1 experimental direct evaluation of functional barriers in pet...
TRANSCRIPT
1
Experimental direct evaluation of functional barriers in PET recycled bottles 1
Comparison of migration behaviour of mono and multilayers 2
3
P. Y. Pennarun a, P. Saillardb, A. Feigenbauma, P. Dolea 4 5
a: INRA UMR FARE, Moulin de la Housse, 51687 Reims, France 6
b: ITOSOPE - CTCPA Technopole Alimentec, 01060 Bourg en Bresse, France 7
8
Abstract 9
10
Recycling of used bottles into new bottles is associated with possible migration of pollutants arising 11
from the previous life of the packages. To reduce or to delay such migration, the recycled resin is 12
depolluted, or a functional barrier layer, made of virgin plastics, is used. Testing migration from 13
such recycled bottles relies on the use of model pollutants (surrogates). In order to enable modelling 14
of migration kinetics, each step of the use of surrogates is carefully investigated here in the case of 15
PET. In a first part, criteria underlying the selection of surrogates are carefully examined; besides 16
volatility, polarity and diffusion behaviour, it is shown here that their solubility in the food simulant 17
and their chemical stability strongly influence migration results. For aqueous test media, 2,4-18
pentanedione and phenol should be used as surrogates. In a second part, a procedure is developed to 19
impregnate surrogates at very large concentrations (several thousands mg/kg PET) which are 20
necessary to monitor migration kinetics. This procedure, which uses dichloromethane as solvent, 21
allows a quick and reproducible impregnation, not sensitive to temperature between 11 and 23 °C, 22
factors which favour its use at a plant scale. In a third part, flakes impregnated with this procedure 23
are processed into bottles, and their physico-chemical properties are compared to those of 24
commercial bottles. In a last part, monolayer and tri-layer polluted bottles (model pollutants in inner 25
layer) are tested for migration for more than 1.5 year. With multilayers, the migration lag time of 26
the fastest surrogates is 6 months with 3 % acetic acid, and 3 months with ethanol as simulant, due 27
to plasticization of PET by ethanol. The sequence of migration of surrogates is different with 28
2
monolayer and with multilayer bottles, which shows that partition effects (solubility) play an 29
essential role, especially with monolayer materials. 30
31
Keywords 32
33
Functional barriers, PET, recycling, surrogates, food simulant, consumer exposure, migration 34
35
Introduction 36
37
Recycling plastics waste has become a major issue in most developed countries. A very good 38
outcome for used food packaging materials is to make new packages from old packages. However, 39
this may raise food safety considerations, as post-consumption collected packages may be polluted 40
by common chemicals available to households or to any consumers (detergents, petrol, garden 41
herbicides or pesticides...) [1]. In order to prevent such chemicals from contaminating foodstuffs 42
packaged in recycled plastics, various processes are available to purify the materials or to reduce 43
migration: washing and reuse [2], depolymerisation [3, 4], post-condensation [5], use of a 44
functional barrier [6, 7] or on line decontamination during processing [8]. 45
46
Testing foodstuffs packaged in such materials for non contamination by these chemical pollutants is 47
an impossible task, since neither the identity nor even the presence of pollutants are known. An 48
interesting approach was proposed years ago by the Food and Drugs Administration [9]. This 49
consists in incorporating model chemicals -or surrogates- into a resin, in order to evaluate the 50
ability of a recycling process to purify the material and of a multilayer packaging structure to delay 51
and to reduce migration. Resins impregnated with surrogates undergo all the steps of manufacture 52
of new containers, including washing, drying, extrusion and/or injection. At the end of the process, 53
the materials are extracted to determine the amount of residual surrogates and the cleaning 54
3
efficiency of the process. In addition, the finished packages can be filled with food simulants, and 55
the migration of residual surrogates is then monitored [10, 11, 12]. 56
57
The efficiency of a functional barrier can be evaluated by the migration lag time, which is the time 58
needed for a surrogate to cross the barrier layer. The lag time depends on several factors, like nature 59
of the polymer and of the pollutant [13, 14], dimension and heterogeneity of the barrier [15], nature 60
of the food in contact, possible pollution of the barrier layer during processing [16]. Two 61
approaches are possible to evaluate the lag time: experimental evaluation of migration on materials 62
spiked with surrogates and prediction using numerical simulation of pollutant migration. Prediction 63
is of course a very good approach; however models require experimental parameters representative 64
of all relevant phenomena. 65
66
In this work, we will evaluate experimentally the migration from recycled mono and tri-layer 67
bottles with a classical geometry. The experimental data obtained will be used in another paper to 68
feed simulation model. We focus here on poly(ethylene terephthalate) (PET), a good candidate for 69
large scale recycling. Migration of model pollutants through PET functional barriers has almost 70
never been detected, and it has been questioned whether the physical properties of surrogates could 71
influence the results and the assessment of the materials [6]. FDA paid attention to polarity and 72
volatility, and surrogates used in most studies have been selected mainly on the basis of these 73
criteria (table 1). 74
75
In the current study, full migration kinetics were obtained for the first time for PET functional 76
barriers, which required migration data well above quantification limits. This was achieved here: 77
- by using a sufficient pollutant concentration in PET: since migration is proportional to the 78
initial concentration in the polymer, high concentrations of surrogates were used: high 79
enough to enable detection in test media, but not too high, to avoid unrealistic modifications 80
4
of PET. The concentrations used are driven by the need to be able to measure kinetics, and 81
are independent on any “realistic” level of pollution. 82
- by choosing pollutants which are likely to migrate in aqueous food simulants (surrogates 83
must be soluble in test media) 84
- by choosing pollutants adapted to the process of artificial pollution (surrogates should not 85
react with the polymer during processing at high temperature) 86
- by recording migration kinetics over very long contact times (over 1.5 year). 87
88
Independently of migration simulation, it will be possible to extrapolate the results obtained here 89
(with concentrations around 1000-1500 ppm surrogate in PET), to predict the migration from PET 90
polluted at “realistic” levels (around 1-3 ppm) considering that migration is proportional to the 91
initial concentration. 92
93
The kinetics described here will allow characterizing experimentally the properties of functional 94
barriers. The influence of the food simulant (ethanol and 3 % acetic acid were used here) will be 95
assessed, both for mono- and tri-layers. The results will provide experimental parameters which will 96
be used in a next paper for migration modelling [17, 18, 19], in order to support prediction of the 97
behaviour of PET materials. 98
5
Materials and methods 99
100
I. Materials 101
17 surrogates (table 2) have been selected on the basis of different volatility, molecular weight, 102
chemical functionality or polarity (especially solubility in water). Most of them are also in usual 103
lists of recommended substances. In order to avoid too large global concentrations of surrogates in 104
PET, they were divided into three sets (A), (B) and (C). Each set of surrogates was incorporated 105
separately, thus minimising alteration of PET physical-chemical properties. The sets were defined 106
in order to prevent reactions between surrogates. 107
108
2,5–Thiophenediylbis(5-tert-butyl-1,3-benzoxazole) (Uvitex OB) (>96%) was supplied by CIBA 109
(France). Phenol (99%), chlorobenzene (99%), 2,6-ditert-butyl-4-methylphenol (BHT) (99%), 110
azobenzene (99%), phenyl benzoate (99%), toluene (99%), nonane (99%), dimethyl sulfoxide 111
(DMSO) (99%), benzophenone (99%), phenylcyclohexane (98%) were supplied by Acros (France). 112
Chlorooctane (99%), dibutyl phthalate (99%), 2,4-pentanedione (99%) and ethyl hydrocinnamate 113
(99%) were supplied by Aldrich (France). 1,1,1-Trichloroethane (99%) was supplied by Fluka. 114
Methyl palmitate (>96%) was supplied by ICN Biomedicals Inc (France). PET flakes from an 115
industrial processing recycling plant were supplied by Amcor PET Packaging Recycling France 116
(Dunkerque, France). 117
118
II. Thermal analysis 119
Modulated Differential Scanning Calorimetry (MDSC) was performed with MDSC2920 (TA 120
Instruments) to evaluate reversible and non reversible heat flow. PET samples (10 mg), taken at a 121
distance X from the bottom of bottles were placed in aluminium pans closed with a sealed cap to 122
prevent modifications of sample contact surface with DSC sensors during experiment. These 123
6
modifications are due to relaxation of orientation during heating [20]. The heating rate was 124
5°C/min, the oscillation period was 60 seconds and the amplitude was 0.796°C (heat only mode). 125
Melt Flow Indexes (MFIs) were determined on a Davenport Melt Viscosimeter (Farcham, UK). 126
127
III. Impregnation by surrogates and processing of model materials: 128
Experimental conditions differ for each experiment, as they were optimised progressively to 129
achieve a target concentration. Changing the concentration of a single surrogate in the 130
dichloromethane solution affects the final concentrations of all the other surrogates in the polymer. 131
However, when the conditions are well defined, the final concentrations are well reproducible. 132
133
Solvent effects on impregnation kinetics 134
PET samples taken from bottle walls (sides of the bottles) were immersed in a solution of 135
surrogates in dichloromethane or heptane. 136
137
Impregnation in dichloromethane: several concentrations in impregnation solution were used (see 138
conditions in following paragraphs). 139
140
Impregnation in heptane: 2.5 g of PET pieces of bottle walls were placed in 20 ml of a solution of 141
toluene and 1,1,1-trichloroethane in heptane at 40°C. This solution contained 66.2% (w/w) heptane, 142
20.5% (w/w) 1,1,1-trichloroethane and 13.3% (w/w) toluene. At regular intervals of time, PET 143
pieces were removed from the mixture, rapidly washed superficially with ethanol (1 min), and 144
extracted by dichloromethane. 145
146
Rinsing of flakes: when flakes are washed by immersion into ethanol on a pilot plant scale, it may 147
be difficult to control the washing time. Therefore, different rinsing times (1-10 minutes) were 148
7
checked on a laboratory scale. It allowed ascertaining that washing did not significantly modify the 149
concentration of the surrogates in the flakes. 150
151
Temperature effect on impregnation 152
A dichloromethane solution of set B [phenol (1.1 % w/w), BHT, chlorobenzene, chlorooctane, 1,3 – 153
butanediol (each 1 % w/w), Uvitex OB (0.54 % w/w)] was used to evaluate the effect of 154
temperature on impregnation. 1 g of flakes was placed in a flask with 50 ml of the dichloromethane 155
solution. The flasks were immersed in a bain-marie at 11, 17 and 23°C. After 5 days, flakes were 156
separated and washed with ethanol at the same temperature for 1 minute, and then extracted as 157
described below. 158
159
Reactivity of surrogates in dichloromethane at 40°C 160
50 ml of solutions of each of the 3 set of surrogates (500 mg each/l) in dichloromethane were stirred 161
for 7 days at 40°C. Samples of 1 µl were analysed by GC-FID at regular interval of time. The 162
internal standard was tetradecane for sets A and B, and toluene (non reactive substances) for set C. 163
164
Effects of drying at 150°C and injection at 280°C processes 165
26 g of PET flakes were introduced in a flask with 92 g of each surrogate solution. The system was 166
stirred regularly. After 5 days at room temperature, the PET flakes were washed twice with 20 ml of 167
ethanol. 168
169
Solutions of surrogates for impregnation into flakes were prepared as shown in table 3: 170
Simulation of drying at 150°C was realised in an oven, under air aspiration. Samples of the flakes 171
were removed periodically, and the surrogates were extracted by dichloromethane and analysed as 172
described below (section IV). 173
8
To evaluate injection process effects, impregnated PET flakes (0.4 g) were first dried for 3 hours at 174
150°C, then placed in a pyrex tube (8 cm long x 6 mm diameter). The top of the tube was sealed 175
under vacuum while the flakes were cooled with ice, to prevent evaporation of the surrogates. The 176
sealed tubes were then heated at 280°C for 5 minutes. After cooling, surrogates were extracted with 177
dichloromethane (32 hours at 40°C), and analysed as described below (section IV). 178
179
Manufacture of impregnated bottles: 180
Impregnation conditions are described in table 3. After 5 days incubation in the solution, the flakes 181
were removed and washed twice with ethanol (2 x 8 l). The flakes were heated at 150°C for 3 hours 182
in a ventilated oven to remove dichloromethane and water. They were then sent to the plant, where 183
they were dried again at 150°C for 3 hours. Preforms were injected by Amcor PET Packaging 184
Recycling France (Dunkerque, France) and bottles were blown. Monolayer (100% of impregnated 185
PET) and tri-layer (25% of impregnated PET) bottles were obtained. The materials were stored at –186
20°C to prevent evaporation of surrogates and their diffusion. 187
188
IV. Extraction and quantification of surrogates in PET. 189
Flakes (less than 0.5g) were extracted with 10 ml of dichloromethane in a closed vial (SVL screw 190
stopper tube, diameter 15) for 32 hours at 40°C. It was checked in separate experiments that this 191
time was needed for a quantitative extraction of the cyclic PET trimer (molecular weight 576 g/mol, 192
larger than that of any surrogate) [21]. 1 ml of tetradecane internal standard solution in 193
dichloromethane (50 mg/l) was then added. Results were expressed as a function of PET mass after 194
flakes drying at 150 °C for 3 h. Extracts were analysed by GC-FID (on column mode) as follows: 195
1,1,1-Trichloroethane, phenylcyclohexane: The column was a DB5-MS J&W Scientific (15 m x 196
0.32 mm x 1 µm). Carrier gas (He) flow was 2 ml/min at 40 °C. FID temperature was 300 °C, H2 197
and air flows were 25 and 250 ml/min respectively. The oven program was: 40 °C for 4 minutes, 198
9
ramp 15 °C/min up to 132 °C, isotherm for 6 minutes, heating 15 °C/min up to 270 °C and isotherm 199
for 3 minutes. 200
DMSO, methyl palmitate, benzophenone, ethyl hydrocinnamate: the column was a DB-WAX J&W 201
Scientific (30 m x 0.25 mm x 0.25 µm). Carrier gas (He) flow was 1.8 ml/min at 40 °C. FID 202
temperature was 240 °C, H2 and air flows were 25 and 250 ml/min respectively. The oven program 203
was: 40 °C for 5 minutes, ramp 15 °C/min up to 230 °C, isotherm for 3 minutes. 204
205
BHT, Uvitex OB: The column was a DB5-MS J&W Scientific (15 m x 0.32 mm x 1 µm). The 206
carrier gas (He) flow was 2 ml/min at 40 °C. FID temperature was 330 °C, H2 and air flows were 25 207
and 250 ml/min respectively. The oven program was: 40 °C for 5 minutes, ramp 15°C/min up to 208
320°C, isotherm for 11 minutes. 209
210
Phenol, chlorobenzene, 1-chlorooctane: The column was a DB-WAX J&W Scientific (30 m x 0.25 211
mm x 0.25 µm). Carrier gas (He) flow was 1.8 ml/min at 40°C. FID temperature was 240°C, H2 and 212
air flows were 25 and 250 ml/min respectively. The oven program was: 40°C for 5 minutes, ramp 213
15°C/min up to 210°C, isotherm for 3 minutes. 214
215
Azobenzene, nonane: The column was a DB5-MS J&W Scientific (15m x 0.32mm x 1µm). Carrier 216
gas (He) flow was 2 ml/min at 40°C. FID temperature was 300°C, H2 and air flows were 25 and 250 217
ml/min respectively. The oven program was: 40°C for 8 minutes, ramp 15°C/min to 170°C, ramp 218
2°C to 180°C, ramp 15°C/min up to 240°C, isotherm for 2 minutes. 219
220
2,4-Pentanedione, DBP, phenyl benzoate, toluene: The column was a DB-WAX J&W Scientific 221
(30m x 0.25mm x 0.25µm). Carrier gas (He) flow was 1.8 ml/min at 40°C. FID temperature was 222
240°C, H2 and air flows were 25 and 250 ml/min respectively. The oven program was: 40°C for 4 223
minutes, ramp 15°C/min up to 230°C, isotherm for 4 minutes. 224
10
225
226
V Migration tests 227
Migration from model bottles into 3 % acetic acid 228
Acetic acid/water (3% w/v) is used as food simulant. Ultrapure water and acetic acid Normapur for 229
analysis (PROLABO) are used. Each bottle is filled with 1.5 l of simulant, closed tightly with a 230
screw stopper and placed at 40°C. Each bottle gives only one migration measurement at time t. 231
Samples of 75 ml are neutralised with 10 M sodium hydroxyde. 20 g of sodium chloride are added 232
and the solution is extracted for 12 hours in a closed vial with 3 ml dichloromethane containing 50 233
mg/l tetradecane. Control extraction tests showed that recovery rate of every surrogates is close to 234
100%, except for DMSO (0%), 2,4-pentanedione (66%) and phenol (33%). 235
236
Extracts are analysed directly by on-column GC-FID (on column mode) as follows: 237
1,1,1-Trichloroethane, phenylcyclohexane: The column is a DB5-MS J&W Scientific (15 m x 0.32 238
mm x 1 µm). Carrier gas (He) flow is 2 ml/min at 40°C. FID temperature is 300°C, H2 and air flows 239
are 25 and 250 ml/min respectively. The oven program is: 40°C for 4 minutes, ramp 15°C/min to 240
132°C, isotherm for 6 minutes, heating 15°C/min up to 270°C and isotherm for 3 minutes. 241
DMSO, methyl palmitate, benzophenone, ethyl hydrocinnamate: The column is a DB-WAX J&W 242
Scientific (30 m x 0.25 mm x 0.25 µm). Carrier gas (He) flow is 1.8 ml/min at 40°C. FID 243
temperature is 240°C, H2 and air flows are 25 and 250 ml/min respectively. The oven program is: 244
40°C for 5 minutes, ramp 15°C/min up to 230°C, isotherm for 3 minutes. 245
BHT, Uvitex OB: The column is a DB5-MS J&W Scientific (15 m x 0.32 mm x 1 µm). The carrier 246
gas (He) flow is 2 ml/min at 40°C. FID temperature is 330°C, H2 and air flows are 25 and 250 247
ml/min respectively. The oven program is: 40°C for 5 minutes, ramp 15°C/min up to 320°C, 248
isotherm for 11 minutes. 249
11
Phenol, chlorobenzene, 1-chlorooctane: The column is a DB-WAX J&W Scientific (30 m x 0.25 250
mm x 0.25 µm). Carrier gas (He) flow is 1.8 ml/min at 40°C. FID temperature is 240°C, H2 and air 251
flows are 25 and 250 ml/min respectively. The oven program is: 40°C for 5 minutes, ramp 252
15°C/min up to 210°C, isotherm for 3 minutes. 253
Azobenzene, nonane: The column is a DB5-MS J&W Scientific (15m x 0.32mm x 1µm). Carrier 254
gas (He) flow is 2 ml/min at 40°C. FID temperature is 300°C, H2 and air flows are 25 and 250 255
ml/min respectively. The oven program is: 40°C for 8 minutes, ramp 15°C/min to 170°C, ramp 2°C 256
to 180°C, ramp 15°C/min up to 240°C, isotherm for 2 minutes. 257
2,4-Pentanedione, DBP, phenyl benzoate, toluene: The column is a DB-WAX J&W Scientific (30m 258
x 0.25mm x 0.25µm). Carrier gas (He) flow is 1.8 ml/min at 40°C. FID temperature is 240°C, H2 259
and air flows are 25 and 250 ml/min respectively. The oven program is: 40°C for 4 minutes, ramp 260
15°C/min up to 230°C, isotherm for 4 minutes. 261
Detection limits (and recovery yields): 1,1,1-Trichloroethane: 10 ppb (100 %), 1-chlorooctane: 5 262
ppb (100 %) , 2,4-pentanedione: 10 ppb (66 %), azobenzene: 15 ppb (100 %), benzophenone: 10 263
ppb (100 %), BHT: 2 ppb (100 %), chlorobenzene: 5 ppb (100 %), DBP: 5 ppb (100 %), DMSO: (0 264
%), ethyl hydrocinnamate: 5 ppb (100 %), methyl palmitate: 5 ppb (100 %), nonane: 3 ppb (100 %), 265
phenol: 10 ppb (33 %), phenyl benzoate: 5 ppb (100 %), phenylcyclohexane: 5 ppb (100 %), 266
toluene: 5 ppb (100 %). 267
268
Migration from model bottles into ethanol 269
Absolute ethanol (pure for analysis, SDS) is used as a simulant. Each bottle is filled with 1.5 l of 270
simulant and placed at 40°C. Samples (10 ml) are regularly taken from bottles. 271
100 µl of ethanol containing 1527 g/l of tetradecane as internal standard are added to the samples. 272
These samples (6 µl) are analysed by GC-FID with Split/Splitless injection technique (Splitless time 273
is 20 s and flow is 20 ml/min) as follows: 274
12
1,1,1-Trichloroethane: The column is a DB1-MS J&W Scientific (15 m x 0.53 mm x 5 µm). Carrier 275
gas (He) flow is 2 ml/min at 40°C. Injector temperature is 230°C, FID temperature 280°C, H2 and 276
air flows are 25 and 250 ml/min respectively. The oven program is: 70°C for 5 minutes, ramp 25 277
°C/min to 280°C, isotherm for 7 minutes. 278
DMSO, methyl palmitate, benzophenone, ethyl hydrocinnamate, phenylcyclohexane: The column is 279
a DB-WAX J&W Scientific (30 m x 0.25 mm x 0.25 µm). Carrier gas (He) flow is 1.8 ml/min at 280
40°C. Injector temperature is 220°C, FID temperature is 240°C, H2 and air flows are 25 and 250 281
ml/min respectively. The oven program is: 70°C for 5 minutes, ramp 15°C/min to 230°C, isotherm 282
for 6 minutes. 283
Phenol, chlorobenzene, 1-chlorooctane, BHT: The column is a DB-WAX J&W Scientific (30 m x 284
0.25 mm x 0.25 µm). Carrier gas (He) flow is 1.8 ml/min at 40°C. Injector temperature is 220°C, 285
FID temperature is 240°C, H2 and air flows are 25 and 250 ml/min respectively. The oven program 286
is: 70°C for 5 minutes, ramp 15°C/min to 230°C, isotherm for 6 minutes. 287
Nonane, 2,4-Pentanedione, toluene: The column is a DB1-MS J&W Scientific (15 m x 0.53 mm x 288
5 µm). Carrier gas (He) flow is 2 ml/min at 40°C. Injector temperature is 230°C, FID temperature 289
280°C, H2 and air flows are 25 and 250 ml/min respectively. The oven program is: 70°C for 6 290
minutes, ramp 25°C/min to 280°C, isotherm for 7 minutes. 291
DBP, phenyl benzoate, azobenzene: The column is a DB-WAX J&W Scientific (30m x 0.25mm x 292
0.25µm). Carrier gas (He) flow is 1.8 ml/min at 40°C. Injector temperature is 230°C, FID 293
temperature is 240°C, H2 and air flows are 25 and 250 ml/min respectively. The oven program is: 294
70°C for 5 minutes, ramp 25°C/min to 230°C, isotherm for 7 minutes. 295
Detection limits: 1,1,1-Trichloroethane: 400 ppb, 1-chlorooctane: 120 ppb, 2,4-pentanedione: 90 296
ppb, azobenzene: 60 ppb, benzophenone: 80 ppb, BHT: 20 ppb, chlorobenzene: 130 ppb, DBP: 50 297
ppb, DMSO: 120 ppb, ethyl hydrocinnamate: 80 ppb, methyl palmitate: 80 ppb, nonane: 50 ppb, 298
phenol: 20 ppb, phenyl benzoate: 50 ppb, phenylcyclohexane: 50 ppb, toluene: 50 ppb. 299
300
13
Results and discussion. 301
302
In the case of mechanical recycling, when collected bottles are washed, chopped and their flakes are 303
mixed with those of clean bottles, the concentration of adventitious pollutants drops to very low 304
amounts, probably well below the detection limits of the most common analytical methods. 305
However, when the purification efficiency of a process has to be determined, or if the aim is to 306
model migration kinetics, much larger concentration of surrogates are needed, well above the 307
detection limits. We decided that a target concentration of 1000 (± 50 %) ppm of each surrogate in 308
the recycled layers was necessary to obtain migration kinetics with a good sensitivity. Since we 309
intended to test a large number of surrogates, this lead to split them into three series, which were 310
incorporated separately, to avoid overload of the PET matrix. Surrogates used should cover a broad 311
range of molecular weight, shape, functionality and solubility in food simulants. 312
313
I. Selection of surrogates. 314
Volatility: Low molecular weight migrants usually migrate the fastest and are therefore those of 315
major concern for safety assessment. On the other hand, the more volatile the surrogates, the easier 316
is their evaporation during the drying of flakes, in an air stream at 150 - 180°C. This was addressed 317
by using surrogates covering a broad range of molecular weights (table 2), up to 431 g/mol (Uvitex 318
OB). Dedicated experiments will be shown in the section “influence of drying” [22]. 319
320
Polarity: Polarity both influences the compatibility of surrogates with the polymeric matrix and the 321
solubility in food simulants. To have surrogates capable to migrate into aqueous test media, we 322
have investigated the behaviour of very hydrophilic compounds, namely triethylamine, 323
dodecylamine, 1,3-butanediol, DMSO, 2,4-pentanedione and phenol. Triethylamine, dodecylamine 324
and 1,3-butanediol were removed from our sets after testing for their reactivity (see below, they 325
graft on PET) and they do not appear in table 2. 326
14
327
Suitability of analytical methods: some surrogates are introduced in the list for specific methods, as 328
azobenzene, to visualize the functional barrier under a microscope. In another work, 2,4-329
dimethoxyacetophenone was used to monitor diffusion in molten polymers [23]. 330
331
Reactivity of surrogates: organic chemicals may be reactive at PET process temperatures. 332
Surrogates may undergo rearrangements, oxidation or grafting on polymer chains. Reactive 333
surrogates would be non detectable in migrates, for reasons other than food safety, and they must be 334
discarded. Hydroxyl and amine functional groups enhance the solubility of surrogates in water. 335
However they are also very reactive and may induce reactions with other surrogates or with the 336
solvent used for impregnation of flakes; at high temperatures, they may also undergo trans-337
esterification or ester-amide exchange reactions with PET, and this must be carefully checked. The 338
reactivity of surrogates was monitored without trying to identify reaction products, only by 339
checking whether the surrogates could be extracted after being submitted to the following 340
conditions: 341
- at 40°C, in solution in dichloromethane (without flakes), for 7 days. Dichloromethane is the 342
solvent used both for impregnation and for extraction of PET flakes. These experiments cover the 343
conditions used for both extraction of surrogates prior to their analytical determination (1.5 days at 344
40 °C) and for the impregnation step (5 days at 20 °C). 345
- at 150°C for 3 hours, simulating the drying 346
- at 280 °C (after drying 150°C for 3 hours) for 5 minutes in a sealed tube, for simulation of the 347
injection conditions (without allowing evaporation of the surrogates). 348
349
a) In dichloromethane at 40°C, all surrogates were stable over the 7 days, except triethylamine, 350
which slowly reacted with dichloromethane (50 % loss). It could therefore not be used as surrogate, 351
as it was incompatible with the use of dichloromethane as impregnation and extraction solvent. 352
15
353
b) Reactivity of surrogates in conditions simulating drying at 150°C followed by injection at 280°C: 354
Impregnated flakes were dried at 150°C for 3 hours and heated at 280°C for 5 minutes (2-3 times 355
longer than the industrial process). Surrogates eliminated because they were not recovered are not 356
in table 4. 357
358
Results are shown in table 4. In all cases, except for Uvitex OB, the concentrations of surrogates are 359
affected mainly by treatment at 150°C. The surrogates are either evaporated, or did undergo 360
degradation by reaction with PET or with another surrogate. 361
362
After heating at 280°C, the DMSO concentration (initial concentration at 20 °C in PET was 4900 363
ppm) was too low to be quantified. However, increasing its initial concentration in the solution (its 364
concentration in PET was 7600 ppm before drying) enabled the incorporation of 1363 ppm of 365
DMSO in monolayer bottles. Only then was its quantification possible in migrates, at the end of the 366
study (our target) (table 4). 367
368
With triethylamine and dodecylamine surrogates, PET became brown and none of these amines 369
could be extracted anymore. The colour did not disappear after extraction, suggesting that amines 370
are probably reacting with PET (grafting or amidation). Similarly, 1,3-butanediol was not recovered 371
after any of the process simulation experiments (esterification, transesterification). These 372
compounds were eliminated from the list of surrogates. 373
374
375
Some surrogates are selected to be used as microscopy UV probes, e.g. azobenzene and Uvitex, in 376
order to measure the thickness of the recycled layer by observation under UV light. After successive 377
16
trials, a target of 1000 mg ± 50% of surrogates /kg finished PET materials was selected for an easy 378
detection of the probe in simulants and bottles. 379
380
To obtain 1000 ppm surrogate in the finished PET article, allowing to monitor migration kinetics, 381
we had to start with much higher initial concentrations (IC in table 4) just after impregnation. We 382
stress again that these levels are well above any realistic level of pollution, but that it is needed to 383
be able to monitor migration kinetics. 384
385
This shows that the normal PET processing conditions already contribute significantly to clean the 386
material, and to remove pollutants, if they were present. 387
388
389
Distribution in the mass: for repeatability and reproducibility of analytical determinations, 390
surrogates should be uniformly distributed over the mass of the material. Moreover, if the aim is to 391
test the efficiency of industrial washing procedures (using aqueous solutions), it is essential to have 392
a distribution of surrogates in the core of the material. Otherwise, if surrogates are only adsorbed on 393
the surface, the efficiency of the washing steps might be overestimated [4, 28], which would reduce 394
the safety margin for risk assessment of consumer’s exposure. 395
396
To check that surrogates do not influence in an unrealistic manner the characteristics of the final 397
PET material studied, a viscosity control at injection temperature is necessary. Reactions (grafting 398
and/or degradation) and plasticization could affect this viscosity, which will also be investigated 399
through glass temperature controls. Orientation and cristallinity should be reasonably close to those 400
of commercial PET bottles. 401
402
403
17
404
II. impregnation of surrogates 405
For impregnation of PET by surrogates, PET flakes were placed in a solution of surrogates in a 406
solvent. The choice of the solvent strongly affects PET resin properties and leads to different 407
impregnation results. 408
409
Solvent effects: 410
Impregnation of surrogates is expected to be more reproducible when the kinetics reach a plateau. 411
The choice of the impregnation solvent is critical in this case. Impregnation kinetics were therefore 412
monitored with two solvents having completely different behaviours: heptane, which is considered 413
as chemically and physically inert [29], and dichloromethane, known for its remarkable swelling 414
and plasticizing effect on PET [30]. 415
Conditions (1): heptane has been used as a vehicle for surrogates at 40°C for 14 days, and uptake 416
kinetics of toluene and 1,1,1-trichloroethane by PET bottle walls pieces was monitored (figure 1). 417
Conditions (2): dichloromethane was chosen, in order to reduce the testing time. It was used at 418
ambient temperature. In order to have measurable kinetics, we monitored uptake of the highest 419
molecular weight surrogate, namely Uvitex OB, whose uptake rate is expected to be the slowest, 420
whatever the solvent (diffusion coefficients are more or less proportional to exp[M]). 421
Using heptane, impregnation is still continuing after 26 days, following apparent Fickian diffusion 422
kinetics (figure 1). In contrast, a plateau was reached with dichloromethane after 2.5 days for the 423
highest molecular weight surrogate: Uvitex OB (figure 1). The data compiled in figure 1 are of 424
different experimental design. No kinetic conclusion is drawn, even though lines are drawn for the 425
kinetics. However there is such a difference between the behaviour of heptane and that of 426
dichloromethane that it can be concluded that use of dichloromethane is as a very fast means to 427
impregnate PET. 428
429
18
Dichloromethane thus appears as a good candidate for large scale treatments of flakes. The effect of 430
dichloromethane on processability was also assessed, using Melt Flow Indexes (MFI). MFIs of both 431
type of virgin flakes and of flakes swollen with dichloromethane, and then dried (150 °C, 3 h) were 432
compared. No significant difference could be observed between the two types of PET (0.75 dl/g). 433
434
It thus appeared that despite the strong physical modifications of the PET structure induced by 435
dichloromethane during impregnation, drying of the flakes removed most of the solvent. TGA data 436
(not shown here) demonstrate that the level of volatiles (water and dichloromethane) was below 1% 437
after this first treatment. A second identical treatment was applied just before injection to remove 438
residual water, as usual in PET processing. The concentration of volatiles was not measured after 439
this second treatment, but from the yields of evaporation (table 4) of other volatile compounds, the 440
residual concentration of dichloromethane is expected to be very low (probably even lower than the 441
other surrogates which are less volatile). 442
443
Temperature effect on impregnation: 444
Since impregnation tests often have be run on a large scale (when a process has to be assessed), it is 445
important to know whether the procedure requires a strict temperature control. Therefore, 446
impregnation of PET flakes by dichloromethane was run at 11, 17 and 23 °C for set (B) surrogates, 447
for 5 days (conditions corresponding to figure 1). No sensible difference was observed in the 448
concentrations of the surrogates in PET flakes at the different temperatures, even for Uvitex OB 449
(the highest molecular weight, corresponding to the slowest sorption kinetics). The impregnation 450
procedure using dichloromethane (5 days, till the plateau) can thus be run at room temperature, and 451
is not sensitive to temperature fluctuations. 452
453
The use of dichloromethane as vehicle for surrogates impregnation at room temperature, allowed 454
reaching a plateau within 2.5 days at 20°C for the sorption of Uvitex OB which has the highest 455
19
molecular weight corresponding to the lowest diffusion coefficient (figure 1). Then, in all 456
experiments reported below, the PET flakes were incubated in the solution of surrogates for 5 days. 457
458
459
III Influence of high temperature drying on residual levels of surrogates 460
461
Typically, PET is dried in any industrial processes for 6 to 8 hours at 150 –180 °C under air 462
circulation. In this section, we investigate about possible evaporation of pollutants during drying. 463
Impregnated flakes were dried in an oven at 150 °C for 3 hours to remove dichloromethane and 464
were then sent to the industrial converter in tightly closed barrels. Once on the plant, a second 465
drying step (150 °C, 3h) was applied just before preform injection. The effect of drying time on 466
surrogates concentrations was monitored (example of set B is given on figure 2). Except Uvitex, a 467
strong initial decrease of the concentrations of all surrogates by a factor 3 [DMSO, chlorobenzene, 468
phenol, chlorooctane] to 8 [2,4-pentanedione, toluene] was observed, mainly during the first 3 hours 469
at 150 °C. These results show that drying contributes considerably to the removal of volatile 470
pollutants from PET, and, finally, to consumers safety. 471
472
473
IV Characterisation of bottles impregnated with surrogates 474
Two types of bottles and of preforms were processed: monolayers (made from 100% impregnated 475
PET) and tri-layers (with 25% impregnated PET). 476
477
Determination of surrogate’s distribution in the bottles: 478
The concentrations of surrogates in monolayer bottles (table 4) effectively correspond to those 479
targeted (1000 ppm ± 50%). Exceptions are 1,1,1-trichloroethane, benzophenone and phenol, which 480
exceeded 2500 ppm. Losses during processing were lower than in model experiments. Our test in 481
20
the sealed tube at 280°C was probably more stringent than reality (5 minutes instead of around 2 482
min). Surrogates were indeed lost by reactivity and by evaporation, but losses were less important 483
than expected from the preliminary model experiments. 484
485
In tri-layer bottles, the concentrations of the surrogates were determined in the neck, in the bottom 486
and at four places in the wall, identified by their distance X (cm) to the bottom (figure 3). Necks do 487
not contain any surrogate. This arises from the fact that in commercial bottles, the manufacturer 488
does not introduce any recycled PET in the neck, as this part of the bottle could be in contact with 489
mouth of consumers. In the bottom of bottles, concentrations of surrogates are lower than in the 490
body because at the end of injection, virgin PET is injected to compensate for thermal retract during 491
cooling. 492
493
Along the walls of the bottles, the concentrations were not homogeneous showing that the thickness 494
of the virgin and impregnated layers is not constant. However, the average value of the four points 495
of measurement in the wall (X=3, 10.5, 18 and 25.5 cm) was equal to 25% of the value of 496
concentrations in the monolayer bottle. This corresponded indeed to the ratio of impregnated/virgin 497
PET in tri-layer bottles. 498
499
Physical characterisation of bottles 500
Samples, taken in the wall of our model bottles (between X=3 and 25.5 cm), were compared to 501
commercial bottles by MDSC. Results are shown in figures 4 and 5. The glass temperature of test 502
bottles (70°C) was lower than that of commercial bottles (80°C). These differences also appeared in 503
reversible and irreversible heat flow thermograms. Transition temperatures of model bottles were 504
lower, showing a plasticization by the surrogates. Since diffusion is accelerated by plasticization 505
effects, these characteristics of model bottles should lead to slightly overestimated migration and 506
21
will then enable evaluation of consumer’s exposure with an additional safety margin for public 507
health protection. 508
509
Tri-layer bottles and preforms were cut into slices (by transversal microtomy) and their cross 510
sections were analysed by visible light microscopy (to observe azobenzene) and fluorescence (for 511
Uvitex). The thickness of the layer was not constant, which was quantitatively consistent with 512
results shown in figure 5. No clear diffusion gradient could be observed for azobenzene and Uvitex 513
OB, then it seemed that limited diffusion of those surrogates occurred during the preform and bottle 514
manufacturing steps. Moreover, it can be concluded that little or no blending of virgin and recycled 515
PET occurs during preform injection. The impregnated layer was not in the middle of the bottle 516
wall. However, the thickness of the functional barrier in bottles was ranging from 60 to 75 µm for 517
these two dying substances. 518
519
IV Migration testing. 520
521
Monolayer bottles 522
Figures 6 and 7 show migration kinetics from monolayer bottles into ethanol and into 3 % acetic 523
acid. The migration rate of surrogates from monolayers is function of the initial pollutant 524
concentration in the polymer, of their diffusion coefficient in PET and of their partition coefficient 525
between packaging and simulant. In a first approximation, to compare the results, we can assume 526
that all initial concentrations in PET (500 – 1500 ppm) are in the same order of magnitude. Partition 527
effects are not expected to play an important role with ethanol, as (i) ethanol is a good solvent of all 528
the surrogates studied and (ii) the large [simulant / packaging] volume ratio favours the transfer into 529
the simulant. As a consequence, migration into ethanol is expected to be mainly dependent on the 530
diffusivity of the surrogates, which itself is known to be more or less a decreasing function of the 531
molar volume [6, 31, 32]. This is indeed the case, as we have the sequence: 532
22
In ethanol (figure 7): Phenol (94) > DMSO (78) ≃ pentanedione (100) > chlorobenzene (113) > 533
toluene (92) ≃ trichloroethane (133) > benzophenone (182) > chlorooctane (149) > nonane (128) 534
≃ ethyl hydrocinnamate (178) > phenyl benzoate (198) > phenylcyclohexane (160) > dibutyl 535
phthalate (278 g/mole). 536
537
The migration behaviour is different in 3 % acetic acid, partition obviously playing an important 538
role, besides diffusivity. Consequently, the largest migration values are obtained for molecules 539
which are both fast diffusing and water soluble. 540
541
In 3 % acetic acid (figure 6): Phenol (94) >> pentanedione (100) > chlorobenzene (113) > toluene 542
(92) > benzophenone > trichloroethane > ethyl hydrocinnamate (178) > nonane (128), chlorooctane 543
(149), phenylcyclohexane (160), phenyl benzoate (198) and dibutyl phthalate (278 g/mole). 544
545
There seems to be an influence of molecular weight, as the lower molecular weight compounds are 546
most likely more soluble in aqueous media. Thus, the hierarchy between migration of phenol, 2,4-547
pentanedione, chlorobenzene and toluene is the same in the case of a contact with ethanol and of a 548
contact with 3 % acetic acid. For higher molecular weight compounds (between 130 and 200 549
g/mol), migration behaviour is totally different in both simulants: polar molecules lead to larger 550
migrations, and partition plays a major role. Despite a 182 g/mol molecular weight, benzophenone 551
migrates more than trichloroethane (133 g/mol), as its solubility is likely to be larger in 3 % acetic 552
acid (the simulant) than in distilled water (see table 2). In the same way, ethyl hydrocinnamate (178 553
g/mol) is detected in 3% acetic acid after 500 days, while the apolar nonane (128 g/mol), 554
chlorooctane (149 g/mol) and phenylcyclohexane (160 g/mol) are not detected. Surrogates with 555
molecular weight above 200 g/mol are not detected, even after 300 days of contact, because both 556
their solubility in water and their diffusivity in PET are too low. 557
558
23
DMSO was not recorded for analytical reasons (easy to analyse in pure water by several GC or LC 559
experimental methods; it was not recovered in the presence of 3 % acetic acid). 560
561
Tri-layer bottles 562
Figures 8 and 9 show migration behaviour of tri-layer bottles in contact with 3 % acetic acid and 563
ethanol. As expected from results above, only fast and water soluble molecules migrate into the 564
aqueous simulant. Surrogates are detected in the aqueous simulant only after 6 months of contact. If 565
we now try to foresee the possible migration of real pollutants, recent studies have shown that the 566
highest realistic levels of pollution of PET flakes are in the 1-3 ppm range [33, 34]. Since migration 567
is proportional to the initial concentration of the pollutants in PET, a worst case realistic migration 568
can be extrapolated from the data in figure 8: with an initial concentration of a pollutant in PET 569
around 1500 ppm, any migration at t=625 days (t0.5= 25 days0.5) is always below 0.016 ppm. With a 570
3 ppm concentration, the migration would be 500 times lower (i.e. 0.32 ppb), which is well below 571
any threshold of concern. With ethanol (figure 9), the largest measured migration at t=6 months is 572
0.5 ppm, which corresponds to 1 ppb migration for a concentration in PET of 3 ppm. Consumer 573
protection from accidental contamination is therefore also at least 6 months with ethanol, and 2 574
years with aqueous acidic beverages. 575
576
Comparing migration of tri-layers and monolayers, a remarkable result is that the hierarchy between 577
surrogates migration rates is different: this is due to the fact that during the lag phase, only diffusion 578
plays a role (diffusion from the inner layer through the barrier), while migration (which occurs after 579
the lag phase) is also connected to packaging / food partition effects, as for monolayers. 580
In 3% acetic acid (figure 8): pentanedione (100) > chlorobenzene (113) > Phenol (94) > toluene 581
(92) >> chlorooctane (149 g/mole). 582
In ethanol (figure 9): pentanedione (100) > chlorobenzene (113) ≃ Phenol (94) > toluene (92) > 583
DMSO >> chlorooctane (149 g/mole). 584
585
24
Higher molecular weight lipophilic surrogates, which are both slow and insoluble in the aqueous 586
test medium, do not migrate in detectable levels. 587
588
This confirms that the solubility of the surrogates in the simulant may play an important role, 589
mainly for monolayer materials. 590
591
Influence of ethanol 592
Migration into ethanol is always much larger than into 3 % acetic acid. This can be attributed to its 593
good solvent properties. However this does not explain all results. Even surrogates which are well 594
soluble in water migrate more and faster in ethanol than into 3 % acetic acid (compare figures 6 and 595
7, 8 and 9). With multilayers, the lag times are always much shorter with ethanol (around 80 days 596
for pentanedione, chloroethane and toluene) than in 3 % acetic acid (around 170 days for these 597
surrogates). Since lag times are influenced by diffusion coefficients, this shows that ethanol 598
plasticizes PET by swelling effects. Swelling effects strongly contribute to the larger migration into 599
ethanol observed both with mono and with multilayers. 600
601
602
603
604
Conclusion. 605
606
Monolayer and tri-layer bottles containing large concentrations of surrogates were obtained. 607
Dichloromethane, a strongly swelling solvent was used as a vehicle for surrogates impregnation 608
into PET. It enables a quick impregnation (≤ 5 days) at room temperature, with a homogeneous 609
concentration in the mass of the flakes. The impregnation levels were not sensitive to temperature 610
25
fluctuations. Drying of these flakes enabled evaporation of dichloromethane and of water prior to 611
injection. 612
613
Surrogates have been chosen with different criteria in order to cover all possible properties of real 614
potential pollutants: volatility, polarity, functionality, stability and solubility in the simulant. 615
Surrogates with amine and hydroxyl groups were tested, since these functional groups tend to 616
increase the solubility in water. Unfortunately, these polar functional groups also react with 617
dichloromethane during impregnation of flakes or with PET in process conditions. Only DMSO, 618
phenol and 2,4-pentanedione were suitable as surrogates highly soluble in water. 619
620
A 1000 ppm concentration was a good compromise between the requirement of a good analytical 621
sensitivity for the migration test and the need not to alter the polymeric matrix properties. To obtain 622
such concentrations of surrogates in finished materials, there was the need to pollute at high levels. 623
60 to over than 90 % of the surrogates were eliminated after processing, which shows that the 624
normal PET handling already cleans the resin, and contributes to safer materials. 625
626
Bottles processed from impregnated PET flakes had physical properties which were reasonably 627
close to those of commercial PET bottles, and the procedure can be considered as acceptable. Due 628
to plasticization by the surrogates, diffusion phenomena should be faster in model bottles than in 629
commercial bottles, which provides again a safety margin to evaluate consumer exposure. 630
631
The surrogates can be classified into groups with different migration behaviour, defined by 632
molecular weight ranges. Besides migration properties, these classes correspond also roughly to 633
volatility scales, and also to the ability of any surrogate to be possibly (M<130g/mol) or not easily 634
(M>130g/mol) removed by a suitable recycling process. Migration kinetics of polluted bottles (with 635
26
and without functional barrier) in contact with 3 % acetic acid and with ethanol at 40 °C show the 636
following: 637
- only low molecular weight compounds (<130 g/mol) migrate into 3 % acetic acid after 1 638
year of contact when a functional barrier is present 639
- before 6 months of contact, any migration is close to zero 640
- only M < 200 g/mol compounds could migrate into 3 % acetic acid, even for large (at least 641
1.5 year) contact times at 40 °C 642
The migration of higher molecular weight lipophilic surrogates is not detectable (most detection 643
limits are below 10 ppb, which illustrates the need to use surrogates soluble in water. As worst case 644
substances, 2,4-pentanedione and phenol should be incorporated in the classical lists of surrogates; 645
646
The sequence of migration of surrogates depends on their diffusion properties and on interface 647
partition effects. Solubility of surrogates in the simulant plays an important role, specially with 648
monolayer materials: surrogates which are insoluble in 3 % acetic acid do not migrate in this 649
medium. With multi-layers, the differentiation of the surrogate occurs during the lag phase, where 650
the surrogates cross the functional barrier, and surrogates reach the surface in the sequence of 651
fastest diffusion behaviour. 652
653
Acknowledgements 654
655
The experimental part of this work was financially supported by ECO-EMBALLAGES, ADEME, 656
INRA and the Région Champagne Ardenne. PYP acknowledges them for a PhD grant. Special 657
thanks to M. B. Vincent, Amcor PET Packaging Recycling France (Dunkerque, France), for 658
manufacturing bottles. 659
27
Bibliography 660
661
1 - Plastics Recycling Task Force, 1995, National Food Processors Association, March, Guidelines
for the safe use of recycled plastics for food packaging applications. The Society of the Plastics
Industry, Inc. 1995
2 - BAYER-FL, 2002, Polyethylene terephthalate recycling for food-contact applications: testing,
safety and technologies: a global perspective, Food Additives And Contaminants. 19, 111-134,
3 - BRANDRUP J., BITTNER M., MICHAELI W., MENGES G., 1996, Recycling and recovery
of plastics. Hanser/Gardner publications.
4 - NIELSEN T., DAMANT A.P., CASTLE L., 1997, Validation studies of a quick test for
predicting the sorption and washing properties of refillable plastic bottles. Food Additives and
Contaminants, 14, N 6-7, 685-693
5 - BAYER F.L., 1997, The threshold of regulation and its application to indirect food additive
contaminants in recycled plastics. Food Additives and Contaminants, 14, N 6-7, 661-670
6 - FEIGENBAUM A., LAOUBI S., VERGNAUD J.M., 1997, Kinetics of diffusion of a pollutant
from a recycled polymer through a functionnal barrier: recycling plastics for food packaging.
Journal of Applied Polymer Science, 66, 597-607
7 - PIRINGER O., FRANZ R., HUBER M., BEGLEY T.H., MCNEAL T.P., 1998, Migration from
food packaging containing a functional barrier: Mathematical and experimental evaluation, Journal
Agricultural and Food Chemistry, 46, 1532-1538. This paper summarises earlier work of the
authors.
8 - FRANZ R., 2002, Ensuring food safety in PET recycling, Edinburgh, Scotland, PIRA 5th
December 2002
9 - Food and Drug Administration, 1992, Points to consider for the use of recycled plastics: Food
packaging, chemistry considerations. FDA Division of Food Chemistry and Technology
publication. HP 410, Washington, May 1992
28
10 - MILTZ J., RAM A., NIR M.M., 1997, Prospects for application of post-consumer used
plastics in food packaging. Food Additives and Contaminants, 14, N 6-7, 649-659. 1997
11 - KOMOLPRASERT V., LAWSON A.R., BEGLEY, T.H., 1997, Migration of residual
contaminants from secondary recycled PET into aqueous ethanol and heptane food simulating
solvents, Food additives and Contaminants, 1997, 14, 491-498
12 - LORIOT C., 12-14 Mars 1997, Evaluation de la qualité et de la sécurité des matériaux
plastiques réutilisés et recyclés destinés au contact alimentaire - Programme AIR2-CT93-1014.
Journées Internationales de l'Emballage, Reims
13 - SIMAL-GANDARA J., SARRIA-VIDAL M., KOOREVAAR A., RIJK R., 2000, Tests of
potential functional barriers for laminated multilayer food packages. Part I: Low molecular weight
permeants, Food Additives and Contaminants, 17 (8) 703-711
14 - SIMAL-GANDARA J., SARRIA-VIDAL M., RIJK R., 2000, Tests of potential functional
barriers for laminated multilayer food packages. Part II: Medium molecular weight permeants,
Food Additives and Contaminants, SEP; 17 (9): 815-819
15 - LAOUBI S., VERGNAUD J.M., 1995, « Process of contaminant transfer through a food
package made of a recycled film and a functional barrier », Packaging Technology and Science, 8,
97-110
16 - PEROU, A. L.; LAOUBI, S.; VERGNAUD, J. M., 1999, Model for transfer of contaminant
during the coextrusion of three-layer food package with a recycled polymer. Effect on the time of
protection of the food of the relative thicknesses of the layers. Journal of Applied Polymer Science
73 (10), 1939-1948
17 - BEGLEY T.H., HOLLIFIELD H.C.: Recycled polymers in Food packaging: migration
considerations. Food Technology, Nov. 1993, 109-112
18 - PENNARUN, P.Y., DOLE, P., FEIGENBAUM A., Functional barriers in PET recycled
bottles. Part I: determination of diffusion coefficients in bi-oriented PET with and without contact
with liquids, Journal of Applied Polymer Science, 2004, in press
29
19 – PENNARUN, P.Y., DOLE, P., FEIGENBAUM A.: Overestimated diffusion coefficients for
the prediction of worst case migration from PET: application to recycled PET and to functional
barriers assessment, 2004, submitted
20 - SUN D.C., MAGILL J.H., 1989, Thermal interactions in oriented polymeric materials:
Shrinkage, crystallization, and melting. Polymer Engineering and Science, 29, N 21, 1503-1510.
21 - BEGLEY T. H., HOLLIFIELD, H. C., High Performance Liquid Chromatographic
Determination of migrating polyethylene terephthalate oligomers in corn oil, Food Additives and
Contaminants, 1990, 7, 339-346
22 - FEIGENBAUM A., DOLE P., PASEIRO P., PAPASPYRIDES C., HANKEMAIER T.,
DAINELLI D., VOULZATIS Y., VITRAC O., SAILLARD P., REYNIER A., PENNARUN P.,Y.,
PAVLIDOU S., PASTORELLI S., N’GONO Y., DE LA CRUZ C., AUCEJO S.: Recycling Plastics
with Functional Barriers, Polymer Processing Society, Athens, September 17, 2003
23 –PENNARUN P.Y., NGONO Y., DOLE P., FEIGENBAUM A., 2004, Functional barriers in
PET recycled bottles. Part II: diffusion of pollutants during processing, Journal of Applied Polymer
Science, in press
24 - MONTGOMERY J.H., WELKOM L.M., 1991, Groundwater Chemicals Desk Reference,
Lewis Publishers, INC., New York 1991
25 - WEAST R.C., ASTLE M.J., BEYER W.H., 1986, CRC Handbook of Chemistry and Physics.
CRC Press. 1986-1987
26 - BUDAVARI S., O'NEIL M.J., SMITH A., HECKELMAN P.E. AND KINNEARY J.F., 1996,
The Merck Index, Twelfth Edition. Merck Research Laboratories, Division of Merck & Co., Inc.
27 - RASCHIG A.G., Technical data of BHT. Sparte Chemie, Mundenheimer Str. 100, D-67061
Ludwigshafen.
28 - DEVLIEGHERE F., DE MEULENAER B., SEKITOLEKO P., ESTRELLA A.A. GARCIA,
HUYGHEBAERT A., 1997, Evaluation, modelling and optimization of the cleaning process of
contaminated plastic food refillables. Food Additives and Contaminants, 14, N 6-7, 671-683
30
29 - FRANZ R., HUBER M., PIRINGER O.-G., DAMANT A.P., JICKELLS S., CASTLE L.,
1996, Study of functional barrier properties of multilayer recycled PET bottles for soft drinks.
Journal of Agricultural and Food Chemistry, 44, 892-897
30 - KWEI T.K., ZUPKO H.M., 1969, Diffusion in glassy polymers. International Journal of
Polymer Science: part A2, 7, 867-877.
31 - REYNIER A., DOLE P., HUMBEL S. FEIGENBAUM A., 2001, Diffusion coefficients of
additives in polymers. I. Correlation with geometric parameters. Journal of Applied Polymer
Science 82 (10), 2422-2433.
32 - BANER, A., BRANDSCH, J., FRANZ, R., PIRINGER, O., 1996, The application of a
predictive migration model for evaluating the compliance of plastic materials with European food
regulations, Food Additives and Contaminants, 13 (5), 587-601
33 - FRANZ R., 2002, Programme on the recyclability of food-packaging materials with respect to
food safety considerations: polyethylene terephthalate (PET), paper and board, and plastics
covered by functional barriers Food Additives and Contaminants, 19 (Suppl) 93-110
34 - FRANZ R., WELLE-F, 2002, Recycled poly(ethylene terephthalate) for direct food contact
applications: challenge test of an inline recycling process, Food Additives and Contaminants, 19
(5) 502-511
662
31
663 664
665 Tables. 666
667
668
669
670
671
672
Surrogates Properties References
Chloroform Polar, volatile [4]
1,1,1-Trichloroethane Polar, volatile [4, 9]
Benzophenone Polar, non volatile [4, 9]
Phenyldecane - [4]
Toluene Non polar, volatile [9, 4, 10, 12]
Benzyl alcool Polar [10]
Phenylcyclohexane Non polar, non volatile [9, 12]
Diazinon [4, 12]
Chlorobenzene Polar, volatile [9, 12]
Phenol Polar, volatile [9]
Iso-amyl acetate - [9]
Cyclohexanone - [9]
n-hexanol - [9]
Propylene glycol Polar, non volatile, water
soluble
[9, 10]
673
674
675
676
677
Table 1: Surrogates most frequently used in literature for PET impregnation 678
32
Surrogates (set) Set Properties (Volatility, Polarity, Mass, Boiling temp °C) Solubility in water
1,1,1-Trichloroethane (A) V, P, M=133 g/mol, BP=75°C < 1000 ppm (20°C) [24]
Dimethyl sulfoxyde (DMSO) (A) NV,P, M=78 g/mol, BP=189°C Very soluble [25]
Methyl palmitate (A) NV, NP., M=270 g/mol, BP= N.c. Not soluble [25]
Benzophenone (A) NV, P, M=182 g/mol, BP=305°C < 1000 ppm (20°C) [26]
Phenylcyclohexane (A) NV, NP, M=160 g/mol, BP=240°C Not soluble [25]
Ethyl hydrocinnamate (A) MV, NP, M=178 g/mol, BP=247°C Not soluble [24]
Phenol (B) NV,P, M=94 g/mol, BP=182°C 93 g/l (25°C) [24]
2,6-Di-tert-butyl-p-cresol (BHT) (B) NV, MP, M=220 g/mol, BP=265°C < 0.01 g/L (20°C) [27]
Chlorobenzene (B) V, P, M=113 g/mol, BP=131°C 500 ppm (20°C) [24]
2,5-Thiophenediylbis(5-tert-butyl-1,3-
benzoxazole) (Uvitex OB)
(B) Fluorescent dye, M=431 g/mol n.d.
1- Chlorooctane (B) MV, NP, M=149 g/mol, BP=183°C n.d.
2,4-Pentanedione (C) NV, P, M=100 g/mol, BP= 133°C >10g/L [26]
Azobenzene (C) Dye, M=182 g/mol ---
Nonane (C) NV, P, M=128 g/mol, BP= 151°C Not soluble
Dibutyl phthalate (C) NV, MP, M=278 g/mol, BP=340°C < 400 mg/L (25°C) [24]
Phenyl benzoate (C) NV, MP, M=198 g/mol, BP=299°C n. d.
Toluene (C) V, NP, M=92 g/mol, BP=110°C 515 ppm (20°C) [24] V: Volatile, NV: Not Volatile, MV: Medium Volatile, P: Polar, NP: Not Polar, MP: Medium Polar, N d.: Not determined.
Table 2: Surrogates used in the current study
33
Weight of chemicals (g) for simulation of Name of series A surrogates [3h at 150°C+5 min at 280°C] Drying at 150°C Bottle processing
1,1,1-Trichloroethane A 5.03 5.30 1658 DMSO A 1.80 4.01 1616 Methyl Palmitate A 0.51 2.62 888 Benzophenone A 0.59 3.04 935 Phenylcyclohexane A 2.05 2.1 679 Ethyl Hydrocinnamate A 0.50 1.02 485 Dichloromethane A 98.0 100.00 32300 PET A - - 10000 Phenol B 0.74 0.84 289 BHT B 0.99 1.36 463 Chlorobenzene B 4.13 4.22 1778 Uvitex B 0.70 0.6 206 1-Chlorooctane B 1.81 4.66 1582 Dichloromethane B 98.0 100.00 34300 PET B - - 10000 2,4-Pentanedione C 5.01 5.58 1805 Azobenzene C 0.76 0.99 323 Nonane C 2.02 4.09 1613 DBP C 0.52 1.59 516 Phenyl Benzoate C 0.51 1.05 340 Toluene C 4.04 5.06 1631 Dichloromethane C 98.0 100.00 32200 PET C - - 10000
Table 3: Quantities of chemicals (surrogate, solvent and PET) used to impregnate the PET for different tests.
34
Drying at 150°C for 3 hours Drying (150°C/3h) + 5 min at 280°C
Surrogate Initial
concentration
(IC) (ppm)
Concentration
(ppm) Recovery %
versus IC Concentration
(ppm) Recovery % versus
IC
Conc. in flakes
after impregnation
(ppm)
Concentrations in
monolayer bottles
(ppm)
1,1,1-Trichloroethane 7648 2105 28 745 10 6840 2 690 DMSO 4911 702 14 < 70 - 7579 1 363 Methyl Palmitate 564 212 38 164 29 3517 704 Benzophenone 819 165 20 155 19 6911 2910 Phenylcyclohexane 4448 1403 32 1033 23 2972 1 285 Ethyl Hydrocinnamate 1064 527 50 436 41 3470 587 Phenol 7361 1020 14 796 11 8263 2616 BHT 1543 892 58 597 39 1908 872 Chlorobenzene 9337 596 6 368 4 7754 1324 Uvitex OB 1090 1176 108 1016 93 591 570 1-Chlorooctane 2952 555 19 323 11 4355 1552 2,4-Pentanedione 8079 556 7 678 8 6819 785 Azobenzene 1378 440 32 536 39 2455 921 Nonane 1631 171 10 148 9 2804 624 DBP 462 205 44 278 60 1070 533 Phenyl Benzoate 1051 442 42 408 39 1782 810 Toluene 5751 340 6 452 8 6931 704
Table 4: Concentrations at the different simulating steps of the process: initial concentration after impregnation (IC), after drying at 150°C for 3 hours, and after drying at 150°C for 3 hours and 5 min at 280°C. Concentration of surrogates in PET during bottle manufacture: concentrations of surrogates in PET flakes after impregnation and in bottles after processing (standard deviation less than 8%)
35
List of figures Figure 1: Impregnation kinetics at 40°C of PET flakes (1 cm x 1 cm x 280 µm) by toluene (13.3% w/w) ( ) and 1,1,1-trichloroethane
(20.5% w/w) ( ) in solution in heptane, and by Uvitex OB (0.54% w/w) in dichloromethane ( ).
Figure 2 : Evolution of serie B surrogates in function of drying time at 150°C: chlorobenzene , chlorooctane , BHT - - ∆
- -, phenol , uvitex OB
Figure 3 : Serie C surrogates concentrations in function of their location in the bottle (x cm is the distance from the bottom B):
toluene , 2,4-pentanedione , azobenzene , phenylbenzoate - - - - , DBP - - ◊- - , nonane .
Figure 4: Mono (100%) and tri-layer (25%) bottle reversible heat flow for series A ( ); comparison with a commercial bottle ( ).
Figure 5: Mono (100%) and tri-layer (25%) bottle non reversible heat flow for series A ( ); comparison with a commercial bottle ( ).
Figure 6: Migration kinetics of monolayer bottles (series A, B C) into 3 % acetic acid. Straight lines suggest apparent fickian behaviour.
Toluene , Pentanedione , Nonane , Chlorobenzene , Chlorooctane , Phenol , DMSO +, Phenylcyclohexane -, Ethyl hydrocinnamate , Methyl palmitate -, Benzophenone , Azobenzene , Phenyl benzoate , DBP , Trichloroethane . Detection limits are below 15 ppb.
Figure 7: Migration kinetics of monolayer bottles (series A, B and C) into ethanol. Straight lines are put to illustrate apparent fickian
behaviour. Same symbols as in Figure 6. Detection limits are below 120 ppb (except trichloroethane, 400 ppb).
Figure 8: Migration kinetics from tri-layer bottles into 3 % acetic acid (series A, B and C). Same symbols as in figure 6. Detection limits
below 120 ppb
Figure 9: Migration kinetics from tri-layer bottles into ethanol (series A, B and C). Same symbols as in figure 6. Detection limits below
120 ppb (except trichloroethane 400 ppb)
36
0
200
400
600
800
1000
1200
0 5 10 15 20 25
Square root of time (hours1/2)Con
cent
ratio
ns o
f sur
roga
tes
in P
ET
(ppm
).t = 26 daysConc = 1000 ppm
Figure 1
37
100
1000
10000
0 5 10 15 20 25Drying time (hours)
Con
cent
ratio
n in
PET
(ppm
)
Figure 2
38
-
50
100
150
200
250
300
Neck B+25.5 B+18 B+10.5 B+3 Bottom
Con
cent
ratio
ns o
f sur
roga
tes
in P
ET (p
pm)
Figure 3
39
Figure 4
40
Figure 5
41
-
0,50
1,00
1,50
2,00
2,50
3,00
3,50
4,00
- 5,00 10,00 15,00 20,00 25,00
Square root of time (days)
Con
cent
ratio
n (m
g/l)
Figure 6
42
0
2
4
6
8
10
12
14
0 5 10 15 20 25
Square root of time (days)
Concentration (mg/L)
Figure 7
43
0
0,02
0,04
0,06
0,08
0,1
0,12
0,14
0,16
- 5,00 10,00 15,00 20,00 25,00
Square root of time (days)
Con
cent
ratio
n (m
g/l)
ChlorobenzenePhenolToluenePentanedione
Figure 8
44
-
0,50
1,00
1,50
2,00
2,50
3,00
3,50
- 5,00 10,00 15,00 20,00 25,00
Square root of time (days)
Con
cent
ratio
n (m
g/l)
ChlorobenzeneChlorooctanePhenolToluenePentanedioneDMSO
Figure 9