Quantitative calibration of spectroscopic signals in combined TG-FTIR system

Article abstract of DOI:10.1016/j.tca.2005.10.018

In thermoanalytical investigations the determination of the composition of the evolved gases is very important, especially when investigating decomposition processes or gas-solid reactions occurring in multi-component systems. The potential of simultaneous techniques, enabling the qualitative analysis of evolved species, such as TG-MS or TG-FTIR has been further improved by the introduction of the pulse thermal analysis (PulseTA). This method provides a quantitative calibration by relating the mass spectrometric or FTIR signals to the injected quantity of probe gas. The influence of several experimental parameters such as concentration of the analyzed species, temperature and flow rate of the carrier gas on FTIR signals has been investigated. The reliability of quantifying FTIR signals was checked by relating them to the amount of evolved gases measured by thermogravimetry. In order to extend the opportunities for quantifying FTIR signals, the possibility of the injection of liquids into the carrier gas stream was studied. The linear dependence between the injected amount of liquids and the integral intensity of spectroscopic signals (peak area) enabled easy quantification of FTIR data. Systematic studies on a new method based on isothermal vaporization of liquids further widen the application range of in situ calibrations.

Full text of DOI:10.1016/j.tca.2005.10.018

Thermochimica Acta 440 (2006) 81–92
Quantitative calibration of spectroscopic signals in
combined TG-FTIR system
F. Eigenmann, M. Maciejewski, A. Baiker∗
Department of Chemistry and Applied Biosciences, ETH-Hoenggerberg, Swiss Federal Institute of Technology, 8093 Zu¨rich, Switzerland
Received 30 August 2005; received in revised form 14 October 2005; accepted 31 October 2005
Abstract
In thermoanalytical investigations the determination of the composition of the evolved gases is very important, especially when investigating
decomposition processes or gas–solid reactions occurring in multi-component systems. The potential of simultaneous techniques, enabling the
qualitative analysis of evolved species, such as TG-MS or TG-FTIR has been further improved by the introduction of the pulse thermal analysis
(PulseTA®). This method provides a quantitative calibration by relating the mass spectrometric or FTIR signals to the injected quantity of probe
gas.
The influence of several experimental parameters such as concentration of the analyzed species, temperature and flow rate of the carrier gas
on FTIR signals has been investigated. The reliability of quantifying FTIR signals was checked by relating them to the amount of evolved gases
measured by thermogravimetry. In order to extend the opportunities for quantifying FTIR signals, the possibility of the injection of liquids into the
carrier gas stream was studied. The linear dependence between the injected amount of liquids and the integral intensity of spectroscopic signals
(peak area) enabled easy quantification of FTIR data. Systematic studies on a new method based on isothermal vaporization of liquids further
widen the application range of in situ calibrations.
© 2005 Elsevier B.V. All rights reserved.
Keywords: TG-FTIR; Hyphenated techniques; Quantitative evolved gas analysis; Calibration techniques; Pulse thermal analysis combined with Fourier transmission
infrared spectroscopy; Thermal analysis
1. Introduction
sulfate. The water molecules after passing over calcium carbide
were transformed into acetylene and the amount of the generated
acetylene was measured.
A variety of efforts [5–7] were made in the following years to
gain benefits from coupling complementary techniques. Com-
pared to other methods of evolved gas analysis (EGA), FTIR
is fast, sensitive and can detect almost all molecules except
homonuclear diatomic gases.
In the strive for extending the opportunities of coupled tech-
niques, successful quantification has been achieved recently,
applying FTIR [8–10] or MS [8,11–14] spectrometry. Such
quantification is especially important when investigating mul-
tistage decomposition reactions, or, when two or more gases
evolve simultaneously. Insuchcases, the thermoanalytical meth-
ods alone fail for the quantitative description of the investigated
process and have to be coupled with MS or FTIR enabling quan-
titative and qualitative determination of the evolved gases.
The conventional method of quantification of FTIR spec-
tra is based on time-consuming calibration, which requires the
preparation and measuring of reference mixtures with varying
The main advantage of the coupled techniques TA-MS
and TA-FTIR is the identification of gaseous products, which
together with the thermal effects (DTA) and mass changes (TG),
allows interpreting the course of the investigated reactions. The
qualitative analysis is routinely done by comparing recorded
spectra with key fragment ions and their relative intensities for
known elements and compounds (MS) and with reference spec-
troscopic signals (FTIR). The development of TG-FTIR and
TG-MS hyphenated techniques have been reviewed in detail by
Materazzi et al. [1–3]. The first application of coupling thermo-
gravimetric and spectroscopic analysis was reported in 1969 by
Kiss [4]. He has monitored the content of water and ammonia in
evolved gases formed during the decomposition of ammonium
paramolybdate and paratungstate and the dehydration of copper
∗
Corresponding author. Fax: +41 1 632 1163.
E-mail address: a.baiker@chem.ethz.ch (A. Baiker).
0040-6031/$ – see front matter © 2005 Elsevier B.V. All rights reserved.
doi:10.1016/j.tca.2005.10.018

82
F. Eigenmann et al. / Thermochimica Acta 440 (2006) 81–92
concentrations. Therefore, several expensive gas standards and
an accurate flow control are required to obtain reliable values
of concentrations in the FTIR gas cell. Moreover, due to the
influence of several experimental parameters on the intensity
of FTIR signal (as e.g. change of the baseline), the calibration
has to be frequently repeated. In order to avoid these prob-
lems, pulse calibration techniques were investigated by de la
Guardia and co-workers [15,16]. They developed a direct and
simultaneous application to quantify different fluids, e.g. butyl
acetate, toluene and methyl ethyl ketone based on the use of
vapor phase FTIR measurements. However, these experiments
were only carried out with FTIR spectrometer without cou-
pling it to the thermoanalyzer. First quantification of evolved
gases based on the pulse technique combined with TG systems
was performed by Maciejewski et al. [12,13]. Quantification
of the FTIR signals in the FTIR-TG system was targeted by
Eigenmann et al. at the beginning of 2000 [17] by decom-
posing a known amount of solids with known stoichiometry
of the decomposition process and by injection of liquids into
the TA-FTIR system. The extension to FTIR-TA coupled tech-
nique was achieved by Marsanich et al. [18]. They applied also
the gas-pulse calibration using ammonia, carbon monoxide and
carbon dioxide. They introduced the vaporization technique for
calibration of the FTIR signals by liquids samples using the
vaporization technique. The applied procedure enabled indirect
quantification of evolved gases. The injections (calibrations)
were not carried out in situ that is during the same experimental
run in which the decomposition of the investigated samples was
studied.
The opportunity of injecting a known amount of a particular
gas or liquid into the carrier gas stream provides a quanti-
tative calibration by relating the FTIR signal to the injected
quantity of probe species. A linear relation between spec-
tral absorbance at a given wavenumber and low concentra-
tion of gaseous compounds is postulated by Lambert–Beer’s
law: Aν = c × d × εν, where A is the integral absorbance over
a wavenumber interval ν (cm−1), c the concentration of the
absorbing species (mol l−1), d the optical path length (cm)
and ε the integral molar absorption coefficient (l mol−1 cm−1).
Three different methods of quantifications in coupled TA sys-
tems have been applied so far. They are based on calibration
by:
liquids and solids. Additionally, the influence of experimental
parameters such as the concentration of the analyzed species,
temperature and flow rate of carrier gas on the FTIR signals
was investigated. Furthermore, the PulseTA® [13] developed
in our group was expanded to new applications, involving liq-
uid components. This new technique, in contrast to the known
vaporization methods, is based on isothermal or non-isothermal
calibration, which allows in situ calibration before the reaction
(decomposition) of the target compound has already started.
2. Experimental
Experiments were carried out on a Netzsch TG 209 and STA
449 analyzer equipped with two pulse devices enabling injection
of a certain amount of one or two different gases or gaseous
mixtures into the carrier gas stream flowing through the system
[13]. The amount of injected gas could be changed from 0.01 to
2.0 ml. Volumes of 0.25, 0.5, 1.0 and 2.0 ml were mainly used.
Additionally two heated injection ports were installed before
and after (on the top of) the thermoanalyzer allowing injections
of liquids with a Hamilton syringe CR-700-20. The amount of
injected liquid was usually in the range from 1 to 10 ␮l.
The flow rate was controlled by mass flow controllers,
Brook’s model 5850E, based on a thermal mass flow sensing
technique. A helium (purity 99.999%, PanGas) carrier gas flow
rate of 50 ml min−1 was used. The thermoanalyzer was con-
nected by a heated (ca. 200 ◦C) transfer-line to a Bruker Vector
22 FTIR spectrometer.
The FTIR apparatus is equipped with a MCT detector and
a specifically developed low-volume gas cell (8.7 ml) with a
123 mm path length and ZnSe windows. To avoid condensation
of low volatile compounds the cell was heated to a constant
temperature of 200 ◦C. The whole FTIR compartment was con-
tinuously purged by nitrogen and additionally molecular sieves
were used to minimize the water and carbon dioxide background
in the recorded spectra. The resolution of the collected spectra
was set to 4 cm−1 and co-addition of four scans per spectrum
was applied. As a consequence spectra were recorded with a
temporal resolution of about 6 s, depending on the integration
methods. The IR acquired interferograms constantly during the
time of the test, and the residence time of the injected species in
the gas cell was about 10 s (flow rate: 50 ml min−1). In this way,
the spectra of all gases were averaged, without cutting peaks of
the calibration pulses, which would lower the accuracy of the
quantification of the evolved species.
A) decomposing solids with well-known stoichiometric reac-
tion e.g. calcium carbonate [12,14], calcium oxalate
[12,14,19,20] or sodium bicarbonate [12,19] in TG-MS sys-
tems,
2.1. Pulse calibration
B) injection of a known amount of the calibration compound
into the carrier gas stream flowing through the thermoan-
alyzer and evolved gas analysis device (MS [12] or FTIR
[17,18]),
C) vaporization of known amounts of liquids in the thermoan-
alyzer itself [18,21].
For the gas injections a home-made device was placed before
the thermoanalyzer. It contains a rotary sample valve enabling a
carrier gas to purge the loop of a given volume, which had been
previously filled with the calibration gas of known composition.
In order to quantify the FTIR signals, pulses of a known volume
were injected before and/or after the decomposition of the inves-
tigated sample. For injections of the liquids another home-made
device consisting of a T-port tube was applied, which allows
quantifying the evolved gases on the same experimental set-up.
In order to extend and validate the opportunities of quantification
ofevolvedgasesinTG-FTIRsystems, wepresenttheresultsofin
situ calibrations applying different techniques and using gases,

F. Eigenmann et al. / Thermochimica Acta 440 (2006) 81–92
83
The T-port tube heated to ca. 100 ◦C was closed with a sep-
3.1. Quantification of FTIR signals using gases
tum on the top, enabling the injections of liquids or dissolved
solids with a syringe. Another injection port was installed on
the top of the thermoanalyzer, allowing injections of liquids
after the thermoanalyzer, therefore, not being in contact with
the sample and passing directly to the FTIR cell. This device
was especially used for dissolved solids and liquids possessing
higher boiling points (>150 ◦C) in order to avoid condensa-
tion on cold spots in the thermoanalyzer. The carrier gas flow
was set to 20 ml min−1. The flow rate has to be relatively low,
because the residence time in the transfer line is very short,
which, in turn, results in a very sharp and narrow FTIR signal
compared to the one obtained upon injection before the ther-
moanalyzer. In that case, the injected gas has to pass through a
much larger volume (TA) and FTIR response has a longer tail-
ing due to prominent backmixing phenomena occurring in the
TA-chamber.
The application of PulseTA® for quantitative interpretation
of FTIR signals is illustrated by comparing results obtained
by means of thermogravimetry and FTIR in a simultaneous
PulseTA®-FTIR experiment. 36.15 mg of NaHCO3 was decom-
posed under helium with a heating rate of 10 K min−1. The
decomposition of sodium bicarbonate occurs according Eq. (1):
2NaHCO3 → Na2CO3 + H2O + CO2
(1)
In order to quantify the FTIR signal of CO2, two 1 ml pulses
of CO2 were injected before and after the decomposition of
NaHCO3 (Fig. 1A).
2.2. Pinhole calibration
This method was used to calibrate the FTIR spectra when
making calibration with organic liquids or water. Usually 25 mg
of the investigated liquid was placed in the crucible closed by the
lidswithpinholeswithdifferentdiameters(0.1, 0.5, 1and2 mm).
The heating and flow rate were 10 K min−1 and 50 ml min−1
respectively.
,
2.3. Differential calibration
Last method allows carrying out the calibration of the FTIR
signal and investigation of the decomposition (or desorption)
process during one run. The set-up consists of two crucibles: the
first contains the liquid used for the calibration and the second
one the investigated sample. The quantification is based on the
FTIR signal obtained during vaporization of the liquid at low,
constant temperature (25–50 ◦C) or during a non-isothermal run
carriedoutwithlowheatingrateuptoatemperaturelyingsignifi-
cantly below the decomposition temperature of the solid sample.
The rate of vaporization can be changed in a very broad range
by varying the temperature and the pinhole diameter. The exper-
iments were normally performed applying an isothermal period
of about 30 min followed by the decomposition period with a
heating rate of 10 K min−1. The sample mass of the calibrated
liquid was in the range of 10–25 mg and the mass of decom-
posed sample was 100–200 mg. In order to avoid overlapping
of the calibration and measuring processes, it was important
that the liquid reference material used for the calibration had
fully evaporated during the isothermal run, before beginning of
the decomposition of the solid. The carrier gas flow was set to
25–50 ml min−1
.
3. Results and discussion
Fig. 1. (A) 3D FTIR-diagram of the calibration (two pulses) and the evolved
gases during the decomposition of NaHCO3. (B) Relationship between amount
of evolved CO2 in the sample and intensity of CO2 traces recorded in TA-FTIR
system. (C) Amount of CO2 found by FTIR as a function of amount of CO2
present in the sample.
WehaveinvestigatedthethreemethodsofcalibrationofFTIR
signals described above. The presented examples describe the
application of both, gases and liquids in the calibration proce-
dure.

84
F. Eigenmann et al. / Thermochimica Acta 440 (2006) 81–92
guaranteed [22]. Best correlation was obtained at 3600 cm−1
for CO2 (r2 = 0.9998), whereas the band at 2350 cm−1 gave also
goodresults(r2 = 0.9952). NH3 measurementsshowedgoodcor-
relation at 1630 cm−1 (r2 = 0.9994), but at 950 cm−1 the linear
dependence was slightly poorer (r2 = 0.9909).
The possibility of the exact quantification of FTIR signals by
means of PulseTA® increases significantly the potential of the
coupled TA-FTIR method. A further application of PulseTA®
for the quantification of evolved gases is illustrated by the results
obtained for the decomposition of zinc oxalate dihydrate (Eq.
(3)):
The mean value of the integral intensities of the injected
pulses was 1727 a.u., while the integral intensity of the signal of
the evolved CO2 was 9236 a.u. The temperature of the injected
gas was 28 ◦C. The amount of CO2 formed during the decompo-
sition calculated from these data corresponds to 9.52 mg, which
agrees well with the stoichiometric value of 9.47 mg, confirming
the accuracy of the quantification method. In order to check the
dependence between the intensity of FTIR signals and amount
of evolved species, different amounts of NaHCO3 were decom-
posed. The resulting CO2 traces are presented in Fig. 1B.
Fig. 1C shows the relation between the amount of evolved
CO2 derivedfromFTIRsignalsandtheamountofCO2 presentin
the sample. The obtained linear dependence between the integral
intensities of CO2 traces (peak areas) and amount of evolved gas
has been hold in a wide range: in the presented case from 0.64
to 12.34 mg of evolved CO2.
ZnC2O4·2H2O → 2H2O + CO + CO2 + ZnO
(3)
To calibrate the FTIR signals, 1 ml pulses of CO and CO2
were injected before decomposition into the carrier gas stream.
The traces of CO and CO2 pulses were normalized in order to
compare the integral intensities of traces of CO and CO2 formed
during decomposition. The resulting curves, presented in Fig. 2,
indicate that the amounts of both evolved species are equal. This
agrees well with the reaction stoichiometry (Eq. (3)). Note that
confirmation of this stoichiometry is difficult when applying
conventional mass spectrometry. Due to the fragmentation of
CO2 species the determination of the composition of CO and
CO2 mixture needs very time consuming calibration of the MS
signals and accurate determination of the fragmentation pattern
of CO2 molecules in the applied mass spectrometer. The use
of tabulated data of CO2 fragmentation for calibration purposes
without their experimental corroboration can lead to uncertain
results due to overlapping of current ions of m/z = 28 resulting
from the presence of CO and fragmentation of CO2.
The gas-pulse calibration method is a single point in situ
calibration, where calibration and investigated reaction are per-
formedinthesameexperimentalrun. Thisenablesquantification
of the spectroscopic signals without taking into account several
parameters such as flow rate and kind of the carrier gas, rate
of the FTIR data acquisition or other experimental factors. In
order to carry out the quantification in the linear range, where
of Lambert–Beer’ law is obeyed, the calibration should be made
with the smallest possible amounts of injected calibration gases.
The influence of the temperature (in the TA chamber) on the
shape and intensity of the FTIR signals was investigated. At each
temperaturetwopulsesofCO2 havebeeninjectedintothecarrier
gas stream; the integral intensity I (Eq. (2)) was independent of
the temperature (50 ◦C: 101/100 a.u. 350 ◦C: 101/99 a.u. 950 ◦C:
101/98 a.u.)
t2
ꢁ
ꢂ
ꢀ
ꢀ
ν2
I =
A(ν) dν dt
(2)
ν1
t1
The results confirmed the conclusion by Maciejewski et
al. [12] for the TA-MS system that in the range room
temperature–1000 ◦C the shape of the spectra is slightly chang-
ing but the integral intensities remain constant. Note that at
higher temperature the backmixing and diffusion in the TA are
enhanced, which in turn affects the residence time distribution
of the calibration gas.
The comparison of the integration of CO2 and NH3 signals
at different wavenumbers presented in Table 1 clearly illustrates
the potential source of errors in the calibration, resulting from
application of too strong vibrational signals. For species giving
very intensive signals as e.g. CO2, it is relatively easy to leave
the range where linearity of the Lambert–Beer relationship is
Table 1
Pulse calibration: FTIR integral intensity recorded during injections of CO2 and NH3
Loops (ml)
CO2 (1)
CO2 (2)
R.S.D. (1) (%)
R.S.D. (2) (%)
0.25
0.50
1.00
2.00
0.28
0.56
1.03
1.96
0.26
0.52
1.00
1.99
10.5
9.3
3.0
3.1
3.1
0.4
−1.5
−0.3
Linear correlation r2
0.9952
0.9998
Loops (ml)
NH3 (1)
NH3 (2)
R.S.D. (1) (%)
R.S.D. (2) (%)
0.25
0.50
1.00
2.00
0.27
0.54
1.10
1.94
0.25
0.52
1.02
1.98
9.6
7.3
9.6
−8.6
−3.1
−6.9
2.3
−3.0
Linear correlation r2
0.9909
0.9994
Slopes of linear regression through 0 were normalized to 1 in order to simplify comparison of the values, gained by different integration methods, for CO2 (1:
2350 cm−1) and (2: 3600 cm−1), for NH3 (1: 950 cm−1) and (2: 1630 cm−1).

F. Eigenmann et al. / Thermochimica Acta 440 (2006) 81–92
85
3.2.1. Pulse calibration
In order to extend the opportunities of quantifying FTIR sig-
nals, the possibility of injection of liquids into the carrier gas
stream was studied. The signals resulting from the injection
of pulses of ethanol into the carrier gas stream are shown in
Fig. 3A. In order to check the dependence between amount
of injected liquid and resulting intensity of FTIR signals, the
amount of ethanol was changed from 5 to 80 ␮l. The traces of
ethanol obtained by applying different integration modes at the
characteristic wavenumbers are shown in Fig. 3B. The averaged
integral intensities of the signals as a function of injected vol-
umes are presented in Fig. 3C. The applied integration modes for
thecharacteristicbandsforethanolrevealedalinearrelationship:
best correlations were obtained at 3670 cm−1 (r2 = 0.9999) and
at 2970 cm−1 (r2 = 0.9999), whereas, the linearity obtained with
the band at 1055 cm−1 was slightly less perfect (r2 = 0.9996).
Other calibrations (injection from 1 to 8 ␮l) were performed
with hexane (b.p. 69 ◦C), 1-propanol (b.p. 97 ◦C) and water (b.p.
100 ◦C). The linear dependence between amount of injected liq-
uid and integral intensity of FTIR signal (peak area) enables the
quantification of FTIR data (Fig. 3D).
It is worth mentioning that the injection of various gases or
liquids allows also creating own libraries facilitating identifica-
tion of gaseous products evolving during reaction or desorption.
The quantification of the FTIR spectra by injecting liquids
was extended to binary organic mixtures. Fig. 4A and B shows
FTIR spectra of binary mixtures of methanol and methyl for-
mate and their characteristic traces due to integration of the
bands at 1030 and 1180 cm−1, respectively. Fig. 4C shows a
linear dependence of the intensity of the FTIR signals on the
amount of injected liquids (methanol and methyl formate), and
finally, Fig. 4D depicts the dependence of the FTIR absorbance
ratio on the mass ratio of methanol in the mixture (wt.%). These
results corroborate that if the selected characteristic FTIR bands
of two or more investigated compounds do not overlap than the
quantification of the resulting spectra is also possible in the case
of binary mixtures. Of course, as mentioned by Barontini et al.
[24], it is necessary to take into account possible factors influ-
encing the composition of the gaseous phase during evaporation
of liquids (as e.g. formation of azeotrops). It is in principle pos-
sible to obtain the vapor–liquid equilibrium data, but a serious
limitation is the poor mixing (natural convection only). Sample
stirring is not possible in the commonly used TG-FTIR systems.
As an example of quantification of FTIR spectra by pulse cal-
ibration with liquids, we present the determination of the amount
of water evolved during the decomposition of sodium bicarbon-
ate and dehydration of copper sulfate pentahydrate.
Fig. 2. (A) 3D FTIR-diagram of the calibration (first pulse: CO2, second pulse:
CO) and the evolved gases during the decomposition of ZnC2O4·nH2O. (B)
Normalized CO and CO2 traces obtained during calibration (injection of 1 ml
CO and CO2) and decomposition of ZnC2O4·nH2O; TG curve is shown in the
lower part of the plot.
Additionally, the amount of the analyzed substance should be
chosen in such away that the amount of evolved gases are similar
to those used for the calibration, especially if adsorption bands
with high intensity are used.
Marsanich et al. [18] also found a linear relationship between
FTIR signals and the amount of injected gases with pulse cali-
bration for NH3, CO2 and CO (0–35 ␮mol). In contrast to our
strategy, they injected the gases not in the carrier gas flow pass-
ing through the TGA system. They applied a separate gas stream
that was directly passed to the FTIR cell.
The in situ method described in this chapter was recently
applied for the quantification of NO2, NO and NH3 during
the selective reduction of NO by NH3 over manganese-cerium
mixed oxides [23].
3.2. Quantification of FTIR signals by liquids
Fig. 5A shows the water traces recorded during two calibra-
tion pulses and decomposition of 106.5 mg NaHCO3 according
to reaction (1). The mean value of the integral intensities of the
injected pulses was 445 a.u., while the integral intensity of the
signal of the evolved H2O was 515 a.u. The amount of injected
water was 10.0 mg (ca. 10 ␮l). The amount of H2O formed dur-
ing the decomposition calculated from these data corresponded
to 10.87 wt.%, which agrees well with the stoichiometric value
of 10.69 wt.%, confirming the accuracy of the quantification
method. The stoichiometry of the decomposition was corrob-
For gases the quantification of FTIR signals is possible either
by conventional calibration (varying concentrations of target
gases) or, as described above, by the injection of a known
amount of the calibrating gas into the carrier gas stream. For
the quantification of FTIR signals by the liquids, we applied
several methods, pulse calibration and two methods based on
the evaporation of liquids: pinhole and differential calibrations.
These methods were studied applying different organic liquids
and water.

86
F. Eigenmann et al. / Thermochimica Acta 440 (2006) 81–92
Fig. 3. (A) 3D FTIR-diagram of the calibration of ethanol (one pulse: 5 ␮l).
(B) Ethanol traces due to different integration methods at three characteristic
wavelengths for ethanol (1055, 2970 and 3670 cm−1). (C) Linear dependence
between the integral intensity of the FTIR signals on the amount of injected
ethanol. (D) Linear dependence between the intensity of the FTIR signals on the
amount of injected liquids (1-propanol, water and hexane).
Fig. 4. (A) FTIR spectra of binary liquid mixtures of methanol and methyl-
formate. (B) Methanol and methylformate traces as a function of the mixtures
composition. (C) The dependence between the integral intensity of the FTIR
signals on the mixture composition. (D) Resulted FTIR absorbance ratio vs.
mass ratio (W) of methanol in the binary mixtures.

F. Eigenmann et al. / Thermochimica Acta 440 (2006) 81–92
87
Fig. 6 shows the mass losses and FTIR signals recorded dur-
ing vaporization of acetone (A), ethanol (B), water (C) and
hexanol (D) from the pans with different pinhole sizes. Due to
the low boiling points of acetone and ethanol remarkable mass
losses were recorded immediately after starting measurement.
In this case, especially when working without lid or using larger
pinhole sizes, it is difficult to determine an accurate mass loss
during evaporation (cf. Fig. 6A) because a certain amount of
the substance will evaporate even at room temperature, which in
turn, will disallow finding the relationship between the amount
of evaporated liquid and intensity of FTIR traces. A lid with
a pinhole size of 0.1 mm was necessary to prevent the acetone
vaporization at room temperature and obtain the FTIR trace,
which started at the baseline and enabled proper integration of
the recorded peak. For ethanol a pinhole size of 2.0 mm was
enough and for water and hexanol even no lids were necessary
to achieve accurate calibration. To avoid uncertainty caused by
the mass loss in the beginning of the experiment due to the high
evaporation rate (vapor pressure) of some liquids, the amount of
the calibrating substance used was higher than 25 mg. During
experiment settling and some isothermal part of the temperature
ramp the mass loss of the sample was not recorded. The cal-
ibration measurements were started exactly when the mass of
the calibration liquid, measured by the internal TA balance, was
precisely 25.0 mg.
Fig. 5. Pulse calibration: water-traces recorded during decomposition of
106.5 mg NaHCO3 (A) and 48.9 mg CuSO4·5H2O (B). Heating rate:
10 K min−1, carrier gas: 50 ml min−1, injected water volumes: 10 ␮l (A), 2.5 ␮l
(B).
The detailed results of the pinhole calibrations are listed in
Table 2, the relative standard deviation (R.S.D.) was smaller than
4% for all four liquids with boiling points ranging from 56 ◦C
(acetone) to 154 ◦C (hexanol). The vaporization at low tempera-
tures at the beginning of the experiment or even before its start,
which can introduce a significant error during the pinhole cal-
ibration procedure, was also reported by Slager and Prozonic
[21]. They found that for highly volatile compounds e.g. ace-
tone and methanol significant evaporation occurred before the
TGA run was actually started.
The carrier gas flow is a crucial factor in the calibration proce-
dure, it affects intensity and shape of the recorded signals. Low
flow leads to higher maximum intensity of the observed spectra,
but increases the residence time of the evaporated molecules in
the TA-FTIR system and leads to more prominent backmixing,
resulting in significant tailing of the signals. This phenomenon
can cause difficulties in the evaluation of the integral intensities
of the recorded signals when the reaction or desorption does not
occur in a single stage due to overlapping signals. The influence
of the flow on the shape and intensity of spectroscopic signals
orated by the TG analysis: the observed mass loss due to the
evolution of CO2 and H2O was 36.85 wt.%, while the stoichio-
metric value amounts to 36.90 wt.%.
Fig. 5B shows the water traces recorded during two calibra-
tion pulses and dehydration of 48.9 mg CuSO4·5H2O. At lower
temperatures the dehydration of the pentahydrate occurred in
two steps: the observed mass loss corresponded to 25.82 wt.%
and agreed well with the evolved amount of water found by
means of FTIR (26.35 wt.%).
During the further dehydration at about 220 ◦C (between
48 and 52 min) the residual water molecules evolved (dehy-
dration of monohydrate), as confirmed by experimental and
stoichiometric mass losses (7.17 and 7.21 wt.%, respectively).
The amount of water determined from the FTIR traces by com-
paring the integral intensity of the injected 2.41 mg (ca. 2.5 ␮l)
water and the intensity of the decomposition peak was 7.24 wt.%
which is in accord with the gravimetric results and the reaction
stoichiometry.
Table 2
Pinhole calibration: FTIR integral intensity recorded during vaporization of
25 mg of different liquids (acetone, ethanol, water and hexanol) from pans cov-
ered by lids with various pinhole sizes
3.2.2. Pinhole calibration
Sample
0.1 mm
0.5 mm
1 mm
2 mm
No lid
R.S.D. (%)
During the pinhole calibration the compound is not injected
into the carrier gas stream, but placed into the thermoanalyzer
and vaporized isothermally or by a linear heating ramp. For
liquid mixtures (cf. binary mixtures) it is important to select
a characteristic wavenumber interval to avoid overlapping of
related absorption bands.
Acteone
Ethanol
Water
98.7
97.2
102.0
104.2
–
–
–
101.1
99.1
–
101.5
102.8
98.8
98.6
100.9
99.2
1.52
2.38
1.78
3.73
Hexanol
102.7
94.8
98.1
100.2
Peak areas were determined at their characteristic wavenumber and normalized
to an average of 100 a.u.

88
F. Eigenmann et al. / Thermochimica Acta 440 (2006) 81–92
Fig. 6. Pinhole calibration of acetone (A), ethanol (B), water (C) and heaxanol (D), sample mass: 25 mg, heating rate: 10 K min−1, flow rate: 50 ml min−1, pinhole
diameters are indicated on the curves.
has been analyzed quantitatively by some of us for the MS-TA
system [11]. In order to study this influence on the calibration by
the pinhole method, experiments were carried out with different
carrier gas flow rates. The real progress of the evaporation is
reflected by the derivative TG curve (DTG) that indicates the
rate of the mass change. This progress has been compared to the
recorded FTIR spectra, which allowed to investigate the changes
of the shape of the recorded spectroscopic signals as a function
of the carrier gas flow rate.
Fig. 7 shows the quantification of the FTIR signals using pin-
hole (2.0 mm) calibration of ethanol when varying the carrier gas
flow rate. 25 mg sample of ethanol was heated with 10 K min−1
.
All DTG and FTIR curves were normalized in order to allow
comparison of flow rate influence on the shape of the FTIR
signals. The results presented in Fig. 7 clearly indicate that low
carrier gas flow leads to increase of the residence time in the TG-
FTIR system and results in a long tailing of the recorded FTIR
traces due to prominent backmixing. (cf. the comparison of the
real rate of the evaporation process represented by DTG curve to
the delayed FTIR traces for the flow of 20 ml min−1). However,
there exists a linear relationship between FTIR absorbances and
the reciprocal values of the flow rate (Fig. 7B). A detailed study
of the influence of mass transfer (by convection and diffusion)
on the relation between thermoanalytical and mass spectromet-
ric curves was reported by Roduit et al. [11].
Although, the best accordance between DTG and FTIR traces
was found for the highest flow rate of 65 ml min−1, in further
experiments generally a flow rate of 50 ml min−1 was used. This
flow rate proved to be a good compromise to achieve high analyt-
ical sensitivity and sampling frequency-allowing discrimination
of different TG-steps in a narrow temperature window. De la
Fig. 7. (A) Pinhole calibration of ethanol with varying carrier flow rate, sample
size: 25 mg, heating rate: 10 K min−1, DTG (thin) and FTIR (thick lines). (B)
The relationship between the integral intensity and the reciprocal flow defined
as the ratio of injected volume (1 ml) to the flow rate of carrier gas (ml min−1),
which are marked on the data.

F. Eigenmann et al. / Thermochimica Acta 440 (2006) 81–92
89
Guardia and co-workers [25] found that the carrier gas flow rate
is a critical parameter in vapor phase FTIR spectrometric anal-
ysis which affects the analytical sensitivity and the sampling
frequency. They selected a nitrogen flow rate of 50 ml min−1
,
which allows a 20 h−1 sampling frequency. Marsanich et al.
[18] used a similar flow rate (60 ml min−1) in his studies with
the same FTIR gas cell volume (8.7 ml) and FTIR path length
(123 mm). Generally, it can be stated that a higher diffusivity
of the evolving species in the carrier gas requires a higher min-
imal carrier gas flow rate to minimize the differences between
thermoanalytical (e.g. TG or DTG) and FTIR signals.
The main disadvantage of pinhole-calibration beside uncon-
trolled mass loss due to evaporation of the sample already during
experiment settling (see Fig. 6A) lies in the fact that the calibra-
tion and decomposition can not be done in one experimental
run. As shown by Slager and Prozonic [21], using this method
temperatures up to the boiling point of the calibrating substance
are necessary to obtain the dependence between mass loss and
integral intensity of the FTIR signal required for the calibration.
Application of such high temperatures can, however, affect the
process to be investigated, if one wants to follow it in situ. There-
fore we developed and applied another method in our laboratory,
which will be discussed next.
3.2.3. Differential calibration
Fig. 8. (A) Differential calibration of ethanol at various temperatures. Interval
time: 5 min, flow rate: 50 ml min−1. (B) Linear dependence between the integral
of absorbance over a specified time (5 min ethanol and 10 min water) on the
amount of evolved liquids during the same time interval. Corresponding tem-
peratures are indicated on the temperature curve. Crucible diameter was 7.5 mm
for ethanol and 17.0 mm for water calibration.
The differential calibration is based on the isothermal or
non-isothermal vaporization of a liquid compound at different
temperatures, with low heating rates and simultaneous moni-
toring, in differential time periods, corresponding mass losses
and intensities of FTIR signals. Fig. 8A shows the principle of
this method applied for isothermal calibration with ethanol at
temperatures between 30 and 70 ◦C. The integral absorbances
weremeasuredduringconstanttimeintervals(5 min)andplotted
against the mass loss ꢀm recorded during the same time inter-
vals by the TG signal. This dependence is presented in Fig. 8B
where additionally the same relationship for water collected dur-
ing the longer time intervals (10 min) is depicted. The very good
regression coefficients obtained for ethanol (r2 = 0.99998) and
water (r2 = 0.9975) indicate the high accuracy of the applied
calibration technique.
The correlation of the integral intensity of the FTIR signal
with the exact amount of evaporated liquid measured by ther-
mobalance in arbitrarily chosen periods of time allowed the
extension of the method for the simultaneous calibration and
measurement of evolved species in one experimental set-up.
Depending on the flow rate, the time lag between the beginning
of the reaction (observed on DTG curve) and the appearance
of the FTIR signals was in the range of 4.3 s (for flow of
65 ml min−1) to 14.1 s (flow of 20 ml min−1). Data collection
occurred during a few minutes after steady state of the mass
loss and FTIR signals had been reached. The 5.6 s delay (flow:
50 ml min−1) between the real course of the reaction (DTG
curve) and that of measured FTIR signals did not influence
the calibration results. At low flows the back mixing became
prominent leading to some tailing of the FTIR traces, but the
calibration was not significantly affected by this phenomenon.
The target sample and the calibration sample are placed in two
separate pans, which are weighed continuously as during con-
ventional thermogravimetric runs. The first calibration period
is performed in the isothermal mode at relatively low temper-
ature. During this period the relationship between the intensity
of the FTIR signal and the amount of evaporated liquid can be
determined in a few differential time periods which increases
the accuracy of the calibration. After total evaporation of the
calibration liquid, which is indicated by the end of the mass loss
on the TG curve, the second experimental stage is started, during
which the temperature in the system is raised according to the
chosen temperature ramp.
This in situ calibration is shown in Fig. 9, which depicts
the quantification by FTIR of the evolved water formed during
dehydration of sodium bicarbonate and copper sulfate pentahy-
drate. Fig. 9A shows the water trace recorded during calibration
and decomposition of 200 mg NaHCO3 according to reaction
(1). The integral of the FTIR absorbance over the interval of
10 min was 2039 a.u., while the integral intensity of the sig-
nal of the evolved H2O was 5045 a.u. The amount of evolved
water during the calibrating stage was 8.90 mg. The amount of
H2O formed during the decomposition calculated from these
data corresponds to 10.98 wt.%, which agrees well with the sto-
ichiometric value of 10.69 wt.%, corroborating the very good
accuracy of the differential quantification method.
Fig. 9B shows water traces recorded during differential cali-
bration and decomposition of 200 mg CuSO4·5H2O. At the first

90
F. Eigenmann et al. / Thermochimica Acta 440 (2006) 81–92
Fig. 9. Differential calibration: water-traces recorded during decomposition of
200 mg NaHCO3 (A) and 200 mg CuSO4·5H2O (B). Heating rate: 10 K min−1
,
carrier gas: 25 ml min−1 (A), 50 ml min−1 (B). The mass of water used for the
calibration at 50 ◦C was ca. 25 mg (A) and 35 mg (B). Interval time: 10 min.
Crucible diameter: 7.5 mm.
Fig. 10. (A) Extracted spectra recorded during decomposition of aspirin at
approximately 260 ◦C under wet (thick line) and dry conditions. The window
of the characteristic wavelength is marked and used for further time-resolved
traces. (B) Acetic acid and salicylic acid traces recorded during decomposition
of 21.5 mg aspirin under wet and dry conditions and calibration pulses injected
before decomposition: Injections were done after the TA, immediately before
the FTIR-transfer-line.
step of decomposition the observed mass loss corresponded to
24.90 wt.% and agreed well with the evolved amount of water
found by FTIR (25.64 wt.%).
During the decomposition starting at about 220 ◦C (45 min),
the water corresponding to the monohydrate evolved: the
observed mass loss agrees well with the stoichiometric value
(7.35 wt.% versus 7.21 wt.%, respectively). The amount of water
determined from FTIR traces based on calibration data was
7.26 wt.% (mass loss due to release of 12.68 mg of water results
in FTIR signal with an integral intensity 1071 a.u).
of the carrier gas by purging through water at 30 ◦C) conditions.
For the calibration a solution of 10 mg SA dissolved in 100 ml
ethanol was used. Two microliters of acetic acid and 2 ␮l of
the SA/ethanol mixture were injected after the thermoanalyzer
in order to avoid the crystallization of SA in the TA chamber.
Gravimetric traces (TG) show clearly a two-step-decomposition
of aspirin under dry conditions (43 and 57 wt.% for the first and
second step, respectively), whereas during the reaction under
wet conditions the decomposition steps are not so well resolved.
Acetic acid is evolved mainly in a first step, at higher tempera-
ture (ca. 260 ◦C) the main decomposition product is SA and at
about 360 ◦C additionally the evolution of phenol is observed.
The amount of evolved SA found by means of FTIR under wet
conditions (29 wt.%) was slightly higher than under dry condi-
tions (23 wt.%). However, the amounts of the acetic acid were
almost the same (24 and 25 wt.%) for both conditions. Asakura
Robeiro et al. [26] reported also a mass loss in two consecu-
tive steps, between 120 and 400 ◦C. The first step up to 260 ◦C
occurred with a loss of 42.8% followed by 57.2% in the second
step, which agrees with our results. They attributed the first mass
loss to the evolution of acetic acid and evaporation of ASA and
SA. They identified the evaporation process by means of chro-
matography but were not able to quantify it.
3.3. Quantification of the FTIR signals using solids
Some of the methods described above for the calibration of
FTIR signals could also be applied with solid calibration mate-
rials. In this case, before injection into the TG-FTIR system the
calibration substance must be dissolved in a solvent that pos-
sesses different characteristic FTIR bands than the calibration
substance.
Fig. 10A shows the spectra recorded during decomposition
of acetylsalicylic acid (ASA or aspirin) at 260 ◦C under dry and
wet conditions, and the spectra of the main products from IR-
library: acetic acid, salicylic acid (SA) and phenol. The bands
used for the quantification were 1800 cm−1 (acetic acid) and
3300 cm−1 (SA).
Fig. 10B depicts the acetic acid and SA traces recorded during
decomposition of 21.5 mg aspirin under dry and wet (saturation

F. Eigenmann et al. / Thermochimica Acta 440 (2006) 81–92
91
To characterize the influence of the gas flow on the FTIR
spectra of the SA in the same TG-FTIR system, salicylic acid
was heated under nitrogen with different flow rates and pres-
sures by Jackson and Rager [27], who found considerable peak
broadening because of the high boiling point of salicylic acid.
With higher flow rate the IR responses would be narrower, but in
TGA system high carrier flow rates can generate instability prob-
lems of the balance. The fact, that we injected the SA directly
on the top of the thermoanalyzer just before the heated transfer-
line (200 ◦C) to the FTIR spectrometer and kept the flow rate
high (50 ml min−1), leads to accurate quantification and avoids
condensation of SA on possible cold spots in the system. This
procedure slightly lowers the accuracy of the data acquisition
due to the sharpness of the peaks caused by the high flow rate
and the short distance between injection position and IR cell (cf.
acetic acid peaks in Fig. 10B), but condensation or resublima-
tion of SA would result in much larger error than some peak
cutting in the calibration segment.
injecting the fluid before or after the thermoanalyzer widens the
application range of the hyphenated TG-FTIR technique. Fluids,
injected before the TA, can react with the sample (allowing e.g.
the investigation of adsorption or gas–solid reactions), whereas
injection after the TA allows simple quantitative analysis of the
evolved species.
The pinhole calibration method significantly extends the
scope of liquid calibration. The results presented in this
study confirmed its applicability for quantification of volatile
compounds. The use of different diameters of the pinhole and
different temperature ramps during calibration provides the
opportunity of calibrating FTIR signals with liquids having very
different evaporation rate (vapor pressure) at room temperature.
Slager and Prozonic [21] achieved calibrations for methanol,
xylene, dimethyl formamide and dimethyl acetamid with this
method using a constant flow rate, however changing experi-
mental parameters (e.g. flow rate) limits this application. We
showed that the amount found by means of FTIR absorbance is
proportional to the reciprocal of the flow rate.
3.4. Comparison of the calibration methods
Another serious limitation of the pinhole method is its appli-
cation with substances possessing very low boiling point as e.g.
acetone. Part of the calibration agent evaporates already during
sample handling and experiment settling, rendering the relation
between the amount of evaporated liquid and the FTIR integral
signal inaccurate.
The differential method, described in this paper, further
extents the opportunities of liquid calibrations by the vaporiza-
tion methods described by Slager and Marsanich et al. [18,21].
Especially high volatile materials (e.g. acetone, ethanol) can
be applied without any limitations for quantification with this
method. For the calibration, the experimental data from only a
certain, arbitrarily chosen differential time interval are taken,
omitting the ill-defined start of the experiment, where already
significant mass losses may occur before the TG-FTIR data are
collected.
Table 3 compares the results of the quantification of the
FTIR spectra using two in situ methods, pulse calibration (PC)
and differential calibration (DC). The comparison is based on
the amounts of water evolved during decomposition of sodium
bicarbonate and copper sulfate pentahydrate. Both compounds
decompose with well-known stoichiometry and the thermo-
gravimetric curves recorded simultaneously allow comparing
the accuracy of the applied procedures used for the quantifica-
tion of the spectroscopic signals.
The relative standard deviation (R.S.D.) is calculated based
on TG and FTIR results, except the decomposition of sodium
bicarbonate, where TG results are not taken into account because
CO2 and water evolved simultaneously, i.e. in the same tem-
perature range. Here the R.S.D. is related to the expected stoi-
chiometric value only. All deviations between gravimetric and
spectroscopic results are lower than 3%, which corroborates the
good accuracy of the applied methods. The fact that both these
methods could be applied in situ, in one experimental run, under-
lines their potential. They offer a much less time-consuming
calibration procedure than generally applied. The opportunity of
4. Conclusions
The methods of quantification of FTIR signals based on the
pulse and differential techniques, described in this work, allow
in situ applications and easy and less time-consuming calibra-
tion procedure which can be used in different studies, such as
the quantification of adsorption phenomena [28], gas–solid reac-
tions [22] and decomposition of solids. The main advantages of
the pulse and differential calibration methods compared to the
commonly used quantification of the FTIR spectra are: simplic-
ity, accuracy and very good reproducibility. The fact that both
the calibration and experiment can be performed in a single run
significantly decreases the influence of artefacts and minimizes
experimental errors (e.g. shift of the base-line in FTIR cell, influ-
ence of changes in the carrier gas flow, etc.), which can bias the
quantification based on FTIR spectra. The described calibration
methods widen the application range of the TG-FTIR technique,
which is a very sensitive and reliable tool for characterizing reac-
tions involving solids and gases such as complex decomposition
reactions. The extension of the quantification by applying liq-
uid and solid–liquid solutions opens up further opportunities in
Table 3
Comparison between gravimetric and spectroscopic results of water evolution
obtained by pulse calibration (PC) or differential calibration (DC) during the
decomposition of sodium bicarbonate and copper sulfate pentahydrate
Samples
Step Method Stoichiometric TG
FTIR R.S.D.
(%)
NaHCO3
NaHCO3
CuSO4·5H2O
CuSO4·5H2O
CuSO4·5H2O
CuSO4·5H2O
1
1
1
1
2
2
PC
DC
PC
DC
PC
DC
10.69
10.69
28.84
28.84
7.21
–
–
10.87 1.68a
10.98 2.71a
25.82 26.35 2.05
24.90 25.64 2.97
7.17
7.35
7.24 0.98
7.26 1.22
7.21
The relative standard deviation (R.S.D.) is calculated based on TG and FTIR
results.
a
R.S.D. is related to the stoichiometric expected value.

92
F. Eigenmann et al. / Thermochimica Acta 440 (2006) 81–92
the area of quantification of spectroscopic signals in coupled
TG-FTIR/MS systems.
[11] B. Roduit, J. Baldyga, M. Maciejewski, A. Baiker, Thermochim. Acta
295 (1997) 59.
[12] M. Maciejewski, A. Baiker, Thermochim. Acta 295 (1997) 95.
[13] M. Maciejewski, C.A. Muller, R. Tschan, W.D. Emmerich, A. Baiker,
Thermochim. Acta 295 (1997) 167.
Acknowledgements
[14] J. Wang, B. McEnaney, Thermochim. Acta 190 (1991) 143.
[15] E. Lopez-Anreus, S. Garrigues, M. de la Guardia, Analyst 123 (1998)
1247.
[16] A. Perez-Ponce, F.J. Rambla, J.M. Garrigues, S. Garrigues, M. de la
Guardia, Analyst 123 (1998) 1253.
[17] F. Eigenmann, M. Maciejewski, A. Baiker, Proceedings of the Third
SKT 2000, Netzsch, Selb, Germany, 2000, p. 183.
[18] K. Marsanich, F. Barontini, V. Cozzani, L. Petarca, Thermochim. Acta
390 (2002) 153.
[19] M. Muller-Vonmoos, G. Kahr, A. Rub, Thermochim. Acta 20 (1977)
387.
[20] B.H. Li, R.D. Gonzalez, Catal. Lett. 54 (1998) 5.
[21] T.L. Slager, F.M. Prozonic, Thermochim. Acta 426 (2005) 93.
[22] F. Eigenmann, M. Maciejewski, A. Baiker, submitted for publication.
[23] F. Eigenmann, M. Maciejewski, A. Baiker, Appl. Catal. B 62 (2005)
311.
[24] F. Barontini, E. Brunazzi, V. Cozzani, Proceedings of the Fifth SKT
2003, Netzsch, Selb, Germany, 2003, p. 37.
[25] A. Perez-Ponce, J.M. Garrigues, S. Garrigues, M. De la Guardia, Analyst
123 (1998) 1817.
The authors would like to acknowledge the continuous
interest, numerous discussion and very valuable suggestions
given by Dr. A. Rager (Bruker), and the help in solving
experimental solutions offered by Netzsch GmbH (Germany).
Special thanks are also due to J. Gast (Bruker) and Dr. W.-D.
Emmerich (Netzsch) for their support in settling the TG-FTIR
system.
References
[1] S. Materazzi, R. Curini, G. D’Ascenzo, A.D. Magri, Thermochim. Acta
264 (1995) 75.
[2] S. Materazzi, Appl. Spectrosc. Rev. 33 (1998) 189.
[3] S. Materazzi, Appl. Spectrosc. Rev. 32 (1997) 385.
[4] A.B. Kiss, Acta Chim. Acad. Sci. Hung. 61 (1969) 207.
[5] C.A. Cody, L. Dicarlo, B.K. Faulseit, Am. Lab. 13 (1981) 93.
[6] P.B. Roush, J.M. Luce, G.A. Totten, Am. Lab. 15 (1983) 90.
[7] J.O. Lephardt, Appl. Spectrosc. Rev. 18 (1982) 265.
[8] M. Mittleman, Thermochim. Acta 166 (1990) 301.
[9] L.C.K. Liau, T.C.K. Yang, D.S. Viswanath, Appl. Spectrosc. 51 (1997)
905.
[26] Y. Asakura Ribeiro, A.C.F. Caires, N. Boralle, M. Ionashiro, Ther-
mochim. Acta 279 (1996) 177.
[27] R.S. Jackson, A. Rager, Thermochim. Acta 367 (2001) 415.
[28] F. Eigenmann, M. Maciejewski, A. Baiker, Thermochim. Acta 359
(2000) 131.
[10] M.F. Cai, R.B. Smart, Energy Fuels 7 (1993) 52.