FRACTIONAL GAS ANALYSIS APPLICATION FOR METALS
AND NANOPOWDERS
Konstantin V. Grigorovitch
Baikov Institute of Metallurgy and Materials Science, Russian Academy
of Sciences; Leninskii pr. 49, Moscow, 119991 Russia
e-mail: grigorov@ultra.imet.ac.ru
Introduction
The quality of metal products is determined by not only their bulk
chemical composition but also the chemical forms of hydrogen,
nitrogen, and oxygen present in the matrix, namely by their distribution
between a solid solution and different species, such as hydrides, nitrides
and oxides. Quality of nanopowders depends significantly the content of
gas forming impurities such as hydrogen, nitrogen carbon and oxygen
because of very significant specific surface value up to 100m2/g. The
carrier gas hot extraction method is one of the thermal evolution
methods routinely used to determine the content of the impurities. The
complete simultaneous extraction of all species from a sample usually
proceeds in the isothermal mode at temperatures above 2500 K .in 30-40
seconds. Fractional gas analysis (FGA) is one of the advanced
possibilities of modern gas analysers. The FGA based on the difference
in thermodynamic stability of oxides and nitrides. It is carried out using
gradually increasing furnace temperature at a given rate. The first
attempts to elaborate a gas fusion technique for the separation of
different species of light elements upon monotonous or step-wise
heating have been realized in 80th. R. Prumbaum et al [1] were the first
who gave account of the temperature ramped inert gas fusion technique
with IR detector. The reduction sequence of oxides was derived from
their values of the standard Gibbs energy of formation. Later, the
successful employment of the commercial LECO RO-316 analyzer has
been reported for oxide separation in steels and slags [2-4]. The progress
of earlier works has not, however, succeeded by an extensive practical
use of the temperature ramping technique. This fact was mainly
attributed to two problems. The first one was the absence of numerical
algorithm and software for processing of raw kinetic data. The second
problem concerned the start temperatures of carbothermal reduction of
oxide species Ts. The above problems have been worked on for last
67
years. At first, an OxSep original software to process the results of
temperature ramped analysis [5] have developed. The numerical
procedure involved consecutive separation and subtraction of individual
peaks from a total evolution curve. The temperature-dependent
background evolutions as well as mixing effects in a gas system of
analyzer were also treated by this model. Then, thermodynamic model
of carbon reduction of oxide inclusions within molten sample, saturated
by graphite during FGA have improved [6]. Then, identification
software, which includes a thermodynamic model of carbon reduction of
oxide inclusions during analysis, has been improved.
The aim of this work was to apply the improved procedure for
oxide and nitride speciation in high-carbon steels, for oxide, nitrides and
carbon speciation in nanopowder materials by temperature ramped
technique on commercially available instruments. The results of the
technique are correlated to those of the chemical analysis and X-Ray
microanalysis.
Experimental
The samples of high-carbon steels in the form of rods (d=10 mm),
referred as CRM SG-4, were involved into the study. The above samples
are supplied as reference materials for gas fusion analysis (Ural Institute
of Metals, Russia). The analysis of nonmetallic inclusions in these
samples using the standard chemical weighting method, X-ray
microanalysis using Leo 430 microscope with an Oxford ISIS EDX
analysis system has been performed. FGA measurements were made on
TC-436 and TC-600 (LECO, USA) commercial analyzers. Double
crucible technique was applied. The outer crucible played a role of a
heater providing a uniform temperature field. It was found that 76% of
the total oxygen is combined with Al in the form of alumna. Oxide
precipitates of different types using the classical procedures of “wet”
chemistry. The results, summarized in Table 1, show that the
deoxidation products consist mainly of alumina and silicates. In SG4
material noticeable amount of Mg- or Ca-containing oxides was found.
The typical oxide species were identified using Leo 430
microscope with an Oxford ISIS EDX analysis system. The SEM
photographs of the alumina and spinel precipitates along with EDX
spectra are shown in Fig.1. In SG4 material two types of high-alumina
inclusions could be distinguished (Table 2).
68
Table 1
Sample Total SiO2 Al2O3
SG4
103
8
68
FeO
3
MgO+CaO MnO
15
-
Others
9
Chemical composition results of high-alumina particles in SG4
reference material with SEM/EDX analysis (oxygen and iron were
excluded before normalization; total 17 particles) are summarized in
Table 2.
Table 2
Particles
min-max
min-max
Size,
m
1-4
7-30
Mg
1-9
18-21
Al
62-82
61-81
Composition, at.%
Si
S
0-3
6-14
0-1
0-10
Ca
1-12
0-7
Mn
3-9
0-3
The first one was composed from alumina, containing Mg as
impurity while the second one corresponded rather to spinel MgO∙Al2O3.
These two groups also differed in particle size, which was noticeably
larger for spinel inclusions. Most of oxide inclusions were covered by
the shell of sulfides. The presence of Mg-containing inclusions in SG4
material is in agreement with the results of chemical analysis. However,
Mg does not form an individual MgO phase, but enters spinel. In
general, the oxide inclusions in SG4 materials are very similar to those,
which have been often found in high-carbon steels and repeatedly
described in literature.
The FGA evolution curve and reproducibility oxygen in peaks
with the characteristic temperatures (TB and Tmax) of the peaks of CRM
SG-4 separated by OxSeP software comprising with calculated by OxId
is presented in Fig 2 and Tab 3. The results of application of software
developed show that the main peak contains more than 60% of total
oxygen in peaks.
69
a)
b)
Fig. 1. The Al2O3 (1) particle inside MnS(2) shell; SG4 material;
(a) view in scattered electrons, (b) EDX spectra of points 1.
The foregoing model suggests that the reduction temperature is
determined by thermal stability of oxide, chemistry of the sample matrix
and pressure in the gas system of analyzer. The sample chemistry affects
the activities of aluminium and silicon in the melt. The lower the activity
the higher is the bubble pressure, and, therefore, the lower is thermal
stability of oxide. The temperature of the reduction start of alumina
increases greatly with aluminium content. This fact must be taken into
account when interpreting the evolution curves from different steel
grades. The following reduction sequence results from thermodynamic
calculations for GSO SG-4 was: (equilibrium values of Ts, K are in
brackets):
SiO2(1560) → 3Al2O3∙2SiO2(1780) → Al2O3(1850)→
→ MgO·Al2O3(1940) → MgO(2110).
70
The numerical procedure involves consecutive separation and
subtraction of individual peaks from a total evolution curve, followed by
minimization of the sum of squared residuals. The profile of individual
peak was modelled in terms of the following empirical formula [6,7]:

 T  Tm T k  E / x 

y (T )  y m  exp E 
 e
dx  ,


 T  Tm Tm

where Tm – temperature of the peak’s maximum, ym- peak’s
height, k, E – model parameters. This function was verified by treating
the single peaks of pure alumina and silica inclusions in iron matrix.
Figure 2 shows peak-separated evolution curve of SG4 materials.
Fig. 2. The FGA evolution curve for CRM-SG-4 and oxide separation
procedure result by OxSeP software
The arrows refer to the reference temperatures of alumina
reduction, calculated with thermodynamic model. The alumina peak of
SG4 material is followed by the smaller peak with the start temperature
close to that, predicted for spinel (1940 K). At higher temperatures
(≥ 2100 K), which are characteristic of MgO species, no peaks are
revealed. Thus, peaks number 3 and 4 of SG4 sample are assigned to
alumina and spinel, respectively. The FGA curve of SG4 samples yield
71
distinct peaks below Tstart of alumina. These are attributed to silicates.
Their reduction starts at about 1650 K. This temperature is higher than
the reference point, predicted for pure SiO2 (1560 K), but lower than that
for 3Al2O3∙2SiO2 (1780 K). The discrepancy is likely due to the fact that
silicate inclusions are of the variable composition, with silicon atoms
partially substituted by aluminium. We, thus, assign peaks below Tstart of
alumina to alumosilicates while omitting the more detailed analysis of
their type. This seems to be difficult with gas fusion technique.
The FGA results of the 10 samples with oxygen results and
characteristic temperatures (Tbeg and Tmax) of the peaks and Standard
Deviation of CRM SG-4 separated by OxSeP software are summarized
in Table 3.
Table 3
Oxygen
total
26.0
1,2
FGA Parameters
Identified
Oxides
Peak
number
ТB, K
Тмax, K
1+2
1570
1780
6.7
Silicates
3
1855
18
1940
1980
10
2035
11.0
1,0
4.9
Alumina
4
cO
Al-Mg Spinel
The reproducibility of oxygen content and peak’s temperatures is
satisfactory. About one half of the total amount of oxygen is found to be
combined with alumina. This is in qualitative agreement with the results
of the chemical analysis. However, these must be considered rather as a
semi-quantitative since they give overestimated content of the total
oxidic oxygen (40-50 mg/g). In general, the results of three different
methods are in reasonable agreement, providing quite consistent picture
of chemical state of oxygen in studied materials.
In general, the results of three different methods are in reasonable
agreement, providing quite consistent picture of chemical state of
oxygen in studied materials.
Tire cord steel quality is especially sensitive to cleanness in nonmetallic inclusions. Hardly deformed non-metallic inclusions of Al2O3 in
72
high strength tire cord steel are one of the principal causes of the fall in
productivity (wire failure during drawing and spin breakage). The tire
cord steel samples were tested by using the fractional gas analysis
method (FGA) and original OxSeP software use. FGA is integral
method, which gives the sum of oxygen in oxide inclusions of different
types (the content oxygen in oxides, ppm). FGA is sensitive to the
composition of inclusions only and rather than to their size. In the next
Figure is presented the cooperative results of the FGA mean values and
Standard deviation for tire cord samples of 4 Heats BEKAERT
production, of Nippon steel – NS, Byelorussian metallurgical plant –
BMZ and Moldavian metallurgical plant (MMZ) production.
silicates
hard deformable silicates
Al, Ca, Mg rich silicates
total oxygen in oxides
22
20
Oxygen content, ppm
18
16
14
12
10
8
6
4
2
0
A
B
C
D
BMZ
NS
MMZ
Heat
Fig. 3. Results of fractional gas analysis of cord steel (mean and standard
deviation).
The testing results has shown that the FGA method and software
developed can be successfully used in quality control of tyre cord steel
production.
Oxygen is the main affected gas impurity that always presented in
powders. There are several forms of the oxygen presence such as
adsorbed, dissolved, and bounded ones. Not only the total oxygen
content but the proportion of oxygen forms predetermines and
73
characterizes the properties of the powders depending on their
production processes and storage conditions. The FGA procedure was
applied to determine the content of oxygen in surface and different
forms in powders and carbon in organic forms, soot and carbides in
nanopowders. The results of the technique are correlated to those of the
chemical analysis and X-Ray analysis. Using the FGA method and
original software developed the contents of different oxygen nitrogen
forms in nanopowders such as W, Mo, Nb, Ta, and Ni; refractory
compounds such as carbides – WC, NbC, TaC, and TiC; nitrides – AlN,
Si3N4, HfN, NbN were determined.
The results obtained are in a good agreement with the
experimental data on the chemical composition and content of oxide
inclusions obtained by materials science methods. The testing of
commercial steels and powder materials has shown that the FGA method
and software developed can be successfully used in different fields of
metallurgy and materials science. It was shown that these methods are
available to certify nanopowder materials according to the oxygen,
nitrogen and carbon amounts in different forms.
1.
2.
3.
4.
5.
6.
7.
References
R. Prumbaum, K. Orths: Gießerei Forsch, 31(1979), (2/3), 71.
D. Sommer, K. Ohls: Fresenius Z. Anal Chem, 313 (1982), 28.
P. Borek, Z. Chizhek, L.L Kunin: Zh. Anal. Khim., XLII (1987), 1,
114.
H. Hocquaux, R. Mieland : Revue de metallurgie CIT, 89(1992),
February, 193.
K.V. Grigorovitch, A.M. Katsnelson, A.S. Krylov, AV. Vvedenskii:
New approach to the fractional gas analysis application in
metallurgy and material control using oxide separation software
developed, Analytical Chemistry in the Steel and Metal Industries,
4th Intern. Conf. Proc., Luxembourg, 1995, p.527.
K.V. Grigorovitch, P.V. Krasovskii, S.A. Isakov, A.A. Gorokhov,
A.S. Krylov: Industrial Lab., 68 (2002), 9, 3.
P.V. Krasovskii, K.V. Grigorovitch: Industrial Lab., 68 (2002), 10,
17.
74