Determination of the liquidus temperatures of ashes from the
biomass gazification for fuel production by thermodynamical and
experimental approaches
Dr. Jérôme Berjonneau
CNRS-CRMHT
1D, avenue de la Recherche Scientifique
45071 Orléans Cedex 2
e-mail : jerome.berjonneau@cnrs-orleans.fr
Laetitia Colombel
PhD student
CNRS-CRMHT
1D, avenue de la Recherche Scientifique
45071 Orléans Cedex 2
e-mail : laetitia.colombel@cnrs-orleans.fr
Jacques Poirier
Professor
CNRS-CRMHT
1D, avenue de la Recherche Scientifique
45071 Orléans Cedex 2
e-mail : jacques.poirier@cnrs-orleans.fr
Dr. Michel Pichavant
Senior Researcher
CNRS-ISTO
1A, rue de la Férollerie
45071 Orléans Cedex 2
e-mail : pichavan@cnrs-orleans.fr
Dr. Françoise Defoort
CEA LPTM DTN/SE2T/LPTM
17 rue des Martyrs
38 054 Grenoble cedex 9
e-mail : francoise.defoort@cea.fr
Dr. Jean-Marie Seiler
CEA LPTM DTN/SE2T/LPTM
17 rue des Martyrs
38 054 Grenoble cedex 9
e-mail : jean-marie.seiler@cea.fr
ABSTRACT:
During the gasification of biomass, inorganic species are produced and constitute an
important obstacle in this process. The aim of this study is to understand the behaviour of the
inorganic species particularly the production of ashes. These ashes will coat the reactor wall
and will play at the same time the role of protection of the refractory layer against external
attacks and of thermal insulation. In order to define the operating conditions and to design the
refractory structure, the melting temperature of slag is one of the most important properties. A
thermodynamic approach is used to obtain this characteristic . The database used for the
calculation is firstly validated by comparison with systems available in the literature and with
an experimental approach. This study reveals the lack of optimisation and the lack of data
concerning alkaline and particularly K2O-rich compositions. An experimental approach has
been used to determine the liquidus temperature of ashes by a classical quenching method.
The volatilization of K2O shows the limits of this method. Experiments using a sealed capsule
largely prevent the volatilisation of this species.
Keywords: ash, biomass gasification; liquidus temperature, thermodynamic, Factsage®
1 INTRODUCTION
The context of this research is the biomass gasification for fuel production in an entrained
flow reactor. During the gasification of biomass, inorganic species are produced and constitute
an important obstacle in this process.
Inorganics are the mineral elements and compounds present in biomass apart from the main
gas and organic species. They play an important role on fluidised bed agglomeration, on the
corrosion of the reactor wall, on the gas phase pollution or on the corrosion of the structure.
In our project, the general objective is the understanding and the mastering of the
behaviour of inorganics: their emission during gasification, their condensation during cooling
down of gas, their effect on the corrosion of the structures and on the production of liquid
ashes (slag).
This last point will be emphasized in this contribution. A layer of solidified and molten ashes
will be condensed on a refractory wall. This wall is externally cooled. This ash layer will play
the role of a thermal insulating lining to minimise the heat losses and to protect the refractory
layer against all external attacks.
In a steady state, the thickness of the slag layer must be constant with no mass transfer. It
depends on the temperature of the gas. On the one hand, if the temperature of the gas is too
low, as a coating of the ashes along the wall could be too thick (thus obstructing the reactor).
On the other hand, if the temperature of the gas is too high, thermal losses will be excessive.
The slag has to be liquid to be able to run down the wall. The melting temperature of the slag
is one, with viscosity, of the most important properties necessary to adjust the wall design.
The objective of this paper is to know this characteristic, particularly the liquidus temperature.
The liquidus temperature is the temperature for which all the material is liquid.
The study will be carried out for two typical compositions of ashes (table 1): the first one
which is rich in SiO2 (C1) is close to the miscanthus ash and the second one with the same
amount of SiO2 and CaO (C2) is close to the straw ash contents.
Table 1. Typical composition of ashes
Wt.%
C1
C2
SiO2
Al2O3
Fe2O3
CaO
MgO
K2O
Na2O
total
60
5
3
13
4
14
1
100
34
1
5
34
3
22
1
100
In the first part of this paper, the thermodynamic approach is described. The
thermodynamic tool used is described and is validated by comparison with existing data from
the literature. The liquidus temperatures calculated for the two typical ashes are then
presented.
Secondly the experimental approach is described. The results for the ternary system (SiO2CaO-K2O) and for the two typical compositions (C1 and C2) are presented. Finally the
thermodynamic and experimental approaches are compared.
2 THERMODYNAMIC APPROACH:
2.1 Description of the thermodynamic calculations
The calculation is based on the minimisation of the free enthalpy of the system. It is thus
possible to deduce the nature of the solid, liquid and gaseous phases as well as their chemical
compositions and their reaction rate at the thermodynamic equilibrium. This knowledge
allows to deduce the liquidus and solidus temperatures and the amount of the solid and liquid
phases.
Calculations have been carried out with Factsage® (version 5.4.1).
Factsage® is a fully integrated database and software package developed jointly between
Thermfact/CRCT (Montreal) and GTT-Technologies (Aachen). It consists of a series of
modules that access and manipulate thermodynamic databases and perform various
calculations. Factsage® is composed of two types of thermo chemical databases - compounds
(i.e. pure substances, over 4,400 compounds) and real solutions.
- Compound databases are mainly for stochiometric solid, liquid and gaseous species.
Allotropes (for example cristobalite, tridymite) are included. Non-stoichiometric
compounds such as FeOx (x = 0.947) are also stored.
- Solution data are for solid and liquid alloys, ceramics, salts, mattes, slags etc. The data are
stored in the form of Gibb’s energy functions for the phase constituents and parameters for
the excess Gibbs energy of mixing between the phase constituents. Factsage® supports
several different solution models including simple polynomial models such as the RedlichKister or the Legendre polynomial combined with different higher order extrapolations
(Muggianu, Kohler, Toop), the quasichemical model, Pitzer parameters and sublattice
models [1].
The associated databases for this study are: ELEM (elements thermodynamic database) FACT
53 (gas species, solid and liquid compounds thermodynamic database) and FT oxid
(compounds and solutions for oxides database).
The expected equilibrium phases were predicted by using the “Equilibrium” module of
Factsage® considering as possible phases: the slag, several solid-solution phases and all liquid
and solid compounds between the inputs. Oxide compositions are entered as initial
compositions in weight percent in the sub-menu “reactants” of “Equilib”.
The FToxid database contains data for pure oxides and oxide solutions of 20 elements. All the
binary and ternary sub-systems have not been evaluated and optimized, nor are all
composition ranges covered. Sub-systems which have not been evaluated and optimized have
been assumed ideal or have been approximated.
Thus, this database contains the following major oxides: Al2O3, CaO, Fe2O3, MgO, SiO2, all
the binary and many ternary and quaternary sub-systems more or less optimized. Interaction
between these 5 oxides should be correctly calculated.
When Na2O and K2O are present, the Factsage® documentation specifies that optimization is
less accurate. Major improvements are in progress for these oxides [2,3].
To conclude, the use of the thermodynamic FToxid database for the calculation of the liquidus
temperature for the two compositions will contain some uncertainties.
In order to ensure the reliability and the pertinence of the results, it seems necessary to
validate the thermodynamic calculations for simpler systems with a less number of oxides
than the multi component ashes. Consequently, the main sub-systems (binary and ternary)
have been tested.
2.2 Validation of the thermodynamic calculations
The thermodynamic database FToxid is carried out from the optimisation of several binary
and ternary systems. A phase diagram is said “optimised” when all the data from the literature
have been analysed, criticized, selected and used to determine the main parameters to carry
out the modelling of the phase diagram. The modelling is not always carried out for all the
composition range of a system.
To validate the FToxid database, calculations have been done in the most pertinent binary,
ternary and quaternary systems.
To assess the validation of the FToxid database, the systems have been checked in two ways:
- For the optimised diagrams, the optimisation has been checked in the studied domain of
ashes C1 and C2. If the range of composition is not optimised, bibliographic research is done
in order to check the existence of other measurements and to reproduce them by calculation
from the database.
- For the non optimised diagram, bibliographic research is performed to find experimental
results which are compared with the calculation. This comparison gives information about the
validation of the database.
For the four main oxides SiO2, CaO, Al2O3 and MgO, all the binary systems have been
checked in the same proportions as the two ashes [4-9]. This has not been done for minor
oxides such as Fe2O3 and Na2O.
For K2O, which is also a major oxide, K2O-SiO2 is the only optimised binary system [10]. It
has been successfully checked for the two ashes compositions C1 and C2. The two other
main binary systems K2O-CaO and K2O-MgO are not optimised in Ftoxid database.
Experimental data has not been found in the literature. It makes the validation of the database
in this domain impossible.
On the other hand, K2O-Al2O3 is optimised [11] but not in the composition range relative to
ashes C1 and C2 (figure 1). Besides, experimental data has not been encountered in the
literature to validate the database in this domain.
Figure 1. Binary diagram K2O-Al2O3 [4]
Concerning the ternary systems resulting from the major oxides SiO2, CaO, K2O, MgO,
only SiO2-CaO-MgO is optimised [11] and successfully checked for the two ashes C1 and C2.
No data has been published for CaO-K2O-MgO system. Experimental data has been
encountered in the literature for SiO2-K2O-MgO [12] and SiO2-CaO-K2O [13]. Validation of
the database has been poorly obtained for these two ternary systems. Particularly the liquidus
temperature is higher than 200-400°C for C1 and C2 as illustrated, for example, for C1 in
SiO2-CaO-K2O (figure 2).
Tliquidus Lit. # 1175°C
C1
Tliquidus Fact.. = 1418°C
Figure 2. Ternary diagram SiO2-CaO-K2O [13]
2.3 Calculation results for the typical ashes
The liquidus temperatures calculated with Factsage® are presented in table 2 for the two
ashes compositions with and without Na2O and K2O.
Table 2. Liquidus temperatures for C1 and C2 with and without Na2O and K2O
C1
C2
Rich in silica
Rich in K2O
(Miscanthus)
(Straw)
With Na2O K2O
Tliquidus 1324°C
1256°C
Without Na2O K2O
Tliquidus 1560°C
1420°C
Composition C1
Without Na2O and K2O, the liquidus temperature is 1560°C (figure 3). The primary
crystallization phase is SiO2.
With Na2O and K2O, the liquidus temperature (1324°C) is lower by around 230°C in
agreement with the existence of a eutectic in the binary system with silica (figure 4). The
calculation predicts the formation of two immiscible liquid phases called slag#1 which is rich
in SiO2 and K2O and slag#2 which is rich in SiO2 and CaO. The primary crystallization phase
is composed of melilite (Ca2MgSi2O7), clinopyroxene (CaMgSi2O6) and wollastonite
(CaSiO3).
100
80
Phase fraction (Wt. %)
Slag
60
SiO2(s)
SiO2(s2)
SiO2(s4)
40
Pyroxene
20
CaAl2Si2O8(s2)
Ca3Fe2Si3O12
Total solid
Wollastonite
Tliquidus = 1560°C
SiO2(s6)
0
500
700
900
1100 T(°C) 1300
1500
1700
1900
Figure 3 : weight % of the phases formed as a function of the temperature for C1 without
Na2O and K2O
100
Phase fraction (Wt.%)
80
Slag 1
60
Slag 2
Total slag
40
Clinopyroxene
20
Wollastonite
Melilite
Ca3Fe2Si3O12(s)
KAlSi2O6(s2)
Tliquidus = 1324°C
0
800
900
1000
1100
T(°C)
1200
1300
1400
1500
Figure 4 : weight % of the phases formed as a function of the temperature for C1 with Na2O
and K2O
Composition C2
Without Na2O and K2O, the liquidus temperature is 1420°C (figure 5). The primary
crystallization phase is wollastonite. With Na2O and K2O, the liquidus temperature (1256°C)
is lower by around 170°C (figure 6). The calculation predicts the presence of two immiscible
liquid phases called slag#1 which is rich in SiO2 and CaO and slag#2 which is rich in SiO2
and K2O. The primary crystallization phase is a spinel phase close to Fe3O4.
For both compositions, the modeling of Na2O and K2O with CaO, which is a major oxide, has
not been optimized. Thus both liquidus temperature is quite uncertain.
Besides it is important to remember that ternary systems with K2O (SiO2-CaO-K2O and SiO2MgO-K2O) are poorly validated by Factsage®. The calculations are probably less reliable
because of the lack of optimization for this oxide and both liquidus temperatures from
calculation are probably too high.
To conclude about the validation of these results it is necessary to compare them with
experimental measurements.
100
Total solid
Slag
Phase fraction (Wt. %)
80
60
CaSiO3(s2)
Wollastonite
Melilite
40
20
Ca3Fe2Si3O12(s)
Tliquidus = 1420°C
Ca3MgSi2O8(s)
Ca3Si2O7(s)
0
600
800
1000 T(°C)
1200
1400
1600
Figure 5 : weight % of the phases formed as a function of the temperature for C2 without
Na2O and K2O
100
Slag 1
Phase fraction (Wt.%)
80
Total solid
60
40
Ca3Si2O7(s)
20
Ca3MgSi2O8(s)
NCSO
0
600
Fe2O3(s)
800
1000
Slag 2
Spinel
Tliquidus = 1256°C
T(°C)
1200
1400
1600
Figure 6 : weight % of the phases formed as a function of the temperature for C2 with Na2O
and K2O
3 EXPERIMENTAL APPROACH
3.1 Experimental procedure
Experiments have been carried out in a vertical quench furnace at ISTO (Institut des
Sciences de la Terre; CNRS laboratory, Orléans) (figure 7). The objectives of the experiments
are to determine an experimental liquidus temperature and to identify the phases formed.
Experiments were performed using the quenching method.
Vertical furnace
(Al2O3 tube)
-+
Al2O3
tubes
Pt. wire 0,5 mm
i
Short-circuit
Pt. wire 0,2 mm
Sample :
Glass drop
# 1 mm
Tthermo
Thermocouple
From high T
to room T
Vertical gas-mixing furnace
ISTO CNRS Orléans
Quench:
Short-circuit at the transition point between
Pt. wire 0,2 mm and pt. wire 0,5 mm
Figure 7. “Quenching” method
The starting sample is made of ground glass mixed with polyvinylalkohl to ensure its cohesion
during the installation in the furnace and is later vaporised at low temperature.
Two different procedure were used : one in open system and the second one in sealed
capsules.
In the first one : open system, the mud obtained is then spread over a platinum hook (Ø : 0,2
mm) hung on a wider platinum wire (Ø : 0,5 mm) inserted into an alumina rod. A
thermocouple (Pt - Pt Rh 10 wt.%) inserted into another alumina rod is located close to the
sample to measure the temperature.
For the second one : sealed capsules, the samples have been enclosed in sealed capsules to
minimize the volatilization of the alkaline oxides (K2O and Na2O) during the experiment [14].
Around 40 mg of powder is prepared. Capsules are made from a platinum tube of two
millimetres in diameter and ten millimetres in length. These capsules are first cleaned, dried,
arc welded (figure 8).
2 Pt wires which
support
the
capsule
Thermocouple
Pt/Pt
Pt. sealed capsule with
sample isnside
Figure 8 : sealed capsule method
For both procedure, the system is homogeneous in temperature. Daily fluctuation is inferior to
2°C and temperature is know to better than ± 5°C. The whole assembly is inserted into a
vertical furnace at a steady temperature for 3 hours, under air.
The quench results from a high intensity generated on the platinum wires which melts down
the thinner wire (Ø : 0,2 mm). The sample falls into the cold part of the furnace, allowing the
melt phases to quench into a glass.
Once the sample has been quenched, the capsule is opened (if sealed capsule procedure).
The glass is extracted, then polished for SEM (Scanning Electron Microscopy) observations
and WDS electron microprobe analysis (Wavelength Dispersive Spectroscopy).
The possible formation of crystals indicates if the sample was completely liquid at the
temperature of the experiment. The liquidus temperature is obtained by bracketing the
desperation of crystals by means of experiments at different temperatures.
The chemical analysis of experimental products is carried out with an electron microprobe.
The experimental approach has been first validated with the ternary system K2O-CaO-SiO2.
Secondly this approach has been applied to the two ash compositions (C1 and C2).
3.2 Validation of the method with the ternary system K2O-CaO-SiO2
To validate this method, one composition called KCS was studied in the ternary system
K2O-CaO-SiO2, in oxide proportion similar to the C1 composition (table 3). This ternary
diagram has not been optimised in the FToxid database. However experimental data exist in
the literature [13]. For this composition, the liquidus temperature, calculated with Factsage®,
is 1418°C (figure 2). It is higher than the one measured in the publication [13], which is
around 1175°C.
Table 3. Studied composition in the K2O-CaO-SiO2 system
Ternary composition
Composition C1
(% wt.)
(% wt.)
SiO2
69
60
CaO
15
13
K2O
16
14
Al2O3
0
5
Fe2O3
0
3
MgO
0
4
Na2O
0
1
The ternary composition has been prepared with K2CO3, CaCO3 and SiO2 oxides. Before
the experiments, the powder was previously decarbonised by heating with plateaus at 700°C
and 900°C. At the end of the experiment, a hollow ball with a diameter of around one
millimetre is obtained, then polished and observed with SEM. The hole comes from the
temperature gradient between the core and the surface during quenching. The surface is cooled
faster than the core and induces the formation of gaps.
The cross section of a glass ball quenched at 1175°C reveals a ring of glass in which crystals,
with a size of around 4 µm, are spread out on the surface (figure 9).
Crystals
Platinum
Hollow of the ball
Glass
Glass ball
500 m
500 m
Hole
Figure 9. Glass balls quenched at 1175 °C for the ternary composition
For samples quenched between 1177°C and 1225°C, crystals are present for this specific
composition. The crystals grow in size as the temperature increases while their number
decreases (figure 10). At 1230°C, no crystal is observed. The liquidus temperature is
bracketed between 1225°C and 1230°C. The experimental temperature is close to the
literature [13] and is distant from the calculation (figure 2). This result confirms the lack of
the Ftoxid database to reproduce mesurements concerning K2O.
The chemical analysis reveals that the crystals are wollastonite (CaSiO3), in agreement with
the literature and the thermodynamic results.
Crystals
Platinum
Crystals
Glass
Glass
500 m
1200°C
200 m
1210 °C
Platinum
Platinum
Glass
Crystals
Platinum
Glass
200 m
1225 °C
200 m
1230°C
Figure 10. Glass balls quenched between 1200 and 1230 °C for the ternary composition
5.3 Composition C1
This composition has been measured by the two procedures (open system and sealed capsule)
- Without the sealed capsule (open system)
This composition has been prepared by Schott (table 4). The experimental composition
(analysed by electron microprobe) is in agreement with the miscanthus ash C1.
An iterative approach has allowed estimating the liquidus temperature with a range of 5°C.
Table 4. Theoretical and experimental compositions of C1
Compo C1 Siliceous slag Compo C1 given by
close to Miscanthus ash Schott and analysed
(% weight)
(% weight)
SiO2
60
61,3
Al2O3
5
4,9
Fe2O3
3
2,6
CaO
13
12,8
MgO
4
3,9
K2O
14
13,5
Na2O
1
1,1
Below the liquidus temperature, crystals in the shape of needles are observed (figure 11).
The number of crystals decreases as temperature rises while their size increases. The chemical
analysis shows that crystals are wollastonite CaSiO3. The glass quenched at 1240°C does not
contain any crystals. The liquidus temperature is then bracketed between 1235 and 1240°C.
500 m
200 m
1169°C
1216°C
500 m
500 m
1235°C
1240°C
Figure 11. Glass balls quenched from 1169°C to 1240°C for composition C1 without sealed
capsule
The evolution of crystal and glass phases is followed with electron microprobe as a function
of the temperature. Data were obtained using a microscope image acquisition and analysis
software (table 5). The amount of crystal decreases as the temperatures rises. Therefore, the
liquidus temperature is getting closer.
Table 5. Evolution of the amount of glass and crystal as a function of the temperature
1170 °C
1216 °C
1235 °C
1240 °C
1278°C
% wt. glass
89.1
96.7
99.8
100
100
% wt. cristal
10.9
3.3
0.2
0
0
% wt. K2O
13
11,75
12,2
11,8
9,6
This analysis shows the evolution of K2O with the temperature (table 5 & figure 12). Up to
1240°C the volatilization of K2O is weak. At 1240°C, 10 wt. % of the K2O is volatilized
which represents a loss of 1 wt.% of the global weight. Even if the experiment could be
improved to avoid volatilization, this measure gives a good estimate of the liquidus
temperature.
K2O final - K2O initial / K2O initial
1160
1180
1200
1220
1240
1260
1280
1300
0
-5
Sealed capsule
T liquidus = 1230°C
-10
-15
%
Pt Wire
T liquidus = 1237°C
-20
-25
-30
-35
Temperature ( °C)
Figure 12. Evolution of K2O with and without capsule as a function of the temperature in
comparison with the initial composition C1
-With sealed capsule
As previously without the sealed capsule, an iterative approach, by means of several
experiments at different temperatures, has made it possible to estimate the liquidus
temperature within a range of 5 °C. The liquidus temperature is around 1230°C (figure 13).
Bubbles
Glass
Crystals
Pt
1228°C
1234°C
Figure 13. Glass balls quenched at 1228°C and 1234°C for composition C1 with sealed
capsule
The evolution of K2O is followed with electron microprobe as a function of the temperature
and is reported on the same graph than the experiments without capsules (figure 12). The loss
of K2O is less important with the use of sealed capsule (less than 0,5 wt. % of the total K2O).
In both cases, the loss of K2O is quite weak. The liquidus temperature is confirmed and is
around 1230°C. This value is one hundred degrees lower than the thermodynamic calculation
with alkaline species which is 1324°C. The primary crystallisation phase predicted by
thermodynamic (the melilite (Ca2MgSi2O7) followed with clinopyroxene (CaMgSi2O6) and
then wollastonite CaSiO3) is only in agreement with experimental results for the formation of
the wollastonite. Besides, the presence of two immiscible liquid phases predicted by the
calculation is not observed.
Composition C2
This composition has been measured by the two procedures (open system and sealed capsule)
-Without sealed capsules (open system)
The composition C2 is richer in CaO, poorer in SiO2 and richer in K2O. In this range of
temperature and without a capsule, the experiments have shown a very important vaporisation
of K2O. For the real composition i.e. without alkaline species, as total K2O vaporization has
occurred, the liquidus temperature has been measured around 1375°C (figure 14).
200 m
1348°C
200 m
200 m
1357°C
200 m
1372°C
1376°C
Figure 14. Glass balls quenched from 1348°C to 1376°C for composition C2 without sealed
capsule
This one is in agreement with the liquidus temperature predicted from the calculation without
alkalines species (1420°C) and the primary crystallization phase which is wollastonite CaSiO3
(figure 5).
-With sealed capsules
To attempt to keep the alkaline species as condensed phases in the glass composition, the
powder is introduced in a sealed capsule. At 1202°C, there is the presence of two immiscible
liquid phases with a crystal phase (figure 15). From 1230°C, there is still the two immiscible
liquid phases but no longer any solid phase. The temperature, above with the C2 composition
is completely molten, is ranged between 1202°C and 1230°C. This result is confirmed by
DSC (Differential Scanning Calorimetry). These results are in agreement with the
thermodynamic calculations which predict that this temperature is 1256°C with the apparition
of two immiscible liquid phases.
Crystals
Slag 2
Slag 1
1202°C
1230°C
(1)
0
Flux de chaleur (mW)
-50
-100
-150
-200
T
liquidus
= 1230°C
-250
600
800
(2)
1000
1200 T(°C)
1400
Figure 15. Glass balls quenched at 1002°C and 1230°C for composition C2 with sealed
capsule (1) and the DSC of C2 composition Tmax = 1400°C à 10K/min m = 390 mg (2).
4 CONCLUSION
Measurements and calculations of liquidus temperatures of two typical ashes from biomass
gasification for fuel production has been compared in this publication and are summarized in
table 6.
KCS
Thermodynamic calculations
Experimental measurements
1418°C
1225 < Tliquidus (°C) < 1230
1560 °C - without alkalines
---
1342°C - with alkalines
1235 < Tliquidus (°C) < 1240 - without capsule
1228 < Tliquidus (°C) < 1234 - with capsule
1420°C - without alkalines
1372 < Tliquidus (°C) < 1376 without capsule
i.e without alkalines
1256°C - with alkalines
Tliquidus (°C) = 1230 with capsule
i.e. with alcalines
C1
C2
Table 6 : liquidus temperatures for KCS, C1 and C2
The liquidus temperature has been calculated with the thermodynamic software Factsage®
for two typical ash compositions. In order to assess the reliability of this calculation a
validation step has been used. It reveals the lack of data concerning the studied system and
particularly alkalines and thus probably an uncertain database. Indeed, with K2O, only one
binary system (K2O-SiO2) [10] and one ternary system SiO2-CaO-MgO [11] are optimised.
For the ternary system resulting from the major oxides, only SiO2-K2O-MgO is well optimised
and correctly validated for the ash composition C1 and C2. For the others, the comparison
with the literature exhibits strong differences which reveal the limitations of the FToxid
database. Thus, the liquidus temperatures obtained from the calculation are not quite certain.
To make up for the lack of thermodynamic data, some updating is required. However, this
procedure depends on the experimental measures available in the literature which are
unfortunately very scarce for these systems and particularly for K2O.
An experimental approach has been used in this paper, in order to get a liquidus
temperature measurement and to compare it with the calculation. This one has been carried
out with and without sealed capsules.
Without capsules, the first system studied is the ternary K2O-CaO-SiO2 (KCS) which
allowed to validate the method used thanks to the comparison with existing literature
measurement. Only few K2O is volatilised. However a 200°C discrepancy with calculation
(table 6) confirm the lack of optimisation of the Ftoxid database as already analysed. This
experimental method (without capsule) has been applied to the multi component composition
C1 containing seven oxides. This result has given a first approximation of the liquidus
temperature for C1 which has been confirmed by measurement with the sealed capsule
method. Owing to the strong differences with the calculation (~100°C) , the lack of
optimisation (for FToxid database) is confirmed. Besides, the primary crystallisation phase
predicted is not in agreement with the one experimentally obtained and no miscibility gap in
the liquid phase is observed experimentally contrary to calculations.
This procedure without capsule could not be applied for the multi component C2 composition
which is reached in K2O than C1. Besides, this phenomenon increases as the temperature
rises: all the K2O is volatilised at the liquidus temperature of C2 (# 1375°C). Nevertheless,
these experimental results are in agreement with thermodynamic predictions for composition
without alkalines. The liquidus temperatures are close (table 6). There is a difference lower
than 50°C and the primary crystallisation phase is wollastonite as calculated. This result
validate the Ftoxid database without alkaline.
For the composition C2, the use of capsule allowed to get reliable results. The liquidus
temperature is around 1230°C and no K2O loss has occurred. The samples exhibit also the
formation of two immiscible liquid phases. This is in agreement with thermodynamic
predictions (1256°C) taking into account alkaline. The
good accordance between
thermodynamic calculations and measurements for this C2 multi component composition is
very surprisingly as this composition is richer in K2O and as the Ftoxid database has been
found poorly validated when alkaline is present. Thus, one have not to conclude that the
Ftoxid is well validated for this C2 ash. Indeed. one must conclude that opposite effect might
be compensated in the database.
In conclusion, this paper has showed that the discrepancy between measured and calculated
liquidus temperature of systems containing alkaline (K2O) is between several hundred °C to
0°C ! Then, if thermodynamic calculations need to be used to predict liquidus temperature of
typical ashes from biomass gasification for fuel production, it is necessary to improve the
FToxid database when alkaline are presents. However, only few measurements are existing in
the literature to be taken into account by the FToxid database developers. Such measurements
with alkaline are not easy to perform because of their high volatility. Nevertheless an
experimental set up has been presented in this paper which is able to produce reliable results
with small order system (binary or ternary) and multi component system (7 oxides). Only the
ternary result presented in this paper could be taken into account to improve the database. The
two multi component compositions C1 and C2 measured in this study can only be used to
validate a thermodynamic database as too many oxides are present and some compensation
error may occur as it is probably the case with the composition C2.
Acknowledgments
Authors thank for ANR (Agence National de la Recherche) and the PNRB ( Programme
National de Recherche sur les Bioénergies ) for their financial helpful contribution in this
research.
REFERENCES
[1] C.W. Bale, P. Chartrand, S.A. Degterov, G. Eriksson, K. Hack, R. Ben Mahfoud,
J.Melançon, A.D. Pelton and S. Petersen, Factsage thermochemical software and
databases, Calphad, 26, 2 (2002) 189-228.
[2] Pelton, A. D. (1999). "Thermodynamic calculations of chemical solubilities of gases in
oxide melts and glasses." Glass Science and Technology-Glastechnische Berichte 72(7):
214-226.
[3] Pelton, A. D. and P. Wu (1999). "Thermodynamic modeling in glass-forming melts."
Journal of Non-Crystalline Solids 253: 178-191.
[4] Eriksson, G., P. Wu, et al. (1993). "Critical-Evaluation and Optimization of the
Thermodynamic Properties and Phase-Diagrams of the MgO-Al2O3, Mno-Al2O3, FeOAl2O3, Na2O-Al2O3 and K2O-Al2O3 Systems." Calphad-Computer Coupling of Phase
Diagrams and Thermochemistry 17(2): 189-205.
[5] Eriksson, G. and A. D. Pelton (1993). "Critical-Evaluation and Optimization of the
Thermodynamic Properties and Phase-Diagrams of the CaO-Al2O3, Al2O3-SiO2, and
CaO-Al2O3-SiO2 Systems." Metallurgical Transactions B-Process Metallurgy 24(5):
807-816.
[6] Eriksson, G., P. Wu, et al. (1994). "Critical evaluation and optimization of the
thermodynamic properties and phase diagrams of MnO-SiO2 and CaO-SiO2." Canadian
Metallurgical Quarterly 33(1): 13-22.
[7] Jung, I. H., S. A. Decterov, et al. (2004). "Critical thermodynamic evaluation and
optimization of the MgO-Al2O3, CaO-MgO-Al2O3, and MgO-Al2O3-SiO2 systems."
Journal of Phase Equilibria and Diffusion 25(4): 329-345
[8] Wu, P., G. Eriksson, et al. (1993). "Critical-Evaluation and Optimization of the
Thermodynamic Properties and Phase-Diagrams of the CaO-FeO, CaO-MgO, CaOMnO, FeO-MgO, FeO-MnO, and MgO-MnO Systems." Journal of the American
Ceramic Society 76(8): 2065-2075.
[9] Wu, P., G. Eriksson, et al. (1993). "Prediction of the Thermodynamic Properties and
Phase-Diagrams of Silicate Systems - Evaluation of the FeO-MgO-SiO2 System." Isij
International 33(1): 26-35.
[10] Wu, P., G. Eriksson, et al. (1993). "Optimization of the Thermodynamic Properties
and Phase-Diagrams of the Na2O-SiO2 and K2O-SiO2 Systems." Journal of the
American Ceramic Society 76(8): 2059-2064.
[11] Jung, I.-H., S. A. Decterov, et al. (2005). "Critical thermodynamic evaluation and
optimization of the CaO-MgO-SiO2 system." Journal of the European Ceramic Society
25(4): 313-333.
[12] Roedder E.W. (1951) “ The system K2O-MgO-SiO2-1” American Journal of Science
249 (2) 81-130
[13] Morey, G., Kracek, F. C. and Bowen, N. L., J. Soc.Glass.Technol., 14, 158 (1930)
[14] Matousek, J. (1998) " Chemical reactions taking place during vaporization from
silicate melts " Ceramics – Silikaty 42 (2) 74-80