: *-- GENCO17230 5 PART II-A 27 CFR 194.11: MEANING OF TERMS (Liquor Bottle) Polyethylene terephthalate (PET) containers are approved for bottling distilled spirits. See ATF Rul. 82- i 2. , page * GENCO17231 6 PART IX-A 27 CFR 250.11: MEANING 0? TERMS (Liquor 3ottle) polyethylene terephthalate (PET) containers are approved for bottling distilled spirits. See ATF Rul. 82- i 2-, page ?= GENC017232 7 PART II-A 27 CFR 251.11: MEANING OF TERMS (Liquor Bottle) Polyethylene terephthalate (PET) containers are approved for bottling distilled spirits. See ATF Rul. 82- 12 , page > GENCO17233 THZ DIFFUSION OF VINYL CHLORIDE IN POLY(VINYL CHLORIDE) by A. R. Berens Corporate Research, The 3. F- Goodrich Co. , Brecksviile, Ohio 44141 (To be presented at American Chemical Society National Meeting, Atlantic City, September 9-13, 1974, Division of Polymer Chemistry; and to appear in Polymer Preprints, Voi. 15, No. 2. qB 9h / ABSTRACT The diffusion of vinyl chloride monomer (VCM) in PVC has been studied by electrobalance sorption/de sorption techniques on polymers in the as-polymerized powder state over the concentration range 0 2000 ppm VCM/PVC and from room temperature to above Tg- The sorption process appears to consist of two stages: 1) Relatively rapid attainment of diffusion equilibrium; 2) Much slower relaxation to swell ing-stress equilibrium. First stage sorption/desorption data on uniformparticle emulsion FVC's and porous mass and suspension PVCs fit a uniform-sphere Fickea.o diffusion model. Diffusion coefficients , D, have been evaluated through this model from sorption half-times using sur face average particle sizes determined^bvmtrc^ "actt'vtrtruii e uT"gy'~OT~'r ~< ca i f mu re r~tJ~sp pear s in sens it rv e and molecular weight in the commercial range type of non-porous, ^gTFssy'' particles causes a slowing of the later part of the sorption or desorption process, in accord with a model for spheres of widely varied size. VCM sorption data thus offer a means of evaluating uniformity of PVC resin particle structure. GENCO17235 THE DIFFUSION OF VINYL CHLORIDE IN POLY(VINYL CHLORIDE) by r A. R. Berens Corporate Research, The B. F. Goodrich Co. , Brecksville, Ohio 44141 INTRODUCTION In the preceding paper (1) we considered the equilibrium aspects of vinyl chloride monomer (VCM) sorption by poly(vinyl chloride) (FVC). Here we turn our attention to the kinetics of the process, i. e. , the measurement and application of the diffusion coefficient (D), which controls the rate of VCM sorption or desorp tion by PVC. Knowledge of D for VCM in PVC has two important applications related to current concerns about possible health hazards of exposure to VCM: First, in the selection of conditions of time, temperature, and pressure for the effective removal of unreacted VCM in PVC at the end of the polymerization, pro cess, and second, in the estimation of the rate at which any residual VCM in fin ished PVC products will escape iato the environment or into other media in con tact with the PVC. We present here our results of VCM sorption and desorption rate measurements from room temnerature to above T^ in VCM/PVC systems containing the low concentrations of VCM likely to be encountered in practice, less than about 2000 ppm. Nearly all published studies of diffusion in polymers (2, 3) have dealt with polymer films or sheet samples, presumably for geometric simplicity and ease of data interpretation. Because of our interest in VCM removal from PVC resins, we chose to work with samples in the powder form obtained directly from the poly merization process. While sacrificing some geometric simplicity, the use of powder samples has some compensating advantages: conversely, sorption studies seem capable of yielding useful information about PVC particle Structure and geometry. EXPERIMENTAL Equipment: All the measurements reported here were made with the record ing electrobalance set-up already described (1). To meet the boundary conditions of the diffusion equations used, the pressure of VCM (Pm) ia the balance chamber should be changed in sharp steps. By bringing a small vessel to a measured VCM pressure, then quickly opening this vessel to the balance chamber, Pm could be increased to a predetermined new value in a few tenths of a second. Step de creases in pressure, especially to zero, are not so easily achieved. With oil diffusion and mechanical pumps, liquid N, traps, and large vacuum lines, we could reduce P^ from 200 mm to -- 30 microns in one minute. In most experi ments, the half-evacuation time was considerably less than the half-desorption time, so no corrections were made for the non-instantaneous pressure change. The response of the electrobalance and recorder provided useful data beginning within 3 to 5 sec. after a pressure change. Materials : The PVC homopolymer resin powders used in this study are listed In Table I. TA3LE I PVC Samples Used mole A c D T G H j K Lj M Tvoe Sus pens ion [T Emuls ion Sus pens ion 11 11 !' 11 Intrms ic V is cos itv 1. 10 0. 70 1. 53 1.10 0.65 0. 65 0. 75 0. 95 0. 95 1 . 10 Surface Area m^ / g 2. 3 < 0. 1 9. 3 0. 32 0. 14 0. 35 0. 21 0. 69 0. 70 1. 3 Glassy-part content ^ ail very high ail high mode rate -- ail high low very low - nil GENC017236 s (' .1 RESULTS AND DISCUSSION General Character of Sorotion Curves: la preliminary interval sorption ex periments, we found that the nature of sorption vs. time curves depends upon the range of and VCM concentration covered. Figure 1 shows two typical curves for Sample D at room, temperature. In the lower Pm range, there is an initial raoid sorption to a VCM content which then remains nearly constant; at a some what highe r ?m range, a similar rapid initial VCM uptake is followed by a slower further sorption which continues on to very long times. Somewhat similar results were reported by Bagley and Long (4) for the system acetone/cellulose acetate, although two-stage sorption was not observed at such a low penetrant concentration. We believe that the Bagley-Long explanation of two-stage sorption is also appli- 1 cable to the VCM/PVC system; In the first stage, a uniform concentration of penetrant is reached throughout the polymer; i. e, , diffusion equilibrium (or ; quasi-equilibrium) is attained quite rapidly. The second stage involves a slow 1 relaxation of the swelling stress, permitting further VCM sorption; since diffusion here is much faster than stress relaxation, the second-stage sorption involves a i very low concentration gradient from surface to center of the PVC particles. Our results seem to suggest that one advantage to the use of finely powdered polymer samples is the attainment of diffusion equilibrium before appreciable stress relaxation has occurred. Thus the two stages may be clearly separated. The extent of first-stage sorption then measures the polymer in a state essentially unperturbed by penetrant swelling action, and may be used to assess effects of prior history (Cf. (1) ). | In the remainder of this paper, we will deal with experiments at sufficiently low Pm that only the first-stage sorption is evident. Experiments at higher Pm and kinetics of the second stage remain for further study. Measurement of Diffusion Coefficient, Uniform Particles: Monodisperse emulsion polymers, such as Sample D, represent the simplest possible geometry among polymer powders. Sorption or desorption experiments on such samples may be treated through a solution of Pick's diffusion equation for spherical geometry (S), -4Dn2v2t/d2 Mf M, (1) where Mf is the weight of penetrant entering or leaving the sphere in time t, M,,, is the total weight change between initial and final uniform concentrations , and d is the sphere diameter. If tg _ 5 is the time when M*/M= = 0. 5, equation (1) gives D = 7.66 x 10 n (2) Thus, knowing d, D can be obtained experimentally from half - sorption or -desorp tion times, provided the diffusion is Fickean. r-- igure Z shows the M^/Mc vs, t I/7" pLot for a typical VCM sorption experi ment an a monodisperse emulsion PVC of 0. 44u particle diameter. The agreement between the experimental points and equation (I) (fitted to the observed tg_ 3) shows that the sorption closely follows Fickean behavior. Equation (2), using the parti cle dia mete r measured by electron micros copy, gives D = 1.9 x 1 0 " 1 ^ cm2 / s ec . Diffusion Coefficients. Suspension PVC s: Several studies by electroa mi croscopy (6, 7, S) have shown that porous particles of suspension PVC are agglom erates of primary or sub-particles of about l to 5 microns diameter. Recent scanning electron micrographs in our laboratories indicate that the primary par ticles in a given sample are fairly uniform in size. This observation suggests the GENCO17237 Dossibility of applying the uniform-s phe re model to diffusion in suspension PVC's. r We also have found that an average size cf the primary particles can be ob tained from specific surface areas measured by N3-adsorption. For a sphere, of density p, the surface-to-wcight ratio is 6/pd. For PVC spheres, p 1.4, and we obtain d,-* = 4. 29/S*O (3) where Sff is the soeciiic surface area in m-/g and ds is the surface-average particle diameter in microns. We find quite reasonable agreement between d3 determined in this way and the primary particle diameter estimated by scanning microscopy on sectioned granules of suspension PVC. Sorption curves on porous suspension PVC's conform quite well to the shape predicted by equation (1), as illustrated in Figure 3. Application of equation (2) to sorption data on Sample A (Hs = 1. o5p) at 25* gives D = 2. Ox 10"^ cm^/sec, in good agreement with our result on monodisperse emulsion PVC. Thus the uniformsphere model seems applicable to porous suspension PVC, and the dimension con trolling sorption rates appears to be the size of the primary particles, not the external diameter of the agglomerate- This means that diffusion of VCM through the pores, and through any pericellular membrane around the gross particle, must be very rapid. Use of the external particle size in calculations of D from sorption data would give values several orders of magnitude higher than ours; this appears to account for the apparent discrepancy between our results and those recently reported by Wolf and Kreter (9). Concentration Dependence of D: In most of our sorption/desorption experi ments, we find that the desorption is somewhat slower than sorption, as illus trated in Figure 3. This is the expected result when D increases with increasing concentration of the penetrant (5). Such an effect seems likely in our system, due to the plasticization of PVC by VCM. Over the limited range of VCM concentra tions considered here, the variation of D is small; an average of sorption and de sorption values for D is probably a satisfactory figure for most practical appli cations. Temperature Dependence of D: We have obtained sorption and desorption curves at temperatures from 25 to 90*C for several suspension PVC resins and calculated D from tg 5 and dg , Samples of varied ds were needed to bring tg 5 into an experimentally accessible or convenient time range. Figure 4 summarizes our results on an Arrhenius plot. The agreement among samples is fairly good; most of the scatter may be attributed to uncertainty in d5 or in particle geometry, so there is no clear evidence that D differs significant!.* from one sample to anoth er. Cur samples covered the range of molecular weights (or intrinsic viscosities) of normal commercial PVC's, so it also anoears that D is insensitive to molecular weight m this range. D varies with cemce ra tar e from about 2 x 10'*" cm"/sec at .10 30* to 2 x 10 at 90*C- The straigr.t line of Figure 4 yields an activation energy of IT kcal/mol. Limited results at I 1 0 1C show no great change in activation ener gy above Tg. G E N C 0 17238 R--esults on---R---ig--i;d PV--C* F...i.l.m... : A single sort*ition exoe rime.1o/7t at 25*C on a speci- men of rigid PVC film Q.05 mm thick gave a linear V.t vs. t ' u plot, but the sys tem was still far from diffusion equilibrium after one week. Assuming that the final solubility (M) was similar to that in PVC powders, we estimate D - 3 x 10"^ cnM/sec. Thus it appears that our values of D may also be applicable to fused (molded or extruded) rigid PVC products, but further experiments arc needed to confirm this. Later Stages of So r.otion/Peso rotion: Plots of XL/M*, vs. tly/" tend lo focus attention on Lite first 90 percent of the sorption or desorption process. Yet, in the removal of unrvucted V'CM after pc 1 y 01 e r lea t io. we are concerned with the final few percent of desorption. Tor this region, it is instructive to examine plots of log(lvs. t. Figure 5 shows such plots calculated from equation (1) using r D = 2 x 10"* cm" / s ec and various values of 3g = 4, 29 /fig - These curves thus represent the predictions of the uniform-sphere, Fickean model for sorption or desorption of VCM from PVC powders at 90"C; they are essentially linear beyond Mt/Ma = 0. 5. Experimental log (1 - Mt/Ma) vs. t plots for several representative suspen sion PVC samples are shown in Figure 6. Deviations from the uniform-sphere model are obvious and quite substantial in some cases. The initial part of the curves , to Mt/M^ = 0.8, show slopes s imilar to those predicted from the Sa values. But .then some of the curves show a pronounced decrease in slope, and others a more gradual curvature. The slope in the later stages does not correlate with Sg. We attribute these results to non-uniformity of particle structure. Under the optical microscope, samples K and L, for example, appear largely to consist of opaque, porous grains but also contain a number of translucent to clear, "glassy" particles. In such granules the pores between the original primary particles have probably been largely filled in with PVC during polymerization; the effective dimension controlling sorption rate then more aearly approximates the external particle size of perhaps 50 to 150 microns than the few-micron primary particle size. The effect of such particles on the sorption/desorption curves can be esti mated through equation (1) simply by adding the curves calculated for various weight fractions of different effective particle size. This procedure is illustrated in Figure 7, where we show the predicted effect of varied weight % of 40 micron ("glassy") particles in a sample otherwise comprising uniform 4 micron particles. Sorption measurements on deliberate mixtures of porous and glassy particles give results closely resembling the calculated curves, as illustrated in Figure 8. For the samples of Figure 6, the extents of sorption at longer times are in good quali tative accord with the relative contents of glassy particles. The more gradually curved lines of Figure 6 may similarly be interpreted as the result of a narrower, more continuous distribution of effective particle sizes. Thus the shape of sorption curves can provide much useful information about the uniformity of particle structure. In principle, at least, once D is known, it seems possible to obtain a rather complete measure of particle size and size distribution from sorption measurements on polymer powders. Practical Implications: Cur results suggest that removal of residual VCM from PVC resins could be accomplished in feasibly short times and low tempera tures, provided the polymer has a relatively high surface area (small primary particle size) and is free from nan-porous or glassy particles. The presence of even a small percentage of such non-porous particles will seriously retard the final stages of VCM removal. Similarly, VCM removal will become very slaw once PVC particles have Become fused into a continuous mass during fabrication. Assuming applicability of our D values to molded or extruded rigid PVC products, it may be shown that escape of any small amounts of residual VCM in such products into the environment will be an extremely slow process. Numerical calculations for products of varied geometry and dimensions can readily be made with available equations (5). A CXNCWCrTDCZMZNTS The author wishes to thank Mr. S. A. Oxenrider for his experimental assistance, and The B.F.Goodrich Company for permission to publish these results. GENC017239 REFERENCES 1. A. R. Berens, PoLyme r Frepri..ts, this issue. 2. J. Crank and C. S. Park, Diffusion in Polyme rs, - A cademic Press, London, 1968. 3. H. B. Hopfenberg and V. Stannett, Chap. 9 in The Physics of Glassy Polvmers, R. N. Haward ed, , Wiley, New York, 1973. 4. E. Bagley and F. A. Long, J. Am. Chem. Soc. , JJ_, 2172 (1955). 5. J. Crank, The Mathematics of p-dfusion, Oxford University Press, London, 1956. 6. A. R. Berens and V. L. Folt, Polym. Eagr. Sci. , , 5 (1968). 7. J. E. Glass and J. W. Fields, J- Appl. Polym. Sci., ljb, 2269 (1972). 8. R. TreganandA. Bonnemayre, Plast. Mod. Elastomers 23_ (7) 220 (1971). 9. F. Wolf and . Kreter, Plast. u. Kautsch, _21_, 27 (1974). Figure!. VCM SCfiPTlCN, SAMPLE 0,23X Figure 2. Mt / Mw vs. t,/2,SAMPLE 0, mg VCM/g PVC GENCO17240 &S. t , . Figure 3. SORPTION /DESORPTION, SAMPLE A, 4QC Figured CALCULATED(I-vs.f CURVES for UNIFORM PVCSPHERES, Figure 4. 0 vs. I/T SORPTION OeSOnFTlON SAHPlE lO * A AA X n 1010. V t'-ftT a o * > 10v'M: N N.** Id'1 .1 1 ____ !_____ L 2.8 3 0 3.2 IOCO/T5K 3.4 80 60 40 TC Figure 6. EXPERIMENTAL!!M-~t). vs. f Figure 7, CALCULATED!1-^1 vs. t CURVES for UNIFORM PARTICLES and Figures. EXPERIMENTAL GENC017241 Diffusion and relaxation in glassy polymer powders: 1. Fickian diffusion of vinyl chloride in po!y(vinyl choride) A. R. Berens Corpora re Research, 3. F. Goodrich Co., Brecksville, Ohio 44141, USA (Received 10 January 1977) A model is presented for diffusion-controlled sorption in polymer powders consisting of uniform spherical particles. Gravimetric sorption-rate data for vinyl chloride in PVC powders are found to * obey this Fickian model for certain types of samples over limited ranges of vinyl chloride pressure increment or temperature. Superficially similar deviations from the model are caused by particle nonuniformity and by the onset of a relaxation-controlled transport mode. These two causes can be dis tinguished by appropriate choice of sample type and experimental conditions. A simple modification of the model satisfactorily accounts for the effect of particle non-uniformitv upon sorption kinetics for conditions under which transport is diffusion-controlled. The use of powder samples in vapoursorption experiments seems to afford several advantages over conventional film specimens, including more convenient measurement of the very low diffusivities characteristic of the glassy state, and more definitive separation of the contributions of diffusion and relaxation processes to the overall transport mechanism. INTRODUCTION' The complex transport behaviour of organic vapours and liquids in glassy polymers has been studied extensively for over twenty years and reviewed by several authors1-1. It is now widely agreed that transport involves both a diffusion process, controlled by a concentration gradient, and a rela xation process, controlled by a nme-aependem response of the poiymer to a swelling stress, .As the relative contribu tions of these r.vo processes change and interact, a wide range of behaviours can be encountered. The observed effects may vary not only with the particular polymer/ penetrant system considered, but aiso with the range of vapour activities or concentrations covered in specific experiments. Rogers-1 has pointed out. for example, that sorption vs. time curves may be pseudo-Fickian, sigmoid, or two-stage is the penetrant concentration is increased. .Aifrey e: :l.5-4 consider nmpie Fickian diffusion (sorption initially linear with :and Case II sorption Qinear with r) as limiting cases representing semination by diffusion and reiaxauan, respectively. Hopfenoerg and Fnsch7 have suggestea that virtually the entire range of possible behaviours may be expectea in a given system when a sufficient range ;f temperatures and penetrant activities are investigated. Hopfenoerg3 has also pointed out, however, that at the low temperatures and activities where simple Fickian behaviour in the glassy state might be expected, experimental times become prohibitively long with conventional polymer film samples. Recently, Vrentas, Jarzebski and Duda9 have proposed a diffusive Deborah number, (DE3)q, as a criterion for pre dicting whether transport is diffusion- or relaxation-con trolled. This dimensionless number is defined as: {DEB)o^_ (1) where \m is a mean relaxation time and dp a characteristic diffusion time given by L-/D, with L the sample thickness and D the diffusion coefficient. In this treatment, then, the sample dimension becomes an important factor, in addi tion to composition and temperature, in determining the transport properties to be anticipated. When \m and dp are similar in magnitude, anomalous diffusion (involving both Fickian and relaxation processes) is probable. For (DE3)p K l, rubbery-staie or viscous Fickian diffusion is expected, while for (DE3)p > 1, elastic, or giassy-state Fickian dif fusion may occur. Vrentas at oL thus suggest adding the Deborah number criterion to the temperature-acavity dia grams of Aifrey5 or Hopfenberg and Frisch7 for a more general prediction of die type ot-transport behaviour to be anticipated. A recent study10 of transport in the system vinyl chioride/ PVC employed powdered polymer samples instead of the film specimens almost universally used in previous vapoursorption work on polymers. In powder samples, diffusion paths may be reduced by orders of magnitude below the practical limit in films, and it is thus feasible to investigate transport behaviour at high values of (DEB)p. The anticipat ed, but little studied range of glassy-state Fickian behaviour, and the interesting transition from Fickian to anomalous transport, can thus conveniently be studied. This paper considers the diffusion of vinyl chionde mono- GENCO17242 POLYMER, 1977, Vol 18, July 697 Utrruzja/7 or /myi cruonaa m v'VC'powewrr M.' n, aenmi mer (VCM) in a variety of ?VC resin powders. A model for Ftckian diffusion in a system of small, uniform spheres a presented, and it is shown that sorption data obeying this model are obtained within certain limits of sample type and VCM activity. Superficially similar deviations from the model are caused by parade non-uniformity and by a con tribution from relaxation-controlled transport. Experimen tal methods for distinguishing these effects are presented. Subsequent papers in this senes will consider other vapour/polymer powder systems and extend these studies to situations (larger parades and higher vapour activity) where relaxation-controlled transport becomes an increasing ly important factor. MODEL FOR FICXlAN DIFFUSION IN UNIFORM-SPHERE POWDERS Comparison of experimental data with Fick's Law requires a solution of the diffusion equation appropriate to the sample geometry and experimental boundary conditions. The simplest geometry for powders ts a collection of spheri cal particles of uniform size. This condition can in fact be closely approached m PVC powders prepared by emulsion polymerization. Consequently, as a model to approximate ricxaan sorption of VCM by PVC resins, we have used Crank's solution1-1 for the non-steady-state diffusion in a sphere m the case of uniform initial concentration through the sphere, a constant concentration at the surface, and constant D. The equation for this case is: EXPERIMENTAL .Vfa tenals Polymer samples used in this study included a number of commercial and experimental PVC homopolymers made by suspension- , mass-, and emulsion-polymerization techniques. All were used m the powder form obtained directly from the polymerization process by normal recovery and drying procedures. Particle structures were characterized by optical and electron microscopy and by nitrogen-adsorption surface area measurements. A few measurements were also made on a sample of rigid ?VC film. Specific samples are described where pertinent in the body of the paper. Equipment and procedures The measurements reported here were made -with a re cording Electrobalance (Model RG, Cahn Division, Ventron Instruments Corp., Paramount, California). The balance was mounted in a glass vacuum chamber with connections to a source of VCM vapour and to vacuum pumps. PVC samples were suspended from the balance in a light liuminium pan near the oottcm of a 40 cm Xovar hang-down tube, which was surrounded by a circulating liquid jacket tor tem perature control. Sample weights were from 100 to 500 mg; use of the oaiance at l tag sensitivity thus could detect weignt changes as small as 2 ppm. The normal procedure was to evacuate the balance chamber'anal the sample reached constant weight, then admit VCM vapour to a selec ted pressure and record weight change with tune on a stnpchan -ecoraet (Hewlett-Packard 7I00B). Pressures were measured to nO.i mmKg with a strain-gauge transducer and digital voltmeter. 3ecause of the large vapour volume and small sample size, sorption mused negligible change in pres sure icier m igciuco of VCM vapour. The diffusion rcuancns used n analysing the data assume tna; chances of oressure, and consequently of surface concen tration, are mace instantaneously. This condition can be closely ascrcacnen for increases of pressure; by bringing l small 'esse: to a measured VCM pressure, then quickly open ing this vesse; to the balance chamber, pressure ocuid be in creased to a rreuetermnaed vaiue m a few tenths of a second. Sucn anarp decreases m pressure, especially to zero, axe not so easily achieved. Using oil iiifusicn and mechanical pump ing, hcuid nitrogen imps, and large vacuum lines, pressure couid oe reduced from, say, 200 mm to ~-30 pan in one minute. In most experiments, the half-evacuation tame was consideraoly less than the half-desorption time, so no cor rections were made for the non-instantaneous pressure mange. The resoonse of the balance and recorder provided useful sate oegmrmmg within 3 to 5 sec after a pressure change. M, S xm 1 M. -t-2 i-->< --nl exv(-iDn-rzt/d*z) ft ~ l (2) where Me is the weight of penetrant entering or leaving the sphere m time r after in instantaneous change in surface con centration, .a the total weight change after infinite time at the new surface concentration, n is the series of integers, D is the diffusion coefficient, and d is the diameter of me sphere. The following assumptions are implicit in applying equation (2) to VCM sorption or desorpuon experiments on PVC powders. (1) Diffusion obeys Fick's laws. (2) D is constant over the concentration range covered by the experiment. (3) The solution for a sphere is applicable to a sample consisting of many spherical particles of uniform size; i.e., diffusion occurs in each particle independently, -with no interparcide diffusion. (4) The PVC powder does, indeed, consist of uiuform- nzed spherical parades. (5) At the start and end of the experiment, the VCM - concentration is uniform through the panicles. (6) The VCM concentration at the particle surface 3 always at equilibrium with the surrounding vapour phase. (7) At time zero, the VCM pressure is instantaneously changed to 1 aew value, thus immediately changing the sur face VCM concentration of the parades. (8) The VCM concentration in the parades 3 zero initi ally m the case of sorption starting at zero VCM pressure, and zero finally for desorptions to zero pressure. 19") There is no concentration gradient m the -vapour pnase, and no surface '.aver effect; i.e., transport is limited oniy ay diffusion within the TVC particles. [f ail these assumptions are vaiid, then equation (2) will describe the weight gam (or loss) as a function of time m 1 VCM sorpaon (or desorption) experiment. This equaacn predicts ihnt the time required for a given fraction of the intimate weight change to occur 3 uadepende.it of the initial or find VCM conesntraaon (or pressure), and .s pro portional to the ntuod-;D. This mao, then, can be deter mined by fitting equation (2) to the experimental data, out evaluation of D requires independent determination of the particle diameter. To test the agreement between experimental data and the model, two graphical representations of equanon (2) are useful. As shown in Figure l, 1 plot of .tyVAIL (or 0iMt) against the square root of time is initially nearly linear, then 598 3ClvVlR 'S77, Vol 13, July GENC017243 ^ |1 mz.;i Figure 2 Plot of ogil -- vvitn p-,0 ' 10* sac Diffusion of vinyl chloride in PVC powders: A. R. 3erens Slope = 4tt-D 2303d2 The determination of d~/D from half-sorption time depends mainiy on the data in the early stages of sorption, while the log(l method emphasizes the later-stage data. r (sec) vs. r. calculated from equation (2) FICKIAN, UNIFORM-SPHERE BEHAVIOUR rN VCM/PVC SORPTION Sorpdon and desorption rate data for VCM in a wide variety of FVC powder samples have been obtained at temperatures from 25 to 110C and VCM pressures from zero to 700 mm. The experimental data thus extend well below and above the glass transition temperature of PVC (7^ = 85aC). Rela tive VCM pressure,Pre( = P/Pq, where P0 is the saturated vapour pressure, ranged from zero to a maximum of 0.23 at 25aC (where Pq = 3000 mm) and to a maximum of 0.04 at 110aC (j?o = 19 200 mm). The form of the Mr vs. r1^2 curves was found to depend both upon the type of PVC sample used and upon the pressure range and temperature of the experiment. Curves closely fitting the Ficldan, uniform-sphere model, equation (2), are found under limited conditions of VCM pressure or temperature, but only for samples which closely approximate the "uniform-sphere' requirement. ' Uniform-sphere'PVC samples By emulsion polymerization, it is possible to prepare PVC powders which consist of spherical particles of diameters from approximately 0.1 to 1 pm, with very narrow size dis tributions. The M, vs.:1/2 curve f7om a typical VCM sorp tion experiment on one such monodisperse sample is shown in Figure 3. Tne experimental data conform very well to the curve calculated from equation (2) using the d-jD value obtained from the half-sorption time and equation (3). Ln the case of PVC resin powders made by suspension or mass polymerization, the gross particles (roughly 100 pm diameter) seen at low optical magnifications are neither spherical nor uniform in size. Electron microscopy has shown, however, that these gross particles are agglomerates of primary panicles (~I to 5 pm diameter) which are appro ximately spherical and fairly uniform m size within a given resin sample. This observation suggested application of the uniform-sphere model ro sorpnon data on the suspensionand mass-polymerized PVC resins. Figure 4 shows an Elus- bends over to apt?roach a wei]-defined limit. A plot of log (1 - vs. as m Figure 2, snows ininai curvature, then approaches a straight line. Each of these plots aiso offers a means of evaluating d-;'D from experimental sorp tion data. From equation (2) it can be shown that the time at which .Wf/AL = 0.5, or the "half-sorption time', is d2 tO.i = 7.66 x 10TM2 -- (3) for : in sec, d in cm, and D m cm2/sec. Thus fgj may be interpolated from a plot of or of.Vfr, vs. r ll~ and d2/D evaluated by equation (3). Alternatively, d2/D may be obtained from the limiting slope of the logigO - vs.: plot: GENC017244 Figure 3 Plot of vs. rJ/* for VCM sorption by 0.44 am emulsion PVC, 23aC, Pret * 0 -`0.C037. O, Experimental:------- equation (2) for r<j.j 7.3 sec POLYMER 1977, VolTS, July 699 Diffusion of vinyl chloride in PVC powders: A. P. 8erens Figure 4 PIor of Vw'A4_ ru ;`n for VCM ionstion by !ujo*nnonOOlvmerizsd PVC, 4QJC. * Q -'Q.CQ43. O, Ejo*nrnnai; ------ quation (21 for r<j_$ " 38.4 sac traove Mr vs. f curve from a VCM sorption experiment on i suspension-polymenzed ?VC; the agreement of `lie data with equation (2) shows that the uniform-sphere model is an adequate approximation for this particular sample. The degree of agreement with the model, however, varies considerably among different samples of mass and suspension types of PVC. It wail be shown below that such variations among samples can be related to the uniformity of their particle structure. Temperature and VCM pressure conditions for Fickian behaviour For samples of known uniform-sphere particle structure (monodisperse emulsion FVCs), VCM sorption or desorption data fitting equation (2) are obtained only within certain conditions of VCM pressure and temperature. At tempera tures below Tp such ample F.ciaan behaviour is observed only for sorption runs covering very small increments of VCM pressure (Prti increments of 0.01 or less), and for de sorption runs to zero pressure. Above Tp equatioa (2) is obeyed m soronon or desorpoon runs over the full ? range covered by the data. Experiments outside these con ditions invariably lead to slow changes in sample weight continuing long beyond the rime required to reach equili brium according to equation (2). Thirsort of deviation from the Fickian uniform-sphere model will also be dis cussed below. Determiruanon of diffusion coefficient'- When rite ?VC sample and experimental conditions pro duce data firing '.he P.dcsn. urmorm-spnertt model, the raao a--D-s most reaaiiy determined from the half-sorption ame. Sorpncn or desorption experiments are run to dif fusion equilibrium to esraoiish.V,, then the ome, 'o.s, at which .Vf, = O.i W_ is -sad from an Mt vs.;11 - plot, or airectiy from the recorder trace of weight vs. rime. D is then calculated via sauaucn (3) from .'o,t andd. The camcie diameters 'used tn caicuiadon of D were de termined from. adsorpaon surface area measurements. For inheres of density o, the sumce-to-wetght raao is oipd, thus far ?VC (p = 1.4): a, = 4.29/^ (5) lere S, is the specific surface area in m2/g and d; is surfaceerage particle diameter in jam. This measure of particle tc agrees reasonably well with electron microscopy for both emulsion PVCs and the primary particles of suspension resins. Diffusion coefficients determined by this-procedure For emulsion, suspension, and mass PVC resms are in good agree ment. This result indicates that the rate-controlling factor in VCM diffusion is the dimension of the primary particles in each case, not the agglomerate size in suspension or mass resins. Use of the agglomerate si2e in calculating D from sorp tion data would increase D by several orders of magnitude; this factor appears to account for the discrepancy between our results and those reported by Wolf and Kreteri: Temperature dependence ofC The diffusion coefficient for VCM in PVC has been de termined from haif-sorpaon or -desorption times at tempera tures from 253 to 110 C. It was necessary to use samples of differing d, in different temperature ranges, since r becomes inconveniently long at low T when d, is relatively large, while .`q.s d too short to measure accurately at the higher Ts whan d, is small. For the determination of D, only data closely approximating the Fickian, uniform-sphere model were used; these data therefore covered low VCM pressures and VCM in PVC concentrations up to --2 mg/g. Results of these measurements are presented as an Arrhenius plot in Figure 5. The agreement among samples is good; most of the scatter may be attributed to uncertainty in o'. Since these samples covered the molecular weight range found among commercial PVCs, it does not appear that D vanes significantly 'with molecular weight in this range. The value of D also appears to oe msensiuve to the type of PVC resin, as emulsion, suspension, and mass resins give very similar results. The straight line cf F'gure 5 indi cates an activation energy for diffusion of 17 kcai/mol. The equation of this line: D - 3.7exp(--17 OQOjRT) (6) may be used to estimate D over the temperature range of the data. ^HrS^^sKis oi sorpnon and desorption runs on the same sample and over a similar VCM pressure increment, it was generally observed that desorpoon was somewhat slower than sorption. Figure 6 illustrates this effect for,a representanve set of sorption and desorpoon data, plotted as .Vf, vs. r1 Consequently, values of D calculated from iC'V ;C-oi lC-.ii I,s w, 5% - * "s' f r 1CT2SL6 'Nt IS V4 , 2-8 30 32 :COC/r!K! 3'4 Figure 3 Plat af ',aq O 'n. MT *qr VCXI dirfutian >n PVC powcers: , sorption; O, desorption 700 PC L VMS P. Voi I 3, Juiv GENCO17245 Figure 5 Pfot of Mf/M^ vs. for VCM sorption and desorption by suspension PVC. 40* C, Pr%\ a 0 "*0.005 -*0. , Sorption; Q, aesorpnon for desorption are lower titan those from sorption runs; this trend is apparent m the data shown in Figure 5. Such an effect ;s characteristic of systems where D increases with increasing concentration of the pentrant11 and thus does not seem unreasonable here, smce VCM does tend to plasticize PVC. To confirm this effect and estimate its magnitude, we have determined D from several ser.es of sorption experi ments ji which the VCM pressure was increased m successive stens .``interval sortccn' experiments'/. Results of such runs it 30" ana cO'C are shown m F'gure 7 as plots off? vs. e. where / is the film thickness1*. From the D values ootainea on powder samples, equation (7) predicts inconveniently long experimental times for measurements on films of any reasonable thickness. Yet it does seem important to estab lish whether or not the D values determined in resins would also apply to fused PVC items. For this purpose, VCM sorp tion measurements were made on a sample of experimental extruded ngid PVC film which contained 2.5 % w/w of a tin stabilizer as the only additive. Sections of thickness 0.005 cm were used in VCM sorption experiments over a pressure interval of 0 to -- 100 mm at 25, 70, and 90C. The initially linear .\ft vs. r ^ plots, shown in Figure 8, indicate Fickian diffusion behaviour. Only at 90"C was equilibrium reached within the time available. For the other tempera tures, and fqj were estimated by assuming the equili brium amount sorbed would be similar to that sorbed by powder samples. The estimated values of D at 25, 70, and 90C were 3.1 x 10~L-, 8.0 x 10~11, and 5.4 x 10~10 cm2/sec, respectively. These values are somewhat higher than the average D% for pure resin powders (see Figure 5), suggesting slight plasticization by the stabilizer. Yet the order-of-magmtude agreement indicates that the pure-resm D values should be applicable, to a fair approximation, in estimating VCM migration rates from rigid PVC products. Other support for this approximation has recently been ob tained from the good agreement between observed VCM mi gration rate into water from PVC pipe and the predictions of a model based on pure resin diffusivity values13. These few measurements on PVC film demonstrate an important experimental advantage to the use of fine poly mer powders Instead of sheet or film specimens; the measurement of very low diffusion coefficients can be brought into in experimentally convenient time-scale with powder samples. For example, the half-sorption time for a 0.005 cm film withZ? = 10-12 cm2/sec is about 14 days; but for a powder of l uni particle diameter and the same D, rgj is only 76 sec. DEVIATIONS FROM FICKIAN, UNIFORM-SPHERE MODEL It was noted above that sorption or desorption vs. time data fitting equation (2) are obtained only when both the Results on rigid PVCfilm For sorpnon by a plane sheet or film, the relation analo gous to equation (3) tor spheres is: tn 5 = 0.049 D (7) Figure 3 Plan of MR/W., vj. pf* for VCM sorption by 0.005 cm rigid PVC film. O, 23jc, X. 7U1C; , 90*C GENC017246 POLYMER, 1977, VoM8, July 701 'Diffusion of vinyl chloric* in PVC powders: A. R, 3rens 40 30 120 160 r'tys'/a} r jgun 9 Deviation from equation (2! aue to nqn-4icitian ionation; U. vs. r1^ for VCM varpoon by 0.44 m mono6tJsn emulsion PVC, :o*C. P,M - 0,02 "*O.OS. . ObrTMs: ---- ,nq---------, equation ill rlgurw 10 Deviation from-equation (2! Oue ro sarnde nonumformitv; ,W. vi. pn' for VCM loqstion by sommercial lusoenjion PVC. 30*C. Pru * 0 -*0.015, 0, Cbearved; --------end------------, equation (2) PVC sample and the VCM pressure range of the experiment meet certain requirements. Experiments outside either set of requirements lead to weight changes which continue long beyond the equilibration time predicted by the model for the ipprconate values ofd~;D. While the results are superficially very stmiiar, it is possibie to distmguisn clearly rwo independent tTM of this behaviour: parade non'mifr.mry ind ncn-Ficiuan diffusian. Disnncron jerween pamcie non-umfarmity md nonFidcan effects These two effects can be separated by appropriate choices of sample and experimental conditions. For example, for a monoaisperse emulsion ?VC (whose sorption--ame curve over i small VCM pressure increment fits equation 2), devia tions from the model m sorption over i larger pressure in crement cannot be attributed to parade non-uniformity; thus a contribution of a non-Fidaan diffusion process is in dicated. Conversely, when slow equilibration is observed in sorption over i small VCM pressure increment (one which gives data fitting (2) for a monodisperse sample), the devia tion from the model can be attributed to parade nonuniformity. Similarly, slow equilibration in sorption or de sorption at T> Tf (where the model is obeyed for uniform- pamcie samples) is also evidence for particle non-uniformiry. Figures 9 and 10 show examples of sorption--time runs in which the deviation from the Fickian, uniform-sphere model can clearly be attributed to particle non-uniformity, in one case, and non-Fickian diffusion in the other. The similarity of the experimental curves shows that these sorp tion data alone would be insufficient to distinguish the cause of the deviation; the particle structure and pressure increment must also be considered. By suitable experimental design, then, it is possible to use sorption--tune data unambiguously to obtain informa tion about either pamcie non-uniformity or about the noaFiciaan diffusion process. Both sorts of experiment will be discussed further m the following sections. EFFECT OF PARTICLE NON-UNIFORMITY Results an commeraal reran samples Sorption or desorption curves on many commensal suspension- and mass-polymerized PVCs, run over small VCM pressure Increments, give Mr vs. t plots approxima ting the uniform-sphere Fictaan model. However, when the same data are plotted as Iog(I -- .Vf./iVL.) vs. f, emphasizing the later stages of surpnon or desorption, substantial devia tions from equation (2) are apparent. Figures 11 and 12 compare the two types of plot for a sorption run at 90C from vacuum to 200 mm VCM pressure (frej = 0,015) for a representative mass-polymerized PVC. Since the high Tem perature and small pressure increment tend to eliminate any ncn-Ficican contribution, the deviation from equation (2) may be attributed to pamcie non-uniformity. Previous knowledge of mass- and suspension-PVC particle structure is consistent with a possible effect of nonunifomity on rate of VCM removal. It has long been recog nizee 'hat certain resms contain a small portion of non-porous pamc.es often identified with `fish-eye' problems. These pamdes are detectable by their dear appearance under a low-power, optical microscope, in contrast to the opaque, white appearance of porous particles. They are believed to arise from excessive knitting-together of primary pamdes during polymerization to form solid PVC regions which may be as large as the gross agglomerated particles of the resin, i.e,, up to about 100 pan. Now if this dimension, rather than the few-micron primary parade diameter, is the controlling i-5r ? i-CF 2 ID > s--m 7yttr* 11 Plot if '.I- vi. " for VCM variation bv commercial lasj-ooJymaniad PVC, SO11* , * 0 ""O.OlS. 3 Obsrved; --, *qu3Ticn (2) ?cr ra^ 5,2 sac, \ ,22 mg/g 702 PCLYMS^, '977, Voi :s, July GENCO17247 Jta Figure 12 5am data and eurva as Figure 11, plorted as I09II ~ MJM^) vs. r Diffusion of vinyl chloride tn PVC powders: A. R. 3enens calculated by applying equation (S) to hypothetical mix tures of 4 and 40 qm particles for desorption of VCM at 90C (D = 2 x 10"10 cm2/sec). Experimental confirmation of the model for non-uniform particles has been obtained through measurements on two resins of different average primary particle size fd, = 1.9 pm, and d: = 20 pm) and on a deliberate 90/10% w/w mixture of the two. The results shown in Figure 14 agree well with the form of the calculated curves of Figure 13. The VCM de sorption curves for many commerdal suspension and mass PVC resins are also in qualitative agreement with their rela tive non-porous partide content judged from microscopy. It thus appears that the effect of non-uniform particle size upon the Ficldan sorption of a vapour by a polymer powder can be modelled rather simply through equation (8). The consistency between experimental data and this model suggests, moreover, that vapour sorption measurements may be a useful method for obtaining information about the particle-size distribution of polymer powders. From calcula ted sorption-time curves for mixtures of two particle sizes (e.g.. Figure 13) it is seen that the slope and level of the curve at the longer times are dependent on the size and weight fraction, respectively, of the larger parades. Similar ly, the curves at short times reflect the size and weight frac tion of the smaller particles. It seems clear that the presence of more than two sizes in the population would cause a more gradual change in slope. In principle, an experimental sorption tune curve seems to contain the information needed to derive a weight-fraction particle size distribution for a polymer powder. Extraction of the size distribution by fitting equation (S) to the experimental data, however, may be a formidable task. A prerequisite for this technique, of course, wouid be the demonstration that the chosen experi mental conditions produce Ficiaan behaviour with a sample known to have a uniform panicle size; without this evidence, it might be impossible to distinguish panicle-size distribution effects from a contribution of non-Fickian sorption, Figure 13 Curves aalculatad from equation (3) for uniform and mixed panicle sues. voids O " 2 X 10~* cnr/iec. A, tOG% 4 jun particles; 3, 95/5 % H/W4^n/40 um; C, 90/10; O. 30/20; Z. 50/50: F. 100% 40 um NON-FICX2AN DIFFUSION EFFECTS Even with monodispene, emuision-PVC samples, sorption experiments below Tg and covering VCM increments greater than about 0.01 show 1 much slower approach to factor m rate of sorption, then the nan-porous particles will sorb VCM much more siowiy than the porous ones. The presence of a tew such parades in an otherwise porous resin thus could /veil account for slow sorption or desorption of the last fraction or the total V CM. Model for 'ion-uniform panicles The affect of a aon-uniform particle size can be predicted through equation (2) by amply adding the curves calculated for various oarncle sizes in proportion to their weight frac tion in the sample. Thus we may write; aa 1 - -- = -tT % 2 -Wp(--4Drera2r/d,2) ir^ ^ni n= 1 (3) where X, is the weight fraction of particles having diameter a,. To illustrate, the curves shown in Figure 13 have been % w/w 1.9 nm/20um GENC017248 POLYMER, 1977, Vol 13, July 703 intrusion of~v<nyt ailonda in PVC powders: A. ft. Serens Figure 15 Schematic separation of diffusion and relaxation con tributions to VCM sorotion by wsoenston PVC. 50"C- , Arel " 0 0.002: 0. P;*i * 0.002 -*0.019 equilibrium than predicted by the Fickian uniform-sphere model. It is suggested that this behaviour represents the on set of a nnn.Fieinan relaxation-controlled mode of sorption superimposed on the Fickian diffusion, as indicated in Figure IS. A somewhat similar `two-stage' sorption process has been observed by Bagiev and Long1** for the sorption of acetone by ethyl cellulose and by Fujita: tor several other systems. Their interpretation also seems applicable to the VCM/PVC system; in the eariy stage of sorpnon. the VCM concentration gradient provides the maior driving force and the transport process is dominated oy Fickian diffusion. When the concentration of VCM resulting from this process is sufficient to develop a significant swelling stress, the slow rescor.se of the glassy poiymer to this stress produces gradual swelling of the polymer structure and permits additional sorption. This swelling-relaxation process may continue even after the Fickian process has produced a virtually uni form concentration of VCM through the polymer; ui this event, the contribution of Fickurn diffusion becomes negli gible, and the later stage of sorption is dominated by the relaxation-controlled process. In the fine PVC powders studied here, the short diffusion oaths result ji emulibntaon tines is short as a few minutes for the Ficxaan aiffusian process. Thus this stage truy oe essentially romoiete before appreciable stress-relaxation has occurred. The net result is a first stage wruen is nearly pure Ficxmn aiffusicn cna a second stage wnach a cimost entirely relaxation-controlled. h letcrsticn e.toenments :o a aon-aero VCM pressure m VCM, ?VC oowder systems, a contribution of die reiuxaucn-conucilec. process aiso is indicated, as :he iporoacn to tree rotten ecudibnum may be very slow12, tn desorations ;o tero VCM pressure, however, 'he sample weiaht follows me Ficman curve to a rapid 'equilibrium' of aero ''CM sontent. Resorption experiments12 have provided evidence mat voiume relaxation of the RVC continues as a slow process even in the aosence of sorbed VOl. It appears that, inder /acuum, ail of :he sorbed VCM escaoes by the rand Fickian process, but volume-relaxation is not neces sarily completed during desorpaon. This benaviour claseiy parallels the observations of Barter, Same, and Slater14 in studies on ethyl cellulose. in contrast to the two-stage serpuon behaviour in the alassy state, sorption and desorption experiments conducted above Tt appear to follow simple Fickian diffusion kmetiui over the VCM pressure range of the present data (f%< * 0 to ~0.05). Thus, in the rubbery state, the relaxation process seems sufficiently rapid to be complete within the time scale of the Fickian diffusion stage. This study of the VCM/PVC system indicates that vapoursorption experiments an polymer powder samples in the glassy state offer a significant advantage over similar experi ments with film samples, in that diffusion and relaxation effects may be conveniently separated. The short diffusion times encountered with fine powders permit the attainment of `diffusion equilibrium' before appreciable relaxation occurs. The extent of first-stage sorpnon then reflects the state of the poiymer essentially unperturbed by penetrantswelling; this has proved to be a useful technique in assessing the effects of poor polymer history on die properties of glassy PVC12, The rate of sorption in the Tint stage is prinmpaily controlled by the Fickian process .and, therefore, may be used to estimate a true diffusion coefficient even when the total sorption does not follow the Ficldan model. Furthermore, since the second-stage sorption is almost en tirely relaxation-controlled, the kinetic and equilibrium parameters describing the slow relaxation processes can be obtained from long-time sorption data. With film specimens, in contrast, diffusion Umes and relaxation times are often comparable, hence separation of relaxation from purely Fickian diffusion is often impossible. Further studies on vapour sorption m glassy polymer powders and mathematical treatments of the data to separate diffusion and relaxation rate constants will be reported in the near future. ACKNOWLEDGEMENTS The author is grateful to Professor H. 3. Hopfenberg for numerous helpful discussions and construcuve comments on ctus work. The permission of The 3. F. Goodrich Co. to publish this work is aiso gratefully acknowledged. REFERENCES 1 `Diffusion in Polymers', (Eds J. Crank and C. 3. Puk), Academic Press, London. 1963 2 Fajita, H. Fonschr, Hccrwoivm. Finch. 1961, 2. I 3 Rogers. C. E. in 'Physics and Chemistry of :he 0ramie solid State', '"Eds For, L-ioes, and Veissoerger), Interscienee, New Yorx. 1963, Ci 6 a Hopfenberg, H. 9- md Stanner*. V In The Physics of Classy Polymers', `Ed. R. N. Hawaii!), 'Vjlev, New Yarx. 1972. Cl 9 3 Alfrey, T, CTie.m. any. .Vyws 1963, S3, 64 > 6 Ufrey, Gurnee. F ind L.aya, W G. /. Polym. Sc-- , C," 1966. 12. 249 Hopfenberg, a. 3. ind Frisco. H. L.J. Poiym. Sea fB) 1969, 7. 405 3 Hopfeneerg, H, 3. In 'Membrane Science ind Technology', (Ed- i. E. Flinn), Plenum Press. New York. 19"T0, ? 16 9 Vrenras, J. 3., iarceosxi. C. M. and Ouda, J. L- AICAE J. 197J, 21, 394 10 3erens, A. R.Aoivm. A*pr 19'!, 13. 203 11 Crank, J. The Mathematics si Diffusion', Clarendon Press, Oxford, 1956 12 'Volf, F, and krerer, E. P'ast. u. Kauucn 1974, 21. 27 13 3erens. A. ind Daniels, C. A. Polym, Frtg. 5cl 1976, 16, 532 14 Bagley, E. and Long, F. A. j. A/n. Otcm. Soc, 1953, 77, 2172 15 3etens. A. R. 2nd Int, 5ymp Poly/Vinyl CniondcI Lyon France July 1976 (in press, J. .'rtaenmoi. Scl 3/ 16 Barret, R. M-, Bame, J. A- and Slater, J. J. Polym. 5cL 1957, 23.315 7Cd TLVVS = 7 Voii 3, July GENCO17249 ETHYL CORPORATION RESEARCH AND DEVELOPMENT DEPARTMENT RESULTS OF EXPERIMENTAL PROGRAM FOR DETERMINATION OF VINYL CHLORIDE MONOMER DIFFUSION IN POLYVINYL CHLORIDE - PHASE II February 23, 19 79 W, E. Burt G, A. Daniels G. C. Gaeke GENC017250 Results of Experimental Program for Determination of "vinyl Chloride Monomer Diffusion in Polyvinyl Chloride - Phase II The results from the first phase of this experimental program were covered by a report dated June 26, 19 78. This report, was reviewed with Food and Drug Administration personnel at a meeting in Washington on August 25, 1973. A need for additional experimental work was defined during this meeting. A proposed program to cover this additional work was submitted with a cover letter dated September 5, 1978, A letter of October 19, 1978 from the FDA approved the proposed program and requested spiking and recovery studies using the FDA methodology to demonstrate the precision and accuracy of the method in our hands. The procedures are described in this report. All analyses were obtained with the FDA method with detection by FID. 1. Objective: To determine the replicability of the FDA analytics method for VCh as a function of time at about the 0,001 ppm level in PVC pellets and bottles. Procedure: Prepare a well mixed sample of about 40 pounds of PVC compound pellets containing about 0.001 ppm VCh. Prepare a batch of about 50C sixteen-ounce bottles from these pellets. Analyze replicate bottles and samples of compound each month for a period of several months. These data will define replicability among bottles and with compound in which all samples have the same VCM concentration. Results: Analysis of the pellets and bottles by month are as follows: Pellets (as ppm VCM) October, 1978 November December January, 1979 0.0018 0.0011 0.0021 0.0007 0.0007 0.0015 age 0.00131 0.0007 0.0008 0.0008 0.0009 0.0012 0.00088 0.0014 0.0013 0.0018 0.0020 0.00163 0.0013 0.0012 0.0010 0.0014 0.0009 0.00116 GENC017251 2. October, 1978 0.0013 0.0020 0.0020 Average 0.00176 Bottles (as ppm VCM) November December January, 1 0.0016 0.0017 0.0016 0.0012 0.0022 0.0014 0.0009 0.0006 0.0014Q 0.0014 0.0016 0.0020 0.00167 0.0014 0.0015 0.0012 0.0014 0.0010 0.00130 The October and November analyses were obtained with all pellet analyses in one time period and all bottle analyses in another time period. As analyses appeared to vary with the time period, the December and January data were obtained by analysis of a pellet and bottle sample at the same time. Conclusion: The standard deviations of the data indicate that replicability of the analytical procedure over a short time period (days) is about 0.0003 ppm and about 0.0004 ppm aver a long time period (months) for VCM in FVC at the non-migration level. 2. Objective: To determine if there is a variation in VCM con centration in PVC bottles with time of blowing for bottles blown from a well mixed compound pellet sample. Procedure: The bottles blown for Part 1 of this program were numerically marked as they were produced. Every tenth bottle was set aside. Bottles for analysis in Part 1 of this program were selected from a numerical distribution represented by the "tenth bottle" samples. Results: A list of the bottle numbers and the analyses are as follows: tie No. ppm VCM Bottle No. pom vcm 11 0.0014 321 0.0020 41 0.0012 331 0.0014 51 0.0015 361 0.0016 91 0.0009 401 0.0016 121 0.0020 431 0.0016 141 0.0014 451 0.0020 151 0.0014 471 0.0010 201 0.0022 482 0.0006 221 0.0012 241 0.0017 281 0.0013 GENC017252 3, Conclusion: There is no apparent variation in VCM concentration in PVC bottles with time of blowing. 3. Objective: To demonstrate that VCM does not diffuse from PVC bottles to the contents with a VCM concentration of about 0.001 ppm in the bottles and at accelerated storage conditions. Procedure: Store eight bottles as prepared in Part 1 of this program filled and sealed with vegetable oil for six weeks at 120F. After completion of storage, analyze in triplicate for VCM, Store twelve bottles as prepared in Part 1 of this program filled and sealed with 50% aqueous ethanol for six weeks at 120F. After completion of storage, analyze bottles in triplicate for VCM. Results: Analyses for six bottles .with corn oil after accelerated storage are 0.0026, 0.0024, 0.0025, 0.0018, 0.0010, and 0.003S ppm VCM. Analyses for five bottles with 50% aqueous ethanol after accelerated storsge are 0.0014, 0.0007, 0.0018, 0.0014, 0.0011 ppm VCM. Conclusions: The average of the analyses of the 19 bottles from Part 1 is 0.00147 ppm VCM. The average for the first five bottles after oil storage is 0.00206 ppm VCM. The average for the five booties after 50% aqueous ethanol is 0.00128 ppm 0.0004 ppm VCM, the three analyses are probably not different. This demonstrates that VCM does not diffuse from PVC bottles to the contents at this VCM concentration in the bottles. 4. Objective: To determine the diffusivity and activation energy of VCM in PVC at low VCM concentrations. Procedure: Repeat the procedure described in the proposed program Part 3 included with a letter to the Hearing Clerk of March 6, 1978. A different drying procedure was used to remove water from the resin samples which has been demonstrated to. remove water without change in the VCM concentration. Results: Initial VCM concentration in the resin was 0.614 ppm. Experimental data are as follows: (all values as ppm VCM) GENC017253 ETHYL CORPOPLTION Rtsr.u^cs i.ND DsvrLOPitcjrr Dzpa^thz^tt February 21, 1979 H15f iOOtH HEH.T TO; A O. SOX 3-.1 BATON naoot, CA. 700*1 Dr. Robert Livingston Food and Drug Administration Division of Chemistry and Physics 200C Street, S.W. Washington, D.C. 20204 Mail Code HFF-144 Dear Bob: We have completed the experimental data in accordance with our proposed program of September 5, 1978. Data complete in December were included in a letter to you of December 12, 1978. This is an informal report of the remainder of the data. Ccmoound and Bottle Analvsis Continued analyses blown from compound from the follows: Pellets- (January) 1.3, Bottles-(January) 1.4, of compound pellets and beetles same batch as the pellets are as 1-2, 1.0, 1.4, 0.9 ppb VCM 1.5, 1.2, 1.4, 1.0 ppb VCM Average 1.16 ppb 1.30 ppb The December and January data were obtained by analysis of a pellet and bottle sample at the same time. This was done because our October and November data were obtained with all pellet analyses in one time period and all bottle analyses in another time period. There was an indication of analytical variation with the time period. Oil Storace Bottles stored with com oil for six weeks at 120 aF were analyzed by FDD. Individual bottle analyses are 2.5, 2.4, 2.5, 1.3, 1.0 and 3.9 ppb VCM for an average of 2.06 ppb for the first five bottles and 2.37 for the six bottles. Activation Fr.ercv Initial VCM concentration in the resin was 0.614 ppm. Analyses after the sequential steam stripping periods are: (analyses reported as odd VCM) GENC017254 ETirVX CORPORATION a*o DW*>r*uaiT DtHjfTwrwt Dr. Robert Livingston February 21, 1979 Page 2 Stripoing Temp. C 50 70 100 Composite After First Strip 9.0, 8.5 15.4, 10.5 3.8, 5.3 7.8, 10.0 After Second Strip 3.1, 3.4 1.9, 1.3 3.0, 1.8 . 3.8, 3.4 After Third Strip 1.7, 2.3 1.4, 1.6 1.8, 2.7, 2.1 -- Temp. Range 100-70 C 70-50*0 A First Strip 18.1 K cal 15.8 Second Strip 15.3 K cal 25.7 Third Strip 9.3 K cal 25.5 Feed for the third strip was a composite that analyzed 3.3 ppb VCM. We are preparing a formal report to cover these data and the data included in the December 12 letter. Sincerely, G. A. Hughmark GAH:mfl cc: Dr. D. Dixler - Keller and Heckman GENC017255  Stripping Temp. C 50 70 100 Composite After First Strip 0.0038 0.0053 0.0154 0.0105 0.0090 0.0085 0.0078 0.010 After Second Strip 0.0031 0.0034 0.0019 0.0013 0.0030 0.0018 0.0038 0.0034 After Third Strip 0.0017 0.0023 0.0014 0.0016 0.0018 0.0027 0.0021 -- " Feed for the third strip was a composite with duplicate analyses of 0.0033 and 0.0033 ppm VCM. The Anoen dix to the report of June 26, 1978 describes the calculation of diff us ion coefficients as D/a . Calculated values from the above data are: Striooina Temo. C First Strip D/a3 (sec*1) Second Strip Third Strip 50 3.02-10"5 4.57.10-6 4.18-10-5 70 1.26*10"" 4.69 -10_s 1.39-IQ-5 100 1.06-10-3 2.8410"" 1.2210* s Calculated activation energies are: Strip 1 2 3 Initial VCM, tom 0.614 0.0089 0.0033 Activated Energy K cal/g Mole X0Q-70C 7Q-50'C 18.1 15.3 9.3 15.8 25.7 25.5 Conclusions: The activation energy for 50- 70C does appear to increase at very low VCM concentrations. The analytical1 reproducabil itv is such that lack of precision of the analytics values represents a large error factor for the activation energies at low VCM concentration. Thus, the third strip activation energies are subject to a large error. GENCO17256 ETHYL CORPORATION Reseaj^c^ a-s-d Deveiopmext Dsxahtoz.'R' February 26, 1979 pce*< aoo*cjj e*vr TOl O. QOY 34) baton nouoe, i_>. 70a2i Dr. Robert Livingston Food and Drug Administration Division of Chemistry and Physics 200C Street, S.W. Washignton, D.C. 20204 Mail Code HFF-144 Dear Bob: The values for the VCM concentrations after the first strip were interchanged for the 50 and 100C strips. Please replace page 2 of my February 21, 1979 letter with the attached corrected copy. Sincerely, ) As) # A/ ^ G. A. Hughmark ' GAH :mfl Attachment CC: Dr. D. Dixler Keller and Heckman bcc : W. E. Burt G. A. Daniels G. C. Gaeke A. J. Haefner J. R. Lees GENC017257 ETHYL COllPORATIOX tmjKu +>c Qcrtiorxuf Dtr4fTH(>T Dr. Robert Livingston February 21, 1979 Page 2 Stripping Temo. C 50 70 100 Composite After First Strip 9.0, 8.5 15.4, 10.5 3.8, 5.3 7.8, 10.0 After Second Strip 3.1/ 3.4 1.9, 1.3 3.0, 1.8 , 3.8, 3.4 After Third Strip 1.7, 2.3 1.4, 1.6 1.8, 2.7, 2.1 -- Temp. Range 100-70 C 70-5QC First Strip 18.1 K: cal 15.8 Second Strip 15.3 K cal 25.7 Third Strip 9.3 K cal 25.5 Feed for the third strip was a compos ite that analyz ed 3.3 ppb V CM. We 3 re p reoarinc a formal zreport to cover these data and the data included in tne December 12 letter. Sincerely, -y G. A. Hughmark GAH:mf1 cc: Dr. D. Dixler - Keller and Heckman GENC017258 Non-Migration Level of Vinyl Chloride Monomer in PVC A. J. HAEFNER and G. A. HUGHMARK Ethyl Corporation Baton Rouge. Louisiana Recent evidence confirms previous findings that VCM in levels of l ppb in PVC bottle resins are nonmigrafory and that such resins can be produced in commercial quantities. his is an update on the paper presented at the TSPE Meeting tn April, 1968. At that time we reported experimental data that indicated a nonmigration level of about one opb VCM in PVC, Our recent work confirms this non-migration level. EXPERIMENTAL PROGRAM A three part program was conducted following discussion and review by FDA scientists. All low level VCM analyses in PVC were determined with the FDA analytical method using flame ionization detection. 1. PVC bottle resin was batch steam stripped as a dilute slurry m water at 100C. Heat up with steam plus strip time at temperature was lVh h. In termediate and final samples were filtered and dried at mild conditions prior to duplicate anal ysis, The experimental data are as follows. Time (min) initial 15 30 60 final (90) VCM (ppm)* 2 0.0031.0.0025 0.0005. 0.0006 0.0006. 0.0005 0.0007, 0.0004 0.0007, 0.0006 2. PVC bottle resin with an initial VCM concen tration of O.T ppm was steam stripped as an aque ous slurry at 50. TO, and 100C to develop diffusion coefficient and activation energy data as a function of temperature and concentration at low VCM levels. Sequential strips for 6 min at 100C, 45 min at T0C, and 3.5 h at 50C were used. Samples were filtered and dried after each strip prior to analysis. Unfortunate!v, we encountered a problem in the drving procedure with the material from the third stnp so that these data are not valid. Similarly, feed analysis for the fourth strip at 50 and 70C are not available. Experimental data are as follows: Stripping After After After After temp, *C first stnp second strip fourth strip fifth strip 50 0.025 ppm 0.0074 ppm 0.0001 ppm 0.0012 ppm 0.0113 0,0004 0.0009 70 0.030 0.0040 0.0000 0.0021. 0,0030 0.0008 0.0010 100 0.012 0.0030 0.0010 0.0019 0.0010 Composite 0.023 0,0120 0.0091 Feed for the fourth strip at 100C analyzed 0.0010 and 0.0004 ppm VCM, so a fifth strip was not re quired at this temperature. 3. Sixteen-ounce PVC bottles with an initial concentration of about 0.0014 ppm VCM were filled with 50 percent aqueous ethanol and stored at 120F for six weeks and 21.5 weeks prior to analysis of the bottle walls. Bottles from the same batch were filled with vegetable oil and stored at 120F for six weeks. Analysis of six bottles from the same batch as was used for storage showed 0.0011, 0.0013, 0.0012, 0.0019, O.OOU, and 0.0013 ppm VCM at the time of filling. Analysis of bottles from this batch six weeks later showed 0,0014 ppm VCM and 0.0018 ppm about twelve weeks later. Also, samples of the ethanol and water for the aqueous ethanol were analyzed to assure that there were no interferences in these on the FID chromatograms. Analysis of six bottles after each of the storage tests showed the following results. 50 Pereent aqueous ethanol Vegetable o/J 6 Weeks 21.3 Weeks S Weeks 0.0004 pom 0.0009 pom 0.0007 0.0012 0.0007 0.0005 0.0007 0.0001 0.0013 Average 0.00063* 0.00087 0.0030 pom 0.0015 0.0040 0.0060 0.0032 0.0009 0.0031 * of 0 IXWTS tf th* 0,0001 ppm attolvv* it j|irna(fd bCaul of apparent V CM lou during umiviti. - 2 JOURNAL OF VINYL TECHNOLOGY. MARCH 1979, VOL. 1, NO. 1 s GENC017259 A. /, Haefnerand G. A. Hughmark DATA INTERPRETATION Results from the first part of the experimental program demonstrate that VCM is reduced to a concentration of about 0,0006 ppm in PVC resin at 1Q0C and that no further reduction in VCM occurs with continued steam stripping at this tempera ture. Results from the second part of the program indicate that the non-migration VCM concentra tion is about 0.00 L pom and is independent at temperature between 50 and 100aC. The results from the third part of the program show that the VCM in the PVC bottles filled with 50 percent aqueous ethanol decreased to slightly less than one ppb after six weeks storage at 120F and then aoparently remained constant tor the next 15 weeV's. This also indicates a non-migration level at about one ppb VCM in PVC. The VCM level in the PVC bottles filled with vegetable oil show a statistically significant increase in apparent VCM in the bottles after six weeks storage at 120F. We suspect that non-VCM components were absorbed by the bottles from the oil and that these represent interference with the FID anal ysis. Interpretation of the data from the second part of the program is complicated by lack of complete data resulting from the problem m the drying pro cedure. Diffusion coefficients and activation ener gies can be estimated from the data. The stnoDing data are analyzed by assuming that the VCM re moval is controlled bv diffusion from uniform spheres to a bounaarv with zero VCM. The rela tionship between the concentration after a period of stripping and the initial concentration is given bv: _C = 6_ v C, rr1 -- m where C. = initial concentration, C = concentra tion at time r, D = diffusmty, and a = particle radius. Generally one or two terms of this equation are adequate for these experiments. The term Dia ls useful m calculating the diffusmty from PVC resin data because the particle radius is unknown. This equation is aoplicable because one week was allowed tor equilibranon ot VCM m the PVC be tween strius. Dia- 's'1' stripping :emo. 'C First smo second itnp Third strip 50 2.2S 1 tO-1 3.74 < i0-` No valid data '0 9 96 * 10~! 3,21 x 10'1 Novaiiddata IG0 1.0 L x ;0t; 4.9 L x 10" No valid data. Prior data obtained with high VCM levels m this PVC resin show correlated D/a- values of 3 x I0'J and 6.5 x 10'* s'1 at TO and 100C, respectively. The values obtained in this work are in good agreement with the prior correlated values. The VCM contents measured otter the fourth6 strip include the effects of both the third strip and the fourth strip for the 50 and T0C runs. For the 100C run the VCM content was not changed by the fourth strip. Likewise, no significant change m VCM content was noted from the fourth to the fifth strip for the 50 and 70aC runs. The results of the fourth and fifth strips at 50 and 70aC and the re sults of the third and fourth strips at 100C are taken as measures of the same residual VCM con centration. A major uncertainty exists m anv attempt to es timate diffusion coefficients and the activation energy for diffusion tor 50 and 70aC for the lowest levels of VCM. For the 100C strips it is clear that the minimum residual level is reached at the end of the third strip but this residual level may have been reached after the third strip or required some or all of the fourth strip to reach this residual level. Values of the diffusion coefficient estimated from the average of all VCM determinations and the as sumption of 1 or 2 strips are as follows: Value of D/a1, s~[ One strip Stripping temp, 'C required Two strips required 50 1.36 x 10-S '.53 x to-' TO 5 79 x :o-` 2.64 < ID'1 100 5.76 x I0"` Not required The values of D'Ct2 are shown in F:g, l. The val ues for the first strip fall very nearly on a straight line and yield a value for the activation energy of 18.2 Kcal. The values for the second strip show a break and yield 19-0 Kcal for the activation energy between 100 and 70C and 29.0 Kcal for the activa tion energy between TO and 50C. Treatment of the data from the third and fourth strips can yield a range for the activation energy between TO and 50C. If it is assumed that one strip is required for both 50 and 70C then the activation energy be tween TO and 50aC is 14.3 Kcal while if it is as sumed that one strip is required at 70aC and two are required at 50aC the activation energy is 24.2 Kcal. In both cases the activation energy for the 100-70C temperature segment is 13.1 Kcal. These values are as follows: Strip 1 -j 3 and 4 \ctivation energv, Kcai/g mole Initial VCM. ppm L00-T0C TO-oO^C 0,* 0,023 0.0108 13.2 19.0 18.1 IS.2 19.0 14.3-24.2 CONCLUSIONS FROM EXPERIMENTAL PROGRAM A non-migration level ot 0.0006 to 0.001 ppm has been demonstrated tor PV C resin that is used in bottle compound. This non-migration level ap pears to be independent of temperature m the range of 50 to 100C. 6 JOURNAL. OF VINYL TECHNOLOGY, MARCH 1979, VOL. 1, .NO 1 GENC017260 Non-Mlymtwn Level of Vinyl Chloride Monomer m PVC FUTURE WORK We plan to repeat the second and third parts of the experimental program including CCMS anal ysis oi the bottles alter vegetable oil storage to separate VCM from intefemng components. This is expected to explain the apparent increase in VCM that was shown with vegetable oil. Repeti tion of the exoenments to obtain data for better definition of diffusion coefficients and activation energies can be accomplished by elimination of the diving problem that occurred with the work reported in this paoer. An experimental program shall also be carried out to determine the replicability of the analytical method for PVC bottles. The entire bottle wall from a bottle is required for a single analvsis so replication data axe not available for what is lenown to be a large, constant VCM composition sample. Similarly, variability in VCM concentration is not known for bottles in the same batch. This shall also be determined. This work is required because of the small differences in VCM concentration that are required with these experimental programs. PVC bottles stored at 120F with 50 percent aqueous ethanol as contents show a non-migration level of about 0.0008 ppm. SUMMARY Experimental data continue to show that there is a non-migration level of about one ppb for VCM in PVC bottle resin and in PVC bottles with food simulating solvents With the possible require ment for a PVC compound at the non-migration level, we have produced a commercial quantity of compound m the one to two ppb VCM range to demonstrate capability of manufacture of this compound. =s~xs*s JOURNAL OF VINYL TECHNOLOGY. MARCH 1979. VOL. I, NO, 1 GENC017261 7 Methods, research, testing* VCSH extraction from FVG bottles For such vinyl containers, tests show that maximum VCM in contents can be heid below 50-ppb level food use, a higher level VCM com pound would be appropriate for other regulated but non-food uses. Fig. 1 shows typical regulated and non-regulated PVC bottles. Statement of problem By G. A. Daniels and . E. Proctor Part of residual VCM in PVC bottles at the tune of filling is extracted by the ecent problems with vinyl bottle contents after time in storage. Rchloride monomer (VCM) ex The VCM concentration in the con traction by alcoholic contents tents is a function of the VCM concen from polyvinyl chloride (PVQ bottlestration in the PVC bottle at the time of nas caused concern about the use of filling, the nme and temperature of this material for food packaging. This storage, the ratio of the oottle weight to paper presents the development and re contents and :he nature of the con sults of a model, which provides tne re- tents. One wouid expect that analysis lauonsmp Detween VCM in ?VC bot- of samples from ?VC bottles with dif des and the maximum concentration in ferent contents and different storage tne contents. periods wouid provide this informa tion. Unfortunately, this direct ap The results predicted from this model show that the maximum VCM proach is not applicable because with low VCM concentrations in the bottles, concentration in oottlc contents can be controlled below the ievei that is ex the concentrations in the contents are pected to oe required by the Food and so low that :ne analytical error may be Drug Administration (FDA) for food- greater than me concentrations. grade PVC containers. These low bot- Extraction of VCM from a bottle is de-content concentrations require a an unsteaov-state diffusion process. This problem can be solved analytically PVC bottle compound with as low as 1 to provide the relationship between the part per million /ppm/ VCM. We are aole to produce this compound at some initial VCM concentration in the bottle and the VCM concentration in the con increase in cost. While this low VCM tents as a function of ume. Equilibrium level compound may be required for data have been ohrained experimental ly as required py the analytically de rived modei. Experimental data for VCM in bottles and contents have been obtained at high VCM levels. Contents represent minimum analytical interfer ence to provide data wncre the analyti cal error is small m comparison with the VCM concentrauon. Agreement of the model and experimental data at these high concentrations provides confidence mat it is also applicable at The authors; Mr Daniels (rignt) and Mr P'octor are associated with tne Research ana Development Dept, of Ethyl Coro., Baton Rouge. LA. Mr, Daniels ;oined Ethyl in 1955 and is an exoert in the aopiication of tne transoort processes. Mr aroctor has 10 years exoenence in plastics processing ana is resoansiofe for technical service tor PVC Olow-moiding compounds. low VCM concentrations. The physical nature of the modei also provides con fidence m the model at these condi tions. Experimental data have also been obtained for VCM bottles and contents at low concentrations and, with consideration of analytical error, are useful as an additional confirmation Figure 1. Typical regulated non-food, regulated food and non-regulated PVC Dotries of the modei. Vinyl chloride in the PVC compound will be distributed uniformly in the compound by the extrusion operation used to process the compound before blowing a bottle. Some VCM will be lost from the compound during the blowing operation. After the bottle is formed and before filling, VCM will diffuse through the bottle wail into the surrounding and interior air. The con tents provide a sink of limited capacity for VCM; after the bottle is filled, VCM wdl diffuse into the concents as weil as into the surrounding air, but at some ume the concentration of VCM in the contents will increase to a max imum value and diffusion wd] be from the contents, chrougn the bottle wail and into the surrounding air. Ultimate ly, the VCM concentrations in tne con tents and in the bottle wail will decay to zero, but this will require a very long Ume, The vmyi-chionde extraction can be stated in mathematical terms, and an expression can be derived which re lates the concentration of VCM in the contents and bottle wail to the initial concentration of VCM in the PVC compound, the ume, the dimensions of the bottle and the partition coefficient GENCO17262 of VCM between the bottle wall and the contents. Several reasonable and quite defensible assumptions are em bodied m the mathematical statement of the problem: 1. Diffusion of VCM through PVC is the rate-determining step. 2. The VCM concentration is uni form through the bottle wall when the bottle is formed. 3. The VCM is removed from the outer surface of the bottle by the sur rounding air as fast as the VCM dif fuses to the outer surface. 4. The concentrauon of VCM in the bottle contents is completely uniform. 5. The concentrauon of VCM at the inside surface of the bottle is in equilib rium with the VCM in the contents. This equilibrium can be described by a paruuon coefficient, which is inde pendent of concentration. 6. The diffusion coefficient is inde pendent of time and space. Differential equations for VCM dif fusion in the bottle wail, accumuiauon in the liquid contents and for equilibri um at the interface between the interior wall and the liquid were then used with the appropriate boundary conditions to obtain solutions for the two cases: Case I-No VCM is lost from the bot tle in the period between bottle forma tion and filling. Case If - VCM loss occurs between formation and filling. This can include loss during blowing as well as loss dur ing storage prior to filling. Experimental procedure, program The analytical methods for VCM are an essenual pan of the experimental program. Flame-ionization-detector gas chromotography was used for bot tles and contents. The method for VCM in liquid involves the use of a long, small-diameter column to obtain good resolution of the liquid's compo nents. The penaiues for using this type of column are long elution and bakeout times between sample injections. The method for VCM m PVC is an adaptation of the prior method and in volves injection of a THF solution of the PVC sample. However, variable baseline interferences at about the l ppm level adversely affect the accu racy. Initial experiments were conducted with standard FDA-recognized foodsimulating solvents and VCM levels of 100 to 400 ppm in the bottle compound to minimize analytical error. Table 1 shows the solvents that were used. This experiment, which includes two brands of vegetable oil. began in January, 1974, to obtain roomtemperature V CM-extraction data, and the study continues. In April, 1974, 2-oz. capacity bottles made from T ppm-VCM PVC were added to the senes; 2-oz. bottles were used (1) to maximize VCM level potential in the product for detection, (2) to reduce quantities of solvent used in migration testing, and (3) to allow filling of several bottles with each solvent so any particular bottle need not be re sampled. Another senes of extraction expenments was started in April, 1974, using weak solvents (water and 8% ethanol in water) with 2-oz. and 16-oz. bottles. Erratic results were obtained as a result of VCM loss during sampling, because of the very low solubility of VCM in these solvents. Subsequent experi ments were conducted with 50% volume ethanol m water to avoid this problem. The experimental procedure that evolved from the succession of expenments is as follows: A small batch of PVC molding compound is thoroughly mixed pnor to bottle blowing. Tnplicate bottles from each batch are then analyzed at the time that the bot tles are filled with solvent. An average of these analyses is used as the initial VCM concentrauon. Immediately after each bottle is filled with solvent to its rated capacity, a foil disc is heat-sealed over the mouth of the bottle, which is then capped to protect the seal. (Ex periments indicate that commercial clo sure losses are negligible with 50% ethanol as the solvent.) The filled and sealed samples are then maintained at the appropriate temperature for the de sired time interval. Tnplicate samples are analyzed for VCM and the sample bottles are emptied and analyzed for VCM concent. Four sets of experiments were con ducted to provide data for comparison with the model. Two-ounce bottles with 50% ethanol were used. Run 4562 - The bottles were filled 78 78 days alter blowing. VCM concentra tion in the bottles at filling was 82 ppm. Data obtained over five months at 120 deg. F. Run 4564 - Bottles filled 118 days after blowing, with VCM concentrauon in bottles of 101 ppm. Data obtained over four monchs at room temperature. Run 4566 - Bottles filled one day after blowing with VCM concentrauon Table 1. Experimental data for 72 deg. F storage Solvent Water Water 8%etnanol 3%etnanoi 3%etnanol 50% etnancl 50% etnanoi 50%ethanol 3% acetic acid 3% acetic acid 3% acetic acid n-Heotane n-rlepcane n-Heotane Soya oil Soya oil Soya oil Cottonseed/ soya Cottonseed/ soya Cottonseed/ soya *rtd-- nondswcudf* Bottle 16-az. oval 2-oz. cyiinoer 16-oz. oval 16-oz. oval 2-oz. cylinder 16-oz. oval 16-oz, oval 2-oz. cylinder 16-oz. oval 16-oz. oval 2-oz. cylinder 16-az. oval 16-oz. oval 2-oz. cylinder 16-oz. oval 16-oz. oval 2-oz. cylinder 16-oz. oval 16-oz. oval 2-oz. cylinder compound 3 mos. 10 <1 393 10 <1 393 10 <1 393 10 <1 393 10 <1 393 10 <1 393 10 <1 0.03 nd 2.90 0.08 nd 4.90 0.11 nd 2.40 0.06 nd 3.90 0.07 nd 2.70 0 065 nd 2.10 .03 nd PPM VCM in contents 8 mos. 9 mos. 0.09 nd 3.CO 0.05 nd 4,70 0.16 -- 1.90 0.06 nd 4.90 -- nd 4,ao 0.115 nd 4.25 .095 nd 0.07 nd 3.20 nd nd 6.02 0.20 nd 3.20 0.06 nd 4 4Q 0.19 nd 6.45 0.15 nd 6.60 .107 nd 12 mos. 0.11 5.28 0.15 3.18 0.20 4.33 0.10 7 53 0.26 7.53 0.26 a.04 0.27 GENCO17263 nt* *rtxw >iuKi '(ifl *t >T* Figure 2. Comparison Qt experimental data from Runs 4565 and 4567 witn model. 'ao -50* 3* Ftgurs 3. Water <31 31 0 n tfflll *u^*_ *T -t* Ia SO QO 0O<-- )a ia ;ai Figure 5. Cooking oil ;j 34 j ta tl if T' u> oa Figure 7. AllowaDle Dottle v/eignt not to exceed 50 ppo VCM m 50% ethanol for oattles made of PVC compound containing i ppmVCM 1BT*14 immv 'VujO AMdl GENCO17264 in bottles of 302 ppm. Data obtained over three months at 120 deg. F. Run 4567 - Bottles filled one day after blowing with VCM concentration in bottles of 304 ppm. Data obtained in same period at room temperature. Data analysis The four sets of data provide a range of variables to determine the diffusivity of VCM in the bottle compound and to check the consistency of the models and the data. The 120-deg.-F condition provides an accelerated test as one week at 120 deg. F. is equivalent to about eight weeks at 72 deg. The difference in time between blowing and filling provides a basis for comparing Cases I and II. Comparison of the data from Runs 4566 and 4567 shows that the rate of increase of VCM concentration in the contents is incon sistent with the Case I model, but is consistent with the Case II model. VCM loss prior to filling was also dem onstrated with another experiment, which showed that about 15% of the VCM is lost during the bottle blowing. Therefore, the Case II model was used to analyze the data. A regression analysis was used to ob tain the parameter D/ l2 and the initial VCM concentration from the data for Runs 4566 and 4567. In the parameter D/ 1J, D is the diffusivity of VCM in the bottle wall and 1 is the bottle-wall thickness. The effective bottle-wall thickness represents a combination of the thicknesses for the bottom, side wail and neck. An excellent fit of the data is ob tained to the curve represenung the Case II model. Fig. 2 shows a com parison of the data from Runs 4566 and 4567 with the two models. The time scale for the 12G-deg.-F-data has been calculated to correspond to the equivalent time at 72 deg. F. A reasona ble comparison of the experimental data for Runs 4566 and 4567 can be made with the no-VCM-loss curve (Case 1) if half of the VCM lost during blowing is added to the observed values (the other half of the VCM dif fuses to the exterior surface and does not appear in the contents) and if the time corresponding to the blowing operation is added to the actual ume after rilling. A similar regression analysis was made wuh the no-VCM-!oss curve 4564. The delay between blowing and filling was included. For Run 4564, the 118 days of storage at room tem perature plus about 12 days, which cor responds to the VCM loss during blow ing. was used as the delay time. For Run 4562, the delay time at 120 deg. F was estimated by divtding the sum of 78 days of room-temperature storage plus 12 days for blowing losses by the ratio of the 120-deg.-F diffusivity to the room temperature (diffusivity deter mined from the data for Runs 4566 and 4567). Reasonable agreement was ob tained between the experimental data and values predicted from the Case II model. Conclusions Experimental extraction data, both the VCM concentration in the 50% ethanol contents and die VCM concen tration in the beetle wall, determined when using 2-oz. bottles, are consistent with the diffusion-controlled extraction model. This extraction model can be used to predict the effect of different bottle sizes and bottle contents on the VCM level in the bottle contents. Ex amples are shown by Figs. 3 to 6 for four different boctle contents and three bottle sizes. The calculated data pre- seated by the figures are based upon an initial value of 1 ppm VCM in the PVC compound and are obtained with the Case I model. As this does not allow for the VCM loss during blowing and prior to filling, actual VCM concentra tions would be less than is shown by these figures. Thus, these figures rep resent the maximum VCM that could be expected at these condiuons. The maximum VCM concentration shown by the figures is always less than 50 parts per billion (ppb) and would ac tually be even lower because of VCM loss prior to filling. This maximum VCM concentration is estimated to oc cur after 2 to 5 years of storage at room temperature, depending upon the con tents and the bottle size. Analysis of VCM m weak solvents always shows substantially less VCM than is predict ed by Figs. 3 and 4 because of losses with closures and sampling. It is apparent that VCM concentra tion in the bottle contents is a function of the ratio of the bottle weight to the volume of the contents. Thus a bottle weight can be calculated for even very small sizes in which a 50-ppb limit will not be exceeded. A 50-ppb maximum VCM concentration (no loss prior to filling) corresponds to about 3.65 g PVC per fluid ounce for 50% ethanol. For average bottle weights, this cor responds to about 5.5 ri. oz. If 15% of the VCM is lost during blowing ftypical loss), a maximum level of 50 ppb can be reached at a size of 3.2 fl. oz. for the average bottle. It is apparent that decreased bottle weight for a specific volume will result in a lower maximum VCM concentration in the bottle con tents. Fig. 7 shows the bottle weight that corresponds to a maximum VCM concentration of 50 ppb in 50% ethanol contents for 1 ppm in the bottle com pound at filling. Reprinted from the April 1975 Issue of MODERN PACKAGING, Copyrighted 1975 by Morgan-Grampian, Inc. All rights reserved, 205 East 42 St., New York, N. Y . 10017 GENC017265 Prediction of Vinyl Chloride Monomer Migration from Rigid PVC Pipe A. R. BERENS Corporate Research The B. F Goodrich Company Research and Development Center Brecksvdle, Ohio 44141 and C. A. DANIELS B F. Coodnch Chemical Company Aeon Lake Technical Center Ation Lake, Ohio 44012 Data on the solubility and diffusion of vinyl chloride monomer (VCM) in PVC resin powders have been combined with published solutions of Fields diffusion equation to vield predictions of the amount and rate of loss of residual VCM (RVCM) from rigid PVC pipe under storage and senoce condi tions, The principal factors controlling VCM migration are the initial V'CM content, thickness of the PVC section, tempera ture, and the age of the PVC product Analytic solutions are presented for RVCM loss from freshly extruded pipe ("uniform VCM concentration) into cither the storage env ironment or the pipe contents. From these solutions, estimates are made tor the real-world situation of closed-system service following vanable storage periods. The validity of this approach for rigid PVC pipe in water-service is supported by reasonable agree ment between its predictions and experimental laboratory data on the VCM content of water stored in PVC pipes. jBothlh3i /-^rredictive model and experimental datS'Indfcafe tffiFPVC_pipeu b containing si rrig'kg_(L_part_per million) residua] VCM *'iVf C^rgrtri^jHrGM exmcernraTionsTn water of less than 0.002 m g- '/j under anv expectedseryice_conditioos. *- - INTRODUCTION i subiect of continuing concern since the discovery of x i.the ootential toxic hazard of vinyl chloride monomer VCM has been the migration of residual VCM from finished PVC products into the environment or into liquids transported in PVC vessels. Many efforts have been made to investigate this problem bv direct analysis for VCM in media ;n contact with PVC vessels. Eariv efforts ;n 1973 were questionable tiecause ot the impre cise analytical methods then available, Vs analytical pro cedures have been improved, the residual VCM levels in commercial PVC products have been sharply re duced. so the direct analysis for VCM migrating from todav > PVC products remains a verv difficult problem. Recent data L show that sensitivitv in the thousandths ot a milligram per kilogram range is needed to anaivze lor VCM in water contained m PVC pipe even at the residual VCM level of 20 mg per kilogram, A reliable model for predicting the amount and rate ot VCM migra tion from available basic transport data and theory thus would be very useful. Our approach to the development of a predictive mode! has been to combine the solubility and diffusion data we have obtained for VCM in uncompounded PVC resin powders (2, 3) with the solutions of the Fickcan diffusion equations given by Crank (4) This report summarizes the assumptions and approximations in volved in applying these diffusion equations to the VCM migration problem and illustrates the possible calcula tions with some numerical examples. The predictions ol our mode! are compared with experimental data on V'CM contents of water stored in PVC pipe , CALCULATIONS OF VCM MIGRATION Background and Assumptions On the basic assumption that migration ol V'CM thru PVC is controlled by the diffusion of V'CM m the l'\ ( ' phase, this process mav be treated bv well-snown theorv The basis of classical diffusion theory is the simple differential equation known as Fick s First law, ax winch states that the amount of diffusing subst.uu. e miss ing a unit plane area in unit time. F, is proportional to Revrinted from polymer engineering and Science, col. 16. no. 8 August 1976 GENCO17266 prediction of Vinyl Chloride Monomer Migration from Rigid PVC Pipe the concentration gradient across the plane, ----. The proportionality constant, D, is called the diffusion coefficient. The minus sign indicates that diffusion al ways occurs toward the region oflower concentration, i e' "downhill" Net diffusive transport ceases when the concentration gradient becomes zero, i.e.-, when the concentration becomes uniform. The use of Fick's law to calculate useful quantities, such as the rate at which a diffusing substance escapes from 3 solid object, or the concentration profile within the object, involves some very complicated mathemat ics. Exact mathematical solutions are generally possible onlv for geometrically simple shapes and for certain specified initial and boundary conditions. Many ot the useful solutions are presented in Crank's text (4), and fortunately it seems that some of the most important problems in VCM migration from PVC can be handled with a few of these equations. Our application of these equations and our diffusion data to VCM migration from PVC products involves several assumptions: (1) The diffusion of VCM in PVC obeys Fick's law. (2) The diffusion coefficient is independent of VCM concentration. (3) The value of the diffusion coefficient in rigid PV'C products is the same as we have determined tor pure PVC resins. Mi The diffusion coefficient is independent of the medium surrounding the PVC; i.e., values we have determined bv vapor sorption/desorption also apply to migration into a liquid phase. Assumptions VT and (2) have been demonstrated to be ierv satisfactory approximations at quite low RVCM concentrations by our work on PVC powders (3). More limited experiments on thin, rigid PVC films also sup port assumption 13), although some variation of D might be expected for varving amounts and types of com pounding additives. The use ofassumption (4) should be considered a tentatively useful approximation; we shall see that it does seem justified by experimental dat3 on VCM migration into water from PVC pipe. To describe the diffusion of RVCM from PVC prod ucts through Cranx's equations, three situations have been considered. Case I RVCM loss during storage of a freshivinumifacrured PVC product. Case II RVCM loss from freshly formed P"VC products into a closed medium. Case III. RVCM loss into a closed medium from previously aged PVC products. In both Cases I and 11, the initial RVCM concentra tion is assumed to be uniform through the thickness of the PVC product, this is probably a valid assumption only at the time of extrusion, as the surface concentra tion in RVCM will quickly decrease upon exposure to a low-VCM environment. Cases 1 and 11 differ m the time-dependence of the surface concentration: In Case 1 the surface concentration of VCM remains essentially zero, as am RVCM escaping is carried away in the cm lrnnment this case may represent storage or service in a continuously renewed environment, such as flowing water. In Case II, the VCM concentration in the t medium builds up with time, and consequently so does the VCM concentration in the surface of the PVC. Case III is the general situation in a real-world application: VCM loss into a closed medium follows a variable stor age period and thus proceeds from a product in which the surface VCM concentration is already depleted. We will see that the aging period between manufacture and closed-svstem service is quite important in determining the rate of VCM migration into the contents of a PVC pipe. Now let us consider the details and some numerical examples of each of these three cases. Case I--VCM Loss During Storage Consider a freshly extruded PVC product, quickly cooled to ambient temperature and stored m an atmo sphere of essentially zero VCM content. The initial RVCM concentration in the product is Ca and may be assumed to be uniform through the product. At the surface, equilibrium is quickly established with the environment and the RVCM concentration is zero We assume that VCM leaving the PVC is earned away je.g , good air circulation/ so that the surface concentration remains zero. We want to calculate (a) the amount of VC.V1 which leaves the PVC and (b) the concentration profile within the PVC, both as functions of time, tem perature, and sample thickness. Crank gives solutions to this problem for several sim ple geometries~plane sheets, solid and hollow cylin ders and solid spheres. The predictions for hollow cylin ders are virtually identical to those for plane sheets, provided the wall thickness is less than the inside diame ter Thus for all practical PVC products (pipes, bottles, sheets, films), we need consider only the mathematical solutions for plane sheets. Equations for the amount of VCM escaping from the sheet may be written in terms of M, the fraction of the original "VCM which escapes m time t The general expression, valid at all tunes, is *o M = 1 - Y yr--------py -0 ;ii ' 11- rr - n-T- i --) where n is the senes of integers .0, 1, 2--. D the diffusion coefficient, and L the sheet thickness. For the late stages oi the process M > -- 0 6i, terms bevon'd n = 0 become insignificant, and Eq 2 becomes M=1- (3) For M < -- 0 6, a verv good approximation is given bv Thus the initial lossofVC.M is proportional to the square root of the storage time .liter extrusion. For numerical calculations, we need oniv the sheet thickness and the diffusion coefficient values. From our measurements on PVC resins (3), the value ol D at various temperatures is given bv GENC017267 A. R. Berens and C. A. Daniels -17,000 D 3.7 exp RT (5) for D in crn'isec, T in and R = 1.987 cal/moltlK. Equations 3, 4 and 5 permit prediction of the frac tional loss of RVCM from rigid PVC products for differ ent sheet thicknesses, times and storage temperatures. Examples of numerical results are given in Figs. 1 and 2 as plots of .W vs tf Figure 1 shows such RVCM-loss curves for several sheet thicknesses at 30C, where D = 2 x 10-IIcm7sec = 1.73 x 10~7 cnr/dav, and Fig. 2, for a I mm sheet thickness at several temperatures. The concentration of RVCM remaining at time t at various distances from the sheet surface (i.e., the con centration profiles) can also be calculated from the same parameters. Crank gives the general solution as Cc* - C'-a _ xp - 1' ! - x 2 (Of)* -T -- 1W - X 1 ;Df)* . l6) where C is the concentration at time t at distance r from the center of the sheet, C0 is the initial (uniform) con centration, C, is the constant concentration at the sur face .zero in our case), and l is the half-thickness of the sheet, "erfc" stands for error function complement de fined as erfc c = l" erf ; (7) The error function, "erf', also called the probability ti 1/2 , d. ays 1/2 F*,n l fraction of untutl VCM content tost i \f) lt, square ruot of t one1 t'f fur PVC sheets of i arums thteknesse t calculated for 3(J*C (D = LT3 * JQ~: cm-.'daq/. Case / Fig 2. Fraction of initial V'C.Vi content lost fM) es square runt of time (f *1 for PVC sheet 1 mm thick, calculated for various tem peratures, Case l. integral is tabulated in standard mathematical tables For times siiort enough that the concentration at the center of the sheet does not decrease significantlv below C0, only the first term of Eq 6 is necessary-, then, using Eq 7, we have simply l -7 1 2{Dtji . ,fi) Using Eq 8. or Eq 6 where necessary, we have calcu lated concentration profiles at 30C ID = 1.73 x J0-r enr/'day) for a PVC sheet l/s in. thick (or pipe with `-'x m. wail). The results at various times are plotted in Fig. 3. Note that the VCM lost in the first month comes only from the 100 microns of PVC near the surface. It takes over 20 years in this case for the V'C.Vi concentration near the center of the sheet to decrease appreciable CASE II--VCM MIC RATIO N FROM FRESHLY/FORMED PVC PIPE INTO CONTENTS The Case 1 calculations may be applied whenes rt VCM leaving the PVC product is earned awav by the environment storage in circulating air, water-pipe ser vice with How mg water, etc. , so that the snrlace con centration of RV CM remiuns essentiallv zero For P\ t ) pipes in ordinary serwice, the outer surface is generaib exposed to a low-VCM environment, i.e., a Case I situa tion. On the inside of pipes with stagnant contents, on the other hand, VCM leaving the PVC builds up in concen tration m the contents. Consequently, the inside surface concentration of RVCM in the PVC, assumed to remain m equilibrium with the contents, also increases with rime. At very long times, RVCM in the inner half of the GENCO17268 Prediction of'Vinyl Chloride Monomer Migration from Rigid PVC Pipe Fte 3 Relative VCM concentration (C/C J vs depth below sheet iunace (l - X) at various times, calculated for 'A m. thick FVC sheet at 30C <D = 1.73 x 10~' cm~!sec). Case 1 PVC wall and in the contents will diffuse outward to the environment. This net outward diffusion of VCM will onlv occur after the VCM concentration at the midltne of the wall falls below its initial value; the time at which this occurs can be estimated from concentration profiles such as those in Fig 3 Since this effect occurs at such long rimes for pipes of normal wall thickness, we have not considered it in our model, but instead have applied Case I and Case II calculations independently to the outer and inner halves of rise PVC walls, respectively The rate and amount of VCM entering the pipe con tents may be calculated through equations given by Crunk. The maximum amount of VCM which will enter the contents is that required to establish equilibrium between PVC and contents, and is governed by the partition coefficient and the ratio of volumes of PVC and contents. The partition coefficient, K, is defined as the ratio oi RVCM concentration m the PVC to that in the contents at equilibrium ,'both expressed in the same 'amts. e. z . Z'lirer:. The volume oi PVC supplving VCM to the container contents, V,,,-c. is one-half the total PVC volume, as RVCM :n the outer half diffuses outward in the situation we are considering. It can be shown that the maximum VCM concentration, C,,,^aT, tppm bv weight in the liquid contents of a PVC pipe originallv containing C0 ppm VCM, is cAp K - V pvc where dPVr and d,,, ire densities of the PVC and con tents. andV,,, is the volume of the contents. From uur data (2) on VCM solubiiitv as a function of VCM pressure over PVC and water, we have estimated a '.ijiie oi K = 49 lor the distribution coefficient ofVCM between PVC and water at 30C. It should be noted that this estimate ol'V assumes that the solubility ofVCM in PVC is not .affected bs contact with water. It also in volves a somewhat arbitrary selection of a value for V( l.M solubility in PVC, since we have shown (2) that this system show's non-ideal and historv-dependent solubil ity, Using this value of K, Eq 9 predicts that the maximum VCM concentration in water in a 1 in. l.D , '/ in. wall, PVC pipe containing L mg/kg residual VCM will be 0.027 mg/kg. The rate ofVCM desorption into the pipe.contents may be obtained from another of Crank's equations; M =a[i - eJsr erfc\Tla.-)i} (10) where T = Dt/l', M is the fraction of the original RVCM desorbed at time f, and at =VnrfVr,t.c, To illus trate, we have applied Eq 10 to the 1 in I. D , Vs in. wall PVC pipe filled with standing water. Figure 4 shows the results, with scales showing both M. the fraction of original RVCM desorbed, and the VCM concentration in the water per original ppm RVCM. Also shown m Fig 4 is the M vs f i plot for Case I Note that the initial rate of VCM desorption is the same for both Cases I and II. but the buildup ofVCM m the water m Case II causes the desorption to slow down as equilibrium is approached. Case LU--VCM Migration in Closed-System Service from Previously Aged PVC Products When a PVC product is put into closed-system service some time after manufacture, the RVCM distribution through the PVC at the time of filling and dosing the container will not be uniform, as was assumed in the Case II calculations. Rather, RVCM will already Indepleted near the surface, and VCM migration into the pipe contents will start from a V'CM distribution .is calculated in Case I ie.g,. Figure 3), An analytical solu tion for this situation has been obtained by Daniels and Proctor ;5, but a useful and simpler estimate of tin- i.ito days 1/1 Fig 4 Fraction of original VCM Content lost < M) vs squmr nu>! of timeit*)for 1 m. I D., ^ in. wall FVC pipe, calculated f>t ,U) V (D = 1.73 x 10~T cm'lday/, Case I and Case II with water m ptju Right-hand scale gives jipb VCM m water per ppm initial \ ( in pipe GENCO17269 A. R, Berens and C. A. Daniels of VCM migration can be made by combining results of our Case 1 and Case II equations. For Case I during desorption of the first 60% ot the VCM. Eq 4 shows that the amount of VCM desorbed is proportional to the square root of storage time. Differentiating Eq 4 gives dM dt Thus the rate of VCM loss is inversely proportional to the square root of storage time. For Case 11. we saw that the initial rate of VCM migration into the medium in a closed system is the same as :n Case I. Thus Eq 11 also gives the initial rate of VCM migration into the closed svstem when t is the storage age of the PVC product at the start of closed-system service. Applying Eq 11 to a PVC product with `4 in. wall thickness at oO^C (D = 1.73 x 10_r cmJ/dav) gives the curve shown m Fig 5. We see that the rate of RVCM desorption drops very sharplv in the first few weeks oi storage after manufacture. It is also possible to estimate the amount of VCM which will migrate from a PVC product during a given period of a closed-system service following various Case I storage periods. This estimation mav be explained with reference to Fig 6, which illustrates the VCM-loss fM) vs f1 curv es ior the three cases. Case III is approximated bx shifting the origin of the Case II curve to point along the Case I line, where t, is the age ot the PVC product at the start of closed-system ser-ice. Then the VCM lost from the PVC in the time interval a during continued open-system storage would be, from Eq 4 EM = 4^ ---j* [if, - a - t,] .,12) The VCM lost, and entering the contents of a PVC container, in Case III service, will be approximately equal to EM for relatively short service periods, and alwavs less than EM. Equation 12 thus is useful for calc ul.inns the maximum fraction of the original RVCM which will migrate into the contents of a PVC container during eiosed-svstem service for anv time period as a junction of the age ot the container at the time of filling and closing F-gure 7 illustrates results calculated trom c 12 for 7 and 30 dav service periods ior a in. wall ttuckness at }0CC as ,'uncnons oi f. \gain we see the .mDortunt effect ot a lew weeks prior aging in reducing the amount of VCM migration into the contents. Comparison of Predictions with Experimental Data The foregoing anaivsis cleariv shows that the age of a P\ C oroduct at the start of an extraction test is an important factor m determining the amount ot VCM extracted. Since this information is seldom available in reported extraction data, direct comparisons between our predictions and experimental data are possible for oniv a few cases. For the data recently obtained by O'Mara and DeCupita C on the VCM content of water stored m 1 in. I D bi in. wall PVC pipes at 2oC, the approximate age of the pipe samples, between extrusion and start of the t. days Fig 5. Race of VCM ioss (dMldt) oj time far Hi in, thick PVC sheet, calculated Jor30C, Case 1. Fig S. Schematic companion offractional VCM lass 'M'i as M curves for Cases !, II, and III. extraction test, was known. A "headspace GC auah ticaJ method was used to provide sensitivity to low VCM levels m the water ^on the order of 1 or 2 thousandths of a milligram per kilogram). \V> have calculated the VCM content expected in the water, using Eq 12 with the parameters appropriate to the experimental conditions, D was obtained from Eq 5. Table 1 compares the calcu lated and experimental results. The agreement must l>e considered quite satisfactory, especially in view ol a) our application of D values obtained from resin powders to pipe compounds, b) the somewhat uncertain age and GENCO17270 prediction of Vinyl Chloride Monomer Migration from Rigid PVC Pipe from the outer half of the wall and into the contents from the inner half may be treated as independent processes over normal service lifetimes. Our simple predictive model mav thus be used with some confidence for estimating the concentrations of VCM in water which might arise during actual service of PVC water pipe systems. To illustrate, Eq 12 has been used to calculate the VCM concentrations resulting from various stagnant exposure times ot water in PVC pipes oi 1 mg'kg original VCM content and varied diameter, warehouse age before installation and service age. Some results of such calculations are given in Table 2 for 2, 6 and 8 in. SDR21 pipes in service at room temperature (-~23C), While these data are presented as though they represent stagnant water situations, from known use conditions (flow rates, pipe dimensions and resultant residence time), these calculations can be shown to model dynamic flowing systems. The figures in Table 2 clearly show that the highest VCM concentrations would be found in new installa Fia 7 Fraction of original VCM content entering pipe contents (Em) for 7 and 30-day extraction periods os pipe age at start of extraction if,;, calculated for Vs in. wall thickness and 30"C, tions of recently manufactured small diameter PVC pipe after long stagnation periods Yet even for these most extreme conditions, the predicted VCM-m-watci con Case III centrations are well below the level of 0 002 mg Y CM/kg storage conditions of the pipe samples, and c' the difficultv of analysis of water for extremely low levels oi VCM. H-,0 when the original residual VCM content of the pipe is 1 mg/kg or less. In actual installations, stagnation times of more than a few days are rarely encountered. A A meal residence time for water in PVC pipes is believed DISCUSSION to oe about 2 days ;6-8' and in this situation. fhe pred.cted VCM-in-water concentration falls in the parts- While further experimental verification would be de pr --tnlhon range. Thus we may conclude that PVC pipe sirable, it appears that the simple approach discussed c-mtainmg <1 ms^kg residual \ CM will result in VCM above is ijmte adequate for describing and predicting concentrations of less than 0 002 mg VC.M/kg H.O die migration of RVCM from PVC pipe into water The under anv expected service conditions, and, therefore, reasonable agreement between predicted .md observed non-dctectabie by present analytical methods. migration data for this application of rigid PVC supports the premises that, a) the diffusion coefficient deter mined for pure PVC resins is applicable to rigid PVC pipe compounds, and b) contact ot PVC with water produces little change in the diffusivitv of VCM com pared to that measured bv vapor sorption/desorption Table 2. Calculated VCM in H.O Concentrations (or Stagnant Storage of Water in PVC Pipes ot Varied Size and Age and Original 1 mg/kg Residual VCM Content techniques Further, for such relatively thick walled products as pices, the diffusion into the environment Pipe size (E0R21) W arehouse age. days Service age, year? VCM in water after qiven storaqe times, mq/kq 2 days 2 weeks 1 month Table i. Comparison of Predicted and Eiperimentai Extraction Results. Water-Pilled 1 in. I.D. in. Wall PVC Pipe 2 m. 30 0( new) 00007 00044 00087 Samples. 23"C 50 0 90 n 00005 00004 00C33 0002 7 OOCpT 00056 RVCM Extraction 90 1 0000130 000125 0CO265 in pipe, Pipe age, Time, VCM in water, (mg/kg) 90 2 00001 00009 000199 s mg/kg t a.days Experimental Calculated 90 5 00001 00006 00013 292 -6 rno* 292 --6 mo. 292 -o mo. 177 --5 mo. 177 -6 mo. 177 -5 mo. 22 ' 5 mo 22 --5 mo. 22 -6 mo. 29 -1 yr. 3 7 14 3 7 14 3 7 14 14 0,021 0.0414 0.113 0.0173 0.0335 0.056 0 0006 0.0022 0.0046 0.0105 0 0257 0 0598 0.118 0,0156 0 0362 0.0717 0,0019 0.0045 0.0089 0.0084 6 m. 3 in. 30 0 60 0 90 0 90 1 90 2 90 5 30 0 so 0 90 0 90 1 90 2 90 5 00002 00002 00001 000006 000004 000003 00015 00011 00009 000042 00003 C0C02 00002 00001 00001 0000045 000003 000002 00011 00008 00007 000031 00002 00001 00029 00022 00019 QC0GS8 000066 00004 00022 00016 00014 0000(36 000050 00003 --i GENC017271 A. R. Berens and C. A. Daniels REFERENCES 1 M. M. O'Mara and E. C. DeCapita. Internal Report, B. F. Goodrich Chemical Company, 6/4/75. 2. A. R. Berens, Am. Chum. Soc. Polym. Prepr., 13, 197 (1974); Angeui. Makromol. Chem., 47, 97 (1973). 3. A. R. Berens, Am. Chem. Soc., Polym. Prepr. 15, 203(1974). 4. J. Crank, "The Mathematics of Diffusion," Oxford University Press, London (1956). 5. G. A. Daniels and D E. Proctor, Modem Packarine n 45 (April 1975). b^ 6. American Water Works Association. C 601-54, Sec. 14 Reten tion Period. 7 "Design Parameters for Rural Water Distribution Systems," J Amer. Water Works Assoc., p. 1595 (December 1966), 8. "A Study of Residential Water Use," U.S.D.H.U.D. (Fel>ruary, 1967). GENCO17272