J Therm Anal Calorim (2015) 122:449–461
DOI 10.1007/s10973-015-4701-2
Characterization of thermal decomposition behavior
of commercial flame-retardant ethylene–propylene–diene
monomer (EPDM) rubber
Ruiyu Chen1,2 • Shouxiang Lu1 • Changhai Li1 • Manhou Li1,2 • Siuming Lo2
Received: 9 March 2015 / Accepted: 7 April 2015 / Published online: 26 May 2015
 Akadémiai Kiadó, Budapest, Hungary 2015
Abstract The thermal decomposition behavior of the
commercial flame-retardant ethylene–propylene–diene
monomer (EPDM) rubber was studied employing cone
calorimeter and simultaneous thermogravimetry (TG)–dif-
ferential scanning calorimetry (DSC). Significant influence
of the external heat flux on the thermal decomposition
behavior of the commercial flame-retardant EPDM rubber
was characterized with the measurement of various pa-
rameters, including the visual observation, the ignition
time, the mass loss, the mass loss rate, the heat release rate
and the effective heat of combustion employing cone
calorimeter. Three different decomposition regions of the
commercial flame-retardant EPDM rubber may be identi-
fied according to the applied external heat flux employing
cone calorimeter: (1) region 1 (external heat flux
B35 kW m-2) with well exertion of effectiveness of fire
retardants to the EPDM rubber; (2) region 2
(35 kW m-2 \ external heat flux B 45 kW m-2) with
gradual loss of effectiveness of fire retardants to the EPDM
rubber; and (3) region 3 (external heat flux [45 kW m-2)
with little effectiveness of fire retardants to the EPDM
rubber. Six thermal decomposition stages of the commer-
cial flame-retardant EPDM rubber were noted in both of
region 2 and 3. However, merely four thermal decompo-
sition stages were demonstrated in region 1. Similar ther-
mal decomposition behaviors of the commercial flame-
retardant EPDM rubber to those of cone calorimeter
& Shouxiang Lu
sxlu@ustc.edu.cn
1
State Key Laboratory of Fire Science, University of Science
and Technology of China, Hefei, China
2
Department of Civil and Architectural Engineering, City
University of Hong Kong, Hong Kong, China
experiments were examined in the TG–DSC tests, thus
validating the cone calorimeter experimental results. The
present study may be of practical use in fire safety design
of the underground rail transportation systems.
Keywords Commercial flame-retardant  EPDM rubber 
Thermal decomposition  Cone calorimeter  TG–DSC
Introduction
Fire safety of passenger trains, especially for those gener-
ally operating underground or inside the tunnels such as
metro trains where the escape and rescue of numerous
passengers are considerably difficult in case of fire, is of
significant concern [1–15]. The floor coverings and seats
are generally the major fixed flammable materials in the
carriage interior of the metro trains in China (the portable
flammable materials such as the luggage from the passen-
gers are not taken into consideration under present study),
and the seats in some metro train carriages are made of
non-combustible stainless steel in China. Thus, the primary
fire hazard in the interior of these metro train carriages is
from the floor coverings. As a consequence, it is of great
importance to evaluate the fire hazard of the floor coverings
in the metro train carriages.
Ethylene–propylene–diene monomer (EPDM) rubber is
a type of synthetic rubber, which has been widely used as
floor coverings in the metro train and other passenger train
carriages [1, 2, 4, 7, 10]. This polymer presents numerous
advantages and excellent performances such as prominent
electric insulation, outstanding elasticity, high abrasive
resistance, remarkable slipperiness resistance, excellent
resistance to aging owing to heat, light, oxygen and
ozone. [16]. However, EPDM rubber has relatively high
123
450
flammability and high emission of smoke. Thus, fire re-
tardants are generally added into EPDM rubber in order to
improve its fire performance and reduce its smoke yield
when it is applied into the metro train and other passenger
train carriages [2, 9, 14]. Therefore, the EPDM rubber in
commercial use in the metro train and other passenger train
carriages is commonly flame retardant. Nevertheless, the
commercial flame-retardant EPDM rubber could still be
ignited and burnt under high external heat fluxes. Large
amount of heat and toxic gases will be released with the
burning of the commercial flame-retardant EPDM rubber.
Other flammable materials in the metro trains may be in-
volved in the burning subsequently, which could lead to
terrible casualties, environmental pollution and financial
loss. Furthermore, the thermal decomposition behavior of
the commercial flame-retardant EPDM rubber probably
becomes quite different from that of the pure EPDM rubber
[17, 18]. Therefore, it is of great significance to investigate
the thermal decomposition behavior of the commercial
flame-retardant EPDM rubber for accessing its potential
fire hazard, providing guidance to fire safety design of the
underground rail transportation systems, such as estimation
of the heat release rate (HRR) of the metro trains em-
ploying Duggan’s method [3] or CFD softwares [7, 10, 11]
with the input of the HRR curves of the flammable mate-
rials in the metro trains obtained in the present study in
case of fire for ventilation design of the operation tunnels
[3, 8, 11].
Much attention has been directed toward how to im-
prove the fire resistance of EPDM rubber by means of
adding various fire retardants and altering the proportion of
the fire retardants into the EPDM rubber [19–21]. The ef-
fectiveness of the fire retardants to EPDM rubber is gen-
erally determined through specific apparatus such as cone
calorimeter [22–28], and the effectiveness of the fire re-
tardants to EPDM rubber is commonly evaluated under the
specific condition employing cone calorimeter, for in-
stance, 35 kW m-2 which represents the condition of the
Fig. 1 General description of
room fire in absence of fire
control [29]
Flashover
Temperature rise
Ignition
Growth
123
R. Chen et al.
growth stage in real fires [1, 11, 29, 30], as shown in Fig. 1.
In general, the fire retardants aim to resist the ignition and
fire spread of the flammable materials, that is, the fire re-
tardants are generally in action at the initial stage and
growth stage of real fires. The effectiveness of the fire
retardants to the EPDM rubber may be disabled under re-
markably high heat flux or temperature conditions such as
the flashover stage and fully developed stage of real fires
such as Daegu subway fire [6], possibly leading to quite
different thermal decomposition behavior of the commer-
cial flame-retardant EPDM rubber from that under low heat
flux or temperature conditions. As a consequence, it is
considerably important to investigate the thermal decom-
position behavior of the commercial flame-retardant
EPDM rubber under remarkably high heat flux or tem-
perature conditions to evaluate its fire hazard, such as ob-
taining its maximum HRR from the complete combustion.
However, to the best knowledge of the authors, no previous
research has been focused on a complete and detailed
thermal decomposition behavior study of the commercial
flame-retardant EPDM rubber employing cone calorimeter
to evaluate and characterize its thermal decomposition
process, especially under high heat flux or temperature
conditions.
Therefore, the present study focuses on the effect of a
large cone calorimeter irradiance level range on the ther-
mal decomposition behaviors of the commercial flame-re-
tardant EPDM rubber, including the visual observation, the
ignition time, the mass loss and mass loss rate (MLR)
evolution, the HRR and effective heat of combustion
(EHC) evolution. The correlation of the MLR and the HRR
allows the description of the physical and chemical phe-
nomena occurring during the thermal decomposition pro-
cess of the commercial flame-retardant EPDM rubber.
Besides, in order to further understand the thermal de-
composition process of the commercial flame-retardant
EPDM rubber, simultaneous thermogravimetry (TG)–dif-
ferential scanning calorimetry (DSC) experiments under
Postflashover
Fully developed fire
Decay
Time
Characterization of thermal decomposition behavior of commercial flame-retardant ethylene… 451

non-isothermal and isothermal conditions in air were Table 2 Compositions and corresponding mass fraction of the
examined. commercial flame-retardant EPDM rubber
Category Mass fraction/%

Pure EPDM rubber 27

Experimental
Microencapsulated red phosphorus 1.5
Material Decabromodiphenyl ethane 6
Magnesium hydroxide 10
Table 1 presents the essential properties of the commercial Aluminum hydroxide 20
flame-retardant EPDM rubber under present study. The Antimonous oxide 4
density and limiting oxygen index (LOI) were provided by Zinc borate 6
the manufacturer. The specific heat, thermal conductivity Fire retardant dedicated for polyolefin (HT-101) 1.8
and thermal diffusivity were measured adopting the hot Zinc oxide 0.6
disk TPS2500s. The compositions and the corresponding Stearic acid 0.2
mass fraction of the commercial flame-retardant EPDM Ozone-resistant antiager (MB) 0.8
rubber are illustrated in Table 2. Oxygen-resistant antiager (SP) 1.2
Precipitated silica 4
Measurement Pulvistalci 6
Titanium dioxide 5
Cone calorimeter Dispersing agent (WB-16) 0.5
Surfactant (PEG 4000) 1.5
Cone calorimeter has been widely used to investigate the Microcrystalline wax 0.5
thermal decomposition behavior of materials [27, 31–36]. Antioxidant (B225) 0.4
An ISO 5660-1 standard Cone Calorimeter [22] with a Yellowing-resistant accelerator (AG 201) 0.4
digital electronic balance (UX6200H) with the accuracy of Vulcanization accelerator (TMTD, TiBTM) 1.8
0.01 g was employed to perform the cone calorimeter ex- Insoluble sulfur 0.8
periments, as shown in Fig. 2a. The dimension of the
specimens was 99 ± 1 mm long, 99 ± 1 mm wide and
3 mm thick. The mass of the specimens was 52 ± 1.6 g.
Aluminum foil was used to wrap the edges and rear surface in the cone calorimeter were adopted to obtain the varia-
of the specimens. A wire grid, which was made of 2-mm tions of the concentration of the CO2 and CO in the gen-
stainless steel rod with all intersections welded, was em- erated gaseous compounds. The ranges of the CO2 and CO
ployed to prevent the specimens from intumescing, as measurement were 0–1 and 1–10 %, respectively [22, 37].
shown in Fig. 2b. Ceramic fiber blanket was used un- The selected heat flux levels were 20, 25, 30, 35, 40, 45,
derneath the specimens for insulation. The specimens as 50, 55, 60 and 65 kW m-2. Tests were carried out in a
well as the specimen holder were positioned horizontally condition with ambient temperature of 298 ± 2 K and
on the lifting platform. The distance between the cone 50 ± 5 % relative humidity. Cone calorimeter was
heater and the top surface of the specimens was 25 mm. A calibrated conforming Ref. [22] before each test. The ex-
paramagnetic oxygen analyzer with a range of 0–25 % in periments were stopped manually if no ignition occurred
the cone calorimeter was used to measure the oxygen after the specimens were exposed to the given heat fluxes
concentration in the generated gaseous compounds from for 32 min in accordance with ISO 17554 [38]. Each ex-
combustion of the specimen [22, 37]. NDIR gas analyzers periment was conducted twice, and the experimental data
used under present study were the average value of the two
Table 1 Essential properties of the commercial flame-retardant
experiments administered.
EPDM rubber
Category Value TG–DSC
Density q/kg m-3 1500
TG–DSC experiments were conducted using a simultane-
Specific heat c/J kg-1 K-1 1020
ous TG–DSC Q Series instrument, SDT Q600 from TA
Thermal conductivity k/W m-1 K-1 0.102
instrument in air. Cuboid specimens with mass of ap-
Thermal diffusivity a/m2 s-1 6.67 9 10-8
proximately 12 mg were heated from 298 up to 1073 K at
LOI/% 35.7
the heating rates of 10, 20 and 40 K min-1 and also heated

123
452
Fig. 2 Experimental schematic of cone calorimeter and specimen
under six isothermal conditions of 723, 773, 873, 923, 973
and 1023 K. The flow rate of the air was 100 mL min-1.
Each experiment was repeated once, and the experimental
data presented in this paper were the average value of the
two experiments performed.
Results and discussion
Cone calorimeter test
Visual observation
Localized swelling and bubbling phenomena were ob-
served at the gaps of the wire grid when the specimens
were thermally decomposed, as shown in Fig. 3. Ripping
sound was heard, and gray volatile gases were released
before ignition. The burning of the specimen at the initial
stage immediately after ignition and the late stage of the
test was more intense than that at other stages of the test
under high external heat fluxes. This may be due to that
higher HRRs at the initial stage immediately after ignition
and the late stage of the test than those at other stages of the
Fig. 3 Appearance of the
residues of the specimens after
test
123
R. Chen et al.
test under high external heat fluxes occurred. Distinct
transitory flaming, which is defined as existence of flame
on or over the surface of the specimen for a period of time
longer than 1 s, but shorter than 10 s [39], occurred before
sustained flaming that denotes persistence of flame on or
over a surface of the specimen for at least 10 s [39] hap-
pened especially for specimens under low external heat
fluxes: 25, 30 and 35 kW m-2. Furthermore, the number of
the transitory flaming decreased with the increase in the
external heat flux. This was likely due to that the fire re-
tardants could be in action to resist the decomposition of
the EPDM rubber under low external heat fluxes, so the
amount of the decomposed flammable volatiles could not
support the sustained flaming within a short time and thus
resulted in the occurrence of the transitory flaming before
the sustained flaming happened. With the increase in the
external heat flux, the effectiveness of the fire retardants to
the EPDM rubber may be gradually lost, enough amount of
the decomposed flammable volatiles could be provided for
the sustained flaming within a short time and the number of
the transitory flaming would consequently decrease [17].
Burning durations were much shorter at lower external
heat fluxes (B35 kW m-2) than those at higher external
Characterization of thermal decomposition behavior of commercial flame-retardant ethylene… 453

heat fluxes, meaning that lower external heat fluxes could (labeled as ‘‘m0’’) of the specimens as a function of ex-
not sustain the combustion up to fuel burnout. It is demon- perimental time, and Fig. 5b shows the variations in mass
strated in Fig. 3b–d that the residues with charred surfaces loss factor / versus external heat flux. The mass loss factor
and apparent crack lines at lower external heat fluxes were / is defined here as the fraction of total mass loss and the
considerably different from those with gray ash at higher initial mass per unit area (the area of the specimens ex-
external heat fluxes, and it was likely that the transition posed to the cone heater was labeled as ‘‘A’’) of the spe-
region of the external heat flux was between 35 and cimens, i.e.,
40 kW m-2, as illustrated in Fig. 3d, e. This difference was m0  m
/¼  100 %: ð1Þ
likely due to the different intensity of combustion of the m0  A
specimens, which would be affected by the function of fire
The acceleration in mass loss of the commercial
retardants in the specimens, possibly suggesting that the fire
flame-retardant EPDM rubber with the rise in external
retardants would lose their effectiveness at high external
heat flux is indicated in Fig. 5a. Longer burning durations
heat fluxes and the transition region of the external heat
are shown for external heat fluxes [35 kW m-2, noting
fluxes at which the effectiveness of fire retardants would
that no ignition occurred at the external heat flux of
begin to lose possibly lay between 35 and 40 kW m-2.
20 kW m-2, which was in accord with the visual ob-
servation as discussed in ‘‘Visual observation’’ section.
Ignition time
This may further indicate that more complete burning
occurred to specimens under external heat fluxes
The ignition time as a function of external heat flux is
[35 kW m-2, and hence, the effectiveness of the fire
presented in Fig. 4. The ignition time corresponded to the
retardants may be gradually lost with the increase in
start time of the occurrence of the sustained flaming instead
external heat flux. More distinct evidence is illustrated in
of the transitory flaming according to the flame image
Fig. 5b. As can be seen from Fig. 5b, the mass loss
record. The left ordinate and right ordinate represent the
factor increased successively up to 35 kW m-2, then a
ignition time and the derivative of the ignition time versus
sharp increase in the mass loss factor occurred when the
external heat flux, respectively. It can be seen from Fig. 4
external heat flux was changed from 35 to 40 kW m-2.
that the ignition time decreased significantly with the in-
After an almost constant mass loss factor at 40 and
crease in external heat flux up to 35 kW m-2 and remained
45 kW m-2, another rapid rise in the mass loss factor is
almost constant at high external heat fluxes (from 45 to
shown from 45 to 50 kW m-2, followed by another
65 kW m-2). It seemed that the effectiveness of fire re-
nearly constant mass loss factor for the external heat
tardants may be in action to resist the ignition of the spe-
fluxes of 50, 55, 60 and 65 kW m-2.
cimens below or equal to 35 kW m-2 and almost
It seemed that the decomposition process of the com-
inoperative under high external heat fluxes [45 kW m-2.
mercial flame-retardant EPDM rubber may be divided into
three regions according to the external heat flux:
Mass loss and mass loss rate
1. Region 1 corresponded to external heat fluxes below or
Mass loss Figure 5a depicts the variations in fraction of equal to 35 kW m-2. In this region, the smallest part
the mass remained (labeled as ‘‘m’’) and the initial mass of the specimens was consumed and the burning
durations were the shortest. It was likely attributed to
Derivative of ignition time/s m2 kW–1

20 that the effectiveness of the fire retardants may be well

Ignition time
600
Derivative of ignition time
exerted to resist the thermal decomposition of the
0
EPDM rubber under low external heat fluxes. The
450 amount of the decomposed flammable volatiles was
Ignition time/s

not enough to sustain the burning of the specimens up

–20
to fuel burnout.
300
2. Region 2 corresponded to external heat fluxes between
–40 35 and 45 kW m-2 (45 kW m-2 was included). In this
150
region, larger part of the specimens was consumed and
–60 longer burning durations were demonstrated in com-
0 parison with those in region 1. However, the specimens
20 30 40 50 60 70
still could not keep burning up to fuel burnout. This
External heat flux/kW m–2
may indicate that the fire retardants would gradually
Fig. 4 Ignition time as a function of external heat flux lose the effectiveness to prevent the thermal

123
454
(a) 100
90 20 kW m–2
Fraction of mass remained
30 kW m –2
m/m0 × 100 %/%
80
40 kW m–2
50 kW m–2
70
–2
60 kW m
60
50
40
0 600 1200 1800 2400
Time/s
Fig. 5 a Fraction of the mass remained and the initial mass as a function of experimental time, and b mass loss factor versus external heat flux
decomposition of the specimens, leading to higher
intensity of burning of the specimens relative to that in
region 1.
3. Region 3 corresponded to external heat fluxes
[45 kW m-2. In this region, the flammable part of
the specimens was totally consumed and the burning
durations were the longest. This phenomenon may be
due to that little effectiveness of the fire retardants to
the EPDM rubber occurred, so the highest intensity of
burning of the specimens was attained among the three
regions.
Mass loss rate The variations in MLR versus ex-
perimental time are shown in Fig. 6. MLR is defined as the
mass rate of solid or liquid fuel vaporized and burned
[40, 41]. Strong dependence of the MLR curves on the
external heat flux can be observed in Fig. 6. The peak MLR
increased with an increase in the external heat flux. Fur-
thermore, the time for the MLR to reach its maximum was
shortened with an increase in the external heat flux. The
MLR curves in the three regions, as noted in ‘‘Mass loss’’
section, were quite different from each other. The MLR
curves at 30, 40 and 60 kW m-2 were selected for illus-
tration, as shown in Fig. 6d. In the case of 30 kW m-2, a
drastic increase in MLR occurred at the outset, followed by
a quasi-steady stage in which transitory flaming occurred,
as discussed in ‘‘Visual observation’’ section. Then, a
second sharp increase in the MLR occurred due to the
ignition and burning of the specimen. After attaining its
peak, significant decline in MLR is shown, followed by a
second quasi-steady stage between 600 and 700 s and rapid
decrease of MLR. However, as shown in Fig. 6c, d, in the
case of 60 kW m-2, the first quasi-steady stage with the
transitory flaming disappeared. In addition, after the MLR
reached its first peak, another quasi-steady stage between
400 s and 525 s and a subsequent local zenith of the MLR
occurred, followed by a rapid decline stage and a similar
123
R. Chen et al.
(b)
70 Mass loss factor
25 kW m–2
Mass loss factor/ % m–2
35 kW m–2 60
45 kW m–2 50
55 kW m–2 Region 1
–2
40
65 kW m
30 Region 2 Region 3
20
10
3000 3600 20 30 40 50 60 70
External heat flux/kW m–2
quasi-steady stage to that at 30 kW m-2. It seemed that the
case of 40 kW m-2 was the transition state between the
cases of 30 and 60 kW m-2. For example, The MLR nei-
ther kept another distinct quasi-steady stage similar to that
at 60 kW m-2 after achieving its maxima nor presented a
direct significant decrease in MLR as the counterpart at
30 kW m-2 after its peak moment, as shown in Fig. 6d.
The different evolution mechanism of the thermal de-
composition process of the commercial flame-retardant
EPDM rubber in different regions, as reported in ‘‘Mass
loss’’ section, may be resulted by whether the effectiveness
of the fire retardants would be exerted on the EPDM rubber
and will be illuminated in the following sections in detail.
Figure 7 shows the variations in average MLR versus
external heat flux. The variations in average MLR were
remarkably consistent with those of the mass loss factor
(see ‘‘Mass loss’’ section). Three regions may also be de-
fined according to the average MLR data, further verifying
the different decomposition behavior of the commercial
flame-retardant EPDM rubber at different external heat
fluxes.
Heat release rate
HRR represents the rate of thermal energy released from
the combustion of the materials and is considered as the
single most important variable in fire hazard evaluation
[34, 42]. The oxygen consumption calorimetry technique
based upon the ISO 5660-1 standard [17, 22, 33, 34, 42–44]
was employed to calculate the HRR with the measurements
of the concentration of gaseous compounds (O2, CO2, CO,
etc.) generated by the combustion of the specimens. Fig-
ure 8 shows the variations in the HRR versus experimental
time. In accord with the MLR evolution, the HRR evolu-
tion with time was considerably influenced by the external
heat flux. Significant differences of the HRR curves at
different regions, as defined in ‘‘Mass loss’’ section, can be
Characterization of thermal decomposition behavior of commercial flame-retardant ethylene… 455

(a) (b)
0.005 0.007

20 kW m–2 0.006 40 kW m–2

0.004
25 kW m–2 45 kW m–2
0.005
MLR/kg s –1 m–2

MLR/kg s–1 m–2

0.003 30 kW m–2
0.004
35 kW m–2
0.002 0.003

0.002
0.001
0.001

0.000 0.000
0 200 400 600 800 1000 1200 1400 0 200 400 600 800 1000 1200 1400
Time/s Time/s

(c) (d) 0.008

0.008
50 kW m–2 First peaks 30 kW m–2
0.007
–2
55 kW m 0.006 40 kW m–2
0.006
MLR /kg s–1 m–2

–2

MLR/kg s –1 m–2
60 kW m 60 kW m–2
0.005
65 kW m–2 0.004
0.004 Local zeniths
0.003

0.002 0.002

0.001
Quasi -steady stages
0.000 0.000
0 200 400 600 800 1000 1200 1400 0 200 400 600 800 1000 1200 1400
Time/s Time/s

Fig. 6 MLR as a function of experimental time

0.0032 may be due to the effects of the fire retardants on the

Average MLR specimens, as discussed in the aforementioned section.
0.0028
Additionally, there were two peak HRRs for the cases of
the external heat fluxes [35 kW m-2, while only one peak
Average MLR/kg s–1 m–2

0.0024
0.0020 Region 1
0.0016
0.0012
0.0008
20
Fig. 7 Average MLR versus external heat flux
demonstrated in Fig. 8. The HRR curves at 30, 40 and
60 kW m-2 were chosen as the typical HRR curves in each
region for comparison, as shown in Fig. 8d. Evident tran-
sitory flaming stages in the HRR curves can be observed at
25, 30 and 35 kW m-2 rather than at higher external heat
fluxes such as 60 kW m-2, as illustrated in Fig. 8a, c. This
HRR was found at the external heat fluxes below or equal
to 35 kW m-2. This phenomenon was probably due to that
at lower external heat fluxes, the formation of insulting
char layer owing to the effectiveness of the fire retardants
Region 2 Region 3
to the specimens resisted the continuing decomposition and
combustion of the unexposed surface. The burning of the
specimens would consequently be extinguished without
sufficient flammable volatiles, indicating that the heat from
30 40 50 60 70
the flame and the cone heater was not sufficient enough to
External heat flux/kW m–2
sustain the continuing burning, which led to shorter burn-
ing durations. Therefore, the specimen behaved as ther-
mally thick materials which are defined as the materials in
which heterogeneous temperature profile occurs when ex-
posed to external heat [45] during the whole thermal de-
composition process. Nevertheless, at higher external heat
fluxes, the effectiveness of the fire retardants may be
gradually lost. The insulating char layer cracked, and then
the unexposed material was heated up and burned subse-
quently, at which the specimens behaved from thermally

123
456 R. Chen et al.

(a) (b)
70 100

60
25 kW m–2 40 kW m–2
80
30 kW m–2 45 kW m–2
50
35 kW m–2
HRR/kW m–2

HRR/kW m–2
60
40

30
40

20 Transitory Transitory
flaming 20
flaming
10

0 0
0 100 200 300 400 500 600 700 800 900 1000 0 200 400 600 800 1000 1200
Time/s Time/s

(c)120 (d)
First peaks
100 30 kW m–2
100 50 kW m–2
40 kW m–2
55 kW m–2 80
80 60 kW m–2
60 kW m–2

HRR/kW m –2
HRR/kW m–2

60 Second peaks
p
60 65 kW m –2

40
40

20 20

0 0
0 200 400 600 800 1000 1200 1400 0 200 400 600 800 1000
Time/s Time/s

Fig. 8 HRR versus experimental time

thick materials to thermally thin materials where tem-
perature distributed approximately homogeneously [45],
resulting in remarkably intense combustion, which could
also explain the phenomena that more intense burning was
observed at the initial stage immediately after ignition and
the late stage of the test at high external heat fluxes, as
described in ‘‘Visual observation’’ section. Note that the
two peak HRRs both increased with an increase in external
heat flux, and the time for the HRR to reach its peaks
declined with the rise of the external heat flux. Moreover,
the difference of the first and second peak HRR decreased
with the increase in the external heat flux. The duration
from the first peak HRR to the second peak HRR was also
shortened with the increase in external heat flux. This was
due to the earlier cracking of the insulating char layer and
more intense combustion.
Effective heat of combustion
Typical MLR, HRR and EHC versus experimental time at
external heat fluxes of 30, 40 and 60 kW m-2 are presented
in Fig. 9. EHC represents the realistic heat release from the
combustion of per unit mass of solid materials and it is a
time and applied external heat flux-dependent parameter
under present study. It can be computed employing the
following equation [33]:
HRR
EHC ¼ : ð2Þ
MLR
As can be observed from Fig. 9d, the variations in EHC
were quite different among the cases of 30, 40 and
60 kW m-2. In the case of 30 kW m-2, a sharp increase in
EHC was found after ignition, and then EHC attained its
maximum, followed by a gradual decrease, after which a
deep decline in EHC occurred due to the extinction of the
flame. However, in the cases of 40 and 60 kW m-2, an-
other significantly higher local zenith of EHC appeared,
following a second progressive increase in EHC. It should
be noted that the second remarkable rise in EHC occurred
with almost constant MLRs, as demonstrated in Fig. 9b, c,
indicating that more effective combustion of specimens
occurred. The occurrence of the second progressive in-
crease and considerably higher peak of EHC was likely due
to that the remained unburned specimen was heated up and
behaved in the manner of thermally thin solid materials
after the char layer above cracked, resulting in relatively

123
Characterization of thermal decomposition behavior of commercial flame-retardant ethylene… 457

(a) 0.004
(b)80
0.005
60 30 kW m –2 40 kW m–2
0.003 60 HRR 0.004
HRR

MLR/kg s–1 m–2

MLR/kg s –1 m–2
HRR/kW m–2
MLR
HRR/kW m–2

40 MLR
0.003
0.002 40

0.002
20
0.001 20
0.001

0 0.000 0 0.000
0 100 200 300 400 500 600 700 800 900 0 200 400 600 800 1000 1200
Time/s Time/s

(c)120 0.008 (d)

32

28 30 kW m–2
100 60 kW m –2
0.006 40 kW m–2
HRR 24

EHC/MJ kg –1
80 60 kW m–2

MLR/kg s–1 m–2

MLR 20
HRR/kW m –2

60 0.004 16

12
Second peaks
40
0.002 8
20 4
First peaks

0 0.000 0
0 200 400 600 800 1000 1200 0 100 200 300 400 500 600 700 800 900

Time/s Time/s

Fig. 9 Typical MLR, HRR and EHC versus experimental time

higher decomposition rate to sustain almost constant MLRs be presented using the MLR and HRR curves at
rather than decrease in MLRs. What is more, burning of 30 kW m-2, as shown in Fig. 10a.
higher concentration of pure EPDM rubber volatiles, 2. Region 2 (35 kW m-2 \ external heat flux
-2
probably owing to the loss of the effectiveness of fire re- B 45 kW m ) Gradual loss of effectiveness of fire
tardants in the remained unburned specimen at this stage, retardants may occur to the specimens. Six typical
may be also responsible for it. thermal decomposition stages of the commercial
flame-retardant EPDM rubber can be presented using
the MLR and HRR curves at 40 kW m-2, as shown in
Summary of the thermal decomposition behavior Fig. 10b.
of the commercial flame-retardant EPDM rubber 3. Region 3 (external heat flux [45 kW m-2) Little effect
employing cone calorimeter experimental results of fire retardants on the specimens may occur in this
region. Six typical thermal decomposition stages of the
All of the parameters, including the visual observation, the commercial flame-retardant EPDM rubber can be
ignition time, the mass loss, the MLR, the HRR and the presented using the MLR and HRR curves at
EHC, showed that the thermal decomposition behavior of 60 kW m-2, as shown in Fig. 10c.
the commercial flame-retardant EPDM rubber was greatly The stages shown in Fig. 10 are interpreted as follows:
dependent on the applied external heat flux which may
considerably influence the effectiveness of the fire retar- • Stage I corresponding to the start of the thermal
dants to the specimens. According to the cone calorimeter decomposition with the beginning of ignition. During
experimental data, three distinct regions may be divided this stage, swelling, bubbling and cracking phenomena
according to the applied external heat flux as follows: at the specimen surface occurred. Fire retardants such
as magnesium hydroxide began to decompose and work
1. Region 1 (external heat flux B35 kW m-2) Effective- on the specimens especially under lower external heat
ness of fire retardants may be well exerted to the fluxes. MLR increased rapidly at the outset, followed
specimens. Four typical thermal decomposition stages by a transitory flaming phase (especially in region 1),
of the commercial flame-retardant EPDM rubber can and then specimen was ignited.

123
458 R. Chen et al.

(a) (b)
80 0.004 80 0.006

30 kW m –2 40 kW m–2
60 60
0.003 0.004
HRR
HRR

MLR/kg s –1 m–2

MLR/kg s–1 m–2

HRR/kW m–2
40 MLR
HRR/kW m–2

40 MLR 0.002 0.002

20
20
0.001 0.000
0
0
0.000 –20 –0.002
I II III VI I II III IV V VI
–20
0 100 200 300 400 500 600 700 800 900 1000 0 200 400 600 800 1000 1200
Time/s Time/s

(c)
120 0.008

100
60 kW m –2 0.006

MLR/kg s–1 m–2

80 HRR
0.004
60 MLR
HRR/kW m –2

0.002
40

20 0.000

0
–0.002
–20
–0.004
–40 I II III IV V VI
0 200 400 600 800 1000 1200
Time/s

Fig. 10 MLR and HRR versus experimental time at 30, 40 and 60 kW m-2

• Stage II corresponding to the accelerated thermal of thermally thin solid materials. MLR consequently
decomposition of the commercial flame-retardant kept almost constant instead of decrease. HRR in-
EPDM rubber after ignition. During this stage, EPDM creased gradually, and the second peak HRR increased
rubber with fire retardants was increasingly decom- with the applied external heat flux, even could match
posed and burned, causing continuous increase in MLR the first peak HRR at high external heat fluxes. A
along with a steep rise in HRR. The first peaks of MLR significant increase in EHC undoubtedly occurred.
and HRR were attained at this stage. The resulting EHC Moreover, a remarkably higher peak EHC was obtained
also underwent a sharp increase and then reached its than that in stage II.
zenith. • Stage V corresponding to the second deceleration of
• Stage III corresponding to subdued burning with the burning with a small quantity of char production.
massive char formation. During this stage, the HRR During this stage, due to the almost complete combus-
declined with a decrease in MLR, possibly due to the tion of specimen, little char would form. The decrease
resistance from the successive increase in the amount of in MLR, HRR and EHC was likely resulted by the
char produced from combustion. The EHC maintained decline in the amount of EPDM rubber and other
a quasi-steady phase, following a slight decrease after organics. Flame was extinguished at this stage.
the peak moment. Effectiveness of fire retardants to the • Stage VI corresponding to the non-flaming oxidation of
EPDM rubber would be possibly lost at the end of this the remained char layer. MLR would be maintained at
stage. Up to stage III, the specimen behaved in the an almost constant level.
manner of thermally thick materials.
• Stage IV corresponding to the further thermal decom-
position of the commercial flame-retardant EPDM TG–DSC test
rubber. During this stage, the effectiveness of fire
retardants may be almost completely lost. The gener- TG coupled with DSC has been widely used to investigate
ated char layer cracked. The remained unexposed the thermal decomposition process of polymers such as the
specimen was heated up and behaved in the manner work of Wang and Zhang [46] and Chen et al. [47]. In the

123
Characterization of thermal decomposition behavior of commercial flame-retardant ethylene… 459

(a) 0.2 (b)

400.31 K Heat flow 200
100 Exo up 150 100 Exo up 0.8
–15.38 % –9.24 %
150
Mass fraction Mass –10.09 %

Heat Flow/mW
–5.79 %
Mass fraction/%
0.1

Heat Flow/mW
80

Mass fraction/%
MLR/ % s –1
fraction 100
80

MLR/ % s –1
742.07 K 0.4
–20.93 %
0 –24.52 % 50
60 630 K
–5.91 % 60
0.0 0
–4.07 % –1.33 %
0.0
40 MLR –3.34 %
Heat flow MLR –50
–150 40
(1) (2) (3) (4) (5) –100
20 –0.1 (1) (2)(3) (4) (5)
–0.4
400 600 800 1000 1200 0 200 400 600 800 1000
Temperature/K Time/s

Fig. 11 TG–DSC curves in air. a 10 K min-1 and b 773 K

present study, TG–DSC tests were performed in air in order effectiveness of the fire retardants in the specimens. The
to further understand the thermal decomposition process of fourth stage corresponded to the crack of the char layer and
the commercial flame-retardant EPDM rubber. Typical thus the oxidation decomposition of the remained virgin
TG–DSC curves under the non-isothermal condition of the specimen. It is noteworthy that the heat flow with lower
heating rate of 10 K min-1 and the isothermal condition of MLR at the fourth stage was much higher than that at the
temperature of 773 K were selected and illustrated in third stage, probably indicating that the effectiveness of fire
Fig. 11. The left ordinate (labeled as ‘‘mass fraction’’ and retardants to the specimen was almost lost at this stage.
marked in black) represents the ratio of the transient mass Further decomposition of the residue occurred at the last
and the initial mass of the specimens versus tem- stage. Similar five thermal decomposition stages are shown
perature/time. The right ordinates (labeled as ‘‘MLR’’ and in Fig. 11b. Furthermore, in comparison with the cases in
marked in blue) denote the rate of variation of the mass Fig. 11a, more prominent phenomenon that higher heat
fraction versus temperature/time, and the right ordinate flow with lower MLR was indicated at the fourth stage than
(labeled as ‘‘heat flow’’ and marked in red) denotes the the counterparts at the third stage in Fig. 11b. The opinion
HRR from the oxidation decomposition of the specimens that the effectiveness of the fire retardants to the specimen
versus temperature/time. may be almost lost at the fourth stage could be tenable.
It is noted in Fig. 11a that approximately five stages Based upon the aforementioned discussion, the thermal
constituted the thermal decomposition process of the decomposition process of the commercial flame-retardant
commercial flame-retardant EPDM rubber. The specimen EPDM rubber obtained employing the TG-DSC instrument
started to decompose at around 400.31 K. The MLR con- was quite similar to that applying the cone calorimeter. It
tinuously increased up to 630 K, followed by a rapid de- was further validated that the loss of the effectiveness of
cline. The occurrence of the peak MLR may be mainly fire retardants to the specimens may occur at the high heat
attributed to the endothermic decomposition of aluminum fluxes or temperatures.
hydroxide, whose decomposition temperature is between
508 and 623 K [48]. Note that the heat flow at the first
stage remained almost zero, this was likely due to the Conclusions
oxidation decomposition of the EPDM rubber and other
organics, which could release heat to compensate the heat The present study has dealt with the thermal decomposition
required for the decomposition of aluminum hydroxide. behavior of the commercial flame-retardant EPDM rubber
Magnesium hydroxide with the decomposition temperature widely used in the metro train carriages in China ex-
between 613 and 763 K [48] underwent thermal degrada- perimentally. Effects of the external heat flux on the ther-
tion at the second stage similar to that of aluminum hy- mal decomposition process of the commercial flame-
droxide. At the third stage, a significant rise in MLR was retardant EPDM rubber were characterized employing
indicated up to 742.07 K, at which the maximum MLR cone calorimeter under piloted ignition. The thermal de-
(0.1 % s-1) was attained. After 742.07 K, the MLR de- composition process of the commercial flame-retardant
creased steeply possibly due to the resistance from the EPDM rubber was further analyzed using TG-DSC in-
generated char layer from the oxidation of the EPDM strument. From the obtained results, the major conclusions
rubber and other organics in the specimen owing to the are summarized as follows:

123
460
1. All the parameters, including the visual observation,
the ignition time, the mass loss, the MLR, the HRR and
the EHC, showed strong dependence upon the applied
external heat flux using cone calorimeter. Two peaks in
the MLR, the HRR and the EHC curves were
demonstrated under high external heat fluxes, and the
second peak of the EHC was even much higher than
the first one (see ‘‘Summary of the thermal decompo-
sition behavior of the commercial flame-retardant
EPDM rubber employing cone calorimeter experimen-
tal results’’ section).
2. It was found that three different decomposition regions
of the commercial flame-retardant EPDM rubber may
be identified according to the applied external heat flux
employing cone calorimeter:
• Region 1 (external heat flux B35 kW m-2) corre-
sponding to well exertion of effectiveness of fire
retardants to the EPDM rubber.
• Region 2 (35 kW m-2 \ external heat
flux B 45 kW m-2) corresponding to gradual loss
of effectiveness of fire retardants to the EPDM
rubber.
• Region 3 (external heat flux [45 kW m-2) corre-
sponding to little effectiveness of fire retardants to
the EPDM rubber.
The phenomena that the effectiveness of fire retardants
to the EPDM rubber under high external heat fluxes or
temperatures may be lost were further validated by the
experimental results from the TG-DSC tests.
3. Six thermal decomposition stages of the commercial
flame-retardant EPDM rubber employing cone
calorimeter were noted in both of region 2 and 3:
initial decomposition before ignition (Stage I), accel-
erated decomposition after ignition (Stage II), subdued
decomposition with the massive char formation (Stage
III), further decomposition with the cracking of the
char layer (Stage IV), the second subdued decompo-
sition with a small quantity of char production (Stage
V) and the non-flaming oxidation (Stage VI). Howev-
er, only four thermal decomposition stages (Stages IV
and V disappeared) of the commercial flame-retardant
EPDM rubber employing cone calorimeter constituted
the thermal decomposition process in region 1 (see
‘‘Summary of the thermal decomposition behavior of
the commercial flame-retardant EPDM rubber employ-
ing cone calorimeter experimental results’’ section).
Similar thermal decomposition stages of the commercial
flame-retardant EPDM rubber were noted in the TG-DSC
experiments, thus verifying the thermal decomposition
behaviors of the commercial flame-retardant EPDM rubber
based upon the cone calorimeter experimental results.
123
R. Chen et al.
It should be noted that the current study merely inves-
tigates the thermal decomposition behaviors of one type of
the floor coverings used in the metro train carriages in
China qualitatively. More comprehensive and quantitative
study will be addressed step by step in the future.
Acknowledgements This work was sponsored by the Research
Fund for the Doctoral Program of Higher Education of China (Grant
Nos. 20123402110048 and 20123402120018) and National Natural
Science Foundation of China (Grant Nos. 51206157 and 51323010).
The authors thank associate Prof. Lei Song and Dr. Qilong Tai from
University of Science and Technology of China and research assistant
Wei Tang from University of Maryland for their help with the
discussion.
References
1. Peacock RD, Braun E. Fire safety of passenger trains, phase I:
material evaluation (cone calorimeter). NISTIR 6132.
Gaithersburg: National Institute of Standards and Technology;
1999.
2. NF F 31-812. Railway rolling stock, rubber floor coverings. Paris:
Association Francaise de Normalisation; 1990.
3. Duggan G. Usage of ISO 5660 data in UK railway standards and
fire safety cases. In: A one-day conference on fire hazards, test-
ing, materials and products. Shrewsbury: Rapra Technology;
1997. p. 1–8.
4. Peacock RD, Bukowski RW, Markos SH. Evaluation of passen-
ger train car materials in the cone calorimeter. Fire Mater.
1999;23(2):53–62.
5. Peacock RD, Reneke PA, Averill JD, Bukowski R, Klote JH. Fire
safety of passenger trains, phase II: application of fire hazard
analysis techniques. NISTIR 6525. Gaithersburg: National Insti-
tute of Standards and Technology; 2001.
6. Hong W. The progress and controlling situation of Daegu
Subway fire disaster. In: Lee SK, editor. Sixth Asia–Oceania
symposium on fire science and technology. Daegu, Korea:
International Association for Fire Safety Science; 2004.
p. 17–20.
7. Chiam BH. Numerical simulation of a metro train fire. Master
dissertation, University of Canterbury, New Zealand. 2005.
8. Dowling V, White N, Webb A, Barnett J. When a passenger train
burns, how big is the fire? Invited lecture. In: Chow WK, editor.
Proceedings of the 7th Asia–Oceania symposium on fire science
and technology. Hong Kong, China: International Association for
Fire Safety Science; 2007. p. 19–28.
9. TB/T 3237. Flame retardant technical specification of decorating
materials for multiple unit train (in Chinese). Beijing: China
Ministry of Railways; 2010.
10. White N. Fire development in passenger trains. Master disserta-
tion, Victoria University, Australia. 2010.
11. Schebel K, Meacham BJ, Dembsey NA, Johann M, Tubbs J,
Alston J. Fire growth simulation in passenger rail vehicles using a
simplified flame spread model for integration with CFD analysis.
J Fire Prot Eng. 2012;22(3):197–225.
12. Lee D, Park W, Jung W, Yang S, Kim H, Hadjishophocleous G,
et al. Estimations of heat release rate curve of railcar fire. J Mech
Sci Technol. 2013;27(6):1665–70.
13. Li Y, Ingason H, Lönnermark A. Fire development in different
scales of a train carriages. In: Hees P van, editor. 11th Interna-
tional symposium on fire safety science. Christchurch: Interna-
tional Association for Fire Safety Science; 2014.
Characterization of thermal decomposition behavior of commercial flame-retardant ethylene…
14. NFPA-130. Standard for fixed guideway transit and passenger rail
systems. Quincy: National Fire Protection Association; 2010.
15. Roh JS, Ryou HS, Park WH, Jang YJ. CFD simulation and
assessment of life safety in a subway train fire. Tunn Undergr Sp
Technol. 2009;24(4):447–53.
16. Cheng M, Chen W. Experimental investigation of the stress
stretch behavior of EPDM rubber with loading rate effects. Int J
Solids Struct. 2003;40(18):4749–68.
17. Babrauskas V. Ignition handbook: principles and applications to
fire safety engineering, fire investigation, risk management and
forensic science. 2nd ed. Issaquah: Fire Science; 2003.
18. Lyon RE, Quintiere JG. Criteria for piloted ignition of com-
bustible solids. Combust Flame. 2007;151(4):551–9.
19. Song L, Zhou S, Wu J, Hu Y. Synergistic effects of lanthanum
oxide on magnesium hydroxide flame-retarded ethylene propy-
lene diene terpolymer composite. Polym Plast Technol.
2009;48(10):1088–93.
20. Shen ZQ, Chen L, Lin L, Deng CL, Zhao J, Wang YZ. Syner-
gistic effect of layered nanofillers in intumescent flame-retardant
EPDM: montmorillonite versus layered double hydroxides. Ind
Eng Chem Res. 2013;52(25):8454–63.
21. Yen YY, Wang HT, Guo WJ. Synergistic effect of aluminum
hydroxide and nanoclay on flame retardancy and mechanical
properties of EPDM composites. J Appl Polym Sci. 2013;
130(3):2042–8.
22. ISO 5660-1. Reaction-to-fire tests-heat release, smoke production
and mass loss rate-part 1: heat release rate (cone calorimeter
method). 2nd ed. Geneva: International Organization for Stan-
dardization; 2002.
23. Tsai KC. Orientation effect on cone calorimeter test results to
assess fire hazard of materials. J Hazard Mater. 2009;172(2):
763–72.
24. Kim J, Lee J, Kim S. Estimating the fire behavior of wood
flooring using a cone calorimeter. J Therm Anal Calorim.
2012;110(2):677–83.
25. Xu Q, Majlingova A, Zachar M, Jin C, Jiang Y. Correlation
analysis of cone calorimetry test data assessment of the procedure
with tests of different polymers. J Therm Anal Calorim.
2012;110(1):65–70.
26. Ma Y, Wang J, Xu Y, Wang C, Chu F. Preparation and charac-
terization of phenolic foams with eco-friendly halogen-free flame
retardant. J Therm Anal Calorim. 2013;114(3):1143–51.
27. Qu H, Liu C, Wu W, Chen L, Xu J. Using cone calorimeter to
study thermal degradation of flexible PVC filled with zinc ferrite
and Mg (OH)2. J Therm Anal Calorim. 2014;115(2):1081–7.
28. An W, Jiang L, Sun J, Liew K. Correlation analysis of sample
thickness, heat flux, and cone calorimetry test data of polystyrene
foam. J Therm Anal Calorim. 2014:1–10. doi:10.1007/s10973-
014-4165-9.
29. Walton WD, Thomas PH. Estimating temperatures in compart-
ment fires. In: DiNenno PJ, editor. SFPE handbook of fire pro-
tection engineering. 3rd ed. Quincy: National Fire Protection
Association; 2002. p. 3.171–88.
461
30. Schartel B, Hull TR. Development of fire-retarded materials-Inter-
pretation of cone calorimeter data. Fire Mater. 2007;31(5):
327–54.
31. Gao L, Zheng G, Zhou Y, Hu L, Feng G. Thermal performances
and fire behaviors of rosin-based rigid polyurethane foam
nanocomposites. J Therm Anal Calorim. 2014:1–14. doi:10.1007/
s10973-014-4192-6.
32. Xu Q, Jin C, Jiang Y. Analysis of the relationship between MCC
and thermal analysis results in evaluating flammability of EPS
foam. J Therm Anal Calorim. 2014;118:687–93.
33. Luche J, Rogaume T, Richard F, Guillaume E. Characterization
of thermal properties and analysis of combustion behavior of
PMMA in a cone calorimeter. Fire Saf J. 2011;46(7):451–61.
34. Luche J, Mathis E, Rogaume T, Richard F, Guillaume E. High-
density polyethylene thermal degradation and gaseous compound
evolution in a cone calorimeter. Fire Saf J. 2012;54:24–35.
35. Quang Dao D, Luche J, Richard F, Rogaume T, Bourhy-Weber
C, Ruban S. Determination of characteristic parameters for the
thermal decomposition of epoxy resin/carbon fibre composites in
cone calorimeter. Int J Hydrog Energy. 2013;38(19):8167–78.
36. Batiot B, Luche J, Rogaume T. Thermal and chemical analysis of
flammability and combustibility of fir wood in cone calorimeter
coupled to FTIR apparatus. Fire Mater. 2014;38(3):418–31.
37. Babrauskas V. The cone calorimeter. In: DiNenno PJ, editor.
SFPE handbook of fire protection engineering. 3rd ed. Quincy:
National Fire Protection Association; 2002. p. 3.63–81.
38. ISO 17554. Reaction to fire-mass loss measurement. Geneva:
International Organization for Standardization; 1998.
39. ISO 13943. Fire safety-vocabulary. 2nd ed. Geneva: International
Organization for Standardization; 2000.
40. Karlsson B, Quintiere JG. Enclosure fire dynamics. New York:
CRC Press; 2000.
41. Chen R, Lu S, Zhang B, Li C, Lo S. Correlation of rate of gas
temperature rise with mass loss rate in a ceiling vented com-
partment. Chin Sci Bull. 2014;59(33):4559–67.
42. Babrauskas V, Peacock RD. Heat release rate: the single most
important variable in fire hazard. Fire Saf J. 1992;18(3):255–72.
43. Thornton W. The relation of oxygen to the heat of combustion of
organic compounds. Lond Edinb Dublin Philos Mag J Sci.
1917;33(194):196–203.
44. Janssens ML. Measuring rate of heat release by oxygen con-
sumption. Fire Technol. 1991;27(3):234–49.
45. Mouritz AP, Gibson A. Fire properties of polymer composite
materials. Dordrecht: Springer; 2007.
46. Wang Y, Zhang J. Thermal stabilities of drops of burning ther-
moplastics under the UL 94 vertical test conditions. J Hazard
Mater. 2013;246:103–9.
47. Chen X, Jiang Y, Jiao C. Smoke suppression properties of ferrite
yellow on flame retardant thermoplastic polyurethane based on
ammonium polyphosphate. J Hazard Mater. 2014;266:114–21.
48. Hu Y, Lei S, You F, Zhong MH. Introduction to fire dynamics (in
Chinese). Beijing: Chemical Industry Press; 2007.
123