Upload
others
View
0
Download
0
Embed Size (px)
Citation preview
AUTHOR QUERIES Journal id: TFAC_A_236020 Corresponding author: V. DUCRUET Title: Sorption of aroma compounds in PET and PVC during the storage of a strawberry syrup Dear Author Please address all the numbered queries on this page which are clearly identified on the proof for your convenience. Thank you for your cooperation
Query number Query
PLEASE PROVIDE EDITABLE FORMAT ARTWORK FOR ALL THE FIGURES
1 Please specify affiliation number “5” in author group 2 Please provide figure captions for all the figures 3 Please provide Publisher and town for Brandrup J, Immergut EH, Grulke
A. 1989. 4 Please provide citation for Nahon DF, Navarro y Koren PA, Roozen
JP, Posthumus MA. 1998. 5 Please provide citation for Van Willige RW, Linssen JPH, Meinders
MBJ, Van der Stege H J, Voragen AGJ. 2002b. 6 Please provide Van Willige 2002a in the reference list 7 Please signify ** in table 2
New XML Template (2007) [1.6.2007–5:57pm] [1–12]{TANDF_FPP}Tfac/TFAC_A_236020.3d (TFAC) [First Proof]
Food Additives and Contaminants, Month? 2007; ??(?): 1–12
Sorption of aroma compounds in PET and PVC during the storageof a strawberry syrup
5 V. DUCRUET1, O. VITRAC2, P. SAILLARD3, E. GUICHARD4, A. FEIGENBAUM4,& N. FOURNIER4
1INRA, Unité Mixte de Recherche SCALE and 2INRA Unité Mixte de Recherche GENIAL, 1 avenue des
Olympiades, F-91744 Massy CEDEX, France, 3INRA Laboratoire de Nutrition et Sécurité Alimentaire, F-78352
Jouy en Josas CEDEX, France, 4INRA, Unité Mixte de Recherche FLAVIC, 17 rue Sully, BV 86510, F-21065 Dijon10 CEDEX, France, and 5INRA, Unité Mixte de Recherche FARE, Moulin de la Housse, F-51687 Reims, France
(Received 11 September 2006; revised 4 March 2007; accepted 23 March 2007)
AbstractThe sorption of 14 aroma compounds into PET and PVC was monitored during storage of a strawberry syrup for 1 year.Concentrations in the syrup and in the polymer were determined during storage and compared with previously published
15 results obtained with glass bottles. Apparent partition coefficients between the polymer and the syrup (noted Kapp) wereestimated from experimental kinetics without reaching equilibrium Kapp values and optimally identified from the kineticdata obtained between 30 and 90 days. They exhibited a similar behaviour for both polymers with values were between2� 10�5 and 2� 10�3, 4� 10�5 and 3� 10�2, respectively, for PET and PVC. The variation of Kapp values in PET wasmainly correlated to the polarity of tested compounds as assessed by their log P values. By contrast, the variations in Kapp
20 values for PVC were mainly related to their chain lengths. Due to slightly higher partition coefficients and diffusioncoefficients in PVC compared with PET, the amount of absorbed aroma was four times higher in PVC than in PET;however, the amount of absorbed aroma compounds was less than 0.1% of the initial amount present into the syrup, exceptfor octyl butanoate. The variation in concentration in the syrup was interpreted as a combination of a degradation processand a transport process into the packaging material. Both effects were particularly noticeable for both PET and unstable
25 aroma compounds.
Keywords: Aroma, PET, PVC, packaging, sorption, strawberry, syrup
Introduction
The absorption of aroma compounds by plasticmaterials may induce both a weakening of flavour
30 and changes to the organic profile of packaged foodproducts. This has mainly been reported for aqueousfood, such as fruit juices, or model solutions, wherethe partition coefficient of aromas between food andpackaging strongly favours sorption into the plastic
35 material (Kwapong and Hotchkiss 1987; Sizer et al.1988). However, there are many discrepancies inreported partition coefficient values (Kutty et al.1994; Hernandez-Munoz et al. 2001; Tehranyand Desobry 2004). The possible reasons
40 include: (1) sorption kinetic studies performedwithout reaching the thermodynamic equilibrium(Gavara et al. 1996; Lebossé et al. 1997;
Feigenbaum et al. 1998); and (2) aroma usedbeyond their concentrations limits, which yield to
45deviations from Henry sorption equilibrium.Besides, partitioning results obtained with modelsolutions are difficult to extrapolate to the behaviourof aroma compounds in real foods, which aregenerally multi-constituent and possibly multi-
50phase. Recent papers have shown that in complexsolutions simulating food matrices the partitioncoefficients between aroma compounds and poly-mers also depend on the interactions with macro-molecules (lipids, polyoligosides or proteins).
55In particular, lipids compete with polymers concern-ing the transfer of aroma compounds into polyolefins(Van Willige et al. 2000a, 2000b). Moreover, duringlong periods of storage of up to 1 year, changes in
Correspondence: V. Ducruet. E-mail: [email protected]
ISSN 0265–203X print/ISSN 1464–5122 online � 2007 Taylor & FrancisDOI: 10.1080/02652030701361283
proofreaderText Box1
New XML Template (2007) [1.6.2007–5:57pm] [1–12]{TANDF_FPP}Tfac/TFAC_A_236020.3d (TFAC) [First Proof]
aroma composition were detected in the food matrix60 without any contact with the polymer (Ducruet et al.
2001; Berlinet et al. 2005). In this case it isimportant to understand how the reactivity of thearoma compound in the matrix can influencethe absorption of the aroma compounds by the
65 packaging material.Although PET and PVC are known to be good
barriers to oxygen, in comparison with polyolefins,PET (one of the most common packaging polymers)and PVC (which is not widely used to package
70 flavoured drinks) have seldom been tested (Ducruetet al. 2001; Van Willige et al. 2002a; Berlinet et al.2005) for their barrier properties to aroma. Theobjective of the present study was to provideadditional results on the sorption of aroma com-
75 pounds in glassy polymers such as PET and PVCduring long storage periods and to compare thekinetic sorption process with the degradationprocess, which also can reduce the amount ofaroma. To make the results valuable for industry,
80 the sorption and degradation kinetics were assessedon real food products and real packaging materials.Since the thermodynamic equilibrium cannot beobtained when the rate constants of both processeshave similar orders of magnitude, the apparent
85 partition coefficients were derived from an approachbased on a new approximated solution proposedby Vitrac and Hayert (2006) to identify transportproperties (including the partition coefficient)from censored kinetics (i.e. without reaching the
90 thermodynamic equilibrium).
Materials and methods
Packaging materials
All the bottles were supplied from Teisseire S.A.France. Strawberry syrups were packaged in
95 rigid PVC bottles (PVC1, PVC2: average thick-ness¼ 1 mm) and in amorphous PET bottles (PET1and PET2; average thickness¼0.75 mm). PVC1was made using GFG 52 D resin from Dorlyl; it issuitable for contact with oils and stabilized with
100 an organotin stabilizer. PVC2 was manufactured byTeisseire. PET1 was manufactured from Ramapet;and PET2 from Shinpet. The surface area/volume(dm2 l�1) ratio was approximately 6. Packagingin glass bottles (GL) was used as a reference to test
105 the stability of the syrup without the influence ofpolymer contact.
Strawberry syrup
Strawberry syrup was obtained by mixing a sucrosesyrup (�Brix: 64.5�1) and fruit juices (strawberry,
110 elder and lemon juices). Nature-identical flavouring
substances were also included. The pH of thestrawberry syrup was 2.65� 0.2; its shelf-life was18 months.
After flash-pasteurization at 105�C for 30 s, syrup115samples were packaged at ambient temperature
under aseptic conditions. All bottles (PET, PVCand glass) were sealed after filling with aluminiumfoil in order to avoid a cap effect. The packagedsyrups were stored at 20�C.
120Solvent extraction of the syrup
A total of 200 ml of syrup, 200 ml of Milli Q waterand 500 ml of tridecane solution (used as an internalstandard, 500 ml l�1) were mixed in a 1-litre flask.Volatile compounds in the strawberry syrup were
125extracted three times using distilled dichloro-methane, with 80, 30 and 30 ml for 30, 15 and15 min, respectively, at 0�C while stirring. Organicphases were separated from the aqueous phase andpooled. The dichloromethane extracts were dried
130over anhydrous Na2SO4 and concentrated usinga Kuderna-Danish column. Analysis were done intriplicate.
Solvent extraction of polymers
At the end of each period of contact the strawberry135syrup and its packaging were separated. The plastic
bottle (PET or PVC) was carefully and rapidlyrinsed twice with ultra-pure water, then withethanol, and finally wiped before cutting. Becauseof the heterogeneous thickness of bottles, samples
140were taken by cutting pieces from three sectionsof a bottle (top, bottom and middle) and thenpooled. Approximately 7 g of each pooled samplewas then plunged immediately into a 100-ml flaskcontaining distilled diethyl ether (95 ml). A total
145of 2 ml of a solution of ethyl pentanoate in diethylether (100 ml l�1) were added as an internal standard.The flasks were tightly plugged with Teflon capsand extraction was carried out for 2 h at 15�C withmagnetic stirring. Solvent extracts were dried with
150sodium sulfate and filtered over glass wool. Diethylether extracts were concentrated using a Kuderna-Danish apparatus.
Two plastic bottles were analysed for eachcontact period. The extraction process was repeated
155twice for each bottle so as to enable analysis induplicate of the plastic material after each periodof contact.
Mass spectra
Identification of the constituents was achieved using160gas chromatography-mass spectrometry (GC-MS).
The GC-MS system consists of a Fisons GC-8000chromatograph and a MD 800 mass spectrometer
2 V. Ducruet et al.
proofreaderText Box6
New XML Template (2007) [1.6.2007–5:57pm] [1–12]{TANDF_FPP}Tfac/TFAC_A_236020.3d (TFAC) [First Proof]
(Fisons Instruments, Les Ulis, France). Separationswere performed on a Supelcowax fused silica
165 capillary column (0.32 mm�30 m; 0.5 mm;Supelco, Bellefonte, PA, USA). The linear flowvelocity of helium was 32 cm s�1. The column wasmaintained at 30�C for 10 min and then pro-grammed at 240�C for SPME analysis and 260�C
170 for solvent extracts at 5�C min�1. The split-splitlessinjector was at 230�C. Solvent extracts were injectedinto the splitless/split system (the split valve wasclosed after 30 s). Electron ionization mass spectrawere recorded under the following conditions:
175 capillary direct interface, 250�C; ion source,200�C; ionization voltage, 70 eV; mass range,29–300 m/z; electron multiplier voltage, 450 V;scan rate, 3 scans s�1.
Mass spectral matches were made by comparison180 with NIST and INRAMASS mass spectra libraries.
Kovat’s Indices of authentic compounds compiledin the INRAMASS library were used to confirmidentification.
Estimation of apparent partition coefficients
185 The partition coefficient is defined from the ratioof concentrations of aroma at thermodynamicequilibrium, noted CP jeq and CF jeq, respectively, inthe packaging material (P) and in food (F ):
K ¼CP jeqCF jeq
ð1Þ
190 Note that equation (1) is valid whatever theconsidered equilibrium. Since the equilibrium wasnot reachable due to insufficient contact time (evenafter 1 year) or due to aroma degradation duringstorage, the equilibrium was extrapolated from the
195 kinetic phase diagram corresponding to the experi-mental sorption kinetic, which was recently pro-posed by Vitrac and Hayert (2006). Only the mainfeatures of KPD relevant for the current applicationare reported here. A kinetic phase diagram, noted
200 KPD, consists in plotting the sorption mass flux orequivalently the variation of the residual concentra-tion in P with time, ðdCP jt=dtÞ vs. the residualconcentration in P, CP jt. During a sorption experi-ment the expected concentration in P at equilibrium
205 corresponds in the KPD space ðCP jt,ðdCP=dtÞjCP jt Þto the value ðCP jeq,0Þ. The interesting feature of theKPD space for the interpretation of sorption kineticsgoverned by diffusion in P and diffusion-convectionin F is that after an initial rapid non-linear decay
210 of ðdCP jt=dtÞ with CP jt, the variation of ðdCP jt=dtÞ isalmost linear with CP jt. As a result, it is possible inthe KPD space to extrapolate linearly the theoreticalequilibrium state. Besides, a possible competitionbetween sorption in P and degradation in F is easily
215 detected by a change in curvature of the KPD.
Thus, a change in the decay of the KPD slopeðd=dCP jtÞðdCP=dtÞjCP jt is interpreted as a degradationprocess, which tends to dominate the sorptionprocess.
220In this work, CP jeq was linearly extrapolated fromthe last experimental point KPD, for whichðd=dCP jtÞðdCP jdtÞjCP jt was monotonous (i.e. increas-ing or constant). The same process was appliedfor the concentration in the liquid. A key step in the
225KPD approach is calculation of the derivatives of theconcentration with time when the experimentaldata are unevenly sampled and noisy. In this work,we use an efficient non deterministic filteringtechnique based on local polynomial approximants
230as detailed and discussed by Vitrac and Hayert(2006). By noting with a superscript ‘extrap’ theextrapolated quantities, the apparent partitioncoefficient, Kapp, corresponding to experimentaldata were calculated from:
Kapp ¼CP jextrapeqCF jextrapeq
ð2Þ
235It is emphasized that Kapp may differ from the realK value; however, it is the best estimator accordingto the data available and our knowledge of sorptionmechanisms. Confidence intervals on Kapp were
240derived by Monte Carlo sampling consisting inadding a white noise similar to the experimentalerror (relative error of 15%) to kinetic data andin repeating the whole analysis. Each confidenceinterval was based on the 2.5th and 97.5th percen-
245tiles of at least 200 Monte Carlo trials.
Statistical analysis
Sorption experiments were analysed using a one-wayANOVA program. When the differences weresignificant (p < 0.05), Duncan’s test was used to
250check the differences between pairs of groups andwas carried out using XLSTAT-Pro 7.0 software(Addinsoft, Paris, France).
Results and discussion
Sorption of the aroma compounds into PET and PVC
255In order to monitor the sorption of aroma com-pounds, the strawberry syrup and the plastic materialwere extracted and analysed at different time pointsduring the 330-day storage period. In the case ofPVC, the presence in solvent extracts of additives
260and small oligomers along with absorbed volatilecompounds produced more complex GC profilesthan those seen with PET. Some of these additiveswere tentatively identified in the extracts of the syrup(Table I).
Sorption of aroma compounds in PET and PVC 3
New XML Template (2007) [1.6.2007–5:57pm] [1–12]{TANDF_FPP}Tfac/TFAC_A_236020.3d (TFAC) [First Proof]
265 These compounds originate from the degradationof additives for example plasticizer, lubricants ormodifiers (Gächter and Muller 1990). Some canmigrate into the syrup such as benzaldehyde,acetophenone and 2-ethylhexanol. By solvent extrac-
270 tion of the syrup or the packaging, these additiveswere extracted and co-eluted with the aromacompounds by GC. Thus, of the 38 aromacompounds previously identified in the syrup at theinitial time of storage (Ducruet et al. 2001), only
275 14 compounds can be quantified in both kinds ofpackaging and in the syrup over the 330-day storageperiod (Table II).
The final concentration of the aroma compoundsinto the syrups stored in both PET and PVC bottles
280 were compared with their initial concentrations(Figure 1). During shelf life, aroma compoundsdecreased by hydrolysis irrespective of the polymer.The decrease was more for short chain esters, and inparticular for acetic acid esters and ethyl hexanoate
285 (Figure 1A) compared with more stable compounds(Figure 1B). This reactivity is not related to thecontact with polymer as it was previously found tooccur during the storage of the syrup in a glass bottle(Ducruet et al. 2001).
290 The greater reactivity was observed for short chainesters (methyl or ethyl esters) and acetic acid estersand can be explained by the low steric hindrancearound their carbonyl group in contrast to linearlonger chain esters (butyl, hexyl or octyl esters) and
295 specially to branched esters (ethyl 2-methylbutano-ate, ethyl 3-methylbutanoate and 3-methylbutyl3-methylbutanaote). These results demonstratedthe instability of the aroma formulation for longperiods of storage in acidic condition such as it was
300 also found in the case of citrus flavoured drinks(Sizer et al. 1988; Berlinet et al. 2005).
The sorption of the aroma compounds fromstrawberry syrup into PET and PVC was monitored
with time. The sorption of the 14 aroma compounds305over 330 days of storage was compared with their
initial concentration into the syrup (Table II). Thesorption into both kinds of packaging materials wasfound to be weak. The sorption was generally lowerthan 0.1% in all cases except for octyl butanoate
310which reached 0.15% in both PET samples, and0.57 or 0.26%, respectively, in PVC1 and in PVC2samples. After 1 year of storage into both samples ofPET and PVC, if we consider the case of the octylbutanoate and hexyl acetate as an example, their
315concentrations decreased by 30–43 and 67–81%,respectively (Figure 1), although the sorption ofthese compounds into PET and PVC only repre-sented 0.15–0.57 and 0.013–0.023% of the initialamount, respectively (Table I). These glassy materi-
320als showed their resistance to sorption by aroma andthus the main parameter which had an impact on thearoma formulation was the hydrolytic process intothe syrup.
Few papers reported quantitative measurement of325the sorption of aroma compounds into PET for real
foods over a long time storage especially for thickmaterials such as bottles. For soft drinks flavouredwith orange oil and stored in PET during 12 weeks,Nielsen (1994) showed that only 2.1 and 1.4% of
330the initial concentrations of myrcene and of limo-nene were respectively sorbed by the polymer.Berlinet et al. (2005) compared the initial amountsof these two aroma compounds in orange juice andthe quantities adsorbed in PET after 5 months of
335storage. They found an adsorption of between 0.2and 0.3% of the initial levels present. These datawere in line with the findings of Van Willige et al.(2003) who stated that only three flavour com-pounds (limonene, �-myrcene and decanal) were
340absorbed by PET. The percentage absorption ofPET from juice containing limonene, �-myrcene,and decanal reached only 0.1, 0.1 and 2.8%,
Table I. Extraneous compounds tentatively identified in the syrup or the packaging material extracts during storage.
PackagingCompounds initially
present in the polymer Migration into the syrup Origin of compounds*
PVC 2-Ethylhexanol Yes after 250 days Thermal degradation of plasticizer2-Ethylhexanol acetate No –Benzaldehyde Yes ABS or MBS impact modifierStyrene No –Isooctyl thioglycolate – –Acetophenone Yes Derived from an organotin stabilizer
PET 2-Ethylhexanol Yes before 225 days Thermal degradation of plasticizerBenzaldehyde Yes –Naphtlalene No –Linear aldehydes No –Acetophenone Yes Possible degradation products of the polyethylene
waxes used as lubricants in PET
*Gächter and Muller (1990).
4 V. Ducruet et al.
New XML Template (2007) [1.6.2007–5:57pm] [1–12]{TANDF_FPP}Tfac/TFAC_A_236020.3d (TFAC) [First Proof]
Tab
leII
.A
mou
nt
of
arom
aco
mp
ou
nd
sso
rbed
into
PE
Tan
dP
VC
afte
r90
day
s**
and
330
day
sof
stora
geat
20� C
.
Vola
tile
com
pou
nd
Cod
eC
o*
(mg
l�1)
PE
T1
(mg
6d
m�
2)
Per
cen
tage
sorp
tion
PE
T2
(mg
6d
m�
2)
Per
cen
tage
sorp
tion
PV
C1
330
day
s(m
g6
dm�
2)
Per
cen
tage
sorp
tion
PV
C2
330
day
s(m
g6
dm�
2)
Per
cen
tage
sorp
tion
Bu
tyl
bu
tan
oat
eB
B24.6
3.7
0.0
15
c3.8
0.0
15
c7.8
0.0
32
a4.9
0.0
20
b
Hex
ylb
uta
noat
eB
H9.6
2.3
0.0
24
c2.6
0.0
27
c11
0.1
15
a5.1
0.0
53
b
Oct
ylb
uta
noat
eB
O8.4
13.1
0.1
56
c12.8
0.1
50
c47.8
0.5
69
a22.1
0.2
63
b
Eth
ylh
exan
oat
eH
E167.5
24.5
**0.0
15
c27.8
**0.0
17
b54.3
0.0
32
a29.1
0.0
17
b
3-M
eth
ylb
uty
l3-m
eth
ylb
uta
noat
e3M
B3M
B134.5
12
0.0
09
c16.8
0.0
12
c146.9
0.1
09
a62.7
0.0
47
b
Eth
yl2-m
eth
ylb
uta
noat
e2M
BE
257.5
8.8
0.0
03
c8.1
0.0
03
c23.2
0.0
09
a16.4
0.0
06
b
Eth
yl3-m
eth
ylb
uta
noat
e3M
BE
166.8
7.3
0.0
04
a7.3
0.0
04
a12.4
0.0
07
a8.6
0.0
05
a
Hex
ylac
etat
eA
cH152.5
23**
0.0
15
b25.8
**0.0
17
b34.6
0.0
23
a19.4
0.0
13
c
3(Z
)-h
exen
ylac
etat
eA
c3H
142.2
11.4
**0.0
08
c13**
0.0
09
b21.3
0.0
15
a12.7
0.0
09
b
Oct
ylac
etat
eA
cO10.4
9.1
**0.0
87
a8.5
**0.0
82
a9.2
0.0
88
a4.3
0.0
41
a
Met
hyl
cin
nam
ate
CIM
E2
530.5
149.5
0.0
28
b164.9
0.0
31
b283.5
0.0
53
a149.7
0.0
28
b
Eth
ylsa
licy
late
2O
HB
zE120.4
45.5
0.0
38
b40.8
0.0
34
b85.7
0.0
71
a42.9
0.0
36
b
Eth
ylb
enzo
ate
BzE
47.5
7.2
0.0
15
bc
6.3
0.0
13
c43
0.0
91
a31.9
0.0
67
ab
�-D
ecal
acto
ne
GD
L339.2
21
0.0
06
c13.6
0.0
05
c132.3
0.0
39
a43.2
0.0
13
b
Dat
aar
eth
em
ean
sof
two
det
erm
inat
ion
s.*C
on
cen
trat
ion
of
the
arom
aco
mp
ou
nd
sin
togl
ass
bott
leat
day
t0.
Dif
fere
nt
lett
ers
inp
erce
nta
geso
rpti
on
colu
mn
sin
dic
ate
sign
ific
ant
dif
fere
nce
sat
p<
0.0
5(D
un
can
)fo
rea
char
om
aco
mp
ou
nd
bet
wee
np
acka
gin
gm
ater
ials
.
Sorption of aroma compounds in PET and PVC 5
proofreaderText Box7
New XML Template (2007) [1.6.2007–5:57pm] [1–12]{TANDF_FPP}Tfac/TFAC_A_236020.3d (TFAC) [First Proof]
respectively, after 1 month of storage. When orangejuice was stored in PET, the sorption of these two
345 compounds was lower. The constituents of the juice(pulp, pectin), missing in the soft drink, may playa competitive role against the sorption into PET.The difference in the behaviour of the aromacompounds whether they are present in a simple
350 solution or a more complex food was noticed byLeufvén and Hermansson (1994). The sorption ofaroma compounds into PET was lower when thesecompounds were into a tomato juice in comparisonwith a model solution. These authors suggested
355 that the aroma compounds probably remain in thetomato juice because they are retained on the naturaljuice constituents instead of being sorbed in thepolymer.
In the case of strawberry syrup, sucrose may360 reduce the diffusion and convection of flavour
compounds in the polymer during the early stages
of storage, as it increases the viscosity of the medium(Van Willige et al. 2000a). Moreover, sucrosemolecules may create local hydrophobic environ-
365ments in the medium which favours inclusion orinteraction with hydrophobic molecules such asaroma compounds. This may lower the sorptionof aroma compounds because of a competition effectbetween the two hydrophobic phases (polymer and
370food matrix) (Nawar 1971; Roberts et al. 1996).
Typical kinetics of sorption into PET and PVC
for the more stable compounds
In both PVC and particularly PVC1, the sorptionof most aroma compounds increased with the
375square-root of time, following a Fickian mode, aswe were able to show for more stable compoundssuch as 3-methylbutyl-3-methylbutanoate. Theirsorption kinetics and their interpretation as kinetic
Figure 1. 2222.
6 V. Ducruet et al.
proofreaderText Box2
New XML Template (2007) [1.6.2007–5:57pm] [1–12]{TANDF_FPP}Tfac/TFAC_A_236020.3d (TFAC) [First Proof]
phase diagrams are shown in Figure 2. Open380 symbols plot experimental values as measured
(duplicate experiments) whereas filled symbols plotnon-deterministically filtered data as describedby Vitrac and Hayert (2006) (Figures 2A–D). Thecontinuous lines depict a cubic spline model, whose
385 values and first derivatives fit the filtered ones.Versus the square-root of time, the kinetics showa subsequent delay related to a possible externalmass transfer resistance, a linear segment followedby a part with a decreasing slope. The presence
390 of a linear segment confirmed that the sorption wascontrolled by the diffusion. The following sharpchange in slope with a possible negative slope is
related to a dynamic modification of sorptionconditions. This modification was created by the
395degradation of the aroma compound in the liquidphase, which modified in return the attainableequilibrium state. The expected equilibriumconcentration in the packaging material was conse-quently decreasing.
400These trends were more discernible in the kineticphase diagram (KPD), which plots the first deriva-tive of the concentration in the packaging materialwith time vs. its primitive (i.e. the concentrationitself) (Figures 2E and F). Filled and open
405symbols plot respectively the filtered values andthe 95% confidence interval as assessed by
Figure 2. 2222.
Sorption of aroma compounds in PET and PVC 7
proofreaderText Box2
New XML Template (2007) [1.6.2007–5:57pm] [1–12]{TANDF_FPP}Tfac/TFAC_A_236020.3d (TFAC) [First Proof]
Monte Carlo sampling. A non-diffusive behavioursuch as the one related to a reaction in the liquidphase is associated to a change in convexity in the
410 diagram. Indeed, a convex shape of the KPD or asharp decrease of the absorbed mass flux when theabsorbed amount increase cannot be related any-more to diffusion (for further details, see Vitrac andHayert 2006). This rule could not however be
415 applied to the earlier stage of desorption kineticssince the mass flux was not estimated with enoughaccuracy. The KPD were used to predict theequilibrium, which would be reached if the sorptionconditions as assessed during the linear part of the
420 concentration vs. the square root of times wereprolonged, since in the parabolic regime of diffusion,the KPD is linear. The concentrations at equilibriumwere linearly extrapolated from the convex part ofthe KPD to the horizontal line dCP jt=dt ¼ 0. The
425 linear extrapolation inferred from the confidenceenvelope of the KPD appears as dashed lines.
Affinity of aroma compounds for PVC and PET
Apparent partition coefficients Kapp derived from theextrapolation of concentrations expected at equili-
430 brium in absence of reaction are plotted vs. log P(octanol–water partition coefficient) (Figures 3Aand B) and vs. the molecular weight (Figures 3Cand D) for PET (Figures 3A and C) and for PVS(Figures 3B and D).
435 It is highlighted that for the set of tested aromacompounds, log P and molecular weight are notindependent quantities. Log(Kapp) appeared almostlinearly correlated to log P values. Whatever thepolymer, sorption was selective and increased with
440 the chain length of the linear butanoic acid esters(butyl butanoate, hexyl butanoate and octyl butano-ate), a behaviour quite similar to that reported byShimoda et al. (1987) with respect to polyethylene.However the correlation was poorer with the
445 molecular mass than with the log P, due tobranching into the molecules. Branching or unsa-turation in the backbone of the molecule (for anequal carbon number) caused sorption to decrease,as can be seen for 3-methylbutyl-3-methylbutanoate
450 by comparison with hexyl butanoate in PET(Figure 3C).
Quite similar partition coefficients were obtainedfor both PET (PET1 and PET2) and both PVC(PVC1 and PV2). Both PET samples sorbed the
455 esters to a lesser extent, PVC1 displayed the highestlevel of sorption, three- to tenfold more than bothsamples of PET mainly for octyl butanoate and3-methylbutyl-3-methylbutanoate. Few studies havedealt with the sorption of aroma compounds
460 into PVC. Koszinowski and Piringer (1987), andDeLassus (1994) showed that diffusion coefficients
in PVC were several orders of magnitude lower thanin polyethylene and polypropylene. In contrast, theirsolubilities (S) were about one order of magnitude
465lower in polyolefins. Barrier behaviour evaluated bythe permeability coefficient P(P¼D�S), resultingfrom the kinetic effect (coefficient of diffusion, D),and the thermodynamic effect (S), only gave a smalladvantage for PVC over polyolefins.
470With PET, the coefficients of diffusion andsolubility of the volatile compounds are lower(Pennarun et al. 2004) than those measured forpolyolefins, so a low interaction can be observedwith this polymer. Figure 3C and D shows that both
475PET samples were more or less equivalent and wereless sorbent than PVC 2, except for cyclic esters suchas ethyl cinnamate and ethyl salicylate The sorptionof these compounds in PET reached the level ofsorption observed in PVC2. These compounds
480exhibited a differential affinity for PET, probablybecause of the similarity of their cyclic structure.This affinity could affect the typical flavour of thesyrup related to the ‘wild strawberry note’ which isassociated with ethyl cinnamate, ethyl salicylate
485and ethyl benzoate.In conclusion, the variation of Kapp values in
PET was mainly correlated to the polarity of testedcompounds as assessed by their log P values.By contrast, the variations in Kapp values for PVC
490were mainly related to their chain lengths.
Kinetics of sorption into PET and PVC for
the unstable compounds
The sorption kinetics of some compounds, (ethylbutanoate, hexyl acetate), showed the same typical
495trends in both types of PET (Figure 4).Sorption increased to a peak at 90 days and
then declined until 150 days, reaching a state ofequilibrium until 330 days.
Several hypotheses can be proposed to explain500this behaviour.
. Durning and Russel (1985) proposed a modelof diffusion with induced crystallization todescribe the sorption kinetics of organicliquids which might trigger PET crystalli-
505zation. Berlinet et al. (2005) showed that theabsorption of aroma compounds (even at lowlevels) may be responsible for increasing thecrystallinity of PET, over 6 months of storage.This ‘solvent induced crystallization’ which
510was described in PET in contact with ethylacetate by Hao and Shore (1999), could leadto a structure modification of PET during3 months of storage and then induce therelease of the aroma compounds from the
515packaging. However this behaviour wouldhave been observed for all the compounds
8 V. Ducruet et al.
New XML Template (2007) [1.6.2007–5:57pm] [1–12]{TANDF_FPP}Tfac/TFAC_A_236020.3d (TFAC) [First Proof]
specially those which are more sorbed suchas octyl butanaoate, but it is not the case andthis behaviour was only noticed for the more
520 reactive compounds.. The sorption behaviour into PET of the more
reactive aroma compounds could be explainedin the following manner: during the first stage
between zero and 90 days, these compounds525were slowly sorbed by PET, and hydrolysis
into alcohols and acids started in the syrup.During the second period (90 < t < 150 days),degradation was the leading parameter, whichresulted in desorption of the volatiles from
530PET, reaching an equilibrium between both
Figure 3. 2222.
Sorption of aroma compounds in PET and PVC 9
proofreaderText Box2
New XML Template (2007) [1.6.2007–5:57pm] [1–12]{TANDF_FPP}Tfac/TFAC_A_236020.3d (TFAC) [First Proof]
phases (syrup and polymer). No alcohol oracid from the corresponding esters was identi-fied in the PET extract. After 150 days, theequilibrium of aroma compounds between
535 packaging and syrup was reached, takingaccount of degradation into the syrup. Thus,after 90 days, PET could play a positive roleas a reserve for some of these unstablecompounds.
540 As previously described (Ducruet et al. 2001),esterified aroma compounds were hydrolysed inglass bottles and this was not influenced by polymercontact. The hydrolysis rate of volatile compoundsdisplayed a first order kinetic. Stronger reactivity was
545 observed for methyl or ethyl esters of acetic acidwhen compared with longer linear chain esters andparticularly branched esters. These compounds wererapidly hydrolysed in the syrup packaged in glassbottles during the first 150 days. Same reactivity was
550 observed in both kinds of polymers. As oxygentransfer through amorphous PET and rigid PVCwere of the same order of magnitude, 4.4 and3.4� 10�11 cm2 s�1 Pa�1, respectively (Brandrupet al. 1989), the oxidation of these compounds
555 cannot explain the observed difference between PETand PVC.
In the case of orange juice in contact with PET,recent results had already shown that, degradationdue to the acid catalysis of aroma compounds
560(�-pinene, limonene, linalool) is the principalparameter controlling exchanges with the glassymaterial (Berlinet et al. 2005) and therefore thearomatic evolution of orange juice was not controlledby the packaging material but by reactions within
565the matrix itself. Polyolefins (polyethylene andpolypropylene) in contact with a model mediumcontaining highly reactive compounds such aslimonene or pinene under acidic conditions mayplay a positive role in stabilizing aroma compounds
570(Lebossé et al. 1997; Feigenbaum et al 1998;Reynier et al. 2004). Because the degradationprocess in the model system is of the same orderof magnitude as diffusion into the polymers, the twophenomena are in competition. In the case of PET,
575transfer is very slow, and many studies, whichconsidered contact times of insufficient duration,were not able to demonstrate this effect. In the caseof contact between real food products during theirshelf life (up to one year) and unstable aroma
580compounds such as short chain esters, the degrada-tion of aroma compounds is more rapid thandiffusion into PET. Release is less rapid than withpolyolefins, and some stabilization effects of PET
Figure 4. 2222.
10 V. Ducruet et al.
proofreaderText Box2
New XML Template (2007) [1.6.2007–5:57pm] [1–12]{TANDF_FPP}Tfac/TFAC_A_236020.3d (TFAC) [First Proof]
may be observed after only 150 days of storage, in585 line with the type trend of sorption kinetics.
In the context of food packaging interactions,the thermodynamic equilibrium of each aromacompound is expressed as the partition coefficient.Because the experimental determination of this
590 partition coefficient may be prohibitive (Tehranyand Desobry 2004), modelling has been proposed asan alternative (Dekker et al. 2003; Tehrany andDesobry 2005). Models are available for polyolefinsbut extensions to glassy polymers such as PET or
595 PVC are still pending. The main limitations consistin the availability of reference data, which are notsubjected to bias due to a competition between slowsorption kinetics (longer than 150 days) and a fastdegradation process. The production of apparent
600 partition coefficient with minimum bias as assessedin this study using a kinetic phase diagram could bean alternative to the lack of published data. From thetechnological point of view, the loss of aroma fora particular food in contact with a glassy material
605 must be envisaged as a consequence of both thesorption and degradation process. In the presentcase, the degradation of the aroma formulationof the syrup is mainly due to hydrolysis of esters.
Conclusions
610 This study showed that the interactions betweenaroma compounds and packaging may result in adynamic and time-dependent change in food qualityduring shelf-life. Two combined mechanisms con-tribute to a loss of aroma during long-term storage:
615 the degradation process in the food product itselfand the sorption process in the packaging material.Both effects were particularly noticeable for bothPET samples and for unstable aroma compounds.It is emphasized that both processes are antagonist.
620 Although the sorption rate is low in glassy polymers,the sorption process tends to prevent the aroma fromthe degradation process. By contrast, the degrada-tion modifies the expected thermodynamic equili-brium of sorption of aroma and tends to limit the
625 sorption of aroma. Both phenomena were simulta-neously studied in this work and a general metho-dology was proposed to extract apparent partitioncoefficients, when the thermodynamic equilibriumcannot be reached due to a coupling with reactions.
630 This methodology was used to derive the partitioncoefficients of 13 aroma compounds found in realfruit juices and two glassy polymers: PET and PVC.Besides, it was highlighted that the stability of thearoma profile in connection with a packaging
635 material cannot be addressed with experiments inshort contact times. Additional studies and dataappear desirable to help the food industry predict the
behaviour of their aroma formulation during long-term storage taking into account both the effects
640of the food matrix and the packaging.
Acknowledgements
This work was supported by the Société Teisseire.The authors would like to thank Mr Rouge andMrs Raynal for their efficient partnership during
645study.
References
Berlinet C, Ducruet V, Brillouet JM, Reynes M, Brat P. 2005.Evolution of aroma compounds in orange juice stored in PET.Food Additives and Contaminants 22:185–195.
650Brandrup J, Immergut EH, Grulke A. 1989. Polymerhandbook, 443.
Dekker M, Van Willige RWG, Linssen JPH, Voragen AGJ. 2003.Modelling the effect of oil/fat content in food systems onflavour absorption by LLDPE. Food Additives and
655Contaminants 20:180–185.DeLassus PT. 1994. Sorption and diffusion of flavors in plastic
packaging. Flavor–food interactions. ACS Symposium Series.Washington, DC: ACS. pp 152–161.
Ducruet V, Fournier N, Saillard P, Feigenbaum A, Guichard E.6602001. Influence of packaging on the aroma stability of
strawberry syrup during shelf life. Journal of Agricultural andFood Chemistry 49:2290–2297.
Durning CJ, Russel WB. 1985. A mathematical model fordiffusion with induced crystallization (2). Polymer 26:131–140.
665Feigenbaum A, Lebossé R, Ducruet V. 1998. Polypropylene asactive packaging material for aroma sorption from modelorange juice. In: Contis ET, Ho CT, Mussinan CJ, ParlimentTH, Shahidi F, Spanier AM, editors. Food flavors: Formation,analysis, and packaging influences. Developments in Food
670Science Series. Amsterdam: Elsevier. pp 743–751.Gächter R, Muller H. 1990. Plastics additives handbook. 3rd ed.
Munich: Hanser.Gavara R, Hernandez RJ, Giacin J. 1996. Methods to determine
partition coefficient of organic compounds in water/polystyrene675systems. Journal of Food Science 61:947–952.
Hao OY, Shore SH. 1999. Ethyl acetate transport in poly(ethylene terephthalate). Journal of Applied Physics85:1148–1152.
Hernandez-Munoz P, Catala R, Gavara R. 2001. Food aroma680partition between packaging materials and fatty food simulants.
Food Additives and Contaminants 18:673–682.Koszinowski J, Piringer O. 1987. Food package compatibility and
migration. Journal of Plastic Film Sheeting 3:96–111.Kutty V, Braddock RJ, Sadler GD. 1994. Oxidation of
685D-limonene in presence of low-density polyethylene. Journalof Food Science 59:402–405.
Kwapong OY, Hotchkiss JH. 1987. Comparative sorption ofaroma compounds by polyethylene and ionomer food-contactplastics. Journal of Food Science 52:761–764.
690Lebossé R, Ducruet V, Feigenbaum A. 1997. Interactionsbetween reactive aroma compounds from model citrus juicewith polypropylene packaging film. Journal of Agricultural andFood Chemistry 45:2836–2842.
Leufvén A, Hermansson C. 1994. The sorption of aroma695components from tomato juice by food-contact polymers.
Journal of the Science of Food and Agriculture 64:101–105.
Sorption of aroma compounds in PET and PVC 11
proofreaderText Box3
New XML Template (2007) [1.6.2007–5:57pm] [1–12]{TANDF_FPP}Tfac/TFAC_A_236020.3d (TFAC) [First Proof]
Nahon DF, Navarro y Koren PA, Roozen JP, Posthumus MA.1998. Flavor release from mixtures of sodium cyclamate,sucrose, and an orange aroma. Journal of Agricultural and
700 Food Chemistry 46:4963–4968.Nawar WW. 1971. Some variables affecting composition of
headspace aroma. Journal of Agricultural and Food Chemistry19:1057–1059.
Nielsen TJ. 1994. Limonene and myrcene sorption into refillable705 polyethylene terephtalate bottles, and washing effects on
removal of sorbed compounds. Journal of Food Science59:227–230.
Nielsen TJ. 1994. Limonene and myrcene sorption into refillablepolyethylene terephtalate bottles, and washing effects on
710 removal of sorbed compounds. Journal of Food Science59:227–230.
Nielsen TJ, Jägerstad M, Öste RE, Wesslén BO. 1992.Comparative absorption of low molecular aroma compoundsinto commonly used food packaging polymer films. Journal of
715 Food Science 57:490–492.Pennarun PY, Dole P, Feigenbaum A. 2004. Functional barriers
in PET recycled bottles. Part I. Determination of diffusioncoefficients in bioriented PET with and without contact withfood simulants. Journal of Applied Polymer Science
720 92:2845–2858.Reynier A, Dole P, Fricoteaux F, Saillard P, Feigenbaum A.
2004. Stabilization of aroma compounds through sorption-release by packaging polymers. Journal of Agricultural andFood Chemistry 52:5653–5662.
725 Roberts DD, Elmore JS, Langley KR, Bakker J. 1996. Effects ofsucrose, guar gum, and carboxymethylcellulose on the releaseof volatile flavor compounds under dynamic conditions.Journal of Agricultural and Food Chemistry 44:1321–1326.
Shimoda M, Matsui T, Osajima Y. 1987. Effect of number730 carbon atoms of flavor compounds on diffusion, permeation
and sorption with polyethylene. Nippon shokuhin kogyogakkaiski 34:535–539.
Sizer CE, Waugh PL, Edstam S, Ackermann P. 1988.Maintaining flavor and nutrient quality of aseptic orange
735juice. Food Technology 42:152–159.Tehrany EA, Desobry S. 2004. Partition coefficients in food
packaging systems: A review. Food Additives andContaminants 21:1186–1202.
Tehrany EA, Desobry S. 2005. Comparison between the different740calculation methods of partition coefficient of aroma com-
pounds with a difference chemical classes in octano/watersystem. Sciences des Aliments 25:23–36.
Van Willige RWG, Linssen JPH, Voragen AGJ. 2000a. Influenceof food matrix on absorption of flavour compounds by linear
745low-density polyethylene: Proteins and carbohydrates. Journalof the Science of Food and Agriculture 80:1779–1789.
Van Willige RWG, Linssen JPH, Voragen AGJ. 2000b. Influenceof food matrix on absorption of flavour compounds by linearlow-density polyethylene: Oil and real food products. Journal
750of the Science of Food and Agriculture 80:1790–1797.Van Willige RW, Linssen JPH, Meinders MBJ, Van der Stege HJ,
Voragen AGJ. 2002b. Influence of flavour absorption onoxygen permeation through LDPE, PP, PC and PET plasticsfood packaging. Food additives and contaminants 19:303–313.
755Van Willige RWG, Linssen JPH, Legger-Huysman A, VoragenAGJ. 2003. Influence of flavour absorption by food-packagingmaterials (low-density polyethylene, polycarbonate and poly-ethylene terephtalate) on taste perception of a model solutionand orange juice. Food Additives and Contaminants 20:84–91.
760Vitrac O, Hayert M. 2006. Identification of diffusion transportproperties from desorption/sorption kinetics: An analysis basedon a new approximation of Fick equation during solid–liquidcontact. Industrial Engineering Chemical Research45:7941–7956.
12 V. Ducruet et al.
proofreaderText Box4
proofreaderText Box5