Experimental Investigation of SF6 Decomposition

under Failure in Gas Insulated Equipment

Xianglian Yan, Chengyu Wang, Yansong Ji, Zhibing Li and Dexin Cao
Department of High Voltage Technology
China Electric Power Research Institute
yanxl@epri.sgcc.com.cn

Abstract
In the presence of arc, partial discharge and thermal
discharge, SF6 decomposes into a wide variety of
chemically active products which possess completely
different properties from SF6. This report studies SF6
decomposition relating to the experiments in test
vessels. Results on the qualitative and quantitative
determination of the by-products are summarized
according to various modes of electrical discharges.
Using the guidance of detection technology for
decomposition products of SF6, the decomposition
mechanism of SF6, the constituent of the decomposed
gases of SF6 and their stability under various failures
in gas insulated switchgear (GIS) are analyzed. The
following decomposed gases such as SO2, H2S, SOF2
and S2OF10 can be used as the gasses characterizing
the fault discharges of GIS. According to the
characteristic products, the condition of internal fault
inside GIS as well as whether the internal fault is
increasingly extending can be determined for fault
diagnosis of SF6 gas equipment.

Keywords
SF6 decomposition, Gas insulated equipment, Failure,
Decomposition mechanism, Experimental investigation
1. INTRODUCTION
The long term aging in the presence of minute
quantities of SF6 decomposition products will continue
to receive considerable attention as utilities are seeking
to stretch the service life of their equipment. Although
the effects of decomposition products on solid
insulating materials have been extensively investigated,
most of the research focused on the short term and
large dosage exposure situations. The extrapolation of
the short term results to the long term situations is not
obvious and may require further investigation.
Much of the earlier work on SF6 decomposition
focuses on the qualitative analysis of the by-products
formed under conditions easily simulated in the
laboratory with a gas cell. The limitations in those
early days include the lack of sophisticated analytical
techniques to determine unambiguously the types of
species formed. As newer techniques will be developed
and researchers will gain a better understanding of the
decomposition environment, research emphasis is
shifted to the quantitative analysis of the by-products
and attempts is made to estimate the amounts formed
in a realistic environment as encountered in operating
gas insulated switchgear (GIS).

A satisfactory understanding of the processes leading

to the final decomposition by-products in SF6 under all
possible circumstances: (1) the identification of the
primary (as opposed to reaction-formed) by-products;
(2) the establishment of reaction mechanisms which
lead through intermediate species to the final longlived by-products; and (3) the quantification of the
results. The goal of the present work is to contribute
toward an improved understanding of by-product
formation.
In this paper, the decomposition mechanism and their
by-products under different failures were introduced in
Section2. Then in section3, the test results of SF6
decomposition products under failures in gas insulated
switchgear were analyzed in detail. Finally the
conclusions were drawn in section 4.
2. DECOMPOSITION MECHANISMS
The decomposition of SF6 can be caused by the various
modes of formation in high current arc, spark, corona
or overheating. In the development of SF6 circuit
breakers, the influence of decomposition products on
the dielectric recovery after current zero is critical to
the breakers function or on the thermal phenomena
during high current interruption in a hot SF6 gas gap.
In gas insulated equipment, decomposition of SF6 by
an electrical discharge is the most common mode of
dissociation. The discharges can be broadly divided
into three types according to the energy dissipated in
the process: arc (current interruption, in-service fault),
spark (capacitive discharge, fault during testing,
disconnector switching), corona or partial discharge
(stress enhancement, electrically floating component
and free conducting particles) and overheating (poor
contact). The most possible scenario for the formation
of the various by-products and the metal fluoride from
contaminants is shown in Fig. 1.

Fig. 1 Decomposition of SF6 by discharge

2.1 SF6 decompositions in power arc

Generally, the arc plasma can be assumed in local
thermodynamic equilibrium. As the temperature
increases beyond 1500K, there is a sharp drop in the
SF6 concentration with a corresponding increase in F
and SF4 density. The equilibrium concentration of SF6
and the decomposition products at temperatures
between 1000K and 3500K have been verified. At
4000K, the molecules are almost completely
dissociated while most of the electrons are still bound
due to the formation of negative ions. At temperatures
beyond 15000K, the particles are mostly positive ions
and electrons. At the center of a high current arc, the
temperature has been measured to be about 20000K.
As the arc cools, recombination of sulphur and fluorine
to form SF6 occurs rapidly, often in the s scale. In the
presence of oxygen, H2O and metal vapor as a result of
electrode heating, the recombination process is altered
which leads to the formation of various arc by-products.
At temperatures below 3000K, the main constituents
are SF2, SF4 and F. As the arc cools from the initial
high temperature, the stable SF4 products will dominate
at temperatures between 1500K and 2500K. Reaction
between oxygen and SF4 will probably take place at
that temperature range to give some of the reactions
outlined above. As the stable SF4 diffuses out from the
arc zone, reaction with H2O will dominate to form the
lower oxy-fluoride. SF5 is not formed in appreciable
quantities due to its instability. The formation of SOF4
will be much less likely because of the lack of parent
compounds. Therefore the formation of SO2F2 or SOF4
is very small in the case of a power arc in comparison
to corona or spark.
The results of prior work with power arcs can be
summarized as follows:
1) SOF2 is the major stable decomposition product
formed in an SF6 power arc.
2) There is ample evidence to indicate that SOF2 is
generated from the primary products SF4 and SF2.
3) The formation rate of SOF2 is about 600nmol/J for
aluminum electrodes. The rate appears to decrease for
low arc current and copper alloy electrodes.
2.2 SF6 decompositions in partial discharge
(1)In corona
In a corona, the electron temperature is much higher
than the gas temperature in which non-equilibrium
conditions prevail. Therefore the corona is the electrondominant process which is a weakly ionized gas. The
formation of negative ions of SF4 and other lower
fluorides of sulphur eventually leads to the formation
of various neutral molecules. Multi-step dissociation
might also contribute to the formation of lower
fluorides of sulphur such as SF4, SF3. In the absence of
contaminants in the gas or on surfaces, the products of
decomposition of SF6 quickly recombine through
relatively rapid processes to explain the thermal
conversion of SF4+F2 into SF6. However, in the
presence of oxygen or water vapor, the recombination
process can be interfered with by the reaction between

the lower fluorides of sulphur and the contaminants to

form sulphur-oxy-fluoride, HF, and metallic fluoride.
While SOF2 is still the dominant species formed in the
decomposition products, the concentration of SO2F2 is
much higher than that found in arcs and to a certain
extent, in sparks. This appears to be a unique
characteristic of corona or low-energy discharges. The
pattern appears that as the discharge temperature
decreases (e.g. from arc to corona), the concentration
of SO2F2 increases. S2F10 is only identified in one
investigation while SOF4 has been identified in three
laboratories. The total formation rates are within an
order of magnitude in spite of the different electrode
materials and diversified experimental conditions. The
estimated formation rate of the total sulphur oxyfluoride in corona appears to be in the neighborhood of
600pmol/C.
(2)In spark
The quantitative ordering of the major long-lived stable
by-products (after times sufficiently long to convert
SF4 to SOF2) is consistently found to be: SOF2 > SOF4
> SiF4 > SO2F2 > SO2.Where, SOF2 constitutes over
half (often 90%) of the total amount. All these byproducts increase in concentration with the number of
sparks. The amount of by-products, SOF2, SiF4 and
SO2F2 increase slowly with time, reaching a maximum
value several hours after sparking, while the amount of
SOF4 is found not to change with time for t>0.5h. Also,
the by-products SOF2 and SiF4 are found to exhibit
energy/pressure dependences, while for SOF4 and
SO2F2 the amount of by-product per unit spark energy
is independent of pressure. It is believed that SOF4 and
SO2F2 formation is linked to each other, acquiring
oxygen from the electrode material rather than from the
gas itself. The addition of oxygen may also influence
the mechanism of SOF4 and SO2F2 formation.
While SOF2 is still the major by-product observed, the
quantities of SO2F2 appear to increase in comparison
with the results from arc discharges. Furthermore,
SOF4 is detected. The other major difference between
decomposition in spark and arc is that the spark
decomposition rate appears to be much lower than that
of an arc discharge. With the exception of the tungsten
case, the copper and stainless steel formation rates are
in the order of 2 to 5nmol/J. The third difference is in
the detection of S2F10 or S2F10O in a spark. With the
exception which detected trace amount of S2F10, there
is no report of its formation in power arc discharges. In
a spark, the detection of S2F10 appears to be
unambiguous and the quantities are substantial.
2.3 Thermal decompositions of SF6
The thermal and chemical stability of SF6 at normal
operating temperature of the gas insulated equipment is
an important issue because it affects the long-term
reliability and aging of the equipment. In comparison
to the electrical decomposition, thermal decomposition
of SF6 has received relatively little attention.

3.1 Arc discharge test

Experiments were carried out with gas mixtures
containing SF6 and moisture content was 50 or
113L/L with absorbents. Arc discharge tests under
0.6MPa SF6 were exposed to 1 ~ 31.5 kA for durations
ranging from 5 to 20 ms. The electrode materials are
Chrome and Copper. Gas analysis of the test results
confirms the formation of SO2 as well as H2S as the
main characteristic products. Traces of CO and HF are
also identified within 24 hours of arc discharges.

SO2/H2S

6
4
2
0
0

10

15

Inspection time for arc discharge in h

Fig. 2

Ratio SO2/H2S of arc discharge in different

currents with inspection time

4
SO2/H2S

3. EXPERIMENTAL INVESTIGATIONS
Gas analysis provides a promising means for a
diagnosis of gas insulated equipment. An experimental
investigation over decomposition by-products at
various abnormal conditions is carried out by use of
simulating models to develop a diagnostic method for
SF6 insulated equipment based on gas sensor and gas
chromatograph
analyses.
Major
components
characteristic to each abnormality such as overheat,
partial discharges, arc discharges, etc. are identified by
the experiments. At the present study attention is also
paid to the identification of SF6 decomposition byproducts, which are a major effect from different
discharge energy occurring in SF6 insulated equipment.
Since the formation of SF6 decomposition by-products
is affected by the amount of containing O2 and H2O,
drying and evacuation level of materials at each model
test is kept at the level of the manufacturing process in
gas insulated equipment. Due to the absorbents, proper
diagnostic time is needed to prevent the disappearance
and conversion of the by-products during transit from
field to laboratory.

The influence of discharge energy and moisture in SF6

arc discharges as well as the formation of SF6 arc byproducts in test vessels are studied. It examines and
proves the effectiveness of adsorbents in removing the
by-products. Even at high moisture contents of
113L/L, the major species detected are still SO2 and
H2S.
The formation ratio of SO2/H2S is obtained from
investigations. Fig. 2 shows that the ratio is about 1~8
in despite of the scattered values. A linear relationship
is obtained on the production ratio of SO2/H2S as a
function of arc energy as shown in Fig. 3 and Fig. 4.
There appears to be a tendency that the formation ratio
depends on the fault type and discharge energy. The
ratio SO2/H2S is about 1 ~ 3 and 2 ~ 6 for high and low
arc energy, respectively. It is possible to evaluate the
internal fault condition by a rough comparison with the
ratio SO2/H2S at SF6 decomposition. Although
absorbents and moisture have the main influence on the
quantity of SO2 and H2S, the ratio of SO2/H2S stays
constant. Therefore, it is suspected that absorbents
have the same effect on these two species and that the
influence of moisture on the formation of SF6 arc byproducts can be neglected due to absorbents.
8

3
2
1
0
0

Inspection time for high arc energy in h

Fig. 3 Ratio SO2/H2S of high arc discharge in

different currents with inspection time
8
SO2/H2S

It showed that the critical temperature of SF6 was

600C, above which SF6 was observed to decompose
into SF4 and SF2. However, it showed that at
temperatures as low as 150C there is evidence of SF6
reaction with silicon steel plates. At temperatures
above 200C, many metals start to react with SF6
forming metal fluorides and sulphides. In addition to
the formation of fluorides and sulphides on the metal
enclosures surface, gaseous decomposition products
such as SO2 were also identified. The influence of
moisture on the formation of SO2 was found to be
significant.
The decomposition rate of SF6 increases markedly as
temperature increases. At temperatures of 650C, it
reported the formation of SOF2, SO2F2 and SO2 when
aluminum and copper were exposed to SF6 for 90h.
Between 700C and 900C, It observed that SF6
decomposed to SF4 in stainless steel, copper and quartz
tubes containing SF6. In the case of quartz tubes, SF4
was observed to react with SiO2 to form SiF4 and SOF2.
Results appeared to confirm that at temperatures up to
1500C, SF6 dissociated principally to SF4. The
subsequent conversion of SF4 to other oxy-fluorides
would be governed by the presence of impurities such
as oxygen and H2O in the gas and on the metal surfaces.

6
4
2
0
0

10

20

30

40

Inspection time for low arc energy in h

Fig. 4 Ratio SO2/H2S of low arc discharge in

different currents with inspection time

50

SOF2/S2OF10

60 kV; 75 L/L

0 kV; 170 L/L

20
15
10
0

10

15

Discharge time in h

Fig. 5

Ratio SOF2/S2OF10 of partial discharge versus

discharge time with absorbent

60 kV; 75 L/L

0 kV; 170 L/L

35 kV; 75 L/L; No absorbent

45 kV; 175 L/L; No absorbent

SOF2/S2OF10

3.2 Partial discharge test

In partial discharge tests, point-plane electrodes and
turn insulation of Teflon insulated wire are selected as
partial discharge models. Although catalytic effect of
metallic materials does not seem to play a major role in
discharge decomposition, two different materials,
including alumina and copper, are used to test whether
there is any catalytic difference between them.
Continuous discharges are necessary to obtain a
conceivable level of gaseous decomposition byproducts. Since the discharge level of test samples
varied widely, the distribution of the discharge level is
recorded by a pulse height analyzer and total charge is
calculated from the measured distribution.
The decomposition products formed of partial
discharge are identified as SOF2, SO2F2, S2OF10 or
S2F10. No H2S is detected. In contrast to the arc
discharge experiment, significant quantities of S2OF10
or S2F10 are detected in the partial discharge which are
completely absent in arc and partial discharges. The
gas analysis indicates that at low energy discharges
SO2F2 and S2OF10 are the dominant product formed,
and as the energy of the discharge is increased, SO2
and H2S are also identified in the thermal discharge. In
some cases, the product SO2 is detected in very small
amounts. With regard to CO2 generation, all of the
experiments show a saturating nature of CF4 on the
dissipated discharges in case of Teflon insulation.
Although the experimented cases are too limited to
confirm this tendency, it is likely that CO2 and CF4 are
the major by-products of Teflon discharges.
In all experiments this decomposition by-product SOF2
is found to be the most abundant. The amount of SOF2
formed is found not to be a simple function of either
the total discharge energy or the gas pressure alone.
Moreover, the amount of SOF2 also changes with
moisture and absorbents. However, the formation ratio
of SOF2/S2OF10 is generally found to be constant
observed in our discharged SF6 samples, shown as in
Fig. 5. Unlike the behaviour found for SOF2, the
formation ratio of SOF2/S2OF10 exhibits no indication
of the moisture, pressure and energy dependence,
shown as Fig. 6. The detected amount of the ratio is
typically about 15 with absorbents and is from 8 to 25
without absorbents, respectively. Since the ratio of
SOF2/S2OF10 appears to be a good indicator of SF6 gas
degeneration, the ratios over discharge magnitude and
calculated energy are plotted to exhibit the effect of
discharge differences on SF6 decomposition and also to
compare the correlations with dissipated total charge
energy. Fig. 5 shows the independence of the ratio on
total charge energy and relatively constant ratio over
total discharge energy. SOF2 appears to be about 15
times as much as that of S2OF10.

30
20
10
0
0

10
Discharge time in h

Fig. 6 Ratio SOF2/S2OF10 of partial discharge versus

discharge energy with and without absorbent
3.3 Abnormal thermal test
In comparison to the electrical decomposition, thermal
decomposition of SF6 receives relatively little attention.
Polymeric materials such as epoxy, polyethylene,
Teflon, and phonetic are frequently used as solid
insulators inside SF6 insulated equipment. The most
common type of support insulators in GIS are made of
filled epoxy. When a flashover occurs at the SF6 solid
interface (e.g. spacer surface flashover), the presence
of polymeric materials in the discharge influences the
formation of the by-products. At the overheat tests of
metallic and insulation materials, the main gaseous byproduct characteristic to material overheat is SO2. The
observation of large amounts of SO2 is consistent with
observations of H2S as the main decomposition product.
The surface adsorbed H2O and organic material are
suspected to play an important role in the production
process. The concentrations of SO2, CF4, SF4 and CO2
are found to be higher when the thermal discharge
occurs near polymeric materials. In addition, CS2 is
identified in overheating near spacers.
Although both heating temperature and heating time
show a slight difference in generation of gaseous byproducts, Fig. 7 shows similar ratio of SO2/H2S.
Though divergent at the data of Fig. 8, Fig. 9 and Fig.
10, whole experimental data from 0.15MPa to
0.45MPa, from 200C to 355C show the similar
tendency in the ratio even with or without absorbents.
In case of overheating with absorbents, general image
of the decomposition is that the decomposing material
is SF6 gas and the absorbent works as a removal. In the
experiment of the overheating material, attention is
paid to removal effect of the absorbents as well as
generating pattern of decomposition by-products. In

15

case of overheating of ceramic or silica insulation,

difference in the decomposition patterns can be
neglected. Comparison of the data between different
temperatures and pressures, suggests that the ratio
SO2/H2S does not change with temperature or pressure.
This leads to a possibility of estimation criteria in an
assessment of fault types with the aid of SO2/H2S ratio
investigated in SF6 insulated equipment.
250C

265C

270C

Comparison of ratio SO2/H2S between arc and thermal

discharges and also between high and low arc energies
is made in Fig. 11. The formation ratios are within one
order of magnitude in spite of the different electrode
materials and diversified experimental conditions. The
pattern appears that as the discharge energy decreases
(e.g. from arc to partial discharge), the concentration
ratio of SO2/H2S increases. The higher ratio SO2/H2S
from 8 to 25 can be attributed to hydrolyse of gaseous
by-product SOF2 and enhancing effect of adsorbed
moisture on insulation materials. In contrast to data
divergence of overheating tests, the ratio of SO2/H2S at
arc tests show relatively defined relation with high and
low arc energy.

280C

SO2/H2S

30
20
10

0
0

10

Fig. 7

Ratio SO2/H2S of thermal discharge versus

heating time with absorbent
0.25 MPa

SO2/H2S

Fault type

Heating time in h

50
40
30
20
10
0
220

0.35 MPa

0.45 MPa

0
0

230

240

250

260

270

Fig. 8 Ratio SO2/H2S of thermal discharge versus

pressure and temperature with absorbent
0.27 MPa

0.39 MPa

305 C

330 C

355 C

SO2/H2S

30
20
10
0
0

10

20

30

40

Heating time in minute

Fig.9 Ratio SO2/H2S of thermal discharge versus

pressure and temperature without absorbent

Thermal discharge

with absorbent

without absorbent

0
0

10

15

20

25

SO2/H2S

Temperature in C

0.15 MPa

10

15

20

SO2/H2S

Fig.10 Ratio SO2/H2S of thermal discharge with and

without absorbent

25

1-high arc discharge; 2-low arc discharge;

3-thermal discharge
Fig. 11 Ratio SO2/H2S of arc and thermal discharges
with absorbent
4. CONCLUSION
(1) SOF2 is the major stable decomposition product in
power arc and partial discharge, the concentration of
SO2F2 in corona is much higher than that found in
power arc and in sparks. The critical temperature of
SF6 is 600C.
(2) The ratio of SO2/H2S is about 1~8 for arc discharge
in despite of the scattered values. The ratio of SO2/H2S
is about 1 ~ 3 and 2 ~ 6 for high and low arc energy,
respectively.
For partial discharge, the formation ratio of
SOF2/S2OF10 exhibits no indication of the moisture,
pressure and energy dependence. The detected amount
of the ratio is typically about 15 with absorbents and is
from 8 to 25 without absorbents, respectively.
Comparison of the data between different
temperatures and pressures, suggests that the ratio
SO2/H2S does not change with temperature or pressure.
The higher ratio SO2/H2S from 8 to 25 can be
attributed to hydrolyse of gaseous by-product SOF2 and
enhancing effect of adsorbed moisture on insulation
materials.
(3) The presence of the gaseous and solid by-products
may lower the dielectric strength of gas insulated
equipment. The identification of SF6 decompositions in
electrical equipment offers a means to diagnose its
internal condition.

Acknowledgements
The authors would like to thank the ICEE Program
Committee for their kind assistance in the preparation
of the guidelines.

References
[1] F. Y. Chu, J. M. Braun, and R. Seethapathy,
Degradation Mechanisms for Epoxy Insulators
Exposed to SF6 Arcing By-Products, Proc. IEEE Intl.
Symp. Electr. Insul., June 1986.
[2] C. Boudene, J-L Cluet, G. Keib, and G. Wind,
Identification and Study of Some Properties of
Compounds Resulting from the Decomposition of SF6
under the Effect of Electrical Arcing in CircuitBreakers, Revue General Electricity Special Issue,
June 1974.
[3] J. M. Braun and F. Y. Chu, Novel Low-Cost SF6
Arcing By-Product Detectors for Field Use in GIS,
IEEE PES Summer Power Meeting, Vancouber, B.C.,
Paper 85 SM369-4, July 1985.
[4] A. Hollinger. Analyzer for Decomposition Products
and Moisture in SF6 Gas, Can. Electr. Assoc. Final
Report 086-T-176, June 1984.
[5] I. Sauers, H. W. Ellis, L. G. Christophorou, G. D.
Griffin, and C. E. Easterly, Spark Decomposition of
SF6; Toxicity of By-Products, Oak Ridge National
Laboratory Report ORNL/TM- 9062, October 1984.
[6] V. G. Arakelian and A. J. Kovalenko, Diagnostics
for the Condition of SF6 Equipment Based on PhysicoChemical Parameters, IEEE Electrical Insulation
Magazine, Vol. 17, No. 2, 2009, pp: 42-51.
[7] X. L. Yan, C. Y. Wang, Y. S. Ji, Y. Y. Guo,
Application of Decomposition Products Detection of
SF6 in Switchgears, Power System Technology, Vol. 34,
No. 9, Sep. 2010, pp:160-165.
[8] Y. W. Liu, L. Y. Wu, Y. P. Gong, Investigation on
SF6 Decomposition Products in GIS and Affecting
Factors, Power System Technology, Vol. 33. No. 5,
Mar. 2009, pp:58-61.
[9] Shotaro Tominaga, Hiroshi Kuwahara, Koichi
Hirooka, et a1. SF6 Gas analysis technique and its
application for evaluation of internal conditions in SF6
gas equipment, IEEE Transactions on Power
Apparatus and Systems, 1981, Vol.100, No.9, pp:41964206.
[10] Yan Xiang-lian, Wang Cheng-yu, Yang Ren, Ji
Yan-song, et al. Fault Diagnosis of High Voltage
Switchgears by Decomposition Products of SF6, Power
System technology, 2011, Vol.35, No.12, pp:120~123.
[11] Belarbi, Pradayrol A C, Casanovas J, et al.
Influence of discharge production conditions, as
pressure, current intensity and voltage type, oil SF6
dissociation under point-plane corona discharges,
Journal of Applied Physics, 1995, Vol. 77, No.4,
pp:1398-1406.
[12] Belmadani, Casanovas B J, Casanovas A M. SF6
decomposition under power arcs, IEEE Transaction on
Chemical Aspects B, 1991, Vol.26, No.6, pp:11771182.

[13] Zhang Xiaoxing, Yao Yao, Tang Ju, Sun Caixin,

Wan Lingyun. Actuality and Perspective of Proximate
Analysis of SF6 Decomposed Products under Partial
Discharge, High Voltage Engineering, 2008, Vol.34,
No.4, pp:664-747.
[14] Junya Suehiro, Guangbin Zhou, Masanori Hara.
Detection of partial discharge in SF6 gas using a carbon
nanotube based gas sensor, Sensors and Actuators,
2005, Vol.105, No.2, pp:164-169
[15] Derdouri A, J Casanovas. 3 Spark decomposition
mixtures of SF6/H2O, IEEE Transactions on Electrical
Insulation, 1989, Vol.24, No.6, pp:1147-1157.