Beruflich Dokumente
Kultur Dokumente
Thermochimica Acta
journal homepage: www.elsevier.com/locate/tca
Review
a r t i c l e i n f o a b s t r a c t
Article history: A review on the experimental studies of the reactivity of ethylene oxide (EO) with various contaminants
Received 4 September 2016 including water, different kinds of iron oxides and alkalis are presented. It is shown that: (1) Pure EO is
Received in revised form 7 March 2017 not prone to initiate a runaway reaction under 200 ◦ C under adiabatic conditions. (2) The possibility of
Accepted 9 March 2017
EO aqueous solution to experience an exothermic reaction depends strongly on the EO concentration,
Available online 18 March 2017
and the onset temperatures could be predicted well by a fourth order reaction rate model. (3) EO exhibits
different sensitivities to various kinds of iron oxides. Although ␥-Fe2 O3 is the most active iron oxide to
Keywords:
EO, its runaway reaction consequence is less severe. The reactivity for EO with iron oxides could be: ␥-
Ethylene oxide
Thermal stability
Fe2 O3 > ␣-Fe3 O4 > hydrated ␣-Fe2 O3 > ␣-Fe2 O3 . (4) Alkalis could reduce the initial reaction temperature of
Calorimetry EO to room temperature with sufficient concentrations, and the proposed hazard potentiality for alkalis
Reactivity is NH4 OH > KOH > NaOH.
Contaminant © 2017 Elsevier B.V. All rights reserved.
Contents
1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 86
2. Basic properties of EO . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 87
2.1. Physical properties of EO . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 87
2.2. Chemical properties of EO . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 87
3. The thermal reactivity of pure EO . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 88
4. The thermal reactivity of EO with water . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 89
4.1. Onset temperatures of EO aqueous solution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 89
4.2. Kinetics for EO reaction with water . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 90
5. The thermal reactivity of EO with iron oxides . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 91
5.1. Reaction of EO with glass beads or ␣-FeO . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 91
5.2. Reaction of EO with iron oxides . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 91
5.2.1. Temperature for EO reaction with iron oxides . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 91
5.2.2. Self-heat rate for EO reaction with iron oxides . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 91
5.2.3. Pressure for EO reaction with iron oxides. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .91
6. The thermal reactivity of EO with alkalis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 92
6.1. Reaction of EO with KOH . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 93
6.2. Reaction of EO with NaOH . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 93
6.3. Reaction of EO with NH4 OH . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 94
6.4. Onset temperature comparison . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 94
7. Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 94
Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 95
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 96
∗ Corresponding authors.
E-mail addresses: liujr.qday@sinopec.com (J. Liu), msj.qday@sinopec.com
(S. Mu).
http://dx.doi.org/10.1016/j.tca.2017.03.008
0040-6031/© 2017 Elsevier B.V. All rights reserved.
86 J. Liu et al. / Thermochimica Acta 652 (2017) 85–96
lyst at about 200 ◦ C and 2.0 MPa. The process parameters of direct
Nomenclature oxidation technology has been summarized in Table 1, depending
on catalysts performance and production technology. The recycle
Notation
gas containing EO and byproduct such as carbon dioxide, water
A Pre-exponential factor, [(m3 )3 kmol−3 s−1 ]
and acetaldehyde is then transported into scrubbers to absorb EO
Cb Specific heat of the bomb, [J (kg ◦ C)−1 ]
and to remove carbon dioxide. The residual recycle gas is reloaded
Cs Specific heat of the sample, [J (kg ◦ C)−1 ]
with oxygen and ethylene through oxygen mixer and returned to
dT/dtmax Maximum temperature rise rate, [◦ C min−1 ]
the reactor. In the liquid phase, the resulting aqueous EO solution
dP/dtmax Maximum pressurization rate, [kPa min−1 ]
is passed to the EO purification unit to collect pure EO product
E Activation energy, [J/mol−1 ]
through a series of stripping towers and distillation towers, where
H Heat of polymerization, [kJ mol−1 ]
low-boiling components are stripped of at the same time. Depend-
k Rate constant, [(m3 )2 kmol−2 s−1 ]
ing on the design of the EO plant, the EO product can be refrigerated
K Equilibrium constant
and transported to the storage tank or subsequently reacted with
Mb Mass of testing cell, [g]
water to produce ethylene glycol.
MEO Mass of EO, [g]
The process hazards of EO production lie mainly in the oxi-
MS Mass sample, [g]
dation reaction and EO purification process [13] (Fig. 1). In the
Mtotal Total mass of sample and cell, [g]
oxidation reaction unit, a significant amount of flammable gases
Pmax Maximum pressure, [kPa]
including ethylene and methane mixes and reacts with oxygen. The
r1 Rate constant of EO with water, [kmol m−3 s−1 ]
flammable gases concentration has to pass through the flammable
r2 Rate constant of EO with glycols, [kmol m−3 s−1 ]
region when diluted down by oxygen to the operating level, requir-
rEG The formation rate of EG, [kmol m−3 s−1 ]
ing a fairly complicated control scheme to assure safety therefore.
rDEG The formation rate of DEG, [kmol m−3 s−1 ]
The oxidation reaction of ethylene is extremely exothermic espe-
t Time, [s]
cially when the complete oxidation of EO to carbon dioxide takes
Tf Final temperature, [◦ C]
place. As a result, the heat removal system must be sufficiently reli-
Tmax Maximum temperature, [◦ C]
able to cope with the heat release in the reactor, otherwise runaway
Tmp Temperature at maximum pressure, [◦ C]
reactions may be induced and the accumulated heat could result in
TMR Time to maximum reaction rate, [min]
damage to assets, human health or the environment [14]. Typical
Ton Onset temperature, [◦ C]
incidents in the oxidation reaction are pre-ignition, temperature
W Mass fraction of EO aqueous solution, [%]
runaway and explosion of flammable cyclic gases. In the EO purifi-
cation unit, high purity EO is combustible and sensitive to various
Greek letters
impurities, prone to cause fires and deflagrations [15–17]. The con-
Tc Adiabatic temperature rise, [◦ C]
sequence of incidents in the EO purification unit is often much more
Thermal inertia factor
severe than that in other units in the EO production process because
of the strong exothermic effect. The disastrous explosion was usu-
ally initiated by the exothermic reaction of EO with inobservant
1. Introduction impurities. Looking through the history of EO industry, it is not dif-
ficult to name a few incidents caused by the existing impurities in
Ethylene oxide (EO) is one of the most versatile chemical inter- the EO purification unit.
mediates because its highly strained ring can be opened easily On July 3rd, 1987, a purification distillation column of BP Chem-
to produce varieties of fine chemicals [1,2]. Besides its direct use icals’ EO plant in Antwerp exploded. It was concluded after accident
as disinfectant, sterilizing agent, and fumigant, EO is frequently investigation that EO leaked from the manhole flange and was
applied in the production of many important surfactants, sol- trapped inside the insulation where it started to react in the pres-
vents, polyesters, and antifreezes such as ethylene glycol (EG), ence of rust, water and the insulation. The temperature gradually
ethanolamine, ethylene glycol ethers etc. Important as EO is in increased because of heat released in the exothermic reactions until
the chemical industry, it is an extremely hazardous material due it reached a critical level, where there was an acceleration in the
to the high reaction sensitivity—it is susceptible to contaminants rate of the exothermic reactions, resulting in a rapid release of heat.
such as alkalis, iron oxides and water, which could lead to runaway The temperature of the resultant hot metal surface was sufficient to
reactions or even spontaneous decomposition [3]. ignite the EO vapor internally [18]. An EO plant of BASF licensed by
Traditionally, EO is produced by the direct oxidation of ethy- Shell in Antwerp also experienced severe fire and explosion damage
lene with oxygen by metallic silver [4,5]. Due to the deflagration in the EO distillation section on March 7th, 1989. It is believed that
and explosion risk of the process, some other inherent safer pro- polyethylene glycols (PEG) were formed by escaped EO and accu-
cesses have been proposed to reduce risk, such as silver-catalyzed
oxidation of ethylene to EO in micro-reactor, or homogeneous liq-
uid phase oxidation with catalyst (methyltriox-orhenium), oxidant Table 1
(H2 O2 ) and reaction medium (methanol/water) [6,7]. Also, some Process parameters in the oxygen-based production of EO [11,12].
researchers conducted optimization of the reaction concept and Parameters Value range
design parameters of EO process in order to save energy and reduce
C2 H4 concentration, vol% 15–40
emission [8–10]. The industrial production technologies for EO O2 concentration, vol% 5–9
based on direct oxidation process over silver catalyst are mainly CO2 concentration, vol% 5–15
patented and licensed by Shell, Scientific Design (SD) and DOW C2 H6 concentration, vol% 0–2
Chemical. As shown in Fig. 1, the direct oxidation process consists CH4 concentration, vol% 1–60
Inlet gas temperature, ◦ C 30–40
of oxidation reaction unit, carbon dioxide removal unit, EO purifi- Reaction temperature, ◦ C 220–275
cation unit, EO storage/transportation unit and possible ethylene Pressure, MPa 1–2.2
glycol production unit. A recycle gas stream containing ethylene space velocity, h−1 2000–4000
and oxygen is first circulated through the fixed bed reactor, partially C2 H4 conversion, % 7–15
Selectivity, % 80–90
oxidizing ethylene into EO over sphere or ring shaped silver cata-
J. Liu et al. / Thermochimica Acta 652 (2017) 85–96 87
Fig. 1. Typical process flow diagram of EO process (green) and main process hazards (yellow): deflagration and explosion at oxygen mixer; pre-ignition/temperature runaway
at EO reactor; EO runaway reaction/decomposition at EO purification section. (For interpretation of the references to colour in this figure legend, the reader is referred to the
web version of this article.)
2. Basic properties of EO The reactions of EO proceed mainly via ring opening and are
highly exothermic. Under appropriate conditions, EO is known to
2.1. Physical properties of EO undergo a variety of reactions, namely isomerization, polymer-
ization, hydrolysis, combustion, and decomposition. Hydroxyethyl
The structural parameters of ethylene oxide determined by compounds could be produced when EO reacts with compounds
microwave spectroscopy are shown in Fig. 2 [22]. Ethylene oxide is containing a labile hydrogen atom, such as HOH, H2 NH, HRNH,
a colorless gas at room temperature and it is miscible with water, R2 NH, RCOOH, RCONH2 , HSH, RSH, ROH, etc. (R = alkyl, aryl).
alcohol, ether, and most organic solvents in all proportions. It is Hydrolysis is the most important hydroxyethylation reaction of EO
highly flammable and poses a dangerous fire and explosion risk—its since about 60% of the total EO produced is converted into ethy-
explosion limits ranging from 2.6 vol% to 100 vol% in air at 101.3 kPa. lene glycol [27,28]. The reaction of EO with water in the absence
The upper flammable limit (UFL) for EO is 100%, since pure EO of catalysts at room temperature is relatively slow. Depending
88 J. Liu et al. / Thermochimica Acta 652 (2017) 85–96
Fig. 3. Variation of temperature and pressure of pure EO in the APTAC [31]: (a) temperature history; (b) pressure history.
4. The thermal reactivity of EO with water using APTAC. These tests were conducted using three different cell
materials including titanium, Hastelloy-C, and glass. The detected
4.1. Onset temperatures of EO aqueous solution onset temperatures measured in glass were higher than those
measured in titanium or Hastelloy-C, especially for higher concen-
In the process of EO production, the yielded EO is absorbed by trations. The reactivity at the cell wall and the thermal inertia of
water, therefore it is necessary to understand the thermal proper- different cell material may affect the thermal parameters of the
ties of EO aqueous solution. EO could react with water and further reaction such as onset temperature [36]. The maximum exothermic
with ethylene glycols to form higher ethylene glycols, determined temperature detected was about 340 ◦ C for 60% mass fraction. EO-
by the concentration ratio between water and EO. The reaction of water solution at this composition has sufficient energy to reach the
EO with water can be catalyzed by acids and bases, and can be glycol decomposition temperature, approximately 300 ◦ C. The gly-
started alternatively from glycol in the absence of water. The EO- col decomposition reactions require special consideration because
water calorimetric tests were conducted by Melhem et al. over a they are very energetic and can produce large amounts of gas.
wide range of phi-factors and EO weight fractions from 10% to 99%
90 J. Liu et al. / Thermochimica Acta 652 (2017) 85–96
E
= 9525 K or E = 4512 J/gmol (5)
R
CROH was total molar concentration of independent hydroxyl
groups in solution and calculated as:
4.2. Kinetics for EO reaction with water If EG is being converted to DEG, either of the two hydroxyls on
the ethylene glycol can undergo the following reaction and reaction
Reaction orders of one, two and three could be served as candi- rate:
dates for EO and water reaction. These previous models performed
I + EG → DEG + 2ROH (13)
reasonably well in the low concentration range, but misestimated
the detected onset temperatures at high EO concentrations. The rDEG = 2k0 CI CEG (14)
overall fourth-order reaction rate expression was proposed by Mel-
hem et al. [32], which considers all species containing hydroxyl According to Eq. (10), the expression of CI could be substituted
groups and takes into account the impact of the reactions of the into Eqs. (12) and (14), which yields the reactions form described
ethylene glycols on the overall kinetics. herein:
According to Melhem et al. [32] EO reacts with water in the liquid 2
rEG = k0 KCEO CH2 O CROH (15)
phase to produce monoethylene glycol according to the following
2
chemical equation: rDEG = 2k0 KCEO CEG CROH (16)
C2 H4 O + H2 OC2 H6 O2 (1) It is convenient by defining k0 K as k’, and the final rate expression
is given by the following fourth order equations:
The rate of reaction is given by the following expression:
2
rEG = k CEO CH2 O CROH (17)
r1 = kCEO CH2 O C2ROH (2)
2
rDEG = 2k CEO CEG CROH (18)
where r1 is reaction rate, kmol/m3 /s, k is rate constant, and defined
as: The fourth-order reaction model accurately predicts the exper-
E
imental onset temperatures, especially at high EO concentration
k = Aexp − (3) compared to the third-order reaction model [32,37]. This model is
RT
useful for the prediction of EO-water contamination incident. The
where pre-exponential factor A and activation energy E were product of EO and water reaction are ethylene glycol, diethylene
determined by fitting the kinetics with obtained ARC/APTAC tests glycol, triethylene glycol and tetraethylene glycol, and the distri-
including the heat-wait-search steps: bution of the above products depends mainly on the initial relative
3 mass fraction of EO water [38]. Both EO and EG could decompose
A = 338 m3 > kmol−3 s−1 (4)
by exothermic, autocatalytic reactions to produce volatiles and
J. Liu et al. / Thermochimica Acta 652 (2017) 85–96 91
non-condensable gases when heated to high temperature approx- which is much greater than other iron oxides and the great pres-
imately 300 ◦ C, which leads to the overpressure or even damage to sure could result in equipment damage in practical plant. Also, the
the equipment [39,40]. However, the pressure variation during EO multiple peak behavior is apparent in pressure curves, particularly
and water reaction is not provided by the previous studies, which for ␣-Fe3 O4 and ␣-Fe2 O3 .
is important for the design of emergency relief systems containing
EO or its aqueous solutions. 5.2.2. Self-heat rate for EO reaction with iron oxides
The self-heat rate for EO with ␥-Fe2 O3 was roughly 1 ◦ C/min
5. The thermal reactivity of EO with iron oxides at room temperature and then accelerated to over 1000 ◦ C/min,
which is the highest in the three types of iron oxides [34] (Fig. 6). A
Iron oxides were believed to be a sensitive contaminant to decrease in the rate was seen when the temperature exceed 300 ◦ C,
induce runaway reactions of EO, which may exists in the pipelines, which may result from the diminishing of the reactant. The next
end sockets of flanges and tubes of heat exchangers. Iron oxides most active iron oxide material appears to be the ␣-Fe3 O4 fines,
have drawn special attention in the production of EO because rust and the self-heat rate exhibits relatively complex trends. A lower
was responsible for several EO distillation explosion incidents. self-heat rate with fluctuation for temperature ranging from room
The reactivity of EO with iron oxide fines was studied by Levin temperature to about 140 ◦ C was observed, and then the reaction
[41] using adiabatic calorimetry APTAC. Liquid EO was collected by seems to vanish shortly thereafter. Sustained reaction appeared
attaching the lecture bottle to a dry ice/acetone-cooled condenser again when the temperature was over 260 ◦ C, and the self-heat
and then was placed in a glove bag, while roughly 0.5 g of fines rate reached as high as 10,000 ◦ C/min in one case. For ␣-Fe2 O3
(or glass beads) were charged into a stainless-steel basket and also and hydrated ␣-Fe2 O3 , the self-heat rate generally increase from
placed into the glove bag. The basket was then fitted around the 0.1 ◦ C/min to 200 ◦ C/min with temperature, except that there is a
sample and wall thermocouples, inserted into the sample cell, and dropping off trend at 140–175 ◦ C for ␣-Fe2 O3 . The complicated vari-
the sample cell was then attached to the vessel head. So a variety of ation of self-heat rate indicates there might be multiple reactions.
commercially-available iron oxides were suspended in the stainless
steel basket above the EO. Finally the heat-wait-search (H-W-S) 5.2.3. Pressure for EO reaction with iron oxides
mode was operated to examine the reactivity of EO with solid fines. The pressure generated in the system was characterized and
compared with the pressure of EO vapor + nitrogen pad gas as
5.1. Reaction of EO with glass beads or ˛-FeO shown in the pressure versus temperature plots (Fig. 7). For EO
with ␣-Fe3 O4 /␣-Fe2 O3 /hydrated ␣-Fe2 O3 , the pressure of the sys-
The mild exothermic activity was observed for EO with empty tem follows the curve of EO vapor and nitrogen pad gas pressure
basket ranging from 60 ◦ C to 165 ◦ C, but it could not be asserted until 170 ◦ C, after which the pressure falls away from the projected
that the onset temperature of pure EO was about 60 ◦ C. Accord- pressure due to consumption of EO and the increasing difference
ing to the self-heat rate, the case of EO with empty basket shows between solids and gas temperature in the cell. For EO with ␣-
only marginal self-heating behavior, so the observed behavior for Fe2 O3 , sharp increases in pressurization rate and pressure were
the empty basket is not actually an exotherm but just experimental observed at about 380 ◦ C, indicating that the ignition of EO vapor
drift. The starting temperature for the reaction of pure EO was iden- might take place. During the cool-down period, the pressure for the
tified to be around 200 ◦ C as the result of Dinh et al. and Melhem system of EO with ␣-Fe3 O4 /␣-Fe2 O3 was below 140 kPa because
et al. [31,32] While for EO with glass beads or ␣-FeO, no discernable the titanium cell ruptured and the rupture disk burst. For EO
exothermic activity was detected and only the typical stair-step with ␥-Fe2 O3 , a deviation of system pressure from the projected
behavior of the heat-wait-search (H-W-S) mode was recorded. The vapor pressure + gas pad line was observed for temperature below
maximum search temperature was above 250 ◦ C in the case of the 30 ◦ C, then the pressure kept almost constant until 350 ◦ C. Dur-
␣-FeO but no exothermic behavior appeared. The pressure history ing the cool-down period, the pressure remained higher than the
show parallel trends with temperature. corresponding pressures during heat-up, suggesting generation of
non-condensable gases. The rapid exothermic reactivity of EO in
the presence of iron oxide could be illustrated by the time-to-
5.2. Reaction of EO with iron oxides
maximum rate. The time to maximum rate generally increase with
initial temperature for all the iron oxides. For EO with ␥-Fe2 O3 , it
5.2.1. Temperature for EO reaction with iron oxides
takes less than 10 min for the rate to reach its maximum from room
In contrast to glass beads and ␣-FeO, the mixture of EO with
temperature, while for ␣-Fe2 O3 /hydrated ␣-Fe2 O3 the time is about
various iron oxide including ␣-Fe3 O4 , ␣-Fe2 O3 , hydrated ␣-Fe2 O3
several hours. For EO with ␣-Fe3 O4 , the time-to-maximum rate also
and ␥-Fe2 O3 display evident exothermic behavior as shown in the
indicate the possibility of multiple reactions as demonstrated in the
temperature history in Fig. 5. For ␥-Fe2 O3 , the exothermic reaction
profiles of temperature, pressure and self-heat rate.
took place at room temperature and the maximum temperature
The calorimetric study suggest the following relative reactive
reached 600 ◦ C, so it is dangerous for EO to be contacted with ␥-
sensitivity of different iron oxides with EO:
Fe2 O3 which could induce the explosive decomposition of EO vapor.
For hydrated ␣-Fe2 O3 , the onset temperature is about 50 ◦ C and ␥-Fe2 O3 > ␣-Fe3 O4 > hydrated␣-Fe2 O3 > ␣-Fe2 O3 >> ␣-FeO ≈ glass
the maximum exothermic temperature is much lower than other
iron oxides. For ␣-Fe2 O3 , though the onset temperature increases EO in combination with all of the above iron oxide fines except
to about 100 ◦ C, the maximum temperature also exceeds 750 ◦ C. FeO displays exothermic activity as low as room temperature. The
The maximum temperature for ␣-Fe3 O4 was even higher, which peak temperature of the fines with EO in many cases ultimately
appears to have encountered a programmed instrument shutdown exceeded 600 ◦ C. The self-heat rates approach 2000 ◦ C/min for ␥-
at 900 ◦ C. The shapes of the curves have more than one peaks sug- Fe2 O3 fines while the rates are in excess of 100 ◦ C/min for the other
gesting occurrence of multiple reactions. The variation of pressure fines. The pressurization rates approach about 7 Mpa/min for ␣-
for EO with iron oxide is in accordance with temperature [34]. Fe3 O4 and ␣-Fe2 O3 . It should be noted that the rust in the practical
The maximum reaction pressure for EO with ␥-Fe2 O3 is less than device may contain not only one forms of iron oxides, and the sur-
1.4 Mpa though the peak temperature could reach 600 ◦ C. The run- face area of the iron oxide is also an important factor to determine
away reaction pressure for EO with ␣-Fe2 O3 was about 18 Mpa, the reactivity with EO.
92 J. Liu et al. / Thermochimica Acta 652 (2017) 85–96
6. The thermal reactivity of EO with alkalis many incidents in history in which the alkali presented as an
impurity initiated exothermic polymerization of EO, and the local
Alkali metal hydroxide such as potassium hydroxide or sodium temperature was increased to the decomposition temperature of
hydroxide can be used as a basic catalyst for many ethoxylation EO. The reactivity of EO with alkali contaminants including potas-
reactions. So alkalis have relatively high possibility to reduce the sium hydroxide (KOH), sodium hydroxide (NaOH), and ammonium
thermal stability of EO and cause runaway reactions. There were
J. Liu et al. / Thermochimica Acta 652 (2017) 85–96 93
hydroxide (NH4 OH) was systematically studied by Dinh et al. using 6.2. Reaction of EO with NaOH
the APTAC [31].
For EO with NaOH, the main parameters such as the onset
temperature, maximum temperature, maximum heat release rate
and maximum pressurization rate generally shown similar trends
6.1. Reaction of EO with KOH with that of KOH. However, the effect of NaOH concentration on
the onset temperature was not obvious. The self-heat rates for
For EO with KOH, the onset temperature decrease with the EO/NaOH was much higher than the ones observed for the KOH,
increase amount of KOH, and the exothermic activities appeared at especially for the 0.1 g NaOH case. For the 0.5 g and 1.0 g NaOH
around 30–60 ◦ C compared to 200 ◦ C for pure EO. While the maxi- tests, the cool-down pressure exceeds the heat up pressure indi-
mum temperature or pressure decrease inversely with the increase cating there might be non-condensable product, which is different
weight of KOH. The measured heat rates with respect to tempera- to the product of EO/KOH. So far, the products of EO with different
tures also indicated that the heat release of the sample starts earlier contaminants have not been determined and need to be further
for higher amount of KOH. Although the self-heat rate attained by studied.
each run was different, no significant difference was observed for The results of ARC analysis on the various EO-NaOH mixtures
the pressurization rates in the investigated range of contaminant are provided by Freeder et al. [42] as shown in Table 4. The onset
concentrations. As shown in Fig. 8 the pressure variation versus temperature decreases with the increase of NaOH concenrations
temperature of EO/KOH samples, the pressure of the mixture fol- and it is as low as room temperature for 1.0 M NaOH molarity. The
lows the vapor pressure of pure EO at heat up stage, and then adiabatic temperature rise produced by the polymerization of EO
become flat due to consumption of EO. During the cool-down stage in the presence of NaOH is calculated as follow:
the slopes were lower than the heating curve slopes indicating that
EO was converted into less volatile components, while this trend is Tc = ˚ · (Tf − Ton ) (19)
inverse to the above mentioned EO with iron oxide case. The time
where Ton is onset temperature, Tf is final temperature, and is
for EO with KOH to reach maximum reaction rate decrease as the
the thermal inertia factor defined as:
increase of initial temperature, and the amount of KOH is also favor-
able to reduce the time to maximum rate. For EO with 1.0 g KOH, Ms Cs + Mb Cb
˚= (20)
the time-to-maximum rate from ambient temperature is about 1 h. Ms Cs
Table 4
Experimental results of ARC analysis for EO/NaOH mixtures [42].
NaOH molarity [mol L−1 ] Onset Final Corrected adiabatic Heat of Maximum Temperature at
temperature Ton temperature Tf remperature rise Tc polymerization H pressure Pmax maximum
[◦ C] [◦ C] [◦ C] [kJ mol−1 ] [kPa] pressure Tmp [◦ C]
where Ms is the sample mass, Mb is the cell mass, Cs is the specific 6.4. Onset temperature comparison
heat of sample, and Cb is the specific heat of cell.
The adiabatic temperature rise appears to be independent of The onset temperature is an important parameter for evaluation
the concentration of alkali apart from the mixture containing 0.5 M of the contamination effects on EO thermal reactivity. The initial
NaOH. The average value of the polymerization heat calculated reaction temperature is much lower when EO is in contact with
from ARC data is about 42.8 kJ mol−1 . The maximum pressures alkalis (Fig. 10). The exothermic onset temperatures decrease with
are achieved at a temperature below the maximum temperature the increase of contamination mass. The effect of contamination
because the reduction in the average vapor pressure of the react- mass on the onset temperature of EO is more significant for NH4 OH
ing mixture as polymerization takes place. According to the plots of than for KOH and NaOH. Obviously, the onset temperature for EO
self-heat rate versus temperature, the initial parts of the plots are with NaOH is higher than that of EO with KOH in the contaminant
approximately linear and the initial gradients have been found to concentration range studied. The presence of NH4 OH with a suffi-
be independent of the concentration of alkali. This is in reasonable cient concentration has the highest impact to decrease the onset
agreement with the results of Dinh et al. [31]. temperature of contaminated EO, so the following relative reactiv-
ity of contamination is suggested: NH4 OH > KOH > NaOH. However,
the thermokinetics of the reactions of EO with KOH, NaOH, and
NH4 OH should be further determined.
tion. For EO with alkalis, three industrial contaminants, KOH, NaOH, It should be noted firstly that the thermo-kinetics parameters
and NH4 OH were investigated to find that NH4 OH with a suffi- obtained from different apparatus under specific operational con-
cient concentration has the highest impact on the thermokinetics ditions could vary a lot, which could be attributed to the thermal
parameters namely onset temperature, pressurization rate and inertia or sensitivity of the instruments. Since the thermal inertia of
maximum reaction pressure etc. The mass of alkalis has a pos- the laboratory instrument is higher than the plant-scale equipment,
itive effect to reduce the initial reaction temperature especially the measured data may underestimate the runway reaction haz-
for NH4 OH. The relative reactivity of alkali contaminations is sug- ards. Besides, it is not explicit what kind of reactions take place and
gested: NH4 OH > KOH > NaOH. what the product distribution is for EO with various contaminants.
Some suggestions could be proposed for inherently safer design The reaction kinetics have not been established yet apart from EO
and operation of EO facility. It is better to keep EO away from incom- with water. For example, the onset temperature for EO/KOH and
patible containments and materials in main equipment, pumps, EO/NaOH differs by 60–70 ◦ C though the properties and molecu-
valves, pipelines, gasket, hoses, insulation, etc. Temperature is the lar structure for KOH and NaOH are similar. Finally, the important
best parameter for the detection and analysis of the evolution of calorimetric data-exothermic heat for EO with various contami-
the situation in case of runaway reaction of EO. For this reason, it nants are not provided by previous study, which is an essential
is recommended to measure permanently this parameter, to set parameter for safety control and equipment design.
adequate alarm and interlock to monitor and control the evolution
of the temperature in case of detected deviation. For EO aqueous Acknowledgments
solution in EO purification unit, a moderate concentration should
be selected based on the relationship between onset temperature The authors gratefully acknowledge the financial support pro-
and EO concentration, and the local concentration and temperature vided by the Ministry of Science and Technology of SINOPEC (No.
need to be detected and controlled. 414085). The support of professor Mannan from Mary Kay O’Connor
96 J. Liu et al. / Thermochimica Acta 652 (2017) 85–96
Process Safety Center, Artie McFerrin Department of Chemical Engi- [20] P. Thibault, L.G. Britton, F. Zhang, Deflagration and detonation of ethylene
neering, Texas A&M University is acknowledged. oxide vapor in pipelines, Process Saf. Prog. 19 (3) (2000) 125–139.
[21] R. Bubbico, G. Dore, B. Mazzarotta, Risk analysis study of road transport of
ethylene oxide, J. Loss Prev. Process Ind. 11 (1) (1998) 49–54.
References [22] C. Hirose, The microwave spectra and r0 , rs , and rm structures of ethylene
oxide, Bull. Chem. Soc. Jpn. 47 (6) (1974) 1311–1318.
[1] M. Ghanta, T. Ruddy, D. Fahey, D. Busch, B. Subramaniam, Is the liquid-phase [23] R. June, R. Dye, Explosive decomposition of ethylene oxide, Plant Oper. Prog. 9
H2O2-based ethylene oxide process more economical and greener than the (2) (1990) 67–74.
gas-phase O2-based silver-catalyzed process? Ind. Eng. Chem. Res. 52 (1) [24] J.P. Dever, K.F. George, W.C. Hoffman, H. Soo, Ethylene Oxide: Kirk Othmer
(2013) 18–29. Encyclopedia of Chemical Technology, vol. 10, 5th edition, John Wiley & Sons,
[2] J.R.H. Carucci, V. Halonen, K. Eränen, J. Wärnå, S. Ojala, M. Huuhtanen, R. Hoboken, NJ, 2005, pp. 632–673.
Keiski, T. Salmi, Ethylene oxide formation in a microreactor: from qualitative [25] R.K. Mcmullan, G.A. Jeffrey, Polyhedral clathrate hydrates. IX. Structure of
kinetics to detailed modeling, Ind. Eng. Chem. Res. 49 (21) (2010) ethylene oxide hydrate, J. Chem. Phys. 42 (42) (2004) 2725–2732.
10897–10907. [26] Y.L. Huang, T. Merker, M. Heilig, H. Hasse, J. Vrabec, Molecular modeling and
[3] J.-L. Gustin, Influence of trace impurities on chemical reaction hazards, J. Loss simulation of vapor-liquid equilibria of ethylene oxide, ethylene glycol, and
Prev. Process Ind. 15 (1) (2002) 37–48. water as well as their binary mixtures, Ind. Eng. Chem. Res. 51 (21) (2012)
[4] R.A. Kemp, Ethylene oxide catalyst and process. US5545603. In US, 1996. 7428–7440.
[5] F. Tamura, M. Saotome, Silver catalyst for production of ethylene oxide and [27] M. Herskowitz, M. Gottlieb, Vapor-liquid equilibrium in aqueous solutions of
method for manufacture thereof. EP, US4645754. In EP, 1987. various glycols and poly(ethylene glycols). 2: tetraethylene glycol and
[6] H. Kestenbaum, A. Lange de Oliveira, W. Schmidt, F. Schüth, W. Ehrfeld, K. estimation of UNIFAC parameters, J. Chem. Eng. Data 2 (2002) 233–234.
Gebauer, H. Löwe, T. Richter, D. Lebiedz, I. Untiedt, H. Züchner, [28] M.R. Altiokka, S. Akyalçin, Kinetics of the hydration of ethylene oxide in the
Silver-catalyzed oxidation of ethylene to ethylene oxide in a microreaction presence of heterogeneous catalyst, Ind. Eng. Chem. Res. 48 (24) (2009)
system, Ind. Eng. Chem. Res. 41 (4) (2002) 710–719. 10840–10841.
[7] H.J. Lee, M. Ghanta, D.H. Busch, B. Subramaniam, Toward a CO2—free ethylene [29] L.D. Chen, G.M. Faeth, Initiation and properties of decomposition waves in
oxide process: homogeneous ethylene oxide in gas-expanded liquids, Chem. liquid ethylene oxide, Combust. Flame 40 (1) (1981) 13–28.
Eng. Sci. 65 (1) (2010) 128–134. [30] G.A. Viera, L.L. Simpson, B.C. Ream, Lessons learned from the ethylene oxide
[8] M.R. Rahimpour, M. Shayanmehr, M. Nazari, Modeling and simulation of an explosion at Seadrift, Texas, Chem. Eng. Prog. 89 (1993) 66–75.
industrial ethylene oxide (EO) reactor using artificial neural networks (ANN), [31] L.T.T. Dinh, W.J. Rogers, M.S. Mannan, Reactivity of ethylene oxide in contact
Ind. Eng. Chem. Res. 50 (10) (2011) 6044–6052. with basic contaminants, Thermochim. Acta 480 (1) (2008) 53–60.
[9] X. Yang, Q. Xu, K. Li, C.D. Sagar, Dynamic simulation and optimization for the [32] G.A. Melhem, A. Gianetto, M.E. Levin, H.G. Fisher, S. Chippett, S.K. Singh, P.I.
start-up operation of an ethylene oxide plant, Ind. Eng. Chem. Res. 49 (9) Chipman, Kinetics of the reactions of ethylene oxide with water and ethylene
(2010) 4360–4371. glycols, Process Saf. Prog. 20 (4) (2001) 231–246.
[10] A. Peschel, A. Jörke, K. Sundmacher, H. Freund, Optimal reaction concept and [33] L.G. Britton, Thermal stability and deflagration of ethylene oxide, Plant Oper.
plant wide optimization of the ethylene oxide process, Chem. Eng. J. 207 Prog. 9 (2) (1990) 75–86.
(2012) 656–674. [34] K.H. Mueller, W.D. Walters, The thermal decomposition of ethylene oxide1,2,
[11] S. Rebsdat, D. Mayer, Ethylene oxide, in: Ullmann’s Encyclopedia of Industrial J. Am. Chem. Soc. 76 (2) (2002) 1458–1461.
Chemistry, Wiley-VCH, Weinheim, 2001. [35] A.A. Pekalski, J.F. Zevenbergen, M. Braithwaite, S.M. Lemkowitz, H.J. Pasman,
[12] B. Fabiano, F. Pistritto, A. Reverberi, E. Palazzi, Ethylene–air mixtures under Explosive decomposition of ethylene oxide at elevated condition: effect of
flowing conditions: a model-based approach to explosion conditions, Clean ignition energy, nitrogen dilution, and turbulence, J. Hazard. Mater. 118 (1–3)
Technol. Environ. Policy 5 (2015) 1261–1270. (2005) 19–34.
[13] F. Wang, Y. Zhao, O. Yang, J. Cai, M. Deng, Process safety data management [36] E. Wilcock, R.L. Rogers, A review of the phi factor during runaway conditions,
program based on HAZOP analysis and its application to an ethylene J. Loss Prev. Process Ind. 10 (5–6) (1997) 289–302.
oxide/ethylene glycol plant, J. Loss Prev. Process Ind. 26 (6) (2013) 1399–1406. [37] P. Virtanen, Kinetics of the hydrolysis reactions of ethylene oxide, Ann. Acad.
[14] F.I. Khan, M.R. Haddara, Risk-based maintenance of ethylene oxide Sci. Fenticae (1963) 9–89.
production facilities, J. Hazard. Mater. 108 (3) (2004) 147–159. [38] T. Tsuji, T. Hiaki, M. Hongo, Vapor-liquid equilibria of the three binary
[15] T.A. Kletz, Fires and explosions of hydrocarbon oxidation plants, Plant Oper. systems: water plus tetraethylene glycol (TEG), ethanol plus TEG, and
Prog. 7 (4) (1988) 226–230. 2-propanol plus TEG, Ind. Eng. Chem. Res. 37 (5) (1998) 1685–1691.
[16] J.L. Woodward, J.W. Wesevich, T.J. Kelly, Q.A. Baker, Analysis of ethylene oxide [39] L.G. Britton, D.A. Taylor, D.C. Wobser, Thermal stability of ethylene at elevated
gas house explosion, Process Saf. Prog. 26 (2) (2007) 150–154. pressures, Plant Oper. Prog. 5 (4) (1986) 238–251.
[17] R.G. Vanderwater, Case history of an ethylene tank car explosion, Chem. Eng. [40] M.A. Villamanan, C. Gonzalez, H.C.V. Ness, Excess thermodynamic properties
Prog. (USA) 85 (1989) 12. for water/ethylene glycol, J. Chem. Eng. Data 29 (4) (1984).
[18] L.G. Britton, Spontaneous fires in insulation, Plant Oper. Prog. 10 (1) (1991) [41] M.E. Levin, The reactivity of ethylene oxide in contact with iron oxide fines as
27–44. measured by adiabatic calorimetry, J. Hazard. Mater. 104 (1–3) (2003)
[19] Ameya Joshi, Xiaoqing You, T.A.B. Barckholtz, H. Wang, Thermal 227–245.
decomposition of ethylene oxide: potential energy surface, master equation [42] B.G. Freeder, T.J. Snee, Alkali-catalysed polymerization of ethylene oxide and
analysis, and detailed kinetic modeling, J. Phys. Chem. A 109 (35) (2005) propylene oxide—hazard evaluation using accelerating rate calorimetry, J.
8016–8027. Loss Prev. Process Ind. 1 (3) (1988) 164–168.