G Model

CHROMA-355449; No. of Pages 12 ARTICLE IN PRESS

Journal of Chromatography A, xxx (2014) xxx–xxx

Contents lists available at ScienceDirect

Journal of Chromatography A
journal homepage: www.elsevier.com/locate/chroma

Assessment of a new method for the analysis of decomposition gases

of polymers by a combining thermogravimetric solid-phase extraction
and thermal desorption gas chromatography mass spectrometry
E. Duemichen a , U. Braun a,∗ , R. Senz b , G. Fabian b , H. Sturm a
a
BAM Federal Institute of Material Research and Testing, Unter den Eichen 87, 12205 Berlin, Germany
b
Beuth University of Applied Sciences, Luxemburger Straße 10, 13353 Berlin, Germany

a r t i c l e i n f o a b s t r a c t

Article history: For analysis of the gaseous thermal decomposition products of polymers, the common techniques are
Received 16 December 2013 thermogravimetry, combined with Fourier transformed infrared spectroscopy (TGA–FTIR) and mass spec-
Received in revised form 15 May 2014 trometry (TGA–MS). These methods offer a simple approach to the decomposition mechanism, especially
Accepted 19 May 2014
for small decomposition molecules. Complex spectra of gaseous mixtures are very often hard to identify
Available online xxx
because of overlapping signals. In this paper a new method is described to adsorb the decomposition
products during controlled conditions in TGA on solid-phase extraction (SPE) material: twisters. Sub-
Keywords:
sequently the twisters were analysed with thermal desorption gas chromatography mass spectrometry
TDS–GC–MS
TGA–FTIR (TDS–GC–MS), which allows the decomposition products to be separated and identified using an MS
TGA–MS library. The thermoplastics polyamide 66 (PA 66) and polybutylene terephthalate (PBT) were used as
Degradation example polymers. The influence of the sample mass and of the purge gas flow during the decomposi-
Polymer tion process was investigated in TGA. The advantages and limitations of the method were presented in
Solid-phase extraction comparison to the common analysis techniques, TGA–FTIR and TGA–MS.
© 2014 Elsevier B.V. All rights reserved.

1. Introduction For FTIR spectroscopy the exhaust of the TGA is fed through a
heated transfer line to a heated measurement cell. The compounds
Analysing the thermal decomposition products of polymers of the decomposition process are transferred without splitting
yields information about the structure of the polymers as well as under atmospheric pressure and are not destroyed during detec-
their additives and a better understanding of thermal behaviour tion. Compounds are identified with reference spectra or by the
during processing, or degradation during decomposition [1–7]. A knowledge of the user. With FTIR spectroscopy small molecules like
standard method for analysing the thermal behaviour of polymers CO2 , CO, NH3 and H2 O can be identified to a high degree of certainty
is thermogravimetry (TGA). TGA yields information about the mass [12]. However, for the complex mixture of thermal decomposi-
change of a sample during heating at a constant heating rate or tion products of polymers, the TGA–FTIR characterisation enables
isotherm from room temperature up to 1100–1500 ◦ C under con- only the assignment of functional groups, but does not clearly ren-
stant conditions (air or inert gases, such as nitrogen or argon, der the decomposition molecules. Furthermore, depending on the
are common as purge gases). Fourier transformed infrared spec- construction of the coupling, the transfer of decomposition prod-
troscopy (FTIR) [8,9] and mass spectrometry (MS) [10,11] are two ucts requires time, such that short-lived decomposition products
common detection methods used for the analysis of TGA gaseous cannot be detected. Thus further reactions of the decomposition
degradation products. They have the advantage of providing simul- products must be considered, for example, water elimination reac-
taneous information about degradation products and mass change. tions from diols or amides as well as decarboxylation reactions from
Both are used for qualitative and quantitative multi-component ester or carbonic acids [9], or when experiments were performed
analysis in gaseous mixtures. under air, the oxidation of species. Decomposition products with
high boiling points (close to the temperature of the transfer line)
tend to condense or adsorb on the inner wall of the transfer system;
for example, terephthalic acid forms during the decomposition of
phthalatesters [6,13].
∗ Corresponding author. Tel.: +49 30 8104 4317; fax: +49 30 8104 1617. TGA is coupled with MS via a heated, fused-silica capillary which
E-mail address: [email protected] (U. Braun). is essential for transfer from atmospheric pressure to the vacuum

http://dx.doi.org/10.1016/j.chroma.2014.05.057
0021-9673/© 2014 Elsevier B.V. All rights reserved.

Please cite this article in press as: E. Duemichen, et al., J. Chromatogr. A (2014), http://dx.doi.org/10.1016/j.chroma.2014.05.057
G Model
CHROMA-355449; No. of Pages 12 ARTICLE IN PRESS
2 E. Duemichen et al. / J. Chromatogr. A xxx (2014) xxx–xxx

in the MS. The reaction products are typically analysed using elec- TDS–GC–MS is a very convenient analytic technique for
tron impact ionisation (EI) with quadruple MS and identified based unknown organic species. The unknown organic compounds are
on the MS library or by the knowledge of the user. However this is first separated through the chromatographic column and then
only possible with high certainty if the gaseous products are sep- analysed with a high sensitivity through a mass selective detec-
arated. Furthermore, because decomposition products with higher tor. Afterwards they can be easily identified by scanning spectra
boiling points and higher molecular masses are not able to pass libraries.
through the transfer line, the method is limited to small decompo- In the present work two common thermoplastic polymers were
sition products, in praxis for masses below m/z = 200. This problem investigated as an example. The first one is polyamide 6,6 (PA
can be overcome using skimmer coupling [14], but this method is 66), which is compounded by condensing hexane-1,6-diamine and
quite expensive and therefore not viable for many studies. hexane-1,6-dioic acid. The other is polybutylene terephthalate
Both methods are effective if there are only a few components (PBT), which is compounded by an ester reaction of terephthalic
in the gaseous mixture. In reality, evolved volatile products of acid and butane-1,4-diol. The literature shows a wide range of
polymer decomposition are very complex and therefore spectra decomposition products found for both materials, depending on
with overlapped signals of various compounds are detected. Con- the analysis system and degradation conditions [38–46]. The sam-
sequently only the main products of degradation can be identified. ples are first analysed with TGA–FTIR and TGA–MS and then
A chromatographic separation is needed to identify the complex compared to the new developed solid-phase extraction method
mixture of products. An established method is pyrolysis coupled TGA–SPE/TDS–GC–MS. The comparison is supposed to demon-
with gas chromatography and mass spectrometry (Py–GC–MS) strate the advantages and disadvantages of the methods used to
[15,16]. In this method, the sample is placed in a furnace or on identify the thermal decomposition products of polymers. Besides
a filament in which the sample is heated to a programmed tem- the comparison of the final results from the three methods, detailed
perature at a more or less controlled heating rate. In many cases insight is also provided into the evaluation procedure for identify-
only helium can be used as carrier gas because of the subsequent ing products.
chromatographic system. The thermal degradation products can be In a next article in the near future, we will present results
trapped by cooling or transferred directly to the chromatographic of polymer decomposition obtained by use of the TGA–FTIR and
column. With Py–GC–MS it is possible to identify and even quantify TGA–SPE/TDS–GC–MS. By this work it will become clear why the
both major and minor thermal degradation products [17–20]. One method is advantageous for understanding molecular decomposi-
disadvantage of Py–GC–MS is the relatively poor reproducibility tion mechanisms.
over the different pyrolysers as result of the different temperature
behaviours [21]. A further disadvantage is the enhanced mainte-
2. Experimental
nance requirements because of blocked capillaries and polluted
injection systems through high boiling compounds [22], especially
2.1. Chemicals and solid-phase extraction agents (twister)
for pyrolysers with a furnace.
Some efforts have been made to couple the TGA to a GC or
The pure polyamide 6,6 (Ultramid® , PA 66) and polybutylene
GC–MS system [23–26]. Nowadays commercial systems are avail-
terephthalte (Ultradur® , PBT) were purchased from BASF AG. Both
able, but they have not found their way into the routine analytics of
were milled with a cryo mill. At least duplicate analyses were car-
polymers because of their high cost and difficulties with coupling.
ried out with all materials using the different methods.
Some efforts have been made to trap the evolved thermal decom-
The polydimethylsiloxane (PDMS) twisters were purchased
position gases of polymers during various decomposition processes
from Gerstel. They are 10 mm long and have a layer thickness of
(not TGA experiments) on solids like Tenax and Carbotrap, or in
0.5 mm. They were fixed in the thermal desorption glass tubes using
liquid adsorbents like dichloromethane [27–34], but with neither
stainless steel sieves. Before use the twisters were cleaned twice
frequency nor significant success.
with 10 mL methanol (HPLC grade, Fisher Scientific) for 10 min in
In this study we describe a new way to trap the volatile
an ultra sonic bath and cured in a tube conditioner TC 2 (Gerstel,
products released by polymers in well-defined thermogravimetric
Mühlheim a/d Ruhr, Germany) at 200 ◦ C for 50 min with a nitrogen
experiments (TGA) on a connected solid-phase adsorber material
purge gas flow of 60 mL min−1 .
(twister). After this solid-phase extraction (SPE) process, the adsor-
All gases (N2 , He) used in the experiments have a purity of
bents on the twister are analysed by established thermal desorption
99.999%.
gas chromatography mass spectrometry (TDS–GC–MS). It will be
shown that this new method (TGA–SPE/TDS–GC–MS) offers some
advantages over the established methods. 2.2. Instrumentation and settings
In the TGA the polymer samples which are easy to handle can be
heated under well defined conditions (gas atmosphere, purge gas 2.2.1. TGA–FTIR
flow, heating programme, sample mass etc.) The connected solid- Thermogravimetric measurements were performed with the
phase extraction materials, the twisters, are small magnetic stirrers single-arm balance TGA/SDTA 851 (Mettler/Toledo, Gießen,
with an adsorbent material on the surface, which are normally Germany) equipped with an auto sampler. A heating rate of
used in liquid analysis. They are normally used for stir bar sorptive 10 ◦ C min−1 from 25 up to 600 ◦ C was used for all measurements.
extraction (SBSE), introduced by Baltussen et al. [35]. The trapp- A purge gas flow of 30 mL min−1 of nitrogen was applied. 10 mg
ing of volatile compounds released in the gas phase by solids on of each sample was weighed in an aluminium oxide crucible. The
twisters has been carried out by Bicchi et al. [36] and Tienpont et al. evolved gas analysis of TGA–FTIR measurements was performed
[37]. The method was called headspace sorptive extraction (HSSE); with a Nexus 470 (Nicolet Instruments, Offenbach, Germany) using
however, heating is possible only up to 200 ◦ C. The most common a glass-lined stainless steel transfer line heated to 250 ◦ C with an
adsorbent is polydimethylsiloxane (PDMS), which is suitable for inner diameter of 1500 ␮m. The TGA–FTIR interface cell was heated
a wide range of compounds (unspecific binding). Other materials to 260 ◦ C. These temperatures are the maximum temperatures that
like polyethylene glycol copolymers are available for highly polar can be realised with this method. At a resolution of 4 cm−1 , 16 scans
compounds. The twisters have the advantage that they are com- were taken for each spectrum from 400 to 4000 cm−1 . The gaseous
mercially viable and have defined surfaces. They are easy to clean decomposition products were identified by comparison with ref-
and can be reused. erence spectra from spectra libraries (Aldrich Vapor Phase Sample

Please cite this article in press as: E. Duemichen, et al., J. Chromatogr. A (2014), http://dx.doi.org/10.1016/j.chroma.2014.05.057
G Model
CHROMA-355449; No. of Pages 12 ARTICLE IN PRESS
E. Duemichen et al. / J. Chromatogr. A xxx (2014) xxx–xxx 3

Table 1
Parameters of the TDS–GC–MS system.

Parameter Value Parameter Value

Thermal desorption system Cooling injection system

Initial temperature (◦ C) 25 Initial temperature (◦ C) −100
Heating rate (K min−1 ) 40 Heating rate (K s−1 ) 12
End temperature (◦ C) 200 End temperature (◦ C) 270
Hold time (min) 5 Hold time (min) 3
Transfer line 250 Oven programme
temperature (◦ C)
Desorption mode Splitless Start temperature (◦ C) 40
Gas flow (mL min−1 ) 50 Heating rate (K min−1 ) 5
End temperature (◦ C) 300
Fig. 1. Schematic view of the coupling of TGA with the thermal desorption glass Hold time (min) 3
tube for solid-phase extraction with a twister.

the mass selective detector (MSD) was an Agilent MSD 5973N.

Library, HR Hazardous Chemicals—Vapor Phase Spectra, HR Nicolet An HP-1MS (30 m × 0.25 mm i.d. × 0.25 ␮m df ) column was used
Vapor Phase). for all measurements. For all measurements a carrier gas flow
(He) of 1 mL min−1 was set. The system is controlled with a
2.2.2. TGA–MS combination of Gerstel Maestro and the MSD ChemStation. A
The TGA–MS system consisted of a double-arm balance TG/DTA summary of all parameters is shown in Table 1. The MSD (EI,
220 (Seiko, Thermo Haake, Karlsruhe, Germany) coupled with a 70 eV) was used in scan mode from m/z = 30 to 350. The temper-
QMG 421C quadrupole mass spectrometer (Balzer, Lichtenstein) ature of the ion source was set to 230 ◦ C and the quadrupole to
and electron impact ionisation (EI). TGA experiments were per- 150 ◦ C.
formed in nitrogen at a flow rate of 200 mL min−1 . About 10 mg The time between the loading of twisters with decomposition
of the sample were heated in a platinum crucible from room products of TGA–SPE experiment and the analysis of twisters with
temperature up to about 800 ◦ C at a heating rate of 10 ◦ C min−1 . adsorbed decomposition products in TDS–GC–MS experiment was
The coupling element was a quartz capillary heated to 200 ◦ C kept as short as possible (below 2 h). However, no strong influence
with an inner diameter of 100 ␮m. Through this coupling only a of time was observed.
small part of the pyrolysis gas flow was transferred to MS. The The GC–MS data were evaluated using spectral libraries (Nist
temperature of the recipient was 200 ◦ C and a basis pressure of 11, Wiley 9th). Only compounds which could be determined with
10−5 mbar was realised during the experiment. The temperatures a high certainty were evaluated.
used are the maximum temperatures that can be realised with
this present method. The chosen conditions restrict the detectable 3. 3 Results and discussion
masses to <m/z = 200. First the thermal decomposition products
were analysed in full-scan mode in the range of 30–200 m/z 3.1. TGA–FTIR
(0.2 s for each mass port). After identification of the characteristic
masses they were monitored over time in single-ion mode (0.5 s In Figs. 2 and 3 the results of TGA–FTIR measurements are pre-
for each mass port). The masses observed were compared with sented for PA 66 and PBT. On top the results of mass loss are shown
the mass fragment patterns of reference spectra (Nist 11, Wiley with the characteristic mass loss curves and the mass loss rates.
9th). In the middle of the figures, the FTIR spectra at their maximum
mass loss rates are illustrated with correlation to reference spec-
2.2.3. TGA–SPE/TDS–GC–MS tres. Because the IR spectra were taken continuously, it is possible
The thermo balance (TGA/SDTA 851) and experiment parame- to generate product release rates by evaluating the area of a charac-
ters were similar to those used for the TGA–FTIR measurements. For teristic absorption band over the temperature as a function of time.
each measurement (except for the mass variation measurements) These were demonstrated below in the figures for the identified
2.0 mg of each sample was weighed in an aluminium oxide crucible. decomposition products.
For each measurement (except for the flow variation measure- PA 66 starts to decompose at 380 ◦ C (Fig. 2A). Two decompo-
ments) a purge gas flow of 30 mL min−1 of nitrogen was applied. sition processes can be identified with a maximum DTG signal
It must be noted that there is a supplementary permanent purge at 430 and at 450 ◦ C. At 600 ◦ C no residue remains. For PA 66
gas stream of about 60 mL min−1 because of the way the TGA is con- the main degradation products at 418 ◦ C and 440 ◦ C are cyclopen-
structed. This additional, constant value will not be considered in tanone, carbon dioxide and ammonia, which could be qualified
the discussion. The TGA interface for connecting the twisters was a exactly over reference spectra because of very sharp and charac-
device of our own construction. Fig. 1 shows a schematic diagram teristic absorption peaks (Fig. 2B). Cyclopentanone was identified
of the coupling. The adapter was constructed in order to connect by the characteristic band of the carbonyl vibration at 1760 cm−1 .
the outlet of the thermo-balance to the thermal desorption glass Furthermore, amine signals at 1600 cm−1 indicate the presence of
tube. The twister was fixed with two stainless sieves, in the same aliphatic amines. A good accordance is given to hexylamine. Inten-
position every time. For all measurements the same kind of twisters sive aliphatic signals between 2750 and 3000 cm−1 can be detected;
was used (PDMS, 10 mm × 0.5 mm). The temperature of the heat- it should be noted that these are present in cyclopentanone and the
ing jacket for the thermal desorption glass tube was set to 120 ◦ C aliphatic amine species. However, no further distinction between
to avoid condensation products, especially water. cyclopentanone and its derivates or hexylamine and its derivates
For thermal desorption a TDS-A2 sampler (Gerstel, Mühlheim is possible. All of the decomposition products identified for PA
a/d Ruhr, Germany) was used, which was mounted on a cooled 66 show formation of a single peak in release rates (Fig. 2C).
injection system (CIS) used as cryotrap in splitless mode with a First the release of CO2 is observed, followed by ammonia and
glass-wool liner. The gas chromatographic system was an Agi- cyclopentanone, which are released at the same time. Finally the
lent 6890+ GC (Agilent Technologies, Palo Alto, CA, USA) and release of hexylamine is observed. The hexylamine signal remains,

Please cite this article in press as: E. Duemichen, et al., J. Chromatogr. A (2014), http://dx.doi.org/10.1016/j.chroma.2014.05.057
G Model
CHROMA-355449; No. of Pages 12 ARTICLE IN PRESS
4 E. Duemichen et al. / J. Chromatogr. A xxx (2014) xxx–xxx

Fig. 2. (A)—Thermogram and DTG signal of PA 66; (B)—spectra of the main decom- Fig. 3. (A)—Thermogram and DTG signal of PBT; (B)—spectra of the main decompo-
position steps of PA 66 in comparison to reference spectra; (C)—product release sition steps of PBT in comparison to reference spectra; (C)—product release rates of
rates of the main degradation products of PA 66, measured with TGA–FTIR. the main degradation products of PBT, measured with TGA–FTIR.

3.2. TGA–MS

although the sample does not lose any more weight. This is In Figs. 4 and 5 the results of the TGA–MS investigation of PA
caused by condensation of hydrocarbon species in the FTIR analysis 66 are shown. Similar investigation of PBT is not feasible, because
cell. formed terephthalate acid crystals will block the transfer line.
PBT shows only one decomposition step (Fig. 3A). It starts to An alternative isothermal investigation of PBT at 375 ◦ C was dis-
decompose at 360 ◦ C and has its maximum DTG signal at 400 ◦ C. missed, because the results would not be relevant for the dynamic
A residue of 7.6% remains until 600 ◦ C. For PBT, 1,3-butadiene and investigations. In TGA–MS the PA 66 decomposes in at least two
CO2 were identified by a very sharp absorption band at 404 ◦ C with decomposition processes, with the maximum of DTG at 430 and
a maximum at 908 cm−1 and a double band between 2400 and 460 ◦ C without leaving residue (Fig. 5A). Compared to the TGA–FTIR
2280 cm−1 (Fig. 3B). An aromatic acid was identified by the absorp- measurement, the second decomposition step occurs at a higher
tion band between 3600 and 3500 cm−1 . The comparison to benzoic temperature. This is caused by the different construction of the
acid revealed a good result. A very sharp absorption band between balance, the different experiment parameters and their resulting
1220 and 1300 cm−1 indicated the C O C vibration of an ester; the sensitivity to the mass loss process.
comparison to dibutyl terephthalate shows an acceptable accor- In Fig. 4 the mass spectra of PA 66 are shown at the main decom-
dance. However, a further distinction between benzoic acid and its positions steps around 430 ◦ C and 460 ◦ C (full-scan mode). It is
derivates or dibutyl terephthalate and its derivates is not possible. obvious that the majority of the triggered mass ports (m/z > 45) do
For the degradation of PBT the formation of terephthalic acid is not show any signal response. Due to the high dwell times for each
expected, but it is known that terephthalic acid condenses in the single mass (∼30 s) in this measurement mode, only major decom-
transfer line [13], making it impossible to detect with the present position products with dominant ions are detected. However, the
conditions of TGA–FTIR. The release of benzoic acid, 1,3-butadiene observed mass pattern was compared with reference data. Charac-
and dibutyl terephthalate takes place in a single process, whereas teristic masses can be observed, such as m/z = 44 for CO2 , m/z = 17
the release of CO2 shows two processes (Fig. 3C). By determin- for NH3 and m/z = 18 for H2 O, as well as characteristic mass patterns
ing product release rates it can be observed that, during the major such as m/z = 27 and m/z = 41 for aliphatic chains.
decomposition step of polymer, the formation of 1,3-butadiene and The subsequent MS analysis in single-ion mode offers higher
CO2 occurs at slightly lower temperatures than the formation of sensitivity. In addition to the decomposition products identified
benzoic acid and dibutyl terephthalate. Similar to PA66, some of in full-scan mode, characteristic masses of degradation products
the signals (benzoic acid, dibutyl terephthalate) remain, although identified by TGA–FTIR measurements were chosen. The ion mass
the sample does not lose any more weight. This is also caused by release rates can be determined as a function of temperature in
condensation of these species in the FTIR analysis cell. the TGA experiment (Fig. 5). In first approximation, masses with

Please cite this article in press as: E. Duemichen, et al., J. Chromatogr. A (2014), http://dx.doi.org/10.1016/j.chroma.2014.05.057
G Model
CHROMA-355449; No. of Pages 12 ARTICLE IN PRESS
E. Duemichen et al. / J. Chromatogr. A xxx (2014) xxx–xxx 5

Fig. 4. Mass spectra of PA 66 at the main decomposition steps at 430 (top) and 460 ◦ C (below), measured with TGA–MS.

similar formation rates are likely to belong to identical decompo- However, the identification of hydrocarbon species by use of
sition products; this means they are the mass fragments of the TGA–MS is related to the species identified by TGA–FTIR mea-
same decomposition product. Therefore in Fig. 5B–E the masses surement, and in both methods we observe a similar ranking of
are sorted by their mass pattern behaviour: in Fig. 5B masses with release rates; first, CO2 is released, afterwards NH3 and cyclopen-
individual releases are presented, in Fig. 5C masses with double tanone and then the release of aliphatic hydrocarbons. These
peak release, in Fig. 5D those with behaviour consisting of a peak results illustrate that TGA–MS is very sensitive in detecting small
with a preceding shoulder and in Fig. 5E masses with a single decomposition molecules (H2 O), but, similar to the TGA–FTIR
peak and very low intensities. Water and CO2 can be identified by measurements, the unambiguous identification of simultaneously
their individual mass patterns. In comparison to TGA–FTIR, where developing, aliphatic decomposition products of polymer is not
it was not possible to verify water with certainty, with TGA–MS possible.
it was clearly identified. The release of ammonia is indicated by
the formation rate of m/z = 17; however, this signal must be cor- 3.3. TGA–SPE/TDS–GC–MS
rected for the formation of water. In reference data for water a
pattern of m/z = 17:m/z = 18 of 0.2:1.0 is observed (according to 3.3.1. Qualitative analysis
a standard conditions, electron ionisation with 70 eV). However, In Fig. 6 the chromatograms of GC–MS analysis are presented
our TGA–MS system shows minor changes in fragmentation due for PA 66 (A) and PBT (B), using TGA–SPE/TDS–GC–MS analysis. It
to small changes in the quadrupole (minor adjustment through can be observed that most of the compounds are well separated,
adsorbed impurities). Thus, the fragmentation factor observed was especially when compared to TGA–FTIR/MS data. For peaks with-
not 5; instead, we use a factor of 3.5, as determined by own out labelling no satisfying contribution to the database pattern was
reference measurements of only components releasing water. Con- observed. The polar compounds, especially the species with car-
sequently our ammonia signal is corrected through multiplication boxyl groups of PBT decomposition, show a broad peak, because the
by 3.5 and subtraction of the mass m/z = 18. The aliphatic masses polar character of these components results in unspecific bonding
are sorted by the relative intensities of their two peaks at 430 at the inner coating of the GC column.
and 460 ◦ C. Intensive masses at m/z = 55 and m/z = 56 indicate Eleven compounds could be identified for PA 66 (Fig. 6A,
the formation of cyclopentanone, but could also be correlated to Table 2). Beside the expected cyclopentanone (1) and hexy-
amine derivatives, such as hexamethylenimine, cyclohexanamine lamine (4), it was also possible to identify their derivates
or aminohexane. The latter could also be attributed to the values unambiguously, such as 2-methyl-cyclopentanone (2), 2-
m/z = 26, 27, 43, 41 and 42. Furthermore, the masses detected are ethyl-cyclopentanone (7), [1,1 -bicyclopentyl]-2-one (9) and
clearly higher than the molecular weight of cyclopentanone, but 2-cyclopentylidenecyclopentan-1-one (10) as well as hex-5-
are also not characteristic of amino-C6 components. enylamine (3), hexamethylenimine (5), cycloazaheptene (6) and

Please cite this article in press as: E. Duemichen, et al., J. Chromatogr. A (2014), http://dx.doi.org/10.1016/j.chroma.2014.05.057
G Model
CHROMA-355449; No. of Pages 12 ARTICLE IN PRESS
6 E. Duemichen et al. / J. Chromatogr. A xxx (2014) xxx–xxx

Table 2
Summary of qualified compounds of PA 66 and PBT with references found in literature.

Rt. (min) Compound CAS Structure Reference TGA–MS/FTIR

PA 66

1 3.94 Cyclopentanone 120-92-3 [43,47,48] X

2 4.90 2-Methyl-cyclopentanone 1120-72-5 [43]

3 5.10 Hex-5-enylamine 34825-70-2 [49]

4 5.32 1-Hexanamine 111-26-2 [50] X

5 6.13 Hexamethylenimine 111-49-9

6 6.70 Cycloazaheptene 998008-63-2

7 7.17 2-Ethyl-cyclopentanone 4971-18-0 [43]

8 11.63 1,6-Hexanediamine 124-09-4 [47]

9 16.73 [1,1 -Bicyclopentyl]-2-one 4884-24-6 [47,51]

10 19.28 2-Cyclopentylidenecyclopentan-1-one 825-25-2

11 21.82 1,2,3,5,6,7-Hexahydro- dicyclopenta[b,e]pyridine 34421-99-3 [47]

12 – 1,4-Dicyanobutane* 111-69-3 [52]

13 – ␧-Caprolactam* 105-60-2 [34,53]

PBT

1 15.83 Benzoic acid 65-85-0 [41] X

2 17.41 4-Methyl benzoic acid 99-94-5

3 18.75 But-3-enyl benzoate – [54]

4 19.28 Biphenyl 92-52-4 [54]

5 21.71 But-3-enyl-4-methylbenzoate –

Please cite this article in press as: E. Duemichen, et al., J. Chromatogr. A (2014), http://dx.doi.org/10.1016/j.chroma.2014.05.057
G Model
CHROMA-355449; No. of Pages 12 ARTICLE IN PRESS
E. Duemichen et al. / J. Chromatogr. A xxx (2014) xxx–xxx 7

Table 2 (Continued)

Rt. (min) Compound CAS Structure Reference TGA–MS/FTIR

6 24.05 Fluorene 86-73-7

7 24.43 4-Vinylbenzoic acid 1075-49-6

8 24.95 Benzophenone 119-61-9

9 33.44 Terephthalic acid di(but-3-enyl)ester – X

*
Found in preliminary study with an HP-5MS column.

1,6-hexanediamine (8). Most of the decomposition products iden- known from the literature [47]. By use of a different GC column in
tified are also described in the literature (Table 2). With very high a preliminary study 1,4-dicyanobutane and ␧-caprolactam were
certainty 1,2,3,5,6,7-hexahydrodicyclopenta[b,e]pyridine (11) was also identified.
found. This decomposition product cannot be related clearly to Water, CO2 and NH3 could not be detected by
derivative species of cyclopentanone and hexylamine, but it is also TGA–SPE/TDS–GC–MS analysis due to the high vapour pressure
and the improper adsorption agent. For these small compounds
the application of TGA–FTIR/MS is more suitable. For PA 66 the
smallest compound that could be detected was cyclopentanone
(1) with a molecular mass of 85 g mol−1 and a boiling point of
131 ◦ C. Therefore, in first approximation the method is limited
to identifying compounds with boiling points higher than 120 ◦ C,
which is the temperature of the twister.
In the chromatogram of PBT nine compounds could be identified
(Fig. 6B, Table 2). Besides the benzoic acid (1) and the terephthalic
acid di(but-3-enyl)ester (9) expected, various derivates contain-
ing acid could be observed, such as 4-methyl benzoic acid (2) and
4-vinylbenzoic acid (7) and various derivatives containing ester,
such as but-3-enyl benzoate (3) and but-3-enyl-4-methylbenzoate
(5). According to our TGA–FTIR experiment three additional, unex-
pected compounds without a carboxylic group could be identified:
biphenyl (4), fluorine (6) and benzophenone (8). Most of the
decomposition products identified could not be found with their
exact structure in the literature (Table 2). CO2 and 1,4-butadiene
could not be detected by TGA–SPE/TDS–GC–MS analysis due to
the high vapour pressure. For these small compounds the appli-
cation of TGA–FTIR/MS is more suitable. The smallest compound
that could be detected was benzoic acid (2), with a molecular
mass of 122 g mol−1 and a boiling point of 250 ◦ C. Similar to the
results of PA 66, the method is estimated to be limited to com-
pounds with boiling points higher than the temperature of the
twister.
The new solid-phase extraction method with PDMS twisters
leads to the improved identification of more detailed and several
new thermal decomposition products for PA 66 and PBT. It was
now possible to distinguish between the derivates of cyclopen-
tanone and hexylamine during PA 66 decomposition and between
the derivatives of aromatic acids and terephthalic esters during
PBT decomposition. The identification could be determined with
a high certainty through the separation of the compounds with
the chromatographic column and comparison of the spectra with
reference spectra. For the investigated samples we found that the
method is limited to decomposition products with boiling points
below 120 ◦ C, which is the temperature of the twister during the
Fig. 5. (A)—TGA and DTG signal of PA 66; (B)–(E)—characteristic ion formation rates,
TGA experiment.
measured with TGA–MS.

Please cite this article in press as: E. Duemichen, et al., J. Chromatogr. A (2014), http://dx.doi.org/10.1016/j.chroma.2014.05.057
G Model
CHROMA-355449; No. of Pages 12 ARTICLE IN PRESS
8 E. Duemichen et al. / J. Chromatogr. A xxx (2014) xxx–xxx

Fig. 6. Chromatograms of PA 66 (A) and PBT (B), measured with TGA–SPE/TDS–GC–MS by using an HP-1MS column, a sample mass of 2 mg and a purge gas flow (TGA) of
30 mL min-1 .

3.3.2. Variation of the purge gas flow It can be concluded that the larger cyclopentanone derivatives
Adsorption and desorption at the twister are complex mech- (9, 10) and the non-cyclic, unpolar hexylamine derivatives (3, 4)
anisms that depend on the chemical and physical abilities of the show increasing adsorption behaviour with increasing purge gas
target compound and the adsorption matrix. An important factor is flow up to 50 mL min−1 . The relative content of large and unpo-
the time the compound has to interact with the matrix. This can be lar species increases with purge gas flow on the twister surface.
varied by changing the purge gas flow in TGA experiment. Purge Small, polar species (1, 2, 7) will be replaced. For the cyclopen-
gas flows of 10, 30, 50 and 100 mL min−1 were used to investi- tanone species a clear correlation between the molecular structure
gate the influence of purge gas flow on the amount of compound and the adsorption behaviour can be concluded.
bonded on the twister matrix. PA 66 was used as an example. The results show that there is no adjustment where an optimum
The response for each compound was determined by integrating was achieved for all compounds. It seems to be a good compromise
the target ion chromatograms. The characteristic target and qual- for detecting all compounds with a good sensitivity and repeat-
ifier ions are summarised in Table 3 for all identified compounds. ability using a gas flow between 30 and 50 mL min−1 ; however,
The integration results for the single components were normalised careful control of purge gas flow is necessary for semi-quantitative
to results of 10 mL min−1 and sorted according to their functional measurements.
groups.
In Fig. 7 the response determined for various components with 3.3.3. Variation of the sample mass
their percental deviation are shown as a function of the purge gas To investigate the influence of sample mass on the amount of
flow. The averaged error of all identified components at purge gas compound bonded on the twister matrix, the sample mass weight
flows higher than 50 mL min−1 is clearly higher (>20%) than the in the crucible of TGA experiment was varied. Masses of 1, 2, 3
averaged error of all identified components at purge gas flows and 4 mg were used. For sample masses higher than 4 mg, black
between 10 and 50 mL min−1 (<12%). Therefore the values with condensation products were observed in the thermal desorption
purge gas flows of 100 mL min−1 will not be discussed, although glass tube. Therefore, no higher masses were investigated. PA 66
some of the components show acceptable deviations even less than was used as an example. The response for each compound was
10%, such as 6 and 11. The results indicate that using a purge gas determined by integrating the target ion chromatograms. The char-
flow of 100 mL min−1 is not useful. At this high purge gas flow acteristic target and qualifier ions are summarised in Table 3 for all
no representative adsorption–desorption conditions exist for all identified compounds. The integration results for the single com-
species. This is probably caused by undefined streaming conditions ponents were normalised to results of 1 mg and sorted according
in the glass tube, which lead to bad repeatability. to their functional groups. They are presented in Fig. 8 with their
In area from 10 to 50 mL min−1 the small cyclopentanone deriva- percental deviations.
tives such as 1, 2 and 7 show a decrease. The bigger cyclopentanone For components 2, 3, 4, 5, 6 and 7 no systematic trend is observ-
derivatives such as 9 and 10 show an increase. The compounds con- able. We can conclude that even with a sample mass of 1 mg
taining nitrogen, 6, 8 and 11 show stagnation or a slight increase. a saturation of species on the twister surface is achieved, The
For 3, 4 and 5 an increase can be observed. present amounts of these components reflect the composition in

Please cite this article in press as: E. Duemichen, et al., J. Chromatogr. A (2014), http://dx.doi.org/10.1016/j.chroma.2014.05.057
G Model
CHROMA-355449; No. of Pages 12 ARTICLE IN PRESS
E. Duemichen et al. / J. Chromatogr. A xxx (2014) xxx–xxx 9

Table 3
Compounds of PA 66 and PBT with their target and qualifier ion as well as their average amounts and standard deviations.

Rt. (min) Compound Target ion/qualifier ion (m/z) x̄ × 106  × 106

PA 66
1 3.94 Cyclopentanone 55/84 63.26 3.33
2 4.90 2-Methyl-cyclopentanone 42/98 5.01 2.57
3 5.10 Hex-5-enylamine 56/82 6.78 0.64
4 5.32 1-Hexanamine 41/101 1.51 0.13
5 6.13 Hexamethylenimine 70/99 1.24 0.26
6 6.70 Cycloazaheptene 68/97 13.74 0.78
7 7.17 2-Ethyl-cyclopentanone 84/112 3.44 1.96
8 11.63 1,6-Hexanediamine 87/100 125.23 14.30
9 16.73 [1,1 -Bicyclopentyl]-2-one 84 17.61 6.67
10 19.28 2-Cyclopentylidenecyclopentan-1-one 150/79 3.36 1.00
11 21.82 Dicyclopenta[b,e]pyridine 158 60.83 1.89
PBT
1 15.83 Benzoic acid 105/122 360.98 46.54
2 17.41 4-Methyl benzoic acid 136/91 5.29 1.89
3 18.75 But-3-enyl benzoate 105/77 475.73 51.06
4 19.28 Biphenyl 154/76 90.07 7.08
5 21.71 Propyl 4-methylbenzoate 119/136 62.70 7.90
6 24.05 Fluorene 167/182 1.51 0.14
7 24.43 4-Vinylbenzoic acid 167/182 1.02 0.13
8 24.95 Benzophenone 182/105 2.08 0.28
9 33.44 Terephthalic acid di(but-3-enyl)ester 203/149 97.23 13.74

the decomposition gas, independent of the absolute amount of of the sample mass is absolutely necessary for semi-quantitative
gases evolved in the TGA experiment. The cyclopentanone com- measurements.
pounds 1, 9 and 10, as well as the nitrogenous 8 and 11 show
an increase in response with an increase in sample mass. These 3.3.4. Repeatability
compounds (aside from 1) have a higher molecular mass (approx- To investigate repeatability, the same material was investigated
imately higher 150 g mol−1 ) than the others. However, in these at least 4 times under the same conditions. The characteristic target
measurements a large deviation can be observed for all compo- ions were integrated. The average (x̄) and the standard deviation
nents, independent of their mass. Only for components 1 and 11 () of the compounds as well as their target and qualifier ions are
is the averaged error <20%. With higher sample masses in the summarised in Table 3. The relative standard deviation in percent
TGA experiment and therefore with higher concentration of these is presented for PA 66 and PBT in Fig. 9.
species in the decomposition gas, compounds with higher molec- For PA 66 the compounds 1, 3, 4, 5, 6, 8, and 11 show an
ular mass will concentrate on the twister surface. acceptable relative standard deviation of below 21%. With the
The results show that there is no adjustment where an opti- exception of cyclopentanone (1), all of them have nitrogen in
mum was achieved for all compounds. It seems to be a good their structure and an aliphatic, cycloaliphatic or aromatic back-
compromise for detecting all compounds with a good sensitivity bone. A relative standard deviation of over 21% was found for
and repeatability using a mass of 2 mg. However, careful control compounds 2, 7, 9 and 10. All of these compounds are derivates

Fig. 7. Normalised responses for each compound in relation to the purge gas flow for PA 66, sorted according to their structure. Measured with TGA–SPE/TDS–GC–MS (2.0 mg
for each sample).

Please cite this article in press as: E. Duemichen, et al., J. Chromatogr. A (2014), http://dx.doi.org/10.1016/j.chroma.2014.05.057
G Model
CHROMA-355449; No. of Pages 12 ARTICLE IN PRESS
10 E. Duemichen et al. / J. Chromatogr. A xxx (2014) xxx–xxx

Fig. 8. Normalised response for each compound in relation to the polymer mass for PA 66, sorted according to their structure. Measured with TGA–SPE/TDS–GC–MS (purge
gas flow of 30 mL min−1 for each sample).

of cyclopentanone. It can be concluded that more polar com- 3.4. Assessment and comparison of the methods
pounds lead to results with higher deviations, possibly due to less
interaction with stationary phases and also to active sites on all The comparison of the three methods reveals clear advan-
systems. tages and disadvantages for the various techniques during polymer
For PBT the relative standard deviation of compounds 1, 3, 4, degradation and decomposition analysis.
5, 6, 7, 8 and 9 is less than 15%. Only 4-methyl benzoic acid (2) Using TGA–FTIR allows gaseous decomposition products with
has a relative standard deviation of about 35%. In comparison to low boiling points to be detected with a high certainty (e.g. CO2 ,
benzoic acid (1) the amount of 4-methyl benzoic acid emitted (2) small alkenes, NH3 ). It is possible to obtain information about mass
is very small. As explained before, the phase chemistry between a changes as well as evolved masses during heating in situ, while
carboxylic group and PDMS on the HP-1 column is not suitable for recording and quantifying the product release rates of different
good analysis, leading to higher sensibility and a higher standard components. According to the construction and the requirements
deviation. of the method, the handling of the technique is simple and the

Fig. 9. Relative standard deviation (RSD) of PA 66 (left) and PBT (right) in percent for each compound.

Please cite this article in press as: E. Duemichen, et al., J. Chromatogr. A (2014), http://dx.doi.org/10.1016/j.chroma.2014.05.057
G Model
CHROMA-355449; No. of Pages 12 ARTICLE IN PRESS
E. Duemichen et al. / J. Chromatogr. A xxx (2014) xxx–xxx 11

danger of a blocked transfer line (ca. 1500 ␮m diameter) is low. the limitation of the method to the detection of molecules with a
The detection of gaseous decomposition products is limited by the boiling point ≥ca. 120 ◦ C, which is the temperature of the twister.
temperature of the coupling device: decomposition products with Therefore, the newly developed method is recommended as a per-
boiling points higher than the transfer line temperature cannot fect supplement to TGA–FTIR or TGA–MS measurement.
be detected (e.g. terephthalic acid); furthermore, highly reactive
decomposition products are able to react further, which could 4. Summary
result in blocking of the transfer line or incorrect conclusions [6,13].
However, especially for small and symmetrical molecules with A new method, TGA–SPE/TDS–GC–MS was developed to analyse
characteristic infrared-active vibration, no sophisticated knowl- the thermal decomposition gases of polymers, and compared to
edge is necessary for identification due to very powerful spectra established TGA–FTIR and TGA–MS investigations. In this method
libraries. the evolved gases of polymers during TGA analysis were trapped
Using TGA–MS also allows gaseous decomposition products on solid-phase extraction material and subsequently analysed in a
with low boiling points to be detected with high certainty. Sim- TDS–GC–MS system. By use of PA 66 and PBT samples the advan-
ilar to TGA–FTIR, it is possible to obtain information about mass tages and limitations of all three methods were evaluated and
changes and product release rates in situ during heating. However, assessed.
according to the construction and the requirements of the method, The major advantage of the new method compared to
the technique is limited to decomposition products smaller than TGA–FTIR/MS is the easy and unambiguous identification of various
200 m/z and the danger of a blocked transfer line (ca. 100 ␮m hydro-carbon species by use of the chromatographic separa-
diameter) is very high, especially through condensation or fur- tion process. In contrast to related methods like Py–GC–MS, this
ther reactions in the capillary. In contrast to evolved gas analysis method shows an extensive reduction in effort through the sep-
with FTIR, the analysis of the mass pattern at a certain point of aration of solid-phase extraction and the following analysis with
decomposition will not result in unambiguous identification of TDS–GC–MS. Because of sample masses which are easy to handle
decomposition products. A high level of user knowledge is neces- it is possible to carry out a direct sample comparison of released
sary to identify characteristic masses, or time-consuming gradation amounts of thermal degradation products.
of the fragmentation patterns are required to detect decomposition It was shown that the sample mass and the purge gas flow of
products. In praxis we observed, and this is also documented in the TGA can have an influence on the chromatographic results.
the literature [55,56], that MS is more sensitive to low concentra- Therefore especially the sample mass (±1 wt.%) and purge gas flow
tions of decomposition products and more stable against deviations (±10 vol.%), should be held constant for semi-quantifications.
in experimental conditions. Furthermore non-infrared-active [1,2] However, the detection of small decomposition products with
molecules can be detected. low boiling points is not possible. Therefore we conclude that this
TGA–FTIR and TGA–MS work very well when FTIR spectra or method is an excellent complement to common analysis meth-
MS fragmentation patterns of the gaseous decomposition prod- ods with TGA–FTIR and TGA–MS for understanding and developing
ucts identified do not overlap. However, this is very common in molecular degradation and decomposition mechanisms.
polymer decomposition analysis. When polymer decomposition
products are very similar in chemical structure, or when many dif-
Acknowledgements
ferent decomposition products are formed, an improved method is
necessary.
The first measurement and the idea of using twisters was born
The new method, TGA–SPE/TDS–GC–MS, offers unambiguous
together with O. Jann, (BAM 4.2) O. Wilke (BAM 4.2), W. Horn
identification of structurally similar hydrocarbon decomposition
(BAM 4.2) and E. Juritsch (BAM 4.2). We want to thank D. Neubert
products during the controlled decomposition of polymers in TGA.
(BAM 6.5) for the TG–MS measurements and the maintenance of
According to the construction and the requirements of the method,
the devices. The TGA interface for the coupling with the TDS-tube
there is no limitation to the molecular size of the decomposi-
was produced by M. Schneider (BAM 7.5) and T. Kukofka (BAM 7.5).
tion products or problems due to impurities (condensation, further
The excellent picture of the TGA coupling was designed by T. Rybak
reaction). A major advantage is the reduced trouble with impu-
(BAM 7.5).
rities (in contrast to Py–GC–MS), because the critical component,
the transfer system consisting of glass tube and twister can be
References
checked und cleaned very easily. Furthermore the twister can be
reused and by varying the adsorption material a choice of specific [1] U. Braun, B. Schartel, J. Fire Sci. 23 (2005) 5.
compounds can be made to increase the sensitivity. The chromato- [2] U. Braun, B. Schartel, Macromol. Chem. Phys. 205 (2004) 2185.
graphic separation of single decomposition products allows for very [3] H. Seefeldt, E. Duemichen, U. Braun, Polym. Int. 62 (2013) 1608.
[4] U. Braun, A.I. Balabanovich, B. Schartel, U. Knoll, J. Artner, M. Ciesielski, M. Dor-
easy identification with a high certainty by use of spectra libraries.
ing, R. Perez, J.K.W. Sandler, V. Altstadt, T. Hoffmann, D. Pospiech, Polymer 47
In comparison to Py–GC–MS, the sample masses used (2–3 mg) are (2006) 8495.
easier to handle. This allows a semi-quantitative comparison of the [5] U. Braun, B. Schartel, M.A. Fichera, C. Jager, Polym. Degrad. Stab. 92 (2007) 1528.
[6] U. Braun, B. Schartel, Macromol. Mater. Eng. 293 (2008) 206.
released thermal degradation products. However the present work
[7] U. Braun, H. Bahr, B. Schartel, e-Polymers 41 (2010) 1.
shows that semi-quantitative evaluation has to be performed care- [8] D.A.C. Compton, D.J. Johnson, M.L. Mittleman, Res. Dev. 31 (1989) 68.
fully, because, depending on the species, the purge gas flow and the [9] D.A.C. Compton, D.J. Johnson, M.L. Mittleman, Res. Dev. 31 (1989) 142.
sample mass used, there is high deviation. For PA 66 only cyclopen- [10] D. Dollimore, G.A. Gamlen, T.J. Taylor, Thermochim. Acta 75 (1984) 59.
[11] H.L. Friedman, Thermochim. Acta 1 (1970) 199.
tanone and cyclopenta[b,e]pyridine always show results with [12] S. Singh, C. Wu, P.T. Williams, J. Anal. Appl. Pyrolysis 94 (2012) 99.
acceptable deviations; for these molecules a semi-quantitative [13] U. Braun, H. Bahr, H. Sturm, B. Schartel, Polym. Adv. Technol. 19 (2008) 680.
evaluation can be performed with high certainty. Therefore for [14] E. Kaisersberger, E. Post, Thermochim. Acta 324 (1998) 197.
[15] T. Shin, O. Hajima, W. Chuichi, in: T. Shin, O. Hajima, W. Chuichi (Eds.),
semi-quantitative measurements a very careful experimental pro- Pyrolysis–GC/MS Data Book of Synthetic Polymers, Elsevier, Amsterdam, 2011,
cedure is necessary. p. 1.
In contrast to TGA–FTIR/MS, no release rates can be determined [16] P. Kusch, Advanced Gas Chromatography—Progress in Agricultural, Biomedical
and Industrial Applications, InTech, Rheinbach, Germany, 2012.
as a function of time or temperature. However, through fraction- [17] F.C.-Y. Wang, J. Chromatogr. A 883 (2000) 199.
ated collection it may be possible to link the thermal decomposition [18] F.C.-Y. Wang, J. Chromatogr. A 886 (2000) 225.
products to a temperature range in the future. Disadvantageous is [19] F.C.-Y. Wang, J. Chromatogr. A 891 (2000) 325.

Please cite this article in press as: E. Duemichen, et al., J. Chromatogr. A (2014), http://dx.doi.org/10.1016/j.chroma.2014.05.057
G Model
CHROMA-355449; No. of Pages 12 ARTICLE IN PRESS
12 E. Duemichen et al. / J. Chromatogr. A xxx (2014) xxx–xxx

[20] F.C.-Y. Wang, W.C. Buzanowski, J. Chromatogr. A 891 (2000) 313. [36] C. Bicchi, C. Cordero, C. Iori, P. Rubiolo, P. Sandra, J. High Resolut. Chromatogr.
[21] K. Robards, P.R. Haddad, P.E. Jackson, Principles and Practice of Modern Chro- 23 (2000) 539.
matographic Methods, Academic Press, London, Boston, MA, 1994. [37] B. Tienpont, F. David, C. Bicchi, P. Sandra, J. Microcolumn Sep. 12 (2000) 577.
[22] T.P. Wampler, Applied pyrolysis handbook, CRC Press, Taylor & Francis, Boca [38] S.V. Levchik, E.D. Weil, M. Lewin, Polym. Int. 48 (1999) 532.
Raton, FL, 2007. [39] F. Samperi, C. Puglisi, R. Alicata, G. Montaudo, Polym. Degrad. Stab. 83 (2004)
[23] H.L. Chung, J.C. Aldridge, Instrum. Sci. Technol. 20 (1992) 123. 11.
[24] L.F. Whiting, P.W. Langvardt, Anal. Chem. 56 (1984) 1755. [40] G. Montaudo, C. Puglisi, F. Samperi, Polym. Degrad. Stab. 42 (1993) 13.
[25] W.H. McClennen, R.M. Buchanan, N.S. Arnold, J.P. Dworzanski, H.L.C. Meuzelaar, [41] G. Botelho, A. Queirós, S. Liberal, P. Gijsman, Polym. Degrad. Stab. 74 (2001) 39.
Anal. Chem. 65 (1993) 2819. [42] P. Gijsman, G. Meijers, G. Vitarelli, Polym. Degrad. Stab. 65 (1999) 433.
[26] L. Costa, G. Camino, L. Trossarelli, J. Chromatogr. 279 (1983) 125. [43] D.H. MacKerron, R.P. Gordon, Polym. Degrad. Stab. 12 (1985) 277.
[27] G. Reggers, M. Ruysen, R. Carleer, J. Mullens, Thermochim. Acta 295 (1997) 107. [44] I. Goodman, J. Polym. Sci. 13 (1954) 175.
[28] W.V. Ligon, M.C. George, J. Polym. Sci., A: Polym. Chem. 16 (1978) 2703. [45] I. Goodman, J. Polym. Sci. 17 (1955) 587.
[29] M. Gröning, M. Hakkarainen, J. Chromatogr. A 1052 (2004) 151. [46] M.A. Schaffer, E.K. Marchildon, K.B. McAuley, M.F. Cunningham, J. Macromol.
[30] W. Klusmeier, K.H. Ohrbach, A. Kettrup, Thermochim. Acta 103 (1986) 231. Sci., C 40 (2000) 233.
[31] J.N. Smith, G.V. White II., M.I. White, R. Bernstein, J.M. Hochrein, J. Am. Soc. Mass [47] L.H. Peebles, M.W. Huffman, J. Polym. Sci., A-1: Polym. Chem. 9 (1971) 1807.
Spectrom. 23 (2012) 1579. [48] H.-R. Schulten, B. Plage, J. Polym. Sci., A: Polym. Chem. 26 (1988) 2381.
[32] G.V. White Ii, R.L. Clough, J.M. Hochrein, R. Bernstein, Polym. Degrad. Stab. 98 [49] F. Wiloth, E. Schindler, Chem. Ber. 100 (1967) 2373.
(2013) 2452. [50] F. Wiloth, E. Schindler, Chem. Ber. 103 (1970) 757.
[33] G.V. White Ii, J.N. Smith, R.L. Clough, J.A. Ohlhausen, J.M. Hochrein, R. Bernstein, [51] M. Gröning, M. Hakkarainen, J. Chromatogr. A 932 (2001) 1.
Polym. Degrad. Stab. 97 (2012) 1396. [52] H. Ohtani, T. Nagaya, Y. Sugimura, S. Tsuge, J. Anal. Appl. Pyrolysis 4 (1982) 117.
[34] M. Gröning, Solid-phase microextraction (SPME) applied to studies of [53] D.H. Mackerron, R.P. Gordon, Polym. Degrad. Stab. 12 (1985) 277.
polyamide 6.6 long-term thermo-oxidation and in plant-recycling PhD Thesis [54] H. Ohtani, T. Kumira, S. Tsuge, Anal. Sci. 2 (1986) 179.
at the KTH Royal Institute of Technology, Stockhom, Sweden, 2002. [55] M. Maciejewski, A. Baiker, Thermochim. Acta 295 (1997) 95.
[35] E. Baltussen, P. Sandra, F. David, C. Cramers, J. Microcolumn Sep. 11 (1999) 737. [56] F. Eigenmann, M. Maciejewski, A. Baiker, J. Therm. Anal. Calorim. 83 (2006) 321.

Please cite this article in press as: E. Duemichen, et al., J. Chromatogr. A (2014), http://dx.doi.org/10.1016/j.chroma.2014.05.057