Thermal Methods in Petroleum Analysis

by. Heinz Kopsch

CopyrightoVCH Verlagsgesellschaft mbH, 1995

2 Methods and instrumentation

Using thermogravimetry (TGA), the dependence of the change in sample weight (mass)
on the temperature during programmed temperature changes in a chosen gas atmosphere
can be measured. The first derivative of the weight (mass) signal with respect to time is
called derivative thermogravimetry (DTG) and is a criterion for the reaction rate. It is usual
to record both the slope of the weight (mass) versus the time or temperature (TGA), and the
differentiatoed curve versus the time or temperature (DTG). The heating rate dictates the
actual position of the TGA and DTG graphs; it is therefore advisable always to use the
same heating rate ( p ) so that different tests may be compared. For small sample weights
(masses), up to approximately 10 rng, a standard heating rate of 10 K/min is practicable.
This heating rate is slow enough to avoid any temperature gradient inside the sample while
permitting a reasonable utilization of the available workmg time. The shift to higher
temperatures of the TGA and DTG curves as a consequence of faster heating rates permits
calculation of the Arrhenius kinetic parameters and hence investigation of the reaction
kinetics (see chapter 3.3). Furthermore, the position of the TGA and DTG curves will be
influenced by the shape of the sample pan, especially by the ratio of surface to volume of
the sample, and lastly by the quantity of gas flowing through the oven (gas flow rate).
Therefore it is important that variations in sample quantity are minimized and that the gas
flow rate is maintained as constant as possible. However, the gas flow rate must not fall
below a certain minimum value in order to avoid condensation of evaporated sample
fractions on the hangdown of the sample holder or in the gas outlet tubes. The minimum
gas flow rate depends on the geometric shape of the oven and the position of the gas inlet
and outlet tubes and therefore differs for different instruments. If the gas flow rate is
sufficient, the evaporated portions of the sample will be discharged immediately and
therefore no equilibrium between liquid and vapor will be attained. As a consequence the
boiling (evaporation)temperature of the sample will decrease adequately.That can be used
to perform a simulated distillation (see chapter 3.1.2). However, the application of ther-
moanalytic methods is limited to substances having a start temperature of evaporation at
atmospheric pressure not far below 200 "C.Otherwise there is the risk that evaporation in
the gas flow will begin at room temperature and thus the correct start temperature of
evaporation (zero point of the TGA curve) cannot be ascertained.
In principle all except very corrosive gases can be passed through a thermobalance; in
practice the inert gases nitrogen, helium, and argon and the reactive gases air, oxygen, and
hydrogen will be used.
The weight calibration of thermobalances is done using standard weights. The tempera-
ture calibration is more difficult. The method using the Curie point temperature, as
4 2 Methods and Instrumentation

described in ASTM E 914-83, does not work if a magnetic field from outside the oven is
prevented from reciprocal action with the standard inside the oven, by the construction or
the material of the oven. Calibration using calcium oxalate monohydrate for standard is
very common, since it has exhibited three clearly-defined steps of weight loss during
heating (Fig. 2-1 to 2-3).:

Reaction Temperature Residue DTG Maximum

Range at Temperature
p = 10 K/min
("C) (%I ("C)
CaC,O,.H,O -+ CaC,04 + H,O t 135.. . 175 87.7 163
CaC,04 + CaCO, + C o t 4 6 3 . . ,502 68.5 49 1
CaCO, CaO+CO,T 660. . .740 38.4 722

As can be seen from the figures, the DTG maximum is found at conversions which are
smaller than the maximum conversion of the reaction step concerned. The onset temperatu-
res as well as the DTG maximum temperatures can be reproduced with coefficients of
variation < 2 % of the corresponding mean value.
The thermogravimetric experiments are run using open platinum sample pans. Pans
made from aluminium, platinum, quartz, glass, stainless steel etc. were also available. The

110 5

0
I00

-5
90

U
-10 :m:
0
no >
U
L -15
PI
a l l N T TABLE TGA
r(
x
70
-20

60
-25

50 -30

TGA
40 -35
__. I I I I
100 too 300 4bO 500 600 700 800 <
Deg c

Fig. 2-1: Thermogravimetry of CaC,O,. H,O

Plot of STA 780: TGA and DTA
Atmosphere: Argon 30 + 20 cm3/min
Heating Rate p: 10 K/min
 2 Methods and Instrumentation 5

110 -

100 -

90 -
-a
U . .3-
c
m 80 - >
4
0
L
m k
0
.2-
a
70 - 4l
.,
m .i-

60 - 0.0 -
-.l -
-
-
50
-.2 -

40 L.--y.r ----. --
9
0 100 200 300 400 500 600 700 800 900
Deg C
Fig. 2-2: Thermogravimetry of CaC,O,. H,O
Plot of STA 780: TGA and DTG
Atmosphere: Argon 30 + 20 cm3
Heating Rate p: 10 K/min/min

110 -
136.13 c
100 -

90 - 171.08 C 463.21 C

U
; 80 -
0
L
m
a 601.36 C
70 - 661.26 C

60 -

50 -

40 -
I 1 I I
0 100 200 300 400 500 600 700 800 I0
6 2 Methods and Instrumentation

catalytic effect of the pan material on the pyrolysis reaction could not be ascertained when
comparing the reaction in platinum and quartz pans, however, it could not be completely
excluded. All thermogravimetric experiments carried out by the author were run in plati-
num pans. Argon was used as the inert atmosphere. Oxidation experiments were run in air
because the reactions are too fast in oxygen.
The first stage of experiments was carried out using a Stanton-Redcroft TG 750 thermo-
balance connected to a three-pen recorder, recording weight (mass) loss (TGA), derivative
thermogravimetry (DTG), and temperature (q. For documentation the graphs of weight
(mass) versus temperature were drawn manually. Later on, the experiments were perfor-
med using a simultaneous thermal analyzer Stanton-Redcroft STA 780 (STA 1 000), which
is equipped with a personal computer for control, data sampling, and data evaluation (Table
2-1). Using this device the curves of TGA, DTG, and DTA (differential thermal analysis)
versus temperature can be plotted. Furthermore, the PC is equipped with extensive softwa-
re to evaluate the results under varying conditions.
Table 2-1: Thermobalances
Instrument: TG 750 Stanton-Redcroft
System: TGA + DTG up to 1000C
Pressure Range: normal pressure and vacuum
Heating Rates: 0.5 . . . 100 K/min
Recording: 3 pen recorder
Evaluation: manually
TGA empirical index numbers
evaporation
pyrolysis
oxidation
simulated distillation
DTG empirical index numbers
kinetics according to ASTM E 698-79

Instrument: STA 780 Stanton-Redcroft (STA 1000)

System: TGA + DTG + DTA simultaneous up to 1000C
Pressure Range normal pressure and vacuum
Heating Rates: 0.5 . 50 K/min
Recording: PC
Evaluation: PC
TGA empirical index numbers
evaporation
pyrolysis
oxidation
simulated distillation
kinetics according to Flynn & Wall
kinetics according to McCarty & Green
DTG empirical index numbers
kinetics according to ASTM E 698-79
DTA specific heat
conversion temperatures
kinetics according to Borchardt & Daniels
kinetics according to ASTM E 698-79
 2 Methods and Instrumentation 7

-5

1- i
i
2a- i
I
2-
t4
3-
1
12-
WALE! ..-
GA S
tN
IN

B- -7

-13
9-

WATER
OUT

10-

\
FURNACE LIFTING SYSTEM 11

Fig. 2-4: Diagram of the Thermobalance Stanton-Redcroft TG 750

1 Balance glass housing 8 Cooling water flow meter
2 Glass protection tube 9 Furnace
2a Brackets 10 Furnace lifting system
3 Glass protection tube 11 Spirit level
4 Counter weight glass housing 12 Support for glass protection tube
5 Gas inlet 13 Lower cover
6 Protection lid (Figure by Stanton-Redcroft Ltd.)
7 Gas flow meter
8 2 Methods and Instrumentation

A schematic digaram of the TG 750 is shown in Fig. 2-4, of the STA 780 in Fig. 2-5.
The recorder script of an experiment with a hydrocarbon using the TG 750 is depicted
schematically in Fig. 2-6. Curve I represents the weight (mass) signal (TGA), curve I1 that
of the first derivative (DTG), and curve I11 the temperature (T> of the thermocouple directly
below the sample pan. Point A marks the start of the weight (mass) loss 1 % and the
corresponding temperature T1 %; point B is the weight (mass) loss 5 % and the correspon-
ding temperature T5 %. Point C corresponds to the weight (mass) loss at 400C (AG400).
This is the temperature limit of the thermal stability of most non-aromatic hydrocarbons
and of the heterocompounds. Point D marks the weight (mass) of the coked residue at
600 "C (R600) or at 800 "C (R800). Point E represents the maximum of the DTG curve

Fig. 2-5: Cross-Section of Water-cooled Furnance for STA 1 000 (STA780)

A Water cooled cold finger
B Ceranuc baffles
C Ceramic tube
D Micro-enviromental cup
E Ceramic stem gas inlet
F Furnace winding
G STA hangdown assembly
(Figure by Rheometnc Scientific, Polymer Laboratories GmbH)
 2 Methods and Instrumentation 9

E D

Fig. 2-6: Schematic Diagram of Recorder Diagram of a Test in Protecting Gas by means of
TG 750.
I TGA signal
II DTG signal
I11 Temperature signal
A Start of weight loss (T1 %)
B Start of weight loss (T5 %)
C Weight loss up to 400 "C (AG400)
D Residue at 600 "C (R600) or at 800 "C (R800)
E Maximum of DTG curve (T-)

with the corresponding temperature T-. The amplitude of the DTG curve corresponds to
the reaction rate. The temperature of the DTG maximum shows whether the reaction
remains in the evaporation (distillation) range (Tmx< 400 "C) or if a pyrolysis (cracking)
reaction has occurred (T- > 400 "C).
An example of rescaling the plot of weight versus time to weight versus temperature is
shown in Fig. 2-7. Here, the point of intersection of the tangents (offset point) represents
the weight (mass) Gw of generated coke at the temperature Tw at the point of inflexion of
the TGA curve. This happens only during experiments in inert gas. Using ash-free substan-
ces in experiments in air, a TGA curve passing through zero weight is obtained, while
ash-containing substances give a constant residual weight.
The DTG graph of the experiment in air always shows more than one maximum, the first
of which can represent vaporization as well as oxidation. In this case the TGA graph in
protecting gas must be consulted for comparison.
Figs. 2-1 to 2-3 demonstrate possible evaluations using the STA 780 in an experiment
with calcium oxalate monohydrate. In Fig. 2-1 the TGA curve is evaluated with respect to
the weight (mass) losses of 1 %, 5 %, 10 %, and further in 10 % steps, whereas in the DTA
curve the peak maximum temperature and the corresponding residual weight (mass) are
plotted. Fig. 2-2 again shows the TGA and DTG curves with peak maximum temperatures
and corresponding residual weights. Fig. 2-3 demonstrates the onset and offset temperatu-
10 2 Methods and Instrumentation

Residue ( % I
100..

80

60.

LO

(c- ___
- - - - - - -9-
z
20.

I
I
I Tw

*
100 300
I
1
500 7M 800
-

res of the three reaction steps. Theoretically all three evaluations could be drawn in one
plot, but that would be very difficult to interpret.
With the help of differential scanning calorimetry (DSC), events can be observed which
are created by energy transfer (take up or delivery) during programmed heating or cooling
of a sample, i.e. melting, crystallization, second order transitions, evaporation, pyrolysis,
oxidation etc. The energetic effect in the sample is compared to a thermally inert reference
substance which undergoes the same temperature programme. The differences between
sample and reference in uptake or delivery of energy will be recorded as energy flows
versus temperature or time. Using DSC it is likweise possible to differentiatethe resulting
data with respect to time (dimensions W/s) or to temperature (dimensionsW/O C). Neither
differential quotient has any meaning in the physical sense. They serve only to elucidate
effects of the graph of energy flows versus temperature or time. Therefore it is not
surprising, that only a very few literature references exist where differentiated curves are
described.
Using DSC, the position of the energy flows versus temperature curve as well as the rate
of an event were influeced by the heating rate, too. Therefore the DSC tests were run
likewise, using a standard heating rate p= 10 K/min with the exception of the investiga-
 2 Methods and Instrumentation 11

tion of reaction kinetics. There the shift of the maxima of the energy flow curve to higher
temperatures as a result of increasing heating rates permits the calculation of the Arrhenius
lanetic parameters (see Chapter 3.3). All the influences, such as oven geometry, shape of
the sample pan, position of gas inlet and gas outlet, on the results of DSC are the same as in
TGA. The gas purge with a minimum flow rate is also necessary in DSC to avoid
condensations,when petroleum and its products or generally volatile substances are tested.
As a consequence, the boiling (evaporation) temperature will decrease in a similar way to
that in TGA. The gases used in TGA can be used also in DSC. Some additional experi-
ments have been carried out in methane to study the influence of a hydrocarbon atmos-
phere.
For calibration, the melting point of indium were measured, which has a temperature of
fusion (MP) = 156.4"C and a heat of fusion Hf= 28.46 J/g (Fig. 2-8). Because reactions at
higher temperatures occur in experiments with petroleum refinery residues, additional
calibration runs were performed using pewter (MP = 231.84"C, H,= 59.61 J/g) and lead
(MP = 327.40C, Hf= 26.47 J/g). If a calibration at higher temperatures is necessary,
potassium perrhenate KReO, (MP = 550C, H,= 294.8 J/g) can be used.
All DSC tests carried out by the author were run using open aluminium pans. For
reference an empty pan were used. Comparative tests with platinum pans gave no indica-

0
Temperature CoC)

Fig. 2-8: Indium Calibration Curve of DSC

MP : 156.4 "C
Hf : 28.46 J/g
12 2 Methods and Instrumentation

tion that the pan material had any influence, neither for pyrolysis nor for oxidation
reactions. The first experiments were carried out with the help of a DuPont 990 Thermo-
analysis System connected to a 910 DSC.This system used a two pen x-y recorder; the
resultant graphs were evaluated manually. Later, a DuPont 9900 Thermoanalysis System
was used, which is equipped with a PC for control, data sampling, and data evaluation
(Table 2-2). A cross-section of the DSC cell is shown in Fig. 2-9.

Table 2-2: Differential Scanning Calorimetry

Instrument: DuPont 9900 Thermal Analysis System
DuPont 910 DSC
Pressure DSC Cell
System: DSC -75 "C . . . +250 "C, normal pressure
DSC RT . . . +650 "C vacuum till bar
pressure up to 70 bar
Heating Rates: 0.5 . . . 50K/min
Cooling Rates: 0.5 . . . 5K/min
Recording: PC
Evaluation: PC
specific heat
conversion temperatures
reaction enthalpy
heat of conversions
kinetics according to ASTM E 698-79

Fig. 2-9: DSC Cell Cross-Section

1 Gas purge inlet 9 Chromel disk
2 Lid 10 Chromel wire
3 Reference pan 11 Alumel wire
4 Silver ring 12 Thermocouple junction
5 Furnace winding 13 Thermoelectric disk (Constantan)
6 Furnace block 14 Samplepan
7 Radiation shield 15 Lid
8 Sample platform (Figure by TA Instruments Inc.)
 2 Methods and Instrumentation 13

Petroleum and its products are multicomponent systems of varying chemical composi-
tion. They are predominantly a mixture of hydrocarbons, usually accompanied by a small
quantity of heterocompounds which contain in addition to carbon (C) and hydrogen (H)
other atoms such as sulfur (S), nitrogen (N), and/or oxygen (0).Metals are present in very
small concentrations, such as vanadium and nickel in organically bound forms. The
average elementary composition of petroleum in weight-% lies between the following
limits [2-11:

C 8 3 . . . 87% 0 0 . . .2 %
H11 . . . 14% N 0.01 . . . 1.7 %
S 0.01.. . 8 % Metals 0 . . . 0.1 %

Petroleum contains four groups of hydrocarbons:

- alkanes (unbranched n- and branched i-alkanes)

- cycloalkanes (naphthenes, unsubstituted and substituted)
- aromatics (unsubstituted and substituted)
- complex hydrocarbons (naphthenoaromatics)

Alkenes (olefins) and alkynes (acetylenes) are not found in petroleum (crude oils). Howe-
ver, they were formed during the processing of petroleum at high temperatures.
With regard to the boiling behavior, the full range of substances occur, from those which
evaporate early during the recovery, as a result of pressure decrease, through to substances
which cannot evaporate without decomposition. It is possible to separate individual chern-
cally-defined substances from the low boiling fractions. From medium and high boiling
fractions and from the non-distillable residues only multicomponent systems can be
obtained, which can be separated into groups characterized by a similar chemical and
physical behavior. Separation into individual compounds is almost impossible.
Under these circumstances, it seems reasonable to study the thermal reactions such as
boiling, pyrolysis, and the oxidation behavior of defined model substances first, in order to
understand the behavior of petroleum and its main products and to draw some analogous
conclusions.