View/Open - Lirias

advertisement
1
Phase Ratio Variation approach for the study of partitioning behavior of volatile organic compounds in
2
polymer sample bags: Nalophan case study
3
4
Jim Van Durme*, Bas Werbrouck
5
6
Research Group Molecular Odor Chemistry, KU Leuven Campus Ghent, Gebroeders De Smetstraat 1, B-9000
7
Ghent, Belgium
8
* Corresponding author: Tel.: +32 (0)9 265 86 39; Fax: +32 (0)9 265 86 38; E-mail address:
9
jim.vandurme@kuleuven.be
10
11
Abstract
12
Sorption of volatile organic compounds on the inner surface of polymer sampling bags leads to important
13
underestimations of the real headspace concentration. Introducing a wide range of volatiles in a two-phase
14
system containing Nalophan, revealed that recoveries decreased down to 57% in a period of 22 hours. In this
15
work a Phase Ratio Variation approach is investigated to quantify the degree of scalping, and thus enabling to
16
compensate for sorption phenomena. This method requires limited measurements, without the need for time-
17
consuming calibrations. By spiking identical amounts of volatiles in three two-phase systems, each having
18
unique polymer volume/mass ratios , individual partitioning coefficients could be
19
experimentally determined for a wide range of compounds. Additionally, a correlation was found between these
20
partitioning coefficients and the liquid molar volume for a number of aliphatic, aromatic and oxygenated
21
compounds.
22
23
Keywords: scalping, sampling, polymer, partitioning, volatile organic compound, environmental analysis
24
25
26
27
28
29
30
1
31
1.
Introduction
32
In the field of environmental analysis, researchers rely on the accurate qualitative and quantitative assessment of
33
odorous compounds in the gas phase. In recent years, attention has been mainly given to advanced hyphenated
34
analytical techniques and the development of new detectors (e.g. e-noses, SIFT-MS, etc.). However, the sample
35
collection step is equally important and strongly influences reproducibility and accuracy. Despite the availability
36
of multiple sorbent media (Solid Phase MicroExtraction, Stirbar Sorptive Extraction, sorbent tubes) and
37
cryocondensation techniques, whole air sampling using polymeric bags is still one of the most frequently used
38
sample collection methods in the field, due to the ease by which they can be manipulated, their reduced cost and
39
the possibility of reuse (Alonso and Sanchez, 2013). In accordance with the European Standard “Air quality –
40
Determination of odor concentration by dynamic olfactometry” (EN 13725; CEN, 2003), Tedlar (polyvinyl
41
fluoride) and Nalophan (polyethylene terephthalate) are frequently used for whole air sampling (Del Pulgar,
42
Carrapiso, Reina, Biasioli, & García, 2013; Heynderickx, Huffel, Dewulf, & Langenhove, 2013, Hansen et al.,
43
2011). Other polymeric materials used for gas sampling are Kynar, Flexfilm, Teflon (Mochalski et al 2013).
44
The degree of scalping, which is defined as sorption of the volatiles on the inner surface of polymeric sampling
45
bag, is often underestimated, in particular in the field of environmental sampling. Today, no simple method is
46
available to estimate this degree of scalping for individual compounds on polymeric materials. Once volatile
47
organic compounds (VOC) are collected inside gas sampling bags, a number of other phenomena can influence
48
the final concentration of volatile compounds (Alonso and Sanchez 2013). Mochalski et al. (2009) reported how
49
Tedlar polymer materials emit residual compounds (e.g. N,N-dimethylacetamide, phenol, carbonyl sulfide,
50
carbon disulfide), making it obligatory to sufficiently rinse the sampling bags with clean air prior to headspace
51
storage. Next, chemical reactions can cause biases such as oxidative reactions initiated by ambient ozone or
52
photochemical conversions when exposed to ambient or artificial light. Also physical phenomena such as
53
leakages or the transfer of volatiles to condensed water vapor in sampling bags strongly influence the final
54
headspace concentration (Groves and Zellers 1996). Most of the abovementioned phenomena can be resolved by
55
adapting the sample collection protocol. An excess of water in the sampled air can for example be reduced by
56
condensation over dry ice, or by forcing it through a packed tube filled with hygroscopic salts (K 2CO3, MgCO3,
57
Mg(ClO4)2) or water-sorbing polymers (e.g. Nafion) (Dewulf et al 1999).
58
As mentioned earlier, the most critical parameter, however, is the tendency of certain polymer materials to sorb
59
volatile compounds. This phenomenon is well described in the food packaging industry (Nielsen and Jägerstad
60
1994). Next to migration of residual volatiles (Helmroth et al 2002) and permeation of gases through the
2
61
packaging, sorption of aroma compounds may explain the change in aroma intensity or the development of an
62
unbalanced flavor profile of packaged foods (Nielsen and Jägerstad 1994; Mentana et al 2009; Pati et al 2010).
63
Pau Balaguer et al (2012) determined the sorption and transport properties in gliadin and chitosan films with
64
respect to four representative food aroma compounds (ethyl caproate, 1-hexanol, 2-nonanone and α-pinene).
65
Laor et al (2010) reported how the average odor concentration of coffee originated volatiles stored both in Tedlar
66
and Nalophan bags, decreased by a factor 4-5 after 24 hours storage.
67
In recent years, growing attention has been given to the abovementioned sorptive and diffusive losses during
68
environmental sampling. Harreveld (2003) determined that the odor concentration of samples in Nalophan bags
69
decreased to about half the initial concentration after 30 hours of storage. Sironi et al (2014) studied the diffusion
70
rate of ammonia through the Nalophan film, considering storage times ranging from 1 to 26 h. Kim & Kim
71
(2013) determined sorptive losses of five volatile fatty acids (acetic, propionic, n-butyric, i-valeric, and n-valeric
72
acid) on the surface of both stainless steel and quartz tubes. Hansen et al (2011) collected humid and dried air
73
samples from a pig production facility in both Tedlar as Nalophan sampling bags. Chemical measurements
74
revealed that concentrations of carboxylic acids, phenols, and indoles decreased by 50 to >99% during the 24 h
75
of storage in both Tedlar and Nalophan bags. The concentration of hydrogen sulfide decreased by approximately
76
30% during the 24 h of storage in Nalophan bags, whereas in Tedlar bags the concentration of sulfur compounds
77
decreased by <5% (Hansen et al 2011).
78
Mochalski, King, Unterkofler, & Amann (2013) evaluated the stability of 41 selected breath constituents in three
79
types of polymer sampling bags; Tedlar, Kynar, and Flexfilm. Findings yielded evidence of better performances
80
of Tedlar bags over the remaining polymers in terms of background emission, species stability (up to 7 days for
81
dry samples), and reusability. From the abovementioned studies, it was concluded that the scalping effects
82
resulted in systematic biases leading to an underestimation of the true concentration of the targeted pollutants.
83
The aim of this study is to introduce the Phase Ratio Variation (PRV) method as a fast and efficient manner for
84
predicting the degree of scalping for individual compounds in a complex gas mixture. Reported methods to
85
quantify the scalping effect are inverse gas chromatography (Boutboul et al 2002) or by quantitative structure–
86
property relationships (QSPR) (Tehrany and Fournier 2006). These methods are however technologically
87
complex and labor-intensive. The proposed PRV method is based on the relationship between the partitioning
88
coefficient and the phase ratio (ratio gas and polymer phase volumes) (Jouquand et al 2004). In food research,
89
the PRV method has been used to determine partitioning coefficients of six migrants (ethyl acetate,
90
acetaldehyde, acetonitrile, methyl ethyl ketone, isopropyl acetate and butyraldehyde) between four food
3
91
simulants (water, 10% ethanol, 3% acetic acid and 95% ethanol) and two polymers (polyamide and polyethylene
92
terephthalate) (Tehrany and Mouawad 2007). In the field of environmental research, this technique has
93
successfully been used to determine Henry’s constants of other volatiles (Jouquand et al 2004). To the best of
94
our knowledge, applications of the PRV method to understand and predict polymer/volatile interactions in
95
environmental sampling have not yet been described.
96
97
2.
98
2.1. Nalophan
Materials and methods
99
Nalophan with 20 µm thickness was purchased from Kalle Gmbh, Wiesbaden, Germany. Small pieces (1 cm²)
100
were cut and cleaned by flushing for 8 hours with pure inert nitrogen gas at 40°C to remove the adsorbed
101
compounds. The cleaned polymer material was stored in a closed vial which was kept in the dark and under
102
atmospheric conditions (1 atm, 25°C).
103
2.2. Standards: Volatiles Organic Compounds
104
The Japanese Indoor Air Standard mix (Sigma Aldrich, Belgium) was used for both calibration and sorption
105
experiments. This standard consisted of 50 volatile organic compounds in methanol at certified concentrations
106
(100 μg/mL each component). To create a stock solution, the reference solution was diluted 1:40 in methanol
107
(99%, Sigma Aldrich, Belgium) resulting in individual solute concentrations of 2.5 µg/mL.
108
2.3. Preparation of two-phase systems
109
From this stock solution 4.5 µL was injected in a preconditioned 20 mL glass vial in which polymeric material
110
was added. In this way 11.25 ng of each compound was introduced in the system, resulting in individual
111
headspace concentrations of 560 µg/m³.
112
It is noteworthy to mention that such spiking method results in methanol (solvent) background headspace
113
concentrations that are much higher than those of the targeted analytes. This is not a problem, since this specific
114
gas matrix is only used as case study for the applicability of PRV to quantify scalping effects in complex gas
115
mixtures. Nevertheless, some additional experiments were performed to verify to which degree the elevated
116
methanol headspace concentrations influenced partitioning of higher molecular weight volatiles on both the
117
SPME fiber material, as well as on the Nalophan polymer (competitive adsorption). As expected from literature,
118
it was observed that SPME extraction efficiencies decreased with increasing methanol headspace concentration
4
119
(results not shown). However, since the amount of injected methanol was identical throughout the experiments,
120
interaction phenomena during extraction are similar and results can be compared with each other.
121
Limam et al (2005) studied sorption and diffusion of organic solvents in polyethylene terephthalate. Although
122
some swelling phenomena due to methanol sorption occurs, it was concluded that sorption dramatically depends
123
on the chemical microstructure of the polymer (amorphous or oriented). It was concluded that highly crystalline
124
polyethylene terephthalate polymers (e.g. Nalophan) show lower sorption of organic solvent and are generally a
125
good barrier to alcohols. Nevertheless, an experiment was conducted to verify the impact of methanol vapor on
126
the partitioning coefficient of volatiles on the used Nalophan material. Both ‘high methanol’ and ‘low methanol’
127
20 mL systems were prepared, each being spiked with 560 µg/m³ of the individual volatiles. The first series of
128
systems was made as abovementioned (injection of 4.5 µL 1:40 standard methanol solution), while the second
129
series of systems was prepared by injecting 1.1 µL 1:10 standard methanol solution. Each series consisted of
130
polymer-free vials (n=3) and Nalophan containing vials (0.025 g, n=3). After a 15 hour incubation at 25°C, peak
131
ratios were measured for a selection of volatiles in both systems. As will be discussed in more detail further in
132
this work, significant scalping was measured (§3.1.) in both series. However, no significant differences were
133
observed in the measured scalping degree between the ‘low methanol’ and ‘high methanol’ systems (p > 0.05). It
134
can be therefore concluded that competitive sorption between methanol and volatiles on the polymer surface are
135
of less importance.
136
Similar as described for methanol, humidity could have an influence on the investigated sorption phenomena.
137
Ajhar et al (2010) found that the headspace concentration of siloxanes in sampling bags were only 4% higher
138
when humidified gas (90% RH at 37°C) was used, compared to experiments in a dry nitrogen atmosphere.
139
Performing the same experiments (90% RH) at 20°C had no significant impact on the total signal compared with
140
experiments performed in dry air. For this study the partitioning coefficients under normal atmospheric
141
conditions (70% RH at 25°C) were determined.
142
2.4. Chemical analytical methodology
143
Pre-concentration of the introduced volatiles was performed by using a Gerstel MPS2 autosampler, equipped
144
with a headspace-solid phase microextraction unit. The conditions were as follow: the required amount of
145
polymer material was hermetically sealed in 20-mL vials and incubated for 30 minutes at 25°C in a thermostated
146
agitator. After equilibration for 30 min, the headspace was extracted on a well-conditioned CAR/PDMS SPME
147
fibre for another 30 minutes.
5
148
The chemical-analytical setup used in this work consisted of a fully automated sample preparation unit (multi-
149
PurposeSampler® or MPS®, Gerstel®, Mülheim an der Rur, Germany) mounted on a 6890/5973 GC-MS
150
system (Agilent Technologies®, palo Alto, CA). Helium was used as carrier gas (1 mL/min). Injector and
151
transfer lines were maintained at 250°C and 280°C, respectively. The total ion current (70 eV) was recorded in
152
the mass range from 40-230 amu (scan mode) using a solvent delay of 2 min and a run time of 5 min. For GC-
153
MS profiling, the cross-linked methyl silicone column (HP-PONA), 50 m x 0,20 mm I.D., 0,5 µm film thickness,
154
Agilent Technology®) was programmed: 40°C (5 min) to 160°C at 3°C/min, from 160°C to 220°C at 5°C/min,
155
held 3 minutes. Identification of volatile organic compounds in the vegetable oil headspace was performed by
156
comparison of the mass spectra with the Wiley ® 275 library. Since analytical standards are used no additional
157
measurements are necessary to confirm the compound identification.
158
Although 50 volatiles were injected in the two phase systems, results are discussed for a selection of 36
159
compounds. The reason for this is that a number of the volatiles were too volatile to be sufficiently extracted
160
using the described SPME method (e.g. ethanol, methylene chloride,…), while others had a low volatility and
161
have to be classified as semi-volatiles (e.g. tridecane, tetradecane, pentadecane, …). Using the chemical–
162
analytical data, paired comparison tests (t-test) were performed using SPSS Statistics 21 software to evaluate if
163
the observed differences in headspace concentrations were significant. Significances for the differences were
164
established at an alpha risk of 5%.
165
2.5. PRV-method
166
The sorptive uptake of a volatile by a polymer can be described by sorption isotherms. In the most simple model
167
the equilibrium headspace concentration (Cg) is directly proportional to the equilibrium concentration on the
168
polymer surface (CA). In this case a partitioning coefficient K can be derived by the ratio between C g and CA. If
169
sorption isotherms are nonlinear, partitioning coefficients depend on the initial concentration (C g, CA). Langmuir
170
sorption isotherm are applicable in the case of a logarithmic decrease of the adsorption enthalpy with increasing
171
sorption on the polymer surface. The Langmuir model describes adsorption on a surface where the monolayer
172
coverage represents the maximum adsorbed concentration on the surface CA,max as;
173
C A,t eq 
174
Where KL the Langmuir sorption coefficient
K L C A,max C g ,t eq
1  K L C g ,t eq
175
Cg,t=eq the gas headspace concentration after equilibrium (in mol/m³)
176
CA,t=eq the concentration of the volatiles on the surface after equilibrium (expressed in mol/m²)
[1]
6
177
CA,t=max the maximum concentration of sorbed volatiles on the surface after equilibrium (in mol/m²)
178
For very low concentrations and KL.Ct=eq <<1 equation [1] predicts a linear relationship and a partitioning
179
coefficient equals K = Cg,t=eq/CA,t=eq can be defined. Assuming that during whole air sampling a linear sorption
180
model can be assumed, the PRV-method can be applied which is based on the following mass balance equation:
181
C g ,t 0  Vg  C A,t eq  S A  C g ,t eq  Vg
182
Where Cg,t=0 the initial gas headspace concentration (expressed in mol/m³)
183
Vg the gas volume (expressed in m³)
184
SA the surface area of the adsorbent (expressed in m²)
[2]
185
Assuming a volume ratio factor  = SA/Vg (m²/m³) and the partition coefficient K = Cg,t=eq/CA,t=eq
186
((mol/m³)/(mol/m³)) and after dividing by Vg, equation [2] can be rewritten as:
187
C g ,t 0  C A,t eq    C A,t eq  K  C A,t eq  (   K )
188
or
189
Taking the reciprocals of both sides, and taken into account that the area of a gas-chromatograph peak
190
(=PeakArea) is proportional to the equilibrium headspace concentration (Peak Areat=eq=fi x Cg,t=eq with fi =
191
proportional factor) equation [4] can be rewritten as (Tehrany et al 2007):
192
C A,t eq 
Cg ,t eq
K

[3]
Cg ,t 0
[4]
 K
1
1
1



PeakAreat eq f i  Cg ,t 0 f i  K  Cg ,t 0
[5]
193
As will be shown further in this manuscript experiments were only done in a concentration range in which a
194
linear decrease of concentration was observed in function of increased amount of introduced polymer surface.
195
Hence, it can be assumed that the partitioning coefficient K is constant in this range. Thus, equation [5] can be
196
corresponds to a linear equation of the following type: 1/PeakArea = a + b . 
197
Where a = 1/(fi.Cg) and b = 1/(K.fi.Cg) and thus K = a/b.
198
Using the above equations K can be calculated from the values of a and b obtained by plotting 1/PeakArea t=eq
199
against different corresponding  values (Jouquand et al 2004).
[6]
7
200
3.
201
3.1. Sorption in function of contact time and Nalophan mass
Results and discussion
202
Figure 1a presents the average relative headspace concentrations in the two-phase system (n=3) as a function of
203
contact time with 0.025g Nalophan for some randomly selected volatiles from different VOC classes. The closed
204
two-phase systems were simultaneously spiked with a wide range of volatiles, each at an identical initial
205
headspace concentration of 560 µg/m³. For n-heptane (aliphatic compounds), toluene (aromatic compounds),
206
limonene (terpenoids), trichloroethylene (chlorinated compounds), and ethyl acetate (oxygenated compounds),
207
the concentration compared to the initial headspace concentration proved to significantly decrease (p < 0.05)
208
over a period of 22 hours incubation at dark and ambient conditions (298.15K, 1 atm, 70% relative humidity). It
209
was chosen only to study the scalping effect at 298.15K since this is a realistic storage temperature of sampled
210
bags. It is noticeable that the degree of scalping after 22 hours differs depending on type of molecule; a recovery
211
of 91% was found for n-heptane and limonene, while only 77% recovery was measured for trichloroethylene and
212
ethylacetate. Also for the other spiked volatiles the raw average peak areas (n=3) obtained at 10 different
213
sampling times over a period of 22 hours were evaluated by one-way ANOVA. In agreement with the results
214
presented in Figure 1a scalping was also measured for the other volatiles (results not shown) with scalping
215
degrees varying between 57% and 102%. These differences in sorption behaviour are expected and can be
216
explained by unique adsorption behavior of a compound depending on electrostatic forces which are influenced
217
by specific hydrogen bonding, polarity, and van der Waal’s interactions (McGarvey and Shorten 2000; Pau
218
Balaguer et al 2012).
219
To verify whether these decreasing concentrations are not attributed to diffusive losses or chemical-physical
220
conversions (e.g. ozone reactions, photolysis) identical experiments were done on similar systems in the absence
221
of polymeric material, showing no significant decrease in headspace concentration (p>0.05). Therefore, it can be
222
assumed that during our experiments, observed effects are mainly related to sorption.
223
Figure 1b represents the relative headspace concentration of the selected volatiles in function of the introduced
224
mass of Nalophan measured after 15 hours incubation. As it was already reported earlier that the ratio of the
225
sample volume to the bag film area is an important factor influencing scalping on polymers (Beghi and Guillot
226
2008). Again for each individual volatile the initial headspace concentration was 560 µg/Nm³. Triplicate two-
227
phase systems were prepared by introducing respectively 0.000g, 0.025g, 0.050g, 0.075g and 0.100g of
228
Nalophan in 20 mL glass vials. Taking into account a surface/mass ratio of 668.9 cm²/g for the used Nalophan
229
material (experimentally determined), calculated gas/polymer volume ratios  for the different two-phase
8
230
systems are respectively 80.9 m²/m³; 161.8 m²/m³; 242.8 m²/m³ and 323.7 m²/m³. Results as represented in Table
231
1 and Figure 1b indicated that the equilibrium headspace concentration for the different volatiles significantly
232
decreases with increasing amount of Nalophan mass. This trend was expected, as Ajhar et al (2010) measured
233
increasing amount of siloxane sorption on Tedlar material with higher surface-to-volume ratios. In agreement
234
with previous results, clear differences in recovery were noticed among the spiked volatiles.
235
Experimental results illustrated in Figure 1b can be explained by the Langmuir Adsorption Isotherm that
236
correlates the partitioning coefficient with the fractional coverage A as:
237
A 
238
The fractional coverage A can be defined as the fraction of total surface adsorption sites that are occupied by the
239
sorbed volatiles. Equation 7 explains that at low headspace concentration C g, the fractional surface coverage is
240
proportional to Cg with a proportionality constant of K (partitioning coefficient), whereas at high headspace
241
concentrations, the surface coverage goes to unity and becomes independent of Cg (i.e. saturation of the available
242
sites). It is clear that sorption kinetics will be different depending on the degree of coverage A. This Langmuir
243
sorption behavior is also reflected in Figure 1b in which initially a significant decrease in equilibrium headspace
244
concentration with increasing amounts of Nalophan is observed. However in the case that more than 0.050 g ( =
245
161,8 m²/m³) is introduced, a different sorption behavior is observed. Since in reality  values are typically
246
below 100 m²/m³ (0.5 L sampling bag (sphere) = 0.03046m² /0.0005m³ = 61), it is chosen only to use data
247
obtained from two-phase systems with the lowest  values, more specific those containing 0.000g ( = 0.0),
248
0.025g ( = 80.9 m²/m³) and 0.050g ( = 161.8 m²/m³) of Nalophan.
K  Cg
1  K  Cg
[7]
249
3.2. Experimental determination of partitioning coefficient using Phase Ratio Variation approach
250
In Figure 2, the inverse peak areas (n=3) for a selection of volatiles are shown in function of varying  values,
251
while other experimental conditions (25°C, 70% RH, initial headspace concentration 560 µg/m³) are maintained
252
constant. Results are shown for a selection of chlorinated compounds (tetrachloroethylene, trichloroethylene,
253
1,2-dichloropropane), aliphatic compounds (n-nonane, n-heptane, 2,4-dimethylpentane), aromatic compounds
254
(1,2,3-trimethylbenzene, o-xylene, toluene) and oxygenated compounds (4-methyl-2-pentanone, 2-propanol,
255
ethyl acetate).
9
256
Figure 2 shows that within each VOC class linear correlations are observed between the equilibrium headspace
257
concentrations and the corresponding  values. Similar, Table 1 represents average reciprocal peak areas (n=3)
258
for all spiked volatiles in function of  value. Again, for the majority of compounds a linear regression was
259
found based on the three data points with correlations R² varying between 1,00 and 0,76. Since for the majority
260
of volatiles R² correlation above 0.95 are observed, it can be concluded that the theory of Phase Ratio Variation
261
can indeed be applied to evaluate and predict sorption behavior of individual volatiles in sampling bag materials.
262
263
Table 1 represents the calculated partition coefficients K for a wide range of volatiles, ranging from 10,73 (n-
264
undecane) to 0,14 (tetramethylbenzene). It is noteworthy that for a limited amount of volatiles (styrene,
265
tetramethylbenzene, 1,4-dichlorobenzene, n-butanol and nonanal) experimentally determined partitioning
266
coefficients are lower than one, indicating that concentrations are increasing with higher amount of introduced
267
Nalophan. This phenomenon might be explained by impurities in the vials or on the fiber, or due to VOC
268
emission from the material into the headspace despite the flushing of the Nalophan material during the sample
269
preparation. Next, Tromelin et al. (2012) described that for some compounds the calculation of partitioning
270
coefficients using the described first order PRV model can be disappointing, in those cases the application of a
271
second-order PRV method could be an interesting alternative.
272
Based on the theory of adsorption potential, Wang et al (2012) derived a correlation between partitioning
273
coefficient, K, and the liquid molar volume, Vl, as: ln(K) = k Vl + b, where, k and b are constants for VOCs with
274
similar function groups and chemical bonds for the same material. Wang et al. (2008) reviewed how for different
275
building materials the natural logarithmic partition coefficients are linearly correlated against the molar volume
276
for various categories of VOCs with more than 0.91 fitting degree. The results obtained in this study revealed a
277
similar behavior between volatiles and polymeric material. In Figure 3 this linear correlation between ln K and
278
the molar volume is shown for aliphatic, aromatic and oxygenated compounds.
279
For the aliphatic (2,4-dimethylpentane; n-heptane, n-nonane, n-decane, n-undecane), aromatic (toluene,
280
ethylbenzene, xylenes, 2- and 3-ethyltoluene, 1,2,4-trimethylbenzene, 4-ethyltoluene, 1,2,3-trimethylbenzene)
281
and oxygenated (2-propanol, ethylacetate, 4-methyl-2-pentanone, n-butylacetate) compounds linear correlations
282
were measured with correlation coefficients ranging between 0.9751 and 0.9195. In the case that linear
283
correlations can be found for specific VOC classes, it is possible to correct measurement data for the degree of
284
scalping after whole air sampling in polymeric bags. No general lineair correlations could be made for terpenes
10
285
since results were only obtained for three compounds (-pinene, -pinene, limonene). Although not presented in
286
Figure 3, a linear correlation was also found for the chlorinated compounds 1,2-dichloropropane,
287
trichloroethylene, 1,2-dichloroethene, however no general trend was observed in our measurements.
288
As confirmed by our measurements, deviations from the general correlation can be expected within each VOC
289
class. Adsorption is the result of both Van der Waals interactions, as specific proton donor-acceptor interactions.
290
The presence of certain functional groups may strongly effect the relative importance of each type of interaction
291
in the overall sorption process, causing deviations from the general sorption behavior in a certain VOC class.
292
Indeed, particular properties such as the presence of an extra double bound or substituded groups might
293
significantly affect the interaction between volatiles and polymeric surface. Next, not all compounds can easily
294
be attributed to a certain class, which is for example the case for 1,4-dichlorobenzene, being both an aromatic as
295
chlorinated compound. Due to its mixed character, no fit was found for 1,4-dichlorobenzene in both classes. The
296
same was observed in the class of chlorinated compounds in which deviating results were for measured for
297
tetrachloroethene. This might be explained by a high density of chloro-groups, possible causing shielding and
298
different intermolecular forces. Similar observations were made by Wang et al (2008) who described a poor
299
correlation between the partition coefficient of seven benzene compounds and polyurethane, with a fitting degree
300
of only 0.67. It was seen that styrene and chlorobenzene deviated from the regression line due to the presence of
301
some unique characteristics such as double bond for styrene or a chloro-group for chlorobenzene.
302
303
4.
Conclusions
304
Measurements using a two-phase system containing polymer material used for whole air sampling, revealed
305
significant scalping for a wide range of volatiles. Results illustrated that scalping of volatiles on Nalophan leads
306
to significant underestimation of effective headspace concentrations. Therefore it is of great importance that the
307
degree of scalping is evaluated systematically to compensate analytical variability and biases during the
308
sampling stage.
309
A relatively easy methodology, the Phase Ratio Variation method, has been studied to quantify the degree of
310
sorption for individual volatiles in a complex gas mixture. It is shown that using the PRV approach, individual
311
partitioning coefficients could be determined. Moreover, for some important classes of volatiles a correlation
312
was illustrated between the individual partitioning coefficient, and the liquid molar volume. Depending on the
313
presence of particular functional groups deviations were observed from this general correlation. Such deviation
314
sorption behavior is the result of interactions between Van der Waals interactions and specific proton donor-
11
315
acceptor interactions. The availability of quantitative information on individual sorption behavior, enables to
316
compensate for the inevitable sorption losses during storage of loaded polymer sampling bags.
317
The advantages of a PRV approach is that based on limited measurements, and without the need of time-
318
consuming calibrations, the partitioning coefficients of individual volatiles can be quantified in a complex gas
319
mixture. As many gas sampling bags are homemade, two-phase systems having different  values but identical
320
initial volatile concentrations can easily be made by enclosing additional masses of Nalophan in the sampling
321
bags prior to collecting equal gas volumes.
322
Our experiments were performed in controlled conditions, with relative humidity, temperature and light
323
intensities held constant throughout the experimental process. As it is accepted that fluctuations in temperature
324
may result in accelerated sample transformation (e.g. chemical reactivity), this is the focus of current ongoing
325
work. Next, it should also be further investigated whether the observed relationship between partitioning
326
coefficient and liquid molar volume can be generalized for specific polymer materials, or if this correlation is
327
dependent on the gas composition and concentration. Finally, it needs to be pointed that a sampling bags consist
328
of different surface materials (e.g. stainless steel) and/or contain chemical species (e.g. particulate matters,
329
reactive chemical compounds, etc.) that can both potentially influence the equilibrium headspace concentrations
330
during the sample post collection.
331
332
333
5.
Acknowledgements
The authors kindly acknowledge dr. Melissa Dunkle for the support of this work.
334
12
335
6.
References
336
337
Ajhar M, Wens B, Stollenwerk KH, et al (2010) Suitability of Tedlar® gas sampling bags for siloxane
quantification in landfill gas. Talanta 82:92–98. doi: http://dx.doi.org/10.1016/j.talanta.2010.04.001
338
339
Alonso M, Sanchez JM (2013) Analytical challenges in breath analysis and its application to exposure
monitoring. TrAC Trends in Analytical Chemistry 44:78–89. doi: 10.1016/j.trac.2012.11.011
340
341
342
Beghi S, Guillot J-M (2008) Use of poly(ethylene terephtalate) film bag to sample and remove humidity from
atmosphere containing volatile organic compounds. Journal of Chromatography A 1183:1–5. doi:
http://dx.doi.org/10.1016/j.chroma.2007.12.051
343
344
Boutboul A, Giampaoli P, Feigenbaum A, Ducruet V (2002) Influence of the nature and treatment of starch on
aroma retention. Carbohydrate Polymers 47:73–82. doi: http://dx.doi.org/10.1016/S0144-8617(01)00160-6
345
346
347
Del Pulgar JS, Carrapiso AI, Reina R, et al (2013) Effect of IGF-II genotype and pig rearing system on the final
characteristics
of
dry-cured
Iberian
hams.
Meat
Science
95:586–592.
doi:
http://dx.doi.org/10.1016/j.meatsci.2013.05.044
348
349
Dewulf J, Langenhove H Van, Everaert P (1999) Determination of Henry’ s law coefficients by combination of
the equilibrium partitioning in closed systems and solid-phase microextraction techniques. 830:353–363.
350
351
352
Groves WA, Zellers ET (1996) Investigation of Organic Vapor Losses to Condensed Water Vapor in Tedlar®
Bags Used for Exhaled-Breath Sampling. American Industrial Hygiene Association Journal 57:257–263.
doi: 10.1080/15428119691014981
353
354
Hansen MJ, Adamsen APS, Feilberg A, Jonassen KEN (2011) Stability of Odorants from Pig Production in
Sampling Bags for Olfactometry. 1096–1102.
355
356
Harreveld A (2003) Odor concentration decay and stability in gas sampling bags. Journal of the Air & Waste
Management.
357
358
359
Helmroth E, Rijk R, Dekker M, Jongen W (2002) Predictive modelling of migration from packaging materials
into food products for regulatory purposes. Trends in Food Science & Technology 13:102–109. doi:
http://dx.doi.org/10.1016/S0924-2244(02)00031-6
360
361
362
Heynderickx PM, Huffel K Van, Dewulf J, Langenhove H Van (2013) Application of similarity coefficients to
SIFT-MS data for livestock emission characterization. Biosystems Engineering 114:44–54. doi:
http://dx.doi.org/10.1016/j.biosystemseng.2012.10.004
363
364
365
Jouquand C, Ducruet V, Giampaoli P (2004) Partition coefficients of aroma compounds in polysaccharide
solutions by the phase ratio variation method. Food Chemistry 85:467–474. doi:
http://dx.doi.org/10.1016/j.foodchem.2003.07.023
366
367
368
Kim Y-H, Kim K-H (2013) Extent of Sample Loss on the Sampling Device and the Resulting Experimental
Biases When Collecting Volatile Fatty Acids (VFAs) in Air Using Sorbent Tubes. Analytical Chemistry
85:7818–7825. doi: 10.1021/ac401385m
369
370
Laor Y, Ozer Y, Ravid U, et al (2010) Methodological Aspects of Sample Collection for Dynamic Olfactometry.
Chemical Engineering Transactions. doi: 10.3303/CET1023010
371
372
373
Limam M, Tighzert L, Fricoteaux F, Bureau G (2005) Sorption of organic solvents by packaging materials:
polyethylene
terephthalate
and
TOPAS®.
Polymer
Testing
24:395–402.
doi:
http://dx.doi.org/10.1016/j.polymertesting.2004.09.004
13
374
375
McGarvey LJ, Shorten C V (2000) The Effects of Adsorption on the Reusability of Tedlar® Air Sampling Bags.
AIHAJ - American Industrial Hygiene Association 61:375–380. doi: 10.1080/15298660008984546
376
377
378
Mentana A, Pati S, La Notte E, Del Nobile MA (2009) Chemical changes in Apulia table wines as affected by
plastic
packages.
LWT
Food
Science
and
Technology
42:1360–1366.
doi:
http://dx.doi.org/10.1016/j.lwt.2009.03.022
379
380
Mochalski P, King J, Unterkofler K, Amann A (2013) Stability of selected volatile breath constituents in
Tedlar{,} Kynar and Flexfilm sampling bags. Analyst 138:1405–1418. doi: 10.1039/C2AN36193K
381
382
383
Mochalski P, Wzorek B, Śliwka I, Amann A (2009) Suitability of different polymer bags for storage of volatile
sulphur compounds relevant to breath analysis. Journal of Chromatography B 877:189–196. doi:
http://dx.doi.org/10.1016/j.jchromb.2008.12.003
384
385
Nielsen T, Jägerstad M (1994) Flavour scalping by food packaging. Trends in Food Science & Technology
5:353–356. doi: http://dx.doi.org/10.1016/0924-2244(94)90212-7
386
387
388
Pati S, Mentana A, La Notte E, Del Nobile MA (2010) Biodegradable poly-lactic acid package for the storage of
carbonic maceration wine. LWT - Food Science and Technology 43:1573–1579. doi:
http://dx.doi.org/10.1016/j.lwt.2010.06.025
389
390
391
Pau Balaguer M, Gavara R, Hernández-Muñoz P (2012) Food aroma mass transport properties in renewable
hydrophilic
polymers.
Food
Chemistry
130:814–820.
doi:
http://dx.doi.org/10.1016/j.foodchem.2011.07.052
392
393
Sironi S, Eusebio L, Capelli L, et al (2014) Ammonia Diffusion Phenomena through Nalophan TM Bags Used
for Olfactometric Analyses. 949–961.
394
395
Tehrany EA, Fournier F (2006) Simple method to calculate partition coefficient of migrant in food simulant /
polymer system. 77:135–139. doi: 10.1016/j.jfoodeng.2005.06.055
396
397
Tehrany EA, Mouawad C (2007) Food Chemistry Determination of partition coefficient of migrants in food
simulants by the PRV method. 105:1571–1577. doi: 10.1016/j.foodchem.2007.01.075
398
399
400
Tehrany EA, Mouawad C, Desobry S (2007) Determination of partition coefficient of migrants in food simulants
by
the
PRV
method.
Food
Chemistry
105:1571–1577.
doi:
http://dx.doi.org/10.1016/j.foodchem.2007.01.075
401
402
403
Tromelin A, Ayed C, Lubbers S, Pagès-Hélary S, Androit I, Guichard E (2012) Proposed alternative phase ration
variation method for the calculation of liquid-vapour partition coefficients of volatiles. Journal of
Chromatography A 1263:158-168. doi: http://dx.doi.org/10.1016/j.chroma.2012.09.039
404
405
406
Wang X, Zhang Y, Xiong J (2008) Correlation between the solid/air partition coefficient and liquid molar
volume for VOCs in building materials. Atmospheric Environment 42:7768–7774. doi:
http://dx.doi.org/10.1016/j.atmosenv.2008.05.030
407
408
409
Wang Z-W, Wang P-L, Hu C-Y (2012) Investigation in Influence of Types of Polypropylene Material on
Diffusion by Using Molecular Dynamics Simulation. Packaging Technology and Science 25:329–339. doi:
10.1002/pts.983
410
411
412
413
14
414
7.
415
Fig. 1 Relative average headspace concentration (n=3) in function of (a) incubation time
416
after introduction of 0.025g Nalophan polymeric material in 20 mL vial (darkness, T = 25°C;
417
relative humidity = 70%) and (b) introduced mass of Nalophan in 20 mL vial (darkness,
418
T=25°C, RH = 70%) measured after 15 hours of incubation (samples analysed in triplicates,
419
RSD <5%)
Figures and Tables
Relative headspace concentration (%)
120
100
80
n-heptane
Toluene
60
Limonene
40
20
0
0
420
5
10
15
20
Incubation time (hours)
8.
a
Relative headspace concentration (a.u.)
120
100
80
n-heptane
Toluene
60
Limonene
40
Trichloroethhylene
Ethyl acetate
20
0
0.00
0.05
0.10
0.15
mass Nalophan added in 20mL vial (g)
421
422
b
15
423
Fig. 2 Inverse average peak area values (n=3) in function of volume ratio factor  (darkness, T =
424
25°C; relative humidity = 70%, incubation time=48 hours)
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
16
441
Fig. 3 Natural logarithmic partition coefficient on Nalophan in relation the molar volume
442
(darkness, T = 25°C; relative humidity = 70%)
443
3
2.5
R² = 0.9329
2
ln K (-)
1.5
R² = 0.9751
aliphatic compounds
1
aromatic compounds
0.5
Oxygenated compounds
R² = 0.9195
0
0.0
50.0
100.0
150.0
200.0
250.0
-0.5
-1
Molar volume (ml/mol)
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
17
459
Table 1. Average repriprocal peak area values in function of volume ratio factor . Intercept,
460
rico and correlation R² after linear regression is represented, which was used to determine
461
individual partitioning coefficients K for each volatile (T = 25°C; relative humidity = 70%,
462
incubation time=48 hours)
Molar
volume
(mL/mol)
=0
 = 80,9
 = 161,9
rico (a.u.)
Intercept
(a.u.)
R²
K
lnK
2,4-dimethylpentane
148,9
9,071E-06
1,017E-05
1,137E-05
1,422E-06
9,053E-06
1,00
6,36
1,85
n-heptane
146,5
4,167E-06
4,512E-06
5,184E-06
6,287E-07
4,112E-06
0,97
6,54
1,88
iso-octane
162,5
1,666E-06
1,435E-06
1,980E-06
1,940E-07
1,537E-06
0,33
7,92
2,07
n-nonane
178,6
5,535E-07
5,550E-07
6,460E-07
5,714E-08
5,386E-07
0,76
9,43
2,24
n-decane
194,9
2,461E-07
2,529E-07
2,845E-07
2,371E-08
2,420E-07
0,88
10,21
2,32
n-undecane
211,2
1,464E-07
1,489E-07
1,680E-07
1,338E-08
1,436E-07
0,84
10,73
2,37
Benzene
89,2
3,767E-06
4,292E-06
5,860E-06
1,293E-06
3,593E-06
0,92
2,78
1,02
Toluene
106,3
1,272E-06
1,566E-06
2,161E-06
5,489E-07
1,222E-06
0,96
2,23
0,80
Ethylbenzene
122,5
4,979E-07
6,416E-07
9,118E-07
2,556E-07
4,768E-07
0,97
1,87
0,62
o-xylene
120,6
4,840E-07
6,442E-07
8,695E-07
2,382E-07
4,731E-07
0,99
1,99
0,69
m-xylene
123,4
4,673E-07
6,476E-07
8,590E-07
2,420E-07
4,621E-07
1,00
1,91
0,65
4,766E-07
1,039E-06
1,911E-06
8,863E-07
4,249E-07
0,98
0,48
-0,74
Compound
Aliphatic Compounds
Aromatic Compounds
Styrene
p-xylene
123,4
3,933E-07
5,173E-07
7,320E-07
2,092E-07
3,782E-07
0,98
1,81
0,59
3-ethyltoluene
135,5
1,986E-07
2,575E-07
4,000E-07
1,244E-07
1,847E-07
0,95
1,48
0,39
2-ethyltoluene
135,5
2,164E-07
2,935E-07
4,490E-07
1,436E-07
2,034E-07
0,96
1,42
0,35
1,2,4-trimethylbenzene
136,6
1,943E-07
2,481E-07
3,891E-07
1,204E-07
1,798E-07
0,94
1,49
0,40
1,751E-07
2,215E-07
2,992E-07
7,667E-08
1,699E-07
0,98
2,22
0,80
1,726E-07
2,599E-07
3,749E-07
1,250E-07
1,680E-07
0,99
1,34
0,30
1,402E-07
2,051E-07
3,239E-07
1,135E-07
1,312E-07
0,97
1,16
0,15
7,449E-08
1,200E-07
1,905E-07
7,167E-08
7,032E-08
0,98
0,98
-0,02
1,013E-05
1,295E-05
1,457E-05
2,745E-06
1,033E-05
0,98
3,76
1,33
4-ethyltoluene
1,2,3-trimethylbenzene
1,3,5-trimethylbenzene
1,2,4,5tetramethylbenzene
136,6
154,6
Chlorinated Compounds
Chloroform
1,2-dichloroethane
79,2
1,852E-05
2,599E-05
4,024E-05
1,342E-05
1,740E-05
0,97
1,30
0,26
1,2-dichloropropane
97,4
6,234E-06
6,869E-06
9,138E-06
1,794E-06
5,962E-06
0,90
3,32
1,20
Trichloroethylene
90,0
4,294E-06
5,473E-06
8,129E-06
2,369E-06
4,048E-06
0,95
1,71
0,54
Tetrachloroethene
102,4
9,998E-07
1,362E-06
1,755E-06
4,667E-07
9,945E-07
1,00
2,13
0,76
2,194E-07
8,214E-07
1,887E-06
1,030E-06
1,422E-07
0,97
0,14
-1,98
1,4-dichlorobenzene
Oxygenated Compounds
Acetone
73,4
4,801E-06
5,798E-06
6,787E-06
1,227E-06
4,802E-06
1,00
3,91
1,36
2-propanol
76,5
1,703E-05
2,264E-05
2,469E-05
4,732E-06
1,762E-05
0,93
3,72
1,31
Ethyl acetate
98,2
n-butanol
4-methyl-2-pentanone
124,9
5,920E-06
7,603E-06
9,004E-06
1,905E-06
4,296E-06
7,882E-06
1,665E-05
7,630E-06
1,847E-06
2,338E-06
3,143E-06
8,003E-07
5,967E-06
3,433E-06
1,795E-06
1,00
0,94
0,98
3,13
1,14
0,45
-0,80
2,24
0,81
18
n-butylacetate
132,0
1,273E-06
1,640E-06
2,315E-06
6,440E-07
1,222E-06
0,97
1,90
0,64
Nonanal
172,0
2,080E-07
2,799E-07
5,154E-07
1,899E-07
1,807E-07
0,91
0,95
-0,05
(1S)-(-)-alpha-pinene
158,8
2,605E-07
2,613E-07
3,231E-07
3,872E-08
2,503E-07
0,76
6,46
1,87
beta-pinene
158,8
2,240E-07
2,275E-07
2,670E-07
2,653E-08
2,180E-07
0,81
8,22
2,11
(R)-(+)-Limonene
162,0
1,543E-07
1,748E-07
2,157E-07
3,792E-08
1,509E-07
0,96
3,98
1,38
Terpenes
463
464
19
Download