chemical engineering research and design 8 8 ( 2 0 1 0 ) 255262

Contents lists available at ScienceDirect

Chemical Engineering Research and Design
j our nal homepage: www. el sevi er . com/ l ocat e/ cher d
Capillary reactor for cyclohexane oxidation with oxygen
Radmila Jevtic

, P.A. Ramachandran, Milorad P. Dudukovic

Chemical Reaction Engineering Laboratory (CREL), Department of Energy, Environmental and Chemical Engineering (EEC), Washington
University in St. Louis (WUSTL), St. Louis, MO 63130, United States
1
a b s t r a c t
Cyclohexane oxidationis the rst stepinthe currently usedtechnology for productionof Nylon-6 andNylon-6,6 which
employs a two-stage process. Inthe rst stage 80%selectivity to two mainproducts, cyclohexanol and cyclohexanone
(KAoil) is obtainedat 48%cyclohexane conversioninstagedbubble columns or stirredtanks. There have beenreports
that increased oxygen concentration in the gas phase or pure oxygen is benecial to cyclohexane oxidation and this
was conrmed in our previous study (Jevtic et al., 2009). To fully utilize this advantage here, we present a novel,
safer capillary reactor for cyclohexane oxidation with pure oxygen. The discrepancy between the experimental and
modeling results was attributed to lower than expected mass transfer achieved in the capillary. With a better design
for gasliquid mixing and contacting this type of a reactor could potentially become attractive for gasliquid reactions
of similar nature.
2010 The Institution of Chemical Engineers. Published by Elsevier B.V. All rights reserved.
Keywords: Hydrocarbon oxidation; Capillary reactor; Kinetics; Reaction engineering
1. Introduction
Cyclohexane oxidation is practiced on a large scale in the
chemical industry worldwide with over 10
6
tons per year of
cyclohexanone and cyclohexanol (known as KA oil, Suresh
et al., 2000) made. KA oil is the starting material in the pro-
duction of adipic acid and caprolactum, which are ultimately
usedinthe manufacture of Nylon-6 andNylon-6,6 (Schuchardt
et al., 1993). The desired cyclohexane oxidation products (i.e.
cyclohexanol and cyclohexanone) are thus intermediates in
a sequence of complicated, multiple, free-radical chain reac-
tions (Berezin et al., 1966; Hendry et al., 1976; Walling, 1969).
Both cyclohexanol and cyclohexanone are more readily oxi-
dized than cyclohexane. Over-oxidation leads to a number
of byproducts, such as succinic, oxalic, caproic, glutaric, and
adipic acids (Wen et al., 1997), and can be minimized only by
keeping cyclohexane conversion low. Typically, in industrial
processes, cyclohexane in the liquid phase is oxidized with air
or a mixture of oxygen and nitrogen where the oxygen con-
centration is less than 21%. In order to approach plug ow
for the liquid reactant, which is needed to maximize the yield
of the desired intermediates, the process is conducted either
in a series of stirred tank reactors or staged bubble columns.

Corresponding author at: One Brookings Drive, Campus Box 1180, Cupples II, St. Louis 63130, United States.
E-mail address: rjevtic@seas.wustl.edu (R. Jevtic).
Received28 February2009; Receivedinrevisedform10 October 2009; Accepted21 December 2009
1
http://crelonweb.eec.wustl.edu.
After about 40min of reaction time at a temperature between
125 and 165

C and at a pressure of 815bar in the presence

of cobalt or manganese salts as a catalyst, cyclohexane con-
version of 4% and the selectivity of 80% to cyclohexanone
and cyclohexanol are typically achieved (Suresh et al., 2000).
Clearly, the process as practiced today demands substantial
capital investments, due to large equipment that is required,
and imparts substantial operating costs for the separationand
recycle of vast amounts of unreacted cyclohexane caused by
its lowconversionper pass. Improvements are thus possible by
increasing the process rate without negatively affecting selec-
tivity, which can then reduce equipment size, or by making
direct oxidation to adipic acid more selective. We focus on the
concept of increasing the reaction rate in this study.
When reaction rate is dependent on dissolved gas reactant
concentration an increase in the gas reactant concentration
has a positive effect on the rate for positive reaction orders
with respect to that reactant. For example, in the oxidation
of p-xylene to terephthalic acid it is reported (Chen, 2004)
that increasing the oxygen concentration by 2% in the feed
air results in a 10% increase in the production capacity of
terephthalic acid. Greene et al. (1998) reported rst cyclohex-
ane oxidation with pure oxygen. The reaction was performed
0263-8762/$ see front matter 2010 The Institution of Chemical Engineers. Published by Elsevier B.V. All rights reserved.
doi:10.1016/j.cherd.2009.12.008
256 chemical engineering research and design 8 8 ( 2 0 1 0 ) 255262
in the Liquid Oxidation Reactor (LOR) designed and patented
by Kingsley et al. (1994). The LOR uses high efciency stir-
ring, which limits the oxygen escape from the liquid phase
into the gas phase, and uses a nitrogen purge in the vapor
space above the liquid thereby keeping the oxygen concen-
tration in the gas phase at the safe level. In comparing the
cyclohexane oxidation with oxygen to that with air it was
established that to reach the desired cyclohexane conversion
of 4% the residence time is reduced from 36min needed with
air to 8min when using oxygen. This was accomplished at
slightly reduced operating temperature when using oxygen.
An improvement in selectivity and yield was also obtained
when using oxygen. In our recent study (Jevtic et al., 2009) in
a stirred autoclave operated in a batch and semi-batch (dead
end) mode, we have conrmed that in the temperature range
of 130160

C an increased oxygen concentration (either via

increased pressure of air or by increased oxygen mole frac-
tion in nitrogen or by combination of the two) has a benecial
effect on the oxidation rate and leads also to improved yield
and selectivity.
This conrmed that the use of pure oxygen or oxygen
enriched air and increased pressure could benet the overall
productivity of the cyclohexane oxidation process. The possi-
bility of deagration has been the major obstacle in assessing
the effect of increased oxygen availability not only on pilot
scale but inlaboratory reactors as well. Agreat concernregard-
ing the oxidation processes and their safety has been present
since the Flixborough accident on June 1st, 1974. It should be
noted that this major incident did not happen due to runaway
reactions but due to the rupture in the pipe that contained
cyclohexane at 150

C and at 1MPa. Subsequently, cyclohex-

ane leaked from the pipe, formed the vapor cloud, and, in
contact with an ignition source, exploded. In fact, it is said
(Berezin et al., 1966) that a mixture of cyclohexane vapor and
air is not capable of spontaneous ignition at temperatures up
to 250

C at any pressure (Berezin et al., 1966). Nevertheless,

considering the use of pure oxygen or oxygen enriched air in
any hydrocarbon oxidation process legitimately raises ques-
tions about safety. Thus, the proper choice of the reactor, that
would enable the safe operation, is of crucial importance.
Herein, we present an assessment of a capillary reactor as
an alternative, novel reactor for safe operation of cyclohex-
ane oxidation with pure oxygen. Its vastly reduced volume
to external wall area ratio makes it inherently safer and
allows the use of higher pressures. Absence of back mixing
due to plug ow should further enhance performance. Our
experimental studies in the batch systems (Jevtic et al., 2009)
provided the data needed for the capillary reactor sizing and
operation. The proposed capillary reactor is not a real micro
reactor, since its diameter is of the order of channel size in a
monolith. Nevertheless, it should share many of the advan-
tages of the small channels in terms of the improved mass
transfer and better temperature control. The objective of this
study was to assess whether the capillary reactor can indeed
live up to its expectations.
2. Experimental work
2.1. Reactor design
The capillary reactor is chosen due to two important reasons:
(a) Improved safety keeping the diameter of the reactor
small while increasing the length allows for more effec-
tive temperature control by wall cooling. The substantially
increased wall to volume ratio compared to those achiev-
able with cooling tubes in stirred tanks or bubble columns
eliminates hot spot possibilities and reduces runaway
chances.
(b) Improved selectivity favorable ow pattern for reactions
in series where desired product is an intermediate, like in
the oxidation of cyclohexane, is plug ow which is achiev-
able in a long two-phase ow capillary reactor.
The diameter of the capillary was chosen to prevent the
occurrence of local hot spots in the reactor. Hot spots can
lead to runaway reactions, which are characterized by uncon-
trolled rise in temperature accompanied by rise in pressure
that can results in an explosion. For a walled cooled Plug Flow
Reactor (PFR) the largest diameter that insures the absence of
the hot spot for cyclohexane oxidation at the reaction condi-
tions was calculated based on established criteria (Barkelew,
1959; Froment and Bischoff, 1990) is 2.5mm.
Another way to check for the occurrence of the runaway
reactionis to calculate Frank-Kamenetskiis parameter (Frank-
Kamenetskii, 1955) given by Eq. (1) below and compare it to
the critical value for various geometries:
=

LT
ud
T
0

,
t
c
t
r
(1)
Runaway reactions occur when a steady state temperature
prole cannot be maintained any more or, in terms of Frank-
Kamenetskiis parameter, when
cr
.
cr
is determined for a
number of different geometries andis equal to 0.88 for a plane-
parallel, 2.00 for a cylindrical, and 3.32 for a spherical vessel
(Frank-Kamenetskii, 1955). In the case of the cyclohexane oxi-
dation in the capillary reactor, can be calculated to be 0.04.
This value is much smaller than critical values for all vessels
mentioned above indicating that runaway reactions are not
likely in this capillary reactor when cyclohexane is oxidized
at the chosenreactionconditions (temperature 130165

Cand
pressure 115bar).
The length of the capillary reactor was determined based
on the residence time necessary to reach about 46% cyclo-
hexane conversion over the liquid owrate range chosen. The
liquid and gas owrate range were chosenbased onthe exten-
sive review of two-phase ow regimes in capillary tubes (Suo
and Grifth, 1964; Triplett et al., 1999; Mishima and Hikibi,
1996) in order to ensure a well characterized ow regime in
the tube while satisfying the gas/liquid stoichiometric ratio
needed for cyclohexane oxidation. Even though bubbly ow
would be preferred due to very large interfacial area between
gas and liquid phase, achieving this ow regime would lead
to oxygen starved conditions because of the very high liquid
to gas ow rates ratios. Thus, the Taylor ow regime was cho-
sen. This ow regime was also conrmed experimentally by
ow visualization carried out using transparent, Nylon tubes
of the same dimensions as the stainless steel tubes used.
Water was used as the liquid and oxygen as gas with liq-
uid and gas velocity of 0.001m/s to 0.0045m/s and 0.0036m/s
and 0.012m/s, respectively. The tests were done at room tem-
perature and pressure ranging from atmospheric to 3bar. In
the Taylor ow regime the reported gasliquid mass trans-
fer coefcients are high (Kreutzer et al., 2001; van Baten and
Krishna, 2004) so that we anticipated the reaction rates to
be kinetically controlled. In our previous studies of uncat-
alyzed cyclohexane oxidation (Jevtic et al., 2009) conducted in
chemical engineering research and design 8 8 ( 2 0 1 0 ) 255262 257
Fig. 1 Schematic of the capillary reactor set-up.
a batch or semi-batch autoclave, which were free from trans-
port effects, we were able to reach 4% cyclohexane conversion
at 1bar of air and 130

C at about 30min and in less time at

elevated pressures and temperatures. We assumed that the
same kinetics will be observed in the capillary reactor and
used this information on the needed residence time to select
its length.
2.2. Experimental procedure
1/8in. stainless steel tubing with 0.085in. inner diameter and
the length of 15m was used as a capillary reactor (Fig. 1).
The reactor tube is coiled with the coil diameter of 7in.
and the coil is placed in the thermal bath lled with silicon
oil. The required pressure is maintained with GO back pres-
sure regulator, BP-66 series. Temperature is measured with
3 Omega

type T quick disconnect thermocouples: at the

inlet, at the exit, and after 15m of the reactor. Cyclohexane
(SigmaAldrich) is pumpedintothe reactor using aHPLCpump
that is rated for ow rates from 0.001 to 12ml/min. Oxygen
and nitrogen (air gas) ow is regulated using two stage reg-
ulators and Brooks

5850EC mass ow controllers that are

calibrated for ow rates of 1100ml/min. Gas and liquid ows
are mixed using a specially designed mixer. The two-phase
ow is preheated to 150

C by owing trough a 4-in. stainless

steel block equipped with a cartridge heater. After the reac-
tor, the gas and liquid phase are separated. The liquid phase
is collected for manual injection into a Gas Chromatograph
(GC) while gas phase is sampled directly into the GC for anal-
ysis.
2.3. Analytical procedure
The quantitative analysis of the liquid reaction mixture is per-
formed using the Agilent 6890 Gas Chromatograph (GC) with
a Flame Ionization Detector (FID). Helium was used as a car-
rier gas. A 30m Alltech

AT-5ms column (0.25mm0.25mm)

was used for separation of oxidation products. The following
measurement conditions were used: injection port temper-
ature was held at 220

C. Initial temperature of the column

was 70

C with immediate rise in temperature of 10

C/min for
10min. After that, when temperature reached 170

C, the nal
temperature ramp was 50

C/min up to 230

C at which point
temperature was held constant for 2min. The MSD Produc-
tivity ChemStation Software was used to record and integrate
the output from the FID.
3. Results and discussion
3.1. Experimental product yields as function of
residence time for non-catalyzed oxidation
Starting from the temperature conditions used in the com-
mercial process for cyclohexane oxidation, cyclohexane is
oxidized with pure oxygen in the capillary reactor at 160

C,
15atm and without the catalyst. Different mean residence
times are achieved by changing gas and liquid owrates while
keeping the feed ratio of cyclohexane to oxygen constant and
equal to stoichiometric ratio. The results are presented in
Fig. 2. Note that time is obtained by dividing the length of
the reactor with the liquid velocity. Unfortunately, achieved
conversions of cyclohexane were so small that they could not
be accurately determined with our chromatographic set-up.
As evident in Fig. 2 the concentration of intermediate cyclo-
hexanol did not reach the maximum value even in 50min of
contact time in the capillary reactor at 160

. It does so in about
60min in the batch reactor at 130

C and in 30min at 140

C in
the semi-batch reactor (Jevtic et al., 2009). This is additional
evidence that the actual process rate is lower than anticipated
in the capillary reactor.
Fig. 2 Experimental results of cyclohexanol (ROH) and
cyclohexanone (RO) in cyclohexane oxidation in the
capillary reactor obtained at 160

C and 15atm with no use

of a catalyst.
258 chemical engineering research and design 8 8 ( 2 0 1 0 ) 255262
3.2. Reactor model and comparison of model
predictions and experimental results
It is always advantageous to view the experimental results in
the framework of an appropriate model. In our previous study
(Jevtic et al., 2009) we have demonstrated the usefulness of
such a model for the catalyzed oxidation of cyclohexane in a
batch and semi-batch autoclave. A simple model for the cap-
illary reactor is based on the plug ow assumption, constant
gas and liquid supercial velocity, steady state, and constant
temperature which leads to the following species, the species
balances in the gas phase:
u
C
V
C
RT
dp
i
dz
= k
L
u

p
i
H
i
c
i

V
L
(2)
Species balances in the liquid phase are given as:
u
L
dc
i
dz
= k
L
u

p
i
H
i
c
i

+
NR

k=1
v
ki
r
k
(3)
where u
G
denotes supercial gas velocity, u
L
denotes super-
cial liquid velocity and z is the axial coordinate along the
length of the reactor.
The assumption of the constant gas supercial velocity
is valid if air is used. However, if pure oxygen is used, and
especially if the conditions are such that total consumption
of oxygen is desired, then this assumption does not hold any
more. Then, the equation that describes the gas phase is given
as:
d
dz
(,
O
2
u
O
2
) = k
L
u

p
O
2
H
O
2
c
O
2

M
O
2
(4)
This however was not the situation in the present
study.
The above reactor model equations require the kinetic
rate expressions for each species and the specication of
kinetic andphysicochemical parameters. For the non-catalytic
oxidation of cyclohexane the most detailed kinetic model
was proposed by Kharkova et al. (1989) based on the free-
radical mechanism. This model was summarized in the thesis
by Jevtic (2008). The rate constants and activation ener-
gies for the 19 key elementary steps in this mechanism are
reportedandthis makes the model applicable at different tem-
peratures. The pertinent reactions were tabulated by Jevtic
(2008).
In the case of catalytic oxidation of cyclohexane (e.g. with
cobalt naphthanate) we have used successfully (Jevtic et al.,
2009; Jevtic, 2008) the kinetic scheme proposed by Spielman
(1964) and Alagy et al. (1964) that consists of lumped reactions
without free radicals, with the rate constants as reported by
Pohorecki et al. (1992) This kinetic scheme can be written in
the form below:
RH+O
2
k
0
ROOH (5)
ROOH
k
1
ROH+
1
2
O
2
(6)
ROOH
k
2
R

O+H
2
O (7)
ROH+
1
2
O
2
k
3
R

O+H
2
O (8)
R

O+
3
2
O
2
k
4
D (9)
where D is assumed to be adipic acid.
The pertinent kinetic constants at the temperature of
160

C were extracted by matching the performance of an

industrial reactor by Pohorecki et al. (1992) and are shown in
the table below:
k
0
(m
3
kmol
1
s
1
)
k
1
(s
1
) k
2
(s
1
) k
3
(m
3
kmol
1
s
1
)
k
4
(m
3
kmol
1
s
1
)
0.0266 0.0048 0.0013 0.5 0.37
We have shown (Jevtic et al., 2009) that these scheme and
constants can be used to interpret the performance of the
batch autoclave and that activation energies can be estimated
from runs at different temperatures.
Besides having the kinetic scheme and kinetic constants,
one has to use appropriate physical parameters for the com-
ponents involved in the process (most notably nitrogen and
oxygen solubility in cyclohexane for the temperature and
pressure ranges used in the experimental investigations) and
mass and heat transfer coefcients for the system. Nitrogen
and oxygen solubility (Suresh et al., 1986; Wild et al., 1978;
Tekie et al., 1997a,b) were obtained from the literature that
studies these solubility incyclohexane at higher temperatures
and pressures. Mass transfer depends on the ow pattern
developed inthe capillary. As already mentioned, the owpat-
tern in the capillary is Taylor ow and the literature on Taylor
owis quite extensive due to its importance for monolithreac-
tors. Contrary to the monolith reactors, which are three phase
catalytic reactors and three steps of mass transfer are present
(gasliquid, liquidsolid, and gassolid directly through thin
lm of liquid), in the capillary, only one step, gas to liquid
mass transfer, exists. There are several correlations available
for this type of mass transfer and all of them represent overall
gasliquid mass transfer (Kreutzer et al., 2001; van Baten and
Krishna, 2004; Bercic and Pintar, 1997). As mentioned earlier,
based on these correlations the predicted the mass transfer
coefcient is in the range of 0.010.06s
1
, mass transfer lim-
itations were not expected in our capillary reactor. Thus, the
mass transfer coefcient between 0.01 and 0.06s
1
depend-
ing on the ow rates was chosen in our model. However, the
comparison of model predictions and experimental results
indicates that actually achieved mass transfer coefcients are
much lower.
First, model predictions and experimental results are com-
pared for the non-catalyzed oxidation of cyclohexane in Fig. 3.
As mentioned above, the mechanistic model from Kharkova
et al. was used to represent the kinetics since that model was
determined for non-catalytic cyclohexane oxidation. The dis-
crepancy between model predictions and data is rather large
as evident in Fig. 3. In addition to poor gasliquid contacting
and mass transfer, more rapid quenching of the free radicals
in the capillary reactor, due to much higher solid wall to liq-
uid volume ratio than in the batch autoclave, may explain
the observed slow down in the oxidation rate compared to
the autoclave. To eliminate this possibility selected runs were
performed with the stirred autoclave by adding the sufcient
amount of small fragments of the capillary wall to the liquid
in the autoclave to achieve the same solid wall to liquid vol-
ume ratio as in the capillary. No slow down in the rates in the
autoclave were observed with the addition of such larger solid
chemical engineering research and design 8 8 ( 2 0 1 0 ) 255262 259
Fig. 3 Comparison of experimental and modeling results
for cyclohexanol (ROH) and cyclohexanone (RO) in
cyclohexane oxidation in the capillary at 160

C and 15atm
without the use of a catalyst.
surface area. Thus, free-radical quenching is not the reason
for the slow down in capillary reactor rates.
Next we compared the experiments carried out at the
same reaction conditions but with the catalyst, cobalt naph-
thanate, with the model predictions based on the lumped
model described above. The modeling and the experimental
results did not follow the same trend (Fig. 4). Note that, in this
case, the lumpedkinetic model was usedsince the kinetic con-
stants available from the literature were determined based on
the operation of the catalytic cyclohexane oxidation process.
3.3. Problems with gasliquid contacting
All our experimental results are reproducible, and we con-
rmed that the temperature and pressure in the reactor were
as specied. Hence, the only explanation for much lower oxi-
dation rates in the capillary reactor for either non-catalyzed or
catalyzedruns thanpredictedby the models for non-catalyzed
and catalyzed oxidations, respectively, and much lower cap-
illary reactor rates than observed in the stirred autoclave
point to inadequate gasliquid contacting and transport in
the capillary reactor. The ow pattern in the capillary was
re-examined. Instead of the stainless steel tubing, the Teon
Fig. 4 Comparison of experimental and modeling results
for catalytic cyclohexanol (ROH) and cyclohexanone (RO) in
cyclohexane oxidation in the capillary at 160

C and 15atm.
Catalyst concentration was 10
5
mol/l and gas and liquid
ow rates were 0.8 and 0.5ml/min, respectively.
tubing of the same size is used. As expected, the ow in cap-
illary appears as slug ow in the Taylor regime. However, the
ow was erratic and irregular. Even with the same gas and
liquid ow rate, non-uniform slug distribution was obtained
(Fig. 5). This has direct implicationonthe obtainedexperimen-
tal results since bubble and slug lengths affect mass transfer
in the capillary. In the model, the value for mass transfer coef-
cient was obtained from the correlations available from the
literature. Those correlations were obtained for uniform Tay-
lor ow, which was not observed in our experimental set up.
Indeed, when the mass transfer in the model was reduced
by an order of magnitude, from 0.02 to 0.002s
1
better agree-
ments between the experimental and modeling results were
obtained (Fig. 6). This conrms that poor mass transfer was
the reason for the disagreement of the model predictions with
experimental results.
There are different possible reasons for the non-uniformity
of the gasliquid ow. It can occur due to pressure uctuations
in the liquid pump, uctuations due to pressure changes dur-
ing bubble formation, or uctuations due to pressure changes
during bubble disengagement in the mixer. Since the pump
was a dual piston pump and the uctuations in the liquid ow
Fig. 5 Flow visualization study in the Teon tubing of the same size as the actual tubing used as a capillary reactor
(D=2.159mm). Gas ow rate was 3.6ml/min and liquid ow rate was 1.2ml/min for all the cases shown above.
260 chemical engineering research and design 8 8 ( 2 0 1 0 ) 255262
Fig. 6 Experimental and modeling results of cyclohexanol
(ROH) and cyclohexanone (RO) concentrations obtained in
the capillary reactor at 160

C and 15atm. Mass transfer

coefcient used in the model was an order magnitude
lower then the predicted one from the correlations available
in the literature.
was not observed, the effort was placed into the design of a
new gasliquid mixer. Three different gasliquid mixers were
designed and tested. The rst one was a simple tee (Fig. 7a).
Second one was a stainless steel block (Fig. 7b) with two chan-
nels for the liquid and the gas phase that meet in a Y-shape.
The channel for the gas phase has two times smaller diame-
ter than the one for the liquid phase. The reason behind this
was the fact that small gas inlet produces smaller bubbles
while the effect of the size of the liquid inlet is not signi-
cant (Amador et al., 2004). Third mixer (Fig. 7c) was a tee with
modied gas inlet. The size of the gas inlet was reduced, again,
to induce the formation of smaller bubbles. When tested, all
three mixers performed similarly and the non-uniformity of
the gas and liquid slugs was not avoided.
3.4. Examination of the potential of the capillary
reactor via model
Our experimental data greatly under-predicted the modeling
results for the capillary reactor due to poor mass transfer
achieved. Unfortunately, time limitations prevented us from
perfecting the gasliquid contacting in our capillary. Clearly,
this is an important problem that must be tackled and solved.
However, due to our success in modeling the cyclohexane
oxidation in stirred vessels (Jevtic et al., 2009; Jevtic, 2008)
it is instructive to report here what benets can one expect
from the capillary reactor with good contacting based on the
model predictions. This seems reasonable since the model
used the kinetics proposed here successfully to interpret auto-
clave data.
Fig. 8 Cyclohexanol (ROH) and cyclohexanone (RO)
concentration vs. residence time in the capillary reactor
obtained with the model at three different temperatures:
140, 150, and 160

C.
Table 1 Modeling results for maximum selectivity in
cyclohexane oxidation in the capillary reactor at
different temperature.
140

C 150

C 160

C 170

C
Max selectivity
(%)
70.0 77.8 80.2 83.5
RH conversion at
max selectivity
(%)
23.1 17.9 15.0 12.1
Residence time
at max
selectivity (min)
86.3 48.0 28.9 17.9
The effect of temperature on the cyclohexanol and cyclo-
hexanone evolutionas functionof meanresidence time, based
on the kinetics of Kharkova et al. (1989), is presented in Fig. 8.
Except for the diminishing induction period with rising tem-
perature, the concentration proles at different temperatures
look similar. The decrease of cyclohexanol and cyclohexanone
with increase in residence time is due to the increase in
production of higher oxidation byproducts. Table 1 shows
how the maximum selectivity for cyclohexanol and cyclohex-
anone changes with changing temperature as well as the time
necessary to achieve said selectivity and at what cyclohex-
ane conversion this occurs. The main difference is the time
needed for maximum selectivity. As temperature increases,
this time decreases, which is expected as there is a longer
induction period needed for the reaction to start at lower tem-
peratures. Also, cyclohexane conversion at which maximum
selectivity occurs, decreased with the increase in temperature
while the maximum selectivity increases. This indicates that
higher temperature has a positive impact on reactions lead-
Fig. 7 Three different mixers tested for gasliquid mixing.
chemical engineering research and design 8 8 ( 2 0 1 0 ) 255262 261
Table 2
u
G
/u
L
=1 u
G
/u
L
=2 u
G
/u
L
=3
Max selectivity
(%)
80.5 80.7 80.8
Residence time
at max
selectivity (min)
59.4 35.2 26.2
ing to two main products, cyclohexanol and cyclohexanone,
whereas lower temperatures are better suited for further oxi-
dation of cyclohexanol and cyclohexanone into byproducts.
Thus, higher temperature, when it comes to non-catalytic
cyclohexane oxidation, is a better choice for the potential
cyclohexane oxidationina capillary reactor. Shorter residence
time at higher temperature would imply use of a shorter
length capillary, which is desirable from practical point of
view.
Another important parameter is the ratio of gas and
liquid ow rate. This ratio was varied and the maximum
selectivity for the desired cyclohexane products was calcu-
lated. Also, the time needed to achieve that selectivity is
obtained. The reaction conditions were: 160

C, 15atm, pure
oxygen, no catalyst, and supercial liquid velocity of 0.002m/s
while gas velocity was varied. The results are presented in
Table 2.
While maximum selectivity is almost the same, the time
necessary to achieve it decreases signicantly with the
increase of the gas to liquid ow rate. This indicates faster
reactionrates withmore oxygenavailable. One couldconclude
that in this case, like in the case with higher temperature,
this would be favorable. However, with more oxygen avail-
able, more oxygen would remain unreacted at the point of
maximum selectivity. One of the argued advantages of this
type of reactor is the fact that the gas and the liquid veloc-
ities could be adjusted so that there is no oxygen left at the
exit point, where gas and liquid get separated, which is one of
the locations of major risk in the oxidation processes. In this
case, the desirable ratio of the gas to liquid velocity is 1 since
then, at the point of maximum selectivity, almost all oxygen
is gone. The conclusion is that, most likely, the combination of
higher reaction temperatures and lower gas to liquid velocity
would give the most desirable results in this type of a capillary
reactor provided excellent mass transfer can be reached and
maintained.
4. Concluding remarks
A novel capillary reactor for hydrocarbon oxidation, cyclohex-
ane oxidation in particular, was designed and tested. This
study proves that the gasliquid reactions of this nature can
be performedinthe small diameter capillary reactor. However,
we failed to design a gasliquid mixer and capillary reactor
with excellent mass transfer properties based on the avail-
able guidance in the literature. Thus, the gasliquid mixing
and contacting has to be resolved in a better fashion and mass
transfer improved. If this is successful, this type of reactor
provides a safer alterative to the current processes. Consider-
ing that cyclohexane oxidation is operated at large scale, the
productivity of this system, even when scaling out or number-
ing up is accounted for, probably will not meet the required
demands of a large capacity plants. However, having in mind
hydrocarbon oxidations in general and their similarity when
it comes to the kinetic mechanism, this approach could be
successfully used for other chemical systems, especially ones
that produce precursors in pharmaceutical and ne chemistry
industry.
Acknowledgements
The authors are grateful for the support of the National
Science Foundation through the Center for Environmentally
Benecial Catalysis (CEBC) at Kansas University (KU) via Grant
EEC-0310689. Fruitful discussions with professors B. Subrama-
niam and R.V. Chaudhari at the University of Kansas were
particularly appreciated. Industrial support of CREL, where
this workwas executed, alsocontributedtoits successful com-
pletion. Our special gratitude to professor P.L. Mills of TAMUat
Kingsville for many useful suggestions regarding experimen-
tal design and techniques and for the analytical equipment
provided on loan.
References
Alagy, J., Defoor, F. and Franckowiak, S., 1964, Oxidation of
cyclohexane to cyclohexanone. Rev Inst Franc Petrole Ann
Combust, 19(12): 13801390.
Amador, C., Salman, W., Sanguanpiyapan, S., Gavriilidis, A. and
Angeli, P., 2004, Effect of gas-inlet conditions on the
mechanism of Taylor ow formation, In 5th International
Conference on Multiphase Flow Yokohama, Japan,
Barkelew, C.H., 1959, Stability of chemical reactors. Chem Eng
Prog Symp, 55: 3749.
Bercic, G. and Pintar, A., 1997, The role of gas bubbles and liquid
slug lengths on mass transport in the Taylor ow through
capillaries. Chem Eng Sci, 52(21/22): 37093719.
Berezin, I.V., Denisov, E.T. and Emanuel, N.M., (1966). The Oxidation
of Cyclohexane. (Pergamon Press).
Chen, J.-R., 2004, An inherently safer process of cyclohexane
oxidation using pure oxygen an example of how better
process safety leads to better productivity. Process Saf Prog,
23(1): 7281.
Frank-Kamenetskii, D.A., (1955). Diffusion and Heat Exchange in
Chemical Kinetics. (Prinston University Press, Prinston, New
Jersey).
Froment, G.F. and Bischoff, K.B., (1990). Chemical Reactor Analysis
and Design. (Wiley, New York).
Greene, M.I., Sumner, C. and Gartside, R.J., 1998, Cyclohexane
oxidation, 5,780,683.
Hendry, D.G., Gould, W.C., Schueltze, D., Syz, M.G. and Mayo, F.R.,
1976, Autoxidation of cyclohexane and its autoxidation
products. J Org Chem, 41(1): 110.
Jevtic, R., 2008, The effect of oxygen on the oxidation of
cyclohexane, Ph.D. Thesis, Washington University in St. Louis.
Jevtic, R., Ramachandran, P. and Dudukovic, M., 2009, Effect of
oxygen on cyclohexane oxidation: a stirred tank study. Ind
Eng Chem Res, 48(17): 79867993.
Kharkova, T.V., Arest-Yakubovich, I.L. and Lipes, V.V., 1989,
Kinetic model of the liquid-phase oxidation of cyclohexane. I.
Homogeneous course of the process. Kinetika i Kataliz, 30(4):
954958.
Kingsley, J.P., Roby, A.K. and Litz, L.M., 1994, Terephthalic acid
production, 5,371,283.
Kreutzer, M.T., Du, P., Heiszwolf, J.J., Kapteijn, F. and Moulijn, J.A.,
2001, Mass transfer characteristics of three-phase monolith
reactors. Chem Eng Sci, 56(2122): 60156023.
Mishima, K. and Hikibi, T., 1996, Some characteristics of air-water
two-phase ow in small diameter vertical tubes. Int J
Multiphase Flow, 22: 703.
Pohorecki, R., Baldyga, J., Moniuk, W., Krzysztoforski, A. and
Wojcik, Z., 1992, Liquid-phase oxidation of cyclohexane
modeling and industrial scale process simulation. Chem Eng
Sci, 47(911): 25592564.
262 chemical engineering research and design 8 8 ( 2 0 1 0 ) 255262
Schuchardt, U., Calvarlho, W.A. and Spinace, E.V., 1993, Why is
interesting to study cyclohexane oxidation? Synlett, 10:
713718.
Spielman, M., 1964, Selectivity in hydrocarbon oxidation. AIChE J,
10(4): 496501.
Suo, M. and Grifth, P., 1964, Two-phase ow in capillary tubes. J
Basic Eng, 86: 576.
Suresh, A.K., Sridhar, T. and Potter, O.E., 1986, Mass transfer
during reaction cyclohexane oxidation. Chem Eng Sci, 41(4):
741746.
Suresh, A.K., Sharma, M.M. and Sridhar, T., 2000, Engineering
aspects of industrial liquid-phase air oxidation of
hydrocarbons. Ind Eng Chem Res, 39: 39583997.
Tekie, Z., Li, J. and Morsi, B.I., 1997, Mass transfer parameters of
O
2
and N
2
in cyclohexane under elevated pressures and
temperatures: a statistical approach. Ind Eng Chem Res, 36(9):
38793888.
Tekie, Z., Li, J., Morsi, B.I. and Chang, M., 1997, Gasliquid mass
transfer in cyclohexane oxidation process using gas-inducing
and surface-aeration agitated reactors. Chem Eng Sci, 52(9):
15411551.
Triplett, K.A., Ghiaasiaan, S.M., Abdel-Khalik, S.I. and Sadowski,
D.L., 1999, Gasliquid two-phase ow in microchannels. Part I.
Two-phase ow patterns. Int J Multiphase Flow, 25(3): 377394.
van Baten, J.M. and Krishna, R., 2004, CFD simulations of mass
transfer from Taylor bubbles rising in circular capillaries.
Chem Eng Sci, 59(12): 25352545.
Walling, C., 1969, Limiting rates of hydrocarbon autoxidations. J
Am Chem Soc, 91(27): 75907594.
Wen, Y., Potter, O.E. and Sridhar, T., 1997, Uncatalysed oxidation
of cyclohexane in a continuous reactor. Chem Eng Sci, 52(24):
45934605.
Wild, J.D., Sridhar, T. and Potter, O.E., 1978, Solubility of nitrogen
and oxygen in cyclohexane. Chem Eng J, 15(3): 209214.