evaluating the potential for recycling all pet bottles into new food packaging

This article was downloaded by: [University of Windsor]On: 02 July 2014, At: 21:24Publisher: Taylor & FrancisInforma Ltd Registered in England and Wales Registered Number: 1072954 Registered office:Mortimer House, 37-41 Mortimer Street, London W1T 3JH, UK

Food Additives & ContaminantsPublication details, including instructions for authors and subscriptioninformation:http://www.tandfonline.com/loi/tfac19

Evaluating the potential for recycling all PETbottles into new food packagingT. H. Begley , T. P. McNeal , J. E. Biles & K. E. PaquettePublished online: 06 Dec 2010.

To cite this article: T. H. Begley , T. P. McNeal , J. E. Biles & K. E. Paquette (2002) Evaluating the potential forrecycling all PET bottles into new food packaging, Food Additives & Contaminants, 19:S1, 135-143

To link to this article: http://dx.doi.org/10.1080/02652030110083720

PLEASE SCROLL DOWN FOR ARTICLE

Taylor & Francis makes every effort to ensure the accuracy of all the information (the “Content”)contained in the publications on our platform. However, Taylor & Francis, our agents, and our licensorsmake no representations or warranties whatsoever as to the accuracy, completeness, or suitabilityfor any purpose of the Content. Any opinions and views expressed in this publication are the opinionsand views of the authors, and are not the views of or endorsed by Taylor & Francis. The accuracy ofthe Content should not be relied upon and should be independently verified with primary sourcesof information. Taylor and Francis shall not be liable for any losses, actions, claims, proceedings,demands, costs, expenses, damages, and other liabilities whatsoever or howsoever caused arisingdirectly or indirectly in connection with, in relation to or arising out of the use of the Content.

This article may be used for research, teaching, and private study purposes. Any substantialor systematic reproduction, redistribution, reselling, loan, sub-licensing, systematic supply, ordistribution in any form to anyone is expressly forbidden. Terms & Conditions of access and use canbe found at http://www.tandfonline.com/page/terms-and-conditions

http://www.tandfonline.com/loi/tfac19

http://dx.doi.org/10.1080/02652030110083720

http://www.tandfonline.com/page/terms-and-conditions

Evaluating the potential for recycling all PET bottles intonew food packaging

T. H. Begley*, T. P. McNeal, J. E. Bilesand K. E. PaquetteFood and Drug Administration, Washington, DC 20204, USA

(Received 5 December 2000; revised 3 July 2001; accepted 19July 2001)

To evaluate the feasibility of recycling all PET bottlesinto food packaging, realistic estimates of the maxi-mum concentration of contaminants that might beexpected in the polymer are needed. To estimate themaximum concentration of a contaminant that might bein PET from the storage of non-food substances,sorption experiments into two types of PET wereperformed. These test materials were 0.8 mm thickamorphous PET (a relative sink for contaminants)and commercial PET bottle wall. Using a commercialshampoo containing 1% lindane (C6H6Cl6), the testmaterials were stored in contact with the shampoo at 20and 40°C for 231 days. This commercial shampoo alsorepresents an extreme case because it contains 7%acetone, a solvent which swells PET, further enhancingsorption of chemicals. Additional sorption experimentsinto PET were performed by preparing solutions of10% toluene in Miglyol (a fractionated coconut oil),10% benzophenone in Miglyol, 5% 2-butoxyethoxyethanol (2-BE) in 50/50 water/ethanol, and 10%methyl stearate in heptane. Sorption data from theshampoo into PET illustrate Fickian behaviour.Speci®cally, the amount of sorption at room tempera-ture is 40 times less than that at 40°C. The amount oflindane sorbed into PET from the shampoo after 231days was 0.1 and 3.7 mg dm 2 at 20 and 40°C respect-ively. These values correspond to 28 and 765 mg kg 1

on a mass/mass basis. All sorptions are within theranges measured and published by other authors usingsurrogate contamination testing schemes. Additionally,actual bottles from recycle bins were analysed for theamount of contamination. Results are discussed in

terms of potential consumer exposure to non-foodcontaminants in food containers made of recycledPET and in relation to the surrogate testing methodsrecommended by the Food and Drug Administration(FDA) for determining the compatibility of a PETrecycling process to produce containers suitable forfood-contact use.

Keywords : recycled polymers, food packaging, sorp-tion

Introduction

In 1999, 309 106 kg PET bottles were recycled in theUSA (Modern Plastics 1999). Of this (309 106 kgbottles), 79% was from soft drinks, and the other20% (of the bottles) was of custom food bottles andbottles for non-food use. Therefore, at most in arandom distribution, one in ®ve bottles may be anon-food bottle. For the purposes of increasing thee� ciency of recycling PET bottles, it would be ben-e®cial not to have to sort bottles into food and non-food categories. Removing this restriction wouldpermit automated bottle sorting on a polymer-typebasis only. The result is commingled PET bottles thatmay have been in contact with food, although all PETbottles produced in the USA were initially producedfrom food grade resins. Given this, the question thenbecomes: does PET have su� cient barrier propertiesto avoid signi®cant contamination from non-foodcontact bottles and to allow a commingled feedstockto be used to make new food packages?

Demertzis et al. (1997) discussed the behaviour ofkinetic sorption of 20 diŒerent chemicals into PET at10, 20 and 40°C for up to 40 days. The chemicals usedin the study ranged in polarity from non-polar topolar or solubility parameters (cal cm 3)1=2 fell be-tween ¯ ˆ 7:4 and ¯ ˆ 14:6. The solubility parameterfor PET was about ¯ ˆ 12. Demertzis et al. describedan extreme in the sorption of chemicals into PETbecause the polymer was placed in solutions made by

Food Additives and Contaminants , 2002, Vol. 19, Supplement, 135±143

* To whom correspondence should be addressed. e-mail: [email protected]

Food Additives and Contaminants ISSN 0265±203X print/ISSN 1464±5122 online # U.S. Government Copyright 2002http://www.tandf.co.uk/journals

DOI: 10.1080/0265203011008372 0

Dow

nloa

ded

by [

Uni

vers

ity o

f W

inds

or]

at 2

1:24

02

July

201

4

combining six pure or undiluted liquids. Generally,Demertzis et al. showed that smaller, generally vola-tile chemicals are more concentrated in PET than arelarger, non-volatile chemicals, which are found inrelatively lower concentrations. For example, 1.4and 5.1 mg dm 2 toluene, a smaller molecule, weresorbed into PET after 30 days at 20 and 40°C,respectively. For a relatively non-volatile chemicallike phenyldecane, 0.29 and 0.62 mg dm 2 weresorbed into PET after 30 days at 20 and 40°C,respectively.

Processing studies on the eŒects of washing anddrying recycled PET, intentionally contaminated withvolatile and non-volatile chemicals, show that non-volatile chemicals are not removed as e� ciently asvolatile chemicals, but they generally do not diŒuseinto PET as fast either. Speci®cally, Komolprasertand Lawson (1995) showed that of the original con-centrations, 2% of the volatile and 21% of non-volatile chemicals remained after washing and dryingPET that was intentionally contaminated PET.However, based on its reviews of numerous PETrecycling processes, the US Food and DrugAdministration (FDA) has found that the levels ofsurrogate contaminants remaining in PET after wash-ing and drying alone still exceed what would result ina dietary concentration of 0.5 ppb, FDA’s level of noregulatory concern (see the `Regulatory Implications’section below) (Federal Register 1995).

Before the work of Demertzis et al. (1997) andKomolprasert and Lawson (1995), Markarewicz andWilkes (1978) showed that the diŒusion of non-reac-tive liquids and vapours into PET obeys Fick’s law.In 1984, Patton et al. also described the sorption andtransport of benzene in PET as Fickian. BecausediŒusion in PET obeys Fick’s law, the amount ofsorption into PET should be predictable, providedgood estimates of physical constants are known.Patton et al. (1984) measured diŒusion coe� cientsfor benzene in PET at 40, 50 and 60°C and from anextrapolation of a ln D versus 1/T plot of this data, itis possible to estimate a diŒusion coe� cient at 20°C.The resulting diŒusion coe� cients for benzene in PETare D ˆ 6:7 10 14 and 3 10 15 cm2 s 1 at 40 and20°C, respectively. On a theoretical basis, the relativeamount of sorption measured for benzene should bepredictable, using these diŒusion coe� cients. Forexample, from the benzene sorption/contaminationexperiments by Komolprasert and Lawson (1995),310 mg kg 1 benzene was measured in 2 litre PETbottles after storing a 10% benzene in hexane solution

for 14 days at 40°C. These data indicate that essen-tially all the benzene remains in the solution, i.e.

0.01% of the benzene originally in the solutionentered the bottle.

With the following equation from Crank (1975),which is the solution to Fick’s law for the case of apolymer sheet in a solution of limited volume, theconcentration of benzene in a 2-litre PET bottle canbe calculated to be 500 mg kg 1 using a diŒusioncoe� cient of 6:7 10 14 cm2 s 1 from Patton et al.(1984), and assuming a solution density of0.66 g cm 3, a bottle mass of 48 g and that no parti-tioning occurs:

Mt

M1ˆ 1

1

n 1̂

2¬…1 ‡ ¬†…1 ‡ ¬ ‡ ¬2q2

n†exp

Dpq2nt

l2…1†

In equation (1), qn is the non-zero positive roots oftan qn ˆ ¬qn, and ¬ ˆ a=Kl, the ratio of the vol-umes of the solution and the polymer sheet, with Kbeing the partition coe� cient. With equation (1),reasonably good agreement between the calculatedand measured values is obtained, considering theuncertainties in the exact bottle con®guration andassuming that no partitioning occurs and that M1related to the initial concentration in the solutionsM1 ˆ 2aC0=…1 ‡ ¬†.Similar results were obtained when Demertzis et al.’smeasured sorption of toluene into PET at 20 and40°C was compared with the calculated sorption.From Sadler et al. (1996), the diŒusion coe� cientfor toluene in PET was determined to beD ˆ 4 10 14 cm2 s 1 at 40°C and 2 10 15 cm2 s 1

at 20°C. The experimental sorption from Demertziset al.’s work versus the calculated values using thesediŒusions, a typical bottle wall thickness of 0.03 cm, atotal surface area of 14.4 cm2, and a polymer mass of0.25 g, is illustrated in ®gure 1. Figure 1 illustrates (1)the sorption measured by Demertzis et al. is Fickian(linear with the square root of time), (2) the calculatedvalues are similar to the experimental values, and (3)the amount of sorption after a year at room tempera-ture is not likely to exceed the amount of sorptionfrom a 40°C, 30-day test. The calculated values againdepend upon the assumption that, ultimately, theequilibrium concentration of toluene in the polymeris the same as the concentration of toluene in solutionand that no partitioning has occurred. This is a veryconservative assumption and in many cases not likelyto happen in reality because chemicals do partition infavour of the solution, otherwise the container wouldfail. For molecules larger than toluene (non-volatile

136 T. H. Begley et al.

Dow

nloa

ded

by [

Uni

vers

ity o

f W

inds

or]

at 2

1:24

02

July

201

4

chemicals), about four times less sorption was foundby Demertzis et al. (1997). This amount of sorptionwould correspond to a diŒusion coe� cient of

2 10 16 cm2 s 1 at 20°C. This level of diŒusion isconsistent with the level of diŒusion predicted bySadler et al. (1996), in their work on evaluating themigration of organic compounds from PET.

In other studies, Nir et al. (1996) measured thesorption and migration of organic liquids into andout of PET. They used neat organic liquids in directcontact with PET, which in some cases resulted inlarge weight gains by the polymer (toluene caused an11% weight gain). This amount of weight gain pro-duces plasticized polymers, which have very diŒerentproperties than standard commercial PET bottles. Asa result, these experiments produced relatively largediŒusion coe� cients of 3:7 10 9 cm2 s 1 for tolueneat 34°C in amorphous PET. If diŒusion coe� cientswere actually this large in commercial PET bottles,then thousands of mg kg 1 would be observed inbottles in contact with solutions, assuming that parti-tioning of the solute between the solution and thepolymer was not a signi®cant factor. The sorption ofthousands of mg kg 1 in PET has not been the case asreported in Komolprasert and Lawson (1995) andFranz and Welle (1999) or in this work. Additionally,if diŒusion was this rapid in real bottles, the bottleswould lose their desired physical properties of highstrength and clarity during their typical shelf life.

Komolprasert and Lawson (1995) also studied thesorption of a commercial pesticide solution into 2-

litre PET bottles. A solution of 5% malathion (acommon household pesticide) in water (the recom-mended commercial preparation of this pesticide) wasprepared. This system represents a non-volatile polarcontaminant in a polar solvent in contact with a polarpolymer. After 14 days at 40°C, 599 mg kg 1 or2.6 mg dm 2 malathion was found in 2-litre PETbottles. This amount of malathion is within the samerange seen for similar chemicals found in PET byDemertzis et al. (1997).

The work by Komolprasert et al. (1996) also showswhat happens when contaminated PET bottles arereprocessed. Bottles that had been contaminated withmalathion were processed by ¯aking, washing, dryingand extruding the PET into new sheets. This pro-cessing, a typical industrial practice, produced disco-loured PET after the extrusion step. This reprocessedmaterial was unacceptable from an appearance stand-point.

Although many studies indicate that sorption intoPET follows Fick’s law and some diŒusion coe� cientdata are known, there are little data on the magnitudeof sorption from actual commercial solutions. Thereis also a lack of information on the maximum amountof contamination that occurs during long-termstorage of real commercial solutions. Surrogate test-ing studies have demonstrated that chemical concen-trations in the hundreds of mg kg 1 can be sorbedinto PET and that large fractions of these chemicalsare removed from PET by washing, drying andextruding (Komolprasert et al. 1995, 1996). The pres-ent paper addresses some of these issues and providesa relative basis for the amount of sorption into PET.

Materials and methods

Sorption experiments were performed on two types ofPET, amorphous sheet (APET) and PET cut frombottle wall ( 0.03 cm thick). The APET sheet wasmade from bottle-grade resin, had a density of1.334 g cm 3 and was 0.08 cm thick. These twomaterials should represent the extremes for theamount of sorption because of the large diŒerencesin crystallinity. It is well documented in the literaturethat diŒusion is faster in materials with lower crystal-linity. For example, Nir et al. (1996) showed thediŒerence in diŒusion to be more than six times fasterfor toluene in APET than in PET from a bottle wall.

137Evaluating the potential for recycling all PET bottles into new food packaging

Figure 1. Experimental and calculated sorption intoPET.

Dow

nloa

ded

by [

Uni

vers

ity o

f W

inds

or]

at 2

1:24

02

July

201

4

The relative diŒerence in the crystallinity of these testmaterials is indicated by the large diŒerence in the160°C exotherm illustrated in ®gure 2. The APET,having the lowest crystallinity, has the highestexotherm.

Five types of solutions were used in the sorptionstudies. A commercial lindane shampoo (containing1% lindane, C6H6Cl6) was purchased at a localpharmacy store. The shampoo was transferred di-rectly from the manufacturer’s container into 20 mlheadspace vials containing 0.75 g PET. For theshampoo test, this con®guration produced three timesmore polymer than the solute lindane. Additionalsorption experiments into PET were performed bypreparing solutions of 10% toluene in Miglyol, 10%benzophenone in Miglyol, 5% 2-butoxethoxy ethanol(2-BE) in 50/50 water/ethanol, and 10% methyl stea-rate in heptane. These additional chemicals representa solvent, a UV-active agent, a surfactant and an oilysubstance. The PET was cut into 1.1 cm 6.0 cmcoupons, weighed and sealed into headspace vialscontaining one of the ®ve solutions. The headspacevials were placed in a thermostated oven set at 40°C,the suggested minimum temperature for evaluatingcontamination of recycled polymers. After incuba-tion, vials were removed from the oven and broughtto room temperature.

The PET was wiped dry with a paper towel, thenrinsed three times with a washing solution to removeresidual surface contamination. A 50/50 isopropylalcohol/water solution was used to remove the sham-poo from the surface, and heptane was used toremove the Miglyol. The polymer coupons werewashed air-dried, at room temperature, weighed and

extracted to determine the total amount of sorption.For extraction, 0.4 g of each PET specimen was cutinto small pieces, put into headspace vials, and ex-tracted with 15 ml methylene chloride. Then the ex-traction vials were placed on a rotating wheel forconstant agitation for 72 h. The resulting methylenechloride extracts from each sorption experiment werediluted 10±20-fold and analysed directly by GC-MS.In addition, some coupons were subjected to totaldissolution by dissolving the polymer in a hexa¯uor-oisopropanol/methylene chloride solution (1 ml/5 ml)to ensure complete extraction by methylene chloridein 72 h. Quantitation was accomplished with externalcalibration. For methyl stearate, the limit of quantita-tion, LOQ, was 0.05 mg methyl stearate g 1 APET,and the limit of detection, LOD, was c. 025 mg g 1.

Analysis of actual post-consumer non-food bottlesinvolved rinsing the inside and outside of the bottleswith water, air drying the bottle at room temperature,followed by extracting 1 g PET in 15 ml methylenechloride as before.

GC/MS apparatus and operating conditions

The GC/MS system consisted of a Hewlett-Packard5890 (HP) Series II gas chromatograph with an HP7673 automated liquid sampler; a capillary split-split-less injector; a 30 m 0.25 mm i.d. HP-5MS, cross-linked 5% diphenyl-95% dimethyl polysiloxane ,fused silica open tubular capillary with a 0.25 mm ®lmthickness (Hewlett-Packard Corp., Wilmington, DE,USA): and an HP 5970B Mass Selective Detector(MSD) with a capillary direct interface to the GC.The GC operating parameters follow: UHP heliumcarrier gas at a constant ¯ow of 1.00 ml min 1;injection volume, 2 ml; split vent opened after1.5 min; temperatures (°C), injector 250, interface280, oven programme, 2 min at 35°C, 10° min 1 to275°. (tR’s: toluene 3.78 min, BEtOH 5.89 min,benzophenone 16.23 min, lindane 17.71 min andmethyl stearate 21.14 min). The MSD was operatedin the selected ion-monitoring mode (SIM); ionsmonitored: toluene m/z 91, BEtOH m/z 57, benzo-phenone m/z 105, lindane m/z 181 and methyl stearatem/z 74. With a dwell time of 50 ms for each ion, 14cycles s 1 were achieved for each ion. The GC/MSsystem was computer controlled using HPChemStation software, Rev. A.03.02.


Figure 2. DiŒerential scanning calorimetry of PET froma carbonated beverage bottle and an amorphous sheet.

Dow

nloa

ded

by [

Uni

vers

ity o

f W

inds

or]

at 2

1:24

02

July

201

4

Control and recovery experiments were performed byfortifying methylene chloride and solutions of methy-lene chloride containing virgin PET. The methylenechloride solutions were diluted as described aboveand analysed by GC-MS. Recoveries were calculatedto be essentially 100% versus the controls at a preci-sion of 58%.

Results and discussion

To determine the potential maximum concentrationof a contaminant that may be in PET from thestorage of non-food substances, sorption experimentsinto two types of PET were performed. These testmaterials were amorphous PET (a relative absorbantfor contaminants because of its low crystallinity) andcommercial blow-moulded PET bottle wall (from 2-litre bottles). The test materials were stored in contactwith a commercial shampoo containing 1% lindane(C6H6Cl6), for various times up to 231 days at 20 and40°C. This commercial shampoo also represents anextreme case because it contains 7% acetone, asolvent that has been shown to swell PET by 30%(Moore and Sheldon 1961), which further enhancesthe capacity of PET to absorb chemicals. Afterregular contact times with the PET in the shampoo,the PET was removed, wiped dry with a paper toweland subjected to three quick washes in a solution of50/50 (v/v) isopropyl alcohol/water. This solution wasexperimentally found to remove all surface contam-ination of lindane from the PET. Although mostcommercial recycling systems use aqueous washingsystems, with their eŒectiveness having been reportedby Komolprasert and Lawson (1995), the focus of thiswork is to determine how much of the solute isactually in the polymer. Therefore, it is important tominimize surface contamination to ensure the amountmeasured in PET is not a surface phenomenon.

The lindane sorption data from the shampoo intoPET is illustrated in ®gure 3, which shows that thesorption of lindane into PET is Fickian. That is, theamount of sorption is initially proportional to thesquare root of time before approaching a maximumuptake. Additionally, ®gure 3 illustrates that there is avery signi®cant diŒerence in the amount of sorptionbetween room temperature and 40°C. The amount oflindane sorbed into PET from the shampoo after 231days was 28 mg kg 1 (0.1 mgdm 2) and 745 mg kg 1

(3.7 mg dm 2) at 20 and 40°C, respectively. All of

these values are within the ranges measured in thesurrogate contamination studies done by Demertziset al. (1997) and Komolprasert and Lawson (1995),indicating that the surrogate testing protocol forevaluating recycled polymers is probably a worst-casescenario because room temperature values are signi®-cantly less than the 40°C values.

The data in ®gure 3 also illustrate a very importantparameter related to the sorption of chemicals intoPET from a commercial shampoo. This is the equi-librium concentration or the maximum concentrationthe polymer will absorb. In the case of lindanesorption into amorphous PET (the worst case), thisconcentration is 800 mg kg 1 or 4 mg dm 2, whichrepresents only 0.3% of the lindane in the shampoo.In other words, 99.7% of the lindane is not absorbedfrom the shampoo, even though there is at least threetimes more polymer than lindane on a mass basis.Therefore, the shampoo essentially remains in®nitelyconcentrated with lindane relative to the concentra-tion in PET and is therefore not the limiting factordetermining the equilibrium concentration. It is alsovery interesting to note that the 800 mg kg 1 or4 mgdm 2 sorption is very close to the amounts ofsorption measured by Demertzis et al. (1997) after40°C and 40 days in the solvent-based systems.Demertzis et al. used solution concentrations 10 timesgreater than the concentration of lindane in theshampoo and obtained similar results. Therefore,


Figure 3. Sorption of lindane into oriented and amor-phous PET from a commercial shampoo containing 1%lindane at 20° and 40°C. Less than 1% of the lindane inthe shampoo is absorbed by PET.

Dow

nloa

ded

by [

Uni

vers

ity o

f W

inds

or]

at 2

1:24

02

July

201

4

800 mg kg 1 could be considered to be the upperrealistic limit for sorption of lindane into PET undernormal conditions of use.

Also illustrated in ®gure 3 are the eŒects of polymermorphology on the sorption of lindane from theshampoo. Here, the diŒerence in sorption at 40°Cinto amorphous PET (APET) and oriented PET(OPET bottle wall) is shown. The amount of lindanesorbed into oriented PET (actual PET bottle wall) isabout nine times less (on a mass/mass basis) than thatfound in APET. This diŒerence in the amount ofsorption is related to the orientation and crystallinityof the two materials. The diŒerence in the crystallinityof the two test materials is illustrated by using diŒer-ential scanning calorimetry shown in ®gure 2. Ingeneral, the diŒerence in the amount of crystallinitybetween these two materials is indicated by the mag-nitude of the exotherm peak at 160°C, the orientedmaterial having a higher crystallinity, therefore asmaller exotherm. Similar eŒects showing less sorp-tion in PET when the crystallinity and orientationincrease have been measured by Markarewicz andWilkes (1978) and Nir et al. (1996).

Additional sorption experiments using higher concen-trations (than lindane) of other penetrants and insolvents that do not interact highly with the polymerwere performed on the amorphous PET test material.The results of these tests, except for methyl stearate,are illustrated in ®gure 4. Methyl stearate was notsorbed from hexane into PET in measurable amounts.By comparing the sorption data in ®gures 3 and 4, itcan be seen that there is much less sorption from the

non-interactive liquids, and the time for equivalentsorption from these liquids is 50 times longer thanfor the lindane shampoo. It is important to note that,even though the concentration of the penetrant in thenon-interactive solutions was 10 times more concen-trated than that used in the lindane shampoo, thetotal amount of sorption into the amorphous PETwas less. This reduced sorption is attributed to thelack of polymer swelling eŒects induced by acetone inthe shampoo on the polymer. Solvent swelling eŒectson PET bottles are easily identi®ed visually becausethese eŒects lead to solvent-induced crystallization ofthe polymer, which creates opaque areas near the topand bottom of the bottles (the inherently amorphousregion of the bottle). In the amorphous PET testmaterials used in this study, noticeable opaquenesswas only observed in samples contacting theshampoo.

The sorption of methyl stearate into PET fromheptane was very low. The amount of sorptionwas not detectable at a LOD of 0.05 mg kg 1

(0.003 mg dm 2). Demertzis et al. found much higheramounts of methyl stearate sorption (3.2 mg dm 2) at40°C in 40 days. This higher sorption is most likelyrelated to swelling of the PET by the other solutesused in the testing mixture. Generally, the sorption oflarge non-polar solutes from non-polar solvents intoPET will be quite low because of partitioning anddiŒusion considerations. Other low sorption results ofnon-polar solutes in solvents have been reported byKomolprasert and Lawson (1995), who measured0.2 mg dm 2 of tetracosane in PET after 14 days at40°C.

Generally, the data show that sorption from solutionsinto PET follows Fick’s law of diŒusion. In the caseof amorphous PET used test for the worst case, theamount of a solute sorbed from a shampoo in theAPET is initially proportional to the square root ofthe contact time, and then the amount of soluteapproaches a limiting concentration (the equilibriumamount). For actual PET from commercial bottlewalls, the equilibrium sorption could not be measuredin the shampoo even though the thickness of thematerial was at least half as thick as the APETmaterial because diŒusion was to slow in the bottlewall material. For sorption of lindane from theshampoo into APET at 40°C, this equilibriumamount appears to be 800 mg kg 1. Extrapolationof the room temperature lindane shampoo data in®gure 3 indicates that this equilibrium value wouldnot be reached in 100 years. The concentration of


Figure 4. Sorption of toluene, benzophenone and 2-butox-ethoxy ethanol into amorphous and oriented PET at 40°C.

Dow

nloa

ded

by [

Uni

vers

ity o

f W

inds

or]

at 2

1:24

02

July

201

4

lindane in APET after actual room temperature in theshampoo for 231 days was 30 mg kg 1. This amountis signi®cantly less than that obtained at 40°C, thesuggested temperature for evaluating the potentialcontamination of polymers intended to be recycledinto food packaging.

Bayer’s (1998) analysis of contaminants in actualpost-consumer non-food PET bottles that had been¯aked suggests that the maximum concentration inraw, non-processed ¯ake, is only 15 mg kg 1 formethyl salicylate. Additionally, Franz and Welle’s(1999) analyses of 150 post-consumer PET ¯akesamples from food bottles originating from 14European PET recyclers/converters showed a maxi-mum concentration of 11 mg kg 1 for limonene (amajor component of ¯avours). In our analysis ofsome speci®c non-food PET bottles obtained directlyfrom collection bins for recycling plastic bottles in theUS, we found 130±204 mg kg 1 of methyl salicylate ina mouth antiseptic wash bottle, 1 mg kg 1 triclosan(an antibacterial agent) in a hand soap bottle, and1.6 mg kg 1 limonene in a paint remover bottle.Methyl salicylate (0.06% in solution) in an antisepticwash is an ideal case for evaluating sorption of acontaminant into non-food PET bottles. This is be-cause the mouth rinse is 21% ethanol, a polar solventthat can aŒect PET and the solubility parameters formethyl salicylate and PET are almost identical(¯ ˆ 21:7 (Mpa)1=2 (Barton 1983) and ¯ ˆ 21:9(Mpa)1=2 (Brandrup and Immergut 1989) for methylsalicylate and PET respectively) . That is, methylsalicylate is very soluble in PET and will causedvisible solvent induced crystallization when placedin contact with APET. It should also be pointed outthat during the complete shelf life and consumer useof these bottles, >98% of the original methyl salicy-late in the mouthwash did not enter the PET bottle.

The magnitude of these data is consistent with thesorption data presented here and surrogate testingstudies done by Komolprasert and Lawson (1995)and Demertzis et al. (1997). Additionally, studiesperformed to evaluate the eŒectiveness of PET clean-ing processes involving surrogate contamination ofPET ¯ake and bottles show that simple washing anddrying reduce the concentration of volatile contami-nants in PET by at least 98% and non-volatile con-taminants by 79% (Komolprasert and Lawson 1995).Consequently, the inclusion of non-food PET bottleswith food bottles in the reprocessing stream to makenew food packages should not signi®cantly alter thepurity of recycled PET. This is because it is not likely

that all bottles in a recycling stream will be non-food(custom) bottles and the data show that PET bottlesdo not absorb large quantities of chemicals. Ideally,though, recycled polymers should be used in non-food applications, such as durable goods, that have along product life (e.g. ®bres).

Regulatory implications

The FDA’s guidance document for industry, the`Points to Consider for the Use of Recycled Plasticsin Food Packaging: Chemistry Considerations’ (FDA1992), was developed to assist recyclers in determin-ing if their processes are capable of producing prod-ucts that are suitable for food-contact uses. The`Points to Consider’ recommend a surrogate testingprotocol that involves soaking (or `challenging’ ) PET¯ake in an appropriate surrogate cocktail for 14 daysat 40°C, running the challenged ¯ake through therecycling process, and then analysing the ®nishedrecycled PET for the individual surrogates. Theamount of each surrogate that can migrate to foodfrom the ®nished recycled PET cannot exceed whatwould result in a dietary concentration of 0.5 ppb,FDA’s level of no regulatory concern (FederalRegister 1995). (The `Points to Consider’ actuallyrefers to FDA’s preliminary thinking of 1 ppb as themaximum exposure corresponding to negligible risk;Kuznesof and VanDerveer 1995.) The dietary con-centration may be determined based on a 100%migration calculation, actual migration tests, or mi-gration modelling (Begley and Holli®eld 1995).

The `Points to Consider’ were developed to addressincidental post-consumer contamination of food con-tainers, not pre-consumer contamination of non-foodcontainers with their intended contents (e.g. house-hold cleaners, furniture polish, shampoos, soaps,pesticides, motor oil, etc.). The conditions for challen-ging the polymer ¯ake were designed with onlyincidental contamination of plastic food containersin mind, i.e., contamination by consumers who mightstore non-food substances in the containers beforereturning them for deposit or recycling them. Thelevel of contamination in PET containers recycledfrom a commingled post-consumer feedstock is likelyto be higher than that in containers recycled exclu-sively from food containers because (1) non-foodcontainers would be introduced deliberately on aregular basis and (2) non-food containers are likely


Dow

nloa

ded

by [

Uni

vers

ity o

f W

inds

or]

at 2

1:24

02

July

201

4

to sorb large amounts of contaminants from theircontents over very long shelf lives ( 1 year).

To determine if the conditions for challenging ¯akerecommended in the `Points to Consider’ adequatelyrepresents he higher level of contamination expectedin post-consumer feedstock containing non-food PETcontainers, we used experimentally measured valuesand calculated values. Calculated values were basedon Crank’s (1975) mathematical expressions derivedfor diŒusion for the case of a polymer in a stirredsolution of limited volume (see equation 1). Weattempted to predict from these calculations howmuch of the surrogate contaminants described inthe `Points to Consider’ will sorb into a PET bottleafter 1 year at 25°C, the shelf life and use temperatureof a typical non-food substance packaged in PET. Weassumed that a typical non-food PET bottle has a 1-litre capacity, a mass of 38 g, a density of 1.37 g cm 3,and a wall thickness of 0.03 cm (12 ml). Additionally,we assumed that a 10% w/w concentration of asurrogate in solution adequately represents the maxi-mum concentration of any given component of a non-food substance packaged in PET. The 10% concen-tration assumption also ensures the total mass to thesolute is greater than the mass of the polymer.Typically, many active ingredients in solutions are1% or less. For example, shampoo contains 1%lindane, soap contains 0.03% Triclosan, and mouthwash contains 0.06% methyl salicylate .

One of the limitations of using calculations to esti-mating concentrations of solutes in the polymer is thelack of accurately measured mass transfer par-ameters. In particular, not many diŒusion coe� cientsare known for solutes in PET because they tend to bevery slow and therefore very di� cult to measure.Therefore, careful consideration must be given tochoosing D used in calculations. Additionally, theshelf life (time) of the packages is not known andtherefore must be assumed to be long times (e.g. 1year). Finally, the partition coe� cients, which are notknown, could favour the polymer and signi®cantlyaŒect the concentration of the solute in the polymer.To minimize these limitations, actual measurementsof some solutes in real PET bottles are needed toensure that the actual extremes are not exceeded. Ingeneral, calculated values for surrogates are muchgreater than measured values of actual chemicalsin PET.

For chloroform (volatile, polar), copper(II) 2-ethyl-hexanoate (heavy metal), and tetracosane (non-vola-tile, non-polar), we modelled concentrations in PET

using diŒusion calculations (equation 1). For chloro-form, we used D ˆ 9:0 10 14 cm2 s 1 (25°C) calcu-lated from the Piringer model (Baner et al. 1996).For tetracosane and copper(II) (from copper(II)2-ethylhexanoate) , we used D ˆ 10 16 cm2 s 1 and10 17 cm2 s 1, respectively, calculated from experi-mental data obtained by Sadler et al. (1996). Using®gures 3 and 4, we extrapolated the 40°C data toobtain concentrations in PET for the surrogatestoluene (volatile non-polar), lindane (non-volatile,non-polar), and benzophenone (non-volatile, polar)at 365 days. We used measurements in actual recycledbottles to determine the concentration in PET ofmethyl salicylate (semi-volatile, polar), which is apolymer-speci®c surrogate for PET. The modelledand experimentally determined sorption results areshown in table 1.

The results in table 1 establish the minimum concen-trations of each surrogate that must be present in thechallenged PET ¯ake at the beginning of surrogatetesting in order for the resultant exposure estimates toapply to a recycling process in which non-food PETcontainers are included in the feedstock. In the reviewof several recycled PET submissions to FDA, we havefound that challenging PET ¯ake with an appropriatesurrogate cocktail for 14 days at 40°C generallyresults in starting concentrations that are comparableto the minimum concentrations given in table 1. Incases in which the starting concentrations are too low,the shortfall may be corrected by multiplying concen-trations in the ®nished recycled PET and any surro-gate migration data by a simple factor.


Table 1. Sorption levels of surrogate contaminants intotypical non-food PET bottles after 365 days at 25°C.

SorptionSurrogate (mgkg 1)

Chloroform 4860a

Toluene 780b

Lindane 750c

Tetracosane 154a

Benzophenone 49b

Copper (II) 2-ethylhexanoate 49a

Methyl Salicylate 200d

a Modelled valueb This value was extrapolated from sorption measurements made after1, 8, 29, 57 and 111 days at 40°C (see ®gure 4).c This is an experimentally measured equilibrium (40°C) used directlyto represent the maximum amount of lindane that might be sorbed bya non-food PET container (see ®gure 3).d Determined from actual recycles bottles.

Dow

nloa

ded

by [

Uni

vers

ity o

f W

inds

or]

at 2

1:24

02

July

201

4

Recycling processes that include non-food PET con-tainers in the feedstock can therefore be used toproduce PET for food containers, provided that (1)the process has been tested using starting levels ofsurrogate contaminants in the PET ¯ake that aregreater than or equal to the levels shown in table 1and (2) testing results on ®nished materials madefrom the recycled surrogate-contaminated PET ¯akeshow that exposure to each of the surrogate contami-nants is 0.5 ppb dietary concentration.

References

Baner, A. L., Brandsch, J., Franz, R., and Piringer, O., 1996, Theapplication of a predictive migration model for evaluating thecompliance of plastic materials with European food regula-tions. Food Additives and Contaminants, 13, 587±601.

Barton, A. F. M. 1983, Handbook of Solubility Parameters and OtherCohesion Parameters (Boca Raton: CRC Press).

Bayer, F., 1998, American Chemical Society Symposium on FoodPackage Interactions, National Meeting, Dallas, TX.

Begley, T. H., and Hollifield, H. C., 1995, Food packaging madefrom recycled polymers. Plastics, Rubber, and Paper Recycling:a Pragmatic Approach, edited by C. P. Rader, S. D. Baldwin,D. D. Cornell, G. D. Sadler and R. E. Stockel. ACSSymposium Series No. 609 (Washington, DC: AmericanChemical Society), pp. 445±457.

Brandrup, J., and Immergut, E. H., 1989, Polymer Handbook, 3rdedn (Chichester: Wiley).

Crank, J., 1975, The Mathematics of DiŒusion, 2nd edn (Oxford:Clarendon), pp. 56±59, equations (4.43, 4.41).

Demertzis, P. G., Johansson, F., Lievens, C., and Franz, R., 1997,Development of a quick inertness test procedure for multi-usePET containersÐsorption behavior of bottle wall strips.Packaging Technology and Science, 10, 45±58.

Federal Register, 17 July 1995, 60(136), 36582±36596.Food and Drug Administration , 1992, Points to Consider for the

Use of Recycled Plastics in Food Packaging: ChemistryConsiderations. Chemistry Review Branch, December(Washington, DC: Center for Food Safety and AppliedNutrition, FDA).

Franz, R., and Welle, F., 1999, Analytical screening and develop-ment of an analytical method for quality assurance checking ofmarket usual PET ¯akes. Deutsche Lebensmittel-Rundschau, 95,94±100.

Komolprasert , V., and Lawson, A. R., 1995, Residual contaminantsin recycled poly(ethylene terephthalate): eŒects of washing anddrying. ACS Symposium Series No. 609 (Washington, DC:American Chemical Society), pp. 435±444.

Komolprasert , V., Lawson, A. R., and Gregor, A., 1996, EŒect ofextrusion remelting on residual contaminants in secondary re-cycled polyethylene terephthalate. Journal of PackagingTechnology and Engineering, 8, 25±31.

Kuznesof, P. M., and VanDerveer, M. C., 1995, Recycled plasticsfor food-contact applications. Plastics, Rubber, and PaperRecycling: a Pragmatic Approach, edited by C. P. Rader, S. D.Baldwin, D. D. Cornell, G. D. Sadler and R. E. Stockel. ACSSymposium Series No. 609 (Washington, DC: AmericanChemical Society), pp. 389±403.

Markarewicz, P. J., and Wilkes, G. L., 1978, DiŒusion studies ofpoly(ethylene terephthalate) crystalilized by non-reactive li-quids and vapors. Journal of Polymer Science, PolymerPhysics Edition, 16, 1529±1544.

Modern Plastics, January 1999, p. 74.Moore, W. R., and Sheldon, R. P., 1961, The crystallization of poly-

ethylene terephthalate by organic liquids. Polymer, 2, 315±321.Nir, M. M., Ram, A., and Miltz, J., 1996, Sorption and migration of

organic liquids in poly(ethylene terephthalate). PolymerEngineering and Science, 36, 862±868.

Patton, C. J., Felder, R. M., and Koros, J., 1984, Sorption andtransport of benzene in PET. Journal of Applied PolymerScience, 29, 1095±1110.

Sadler, G., Pierce, D., Lawson, A., Suvannunt, D., and Senthil,V., 1996, Evaluating organic compound migration in poly(ethy-lene terephthalate): a simple test with implications for polymerrecycling. Food Additives and Contaminants, 13, 979±999.


Dow

nloa

ded

by [

Uni

vers

ity o

f W

inds

or]

at 2

1:24

02

July

201

4

evaluating the potential for recycling all pet bottles into new food packaging

Documents