SSSLINDE concsponvence
lent |
HOB _ryr Gm excel det by.
a
cc BH Torn Bergman!
Strom Dep (OLD SAW MILL FIVER ROAD, TARRYTOWN, WY 10591
mob. M. Lite ose November 1, 1989
oreweira deer Industrial Gases Development
coy D.C. decastro Swe! Cumene Literature
D. A. Haid
F. S. Iacopelli
3. P. Kingsley
K. A. Kuberka
F. D. Lauricella
I have surveyed the literature on the production of phenol
from cumene that we have in our files and have prepared the
following summary:
THE PRODUCTION OF PHENOL FROM CUMENE
1. WHAT PROCESSES AND REACTIONS OCCUR IN THE PRODUCTION OF
PHENOL FROM CUMENE?
The only industrially signiticant method ror producing
phenol is currently via the oxidation of cumene (isopropyl
benzene) (1). In this process, cumene is first oxidized to cunyl
hydroperoxide (CHP). The CHP is then concentrated (and the
unreacted cumene recycled to the oxidation reactors) and cleaved
to form phenol and acetone in the presence of sulfuric acid (2).
Several by-products are also produced (the major by-products are
acetophenone and dimethylphenyl carbinol). at the end of the
process, the acid is neutralized with caustic and the resulting
salts are removed. The phenol is then separated from the acetone
by distillation.
The mechanism of the cumene to phenol reaction has been
determined (3); the production of CHP from cumene occurs via a
free radical mechanism while the production of phenol from CHP
occurs via a cationic reaction (hence requiring the presence of
an acid).
It is clearly the oxidation step that is of interest to
Linde. The oxidation reaction is zero order with respect to
BUSINESS CONFIDENTIAL2
oxygen. This is demonstrated by the rate expression for the
reaction
rate==k,_ [Cul] (2k; [cHp]) 7? ()
On
where k, is the rate constant for the propagation reaction, k, is
the rat® constant for the termination reaction and k, is the rate
constant for the initiation reaction (this expression has been
rigorously derived (13)). Therefore, as long as the reactor is
not oxygen transfer limited, the reaction will always be op-
erating at its optimal rate with respect to the amount of oxygen
available. However, if the rate at which oxygen dissolves is
lower than the rate at which the oxygen reacts, the reactor is
mass transfer limited and the rate can be improved by dissolving
more oxygen. Most cumene reactors operate in this regime. If
the reactor is oxygen transfer controlled, the appropriate rate
expression is:
Rate = a k,°[0,), + ar (2)
2
where a is the mass transfer area per unit volume of liquid, k,°
is the liquid phase physical absorption coefficient, (0,], is the
oxygen concentration at the gas/liquid interface, r is the
intrinsic rate (from equation 1) and D, is the diffusivity of
oxygen in cumene. The over all production rate may then be
calculated for any given reactor by multiplying the reaction rate
by the liquid volume.
It is also interesting to note the rate expressions for the
formation of the most prevalent by products, acetophenone and
dimethylphenyl carbinol. These expressions are as follows:
Ratey, = Kyp [CuH] [CHP] Q)
Ratepyrc = Koype [CHP]
where
Ky = 0.54389 x 10° % exp(-11395.4/T) — (1/gmole-min) (4)
and Kiyec = 0529905 x 10’ x exp(-91284.41/T) (1/min)
(Note that the rate, constant for the production of CHP is
Koy = 0+276633 x 10° x exp(-7599.29/T) (1/gmole) #(1/min))
An expression for the reaction efficiency has also been
derived and ie ac follows:
1 - (k,(2k,[cHPy) V2
Efficiency = Kk, [Cuz (5)
1 + 20k, fax, (CHP)
K, [Cai]
BUSINESS CONFIDENTIAL3
values for the constants in Equations 1 through 2 obtained
from the literature and from Bound Brook reports are listed in
Table 1.
The oxidation reaction is heat initiated, autocatalytic and
has a heat of reaction of -328 BTU/pound of CHP (exothermic).
Both the reactants and product, CHP, are insoluble in water but
readily soluble in most organic solvents. ‘The reaction is
inhibited by phenol and sulfur dioxide as well as several
corrosion inhibitors (16). Certain alkyl heterocycles, in
conjunction with hydrochloric acid, are very active promoters tor
the oxidation reaction. It was also discovered that a wide
variety of ‘onium halides ranging from amine hydrochlorides to
quaternary ammonium and phosphonium salts have catalytic activity
at low concentrations. For example, as little as 50 ppm of
lauryltrimethylammonium chloride leads to a rate increase of from
50-75% over uncatalyzed oxidation. The only penalty is a slight
(0.5-1%) drop in reaction efficiency. These catalysts act by
causing a large amount of the CHP to degrade to the cumyl hydro-
peroxide radical, which causes the oxidation reaction to pro-
pagate.
2. HOW IS THE REACTION CARRIED OUT?
A. HISTORY OF CHP PRODUCTION
The first commercially feasible technique for oxidizing
cumene to CHP was invented in 1953 (12). This technique made use
of heavy metal oxide or carbonate catalysts (which can be ex-
pensive) and gave reaction efficiencies that were better than
those observed in previous processes, although they were much
lower than those observed today. If these catalysts were not
used, very long reaction times (on the order of 20 days) were
required. Another early technique involved the oxidation of
cumene in an organic in water emulsion, which required the use of
an emulsifying agent, such as sodium stearate (11). Here, the
reaction times involved were reduced over previous processes.
B. PRESENT DAY PROCESS
Today, the oxidation step is typically carried out in a
reactor containing only cumene, cumyl hydroperoxide and a small
concentration of reaction by products. No catalysts are re-
quired and an aqueous media is not used. Formerly, a sodium salt
of cumyl hydroperoxide was used as an accelerator in the oxi-
dation reaction. However, this salt has been eliminated due to
process improvements. A Schematic diagram ot this process (as it
BUSINESS CONFIDENTIAL4
was used at Union Carbide's Bound Brook plant) is shown in Figure
1.
In the process used commercially today, the oxidation
reaction is carried out ina series of reactors in which the
temperature is lowered from one reactor to the next (4). The
temperatures used in this step are relatively low (on the order
of 100°C) to minimize the production of by-products by de-
composition of CHP. The temperature is maintained at a very
constant level by circulating the reactor contents through a heat
exchanger al 4 high rate.
This process is used in the former Bound Brook plant that
has been moved to Texas as part of Georgia Gulf (Note: Union
Carbide also has (or had) a plant to produce cumene from phenol
at Ponce, Puerto Rico). The plant has been producing phenol from
cumene at a rate of 180 MM 1b/year by a process licensed fron
Allied Chemical (5) (note that Allied also uses approximately the
same process). The reaction is carried out in a series of five
reactors at a temperature of 99-112°C (the temperature being
highest in the first reactor in the series and lowest in the
last) and a pressure slightly above 1 atmosphere.
Air is added to these reactors through a simple sparger with
slot-like holes. In the original process design, the rate at
which air was sparged decreased from one reactor to the next.
The purpose of doing this was to minimize the amount of oxygen in
the headspace. It was thought that since the cumene concen-
tration decreases from one reactor to the next, the amount of
oxygen required in each consecutive reactor decreases. There-
fore, the amount sparged was decreased to maintain the same
headSpace oxygen concentration in every reactor. The sparger
slot size also decreases from one reactor to the next to main-
tain a constant gas injection velocity in all reactors. More
recently, the reactor is run with the same air flow rate in each
reactor, which has been shown to increase the rate of CHP pro-
duction without resulting in headspace oxygen levels above the
explosive limits.
The reactors are not mechanically agitated; the only source
of agitation is a pump-around-loop that turns over each reactor
every 35 minutes (the loops also circulates the reactor contents
through an external heat exchanger as mentioned above). This
process produces 70.1 1b of phenol for each 100 ibs of cumene
with a reaction efficiency of 89.5%. The conversion realized in
the oxidation process is typically 25-30% (8), although the
Georgia Gulf plant has been able to approach 40%. It is not
desireable to achieve high conversions in these reactors because
BUSINESS CONFIDENTIAL5
high levels of CHP in an oxidation can cause a potentially
explosive situation.
WHAT ARE THE HYDRODYNAMICS OBSERVED IN BUBBLES COLUMNS, SUCH
AS THOSE USED IN CUMENE REACTORS?
‘The two phase flow observed in bubble column reactors is
usually divided into three regimes. At the lowest gas flow
rates, the separate bubbling regime is observed. Here, the gas
flow rate is so low that individual bubbles form at the inlet
orifice and a nearly straight path is tollowed to the to top of
the liquid phase (13). Very little mixing of the liquid occurs
in this regime.
The second regime is termed the quiescent regime and has
been defined in terms of superficial gas velocities below 0.15
and 0.2 ft/second. In this regime, the gas provides considerable
mixing of the liguid but still retains its ordered bubble charac-
ter. Thus, the system is approximated by assuming a plug flow
gas phase and a well-mixed liquid phase.
The last regime is called the turbulent regime. It exists
for gas superficial velocities greater than 0.25 ft/s. Here,
both the gas and liquid phases can be approximated as well-
mixed.
The superficial gas velocity observed at Bound Brook was 0.1
ft/s. Therefore, the observed flow regime is quiescent. Gas
enters the reactor and rises as individual bubbles. However, the
liguid phase is well mixed.
Some of the work done at Bound Brook described the effects
of fluid properties on bubble formation and size. It was shown,
for example, that surface tension has its greatest influence on
bubble size for low gas flow rates, large orifice diameters and
with liquids of low viscosity. Under these conditions, the
bubble size is determined by a balance between buoyancy and
surface tension forces. However, as gas flow rates increase, the
effects of surface tension decrease. In fact, at the gas flow
rates used in most industrial bubble columns, surface tension has
an insignificant effect on bubble diameter.
With higher gas flow rates, low surface tension and small
orifice diameters, the liquid viscous forces become the primary
resistance to bubble formation compared to the surface tension
force.
The size of the orifice used in the sparger also has an
effect on the bubble diameter. Figure 2 shows the bubble dia~
BUSINESS CONFIDENTIAL6
meter resulting from various orifice Reynolds numbers for quies-
cent flow. Bubble size increases significantly with both orifice
diameter and gas velocity in the laminar region. However, for
Reynolds numbers greater than 10,000, orifice diameter and
velocity have a negligible effect on bubble size.
It was indicated that no study had been completed to compare
the effect of orifice hole geometry on bubble size. Bound Brook
used long, thin slots instead of round hole spargers.
Models for predicting fractional gas holdup using the
terminal bubble velocity and superficial gas velocity. Both
bubble diameter and fractional gas holdup are important in
determining the available interfacial area between the gas and
the liguid and the amount of oxygen that is transferred.
It has also been demonstrated that the average air bubble
diameter at Bound Brook was 0.625 cm (19).
4. WHAT EXPERIMENTAL WORK HAS BEEN DONE TO DATE TO STUDY THIS
REACTION?
A study was conducted at Bound Brook (6) in which the
conventional reactor configuration described above was compared
with a mechanically agitated reactor on the bench scale level to
determine the effect of reactor type on rate and efficiency. A
comparison was also made between air and oxygen enrichment. It
was found that both agitation and the use of oxygen-enriched
streams increased the reaction rate without decreasing the
reaction efficiency. With agitation and enrichment, it is
possible to obtain rates almost double those presently achieved
in the plant.
In this study, a 3 liter, bench scale, stirred tank reactor
was used. A detailed description of this reactor and the pro-
cedures used was given in Reference 6. A titration technique for
measuring CHP was also given. Data on reaction rate was given as
a function of percent oxygen injected (see Figure 3). A later
letter (7) also showed the distribution of products over time in
the laboratory reactor and showed very little by product forma-
tion with both air and oxygen.
It was shown that the best method for analyzing samples of
the reactor contents for cumene and cumene oxidation products is
liquid chromatography (LC). Analysis by gas chromatography is
not reliable because the unstable hydroperoxides tend to form
other products not seen in the reactor (10). Liguid chromato-
graphy offers advantages over gas chromatography in that it can
be used to analyze for nonvolatile and thermally unstable con-
pounds.
BUSINESS CONFIDENTIAL7
Another study was conducted at Bound Brook in which Equa-
tions 1 through 3 and Bound Brook's operating data were used to
model the cumene reactor. The purpose of this study was to
determine (qualitatively) the effects of percentage of oxygen
injected, temperature, bubble size and CHP concentration on the
point at which the reaction undergoes the transition from ki-
netically controlled to mass transfer controlled. The results of
this study are summarized in Figures 4 through 6. Remember that
these graphs were made from theoretical equations and, therefore,
should be used solely for qualitative observations. Note that
the predicted rates are much lower than those observed in Figure
2. This is, once again, a result of the lack of experinental
Gata at a variety of operating conditions to back up the data.
Note that in the kinetically controlled regime, the bubble size
and mole fraction oxygen do not effect the rate. This is because
the reaction is zeroth order in oxygen. Note also that in the
mass transfer controlled regime, the temperature has little
effect on the rate. Increasing the concentration of CHP causes a
decrease in the transition temperature from the mass transfer to
kinetically controlled regime.
The effect of temperature and cumene concentration on
reaction efficiency has also been studied. The reaction ef-
ficiency is a function of both the temperature of the reaction
and the concentration of CHP in the oxidizer. The efficiency
decreases as both the temperature and CHP concentration increase.
Thus, when operating the oxidizers in the kinetic regime, in-
creasing the temperature will increase CHP production, but will
dcerease the reaction efficiency; operations in the mass transfer
regime will only decrease reaction efficiency as temperature is
increased and have no corresponding increase in CHP production
(as).
Data was also collected at Bound Brook and at Ponce to
determine the best parameters to use in Bound Brook optimization
computer program, called OXOPT, of which the author is in posses-
sion. Using the program, the productivity of the plant was
increased from 150 MM pounds of phenol per year to 180 MM pounds
of phenol per year.
Other experiments showed that it is desireable to keep the
liguid level in the oxidizers as high as possible to get the best
efficiency. The higher the liguid level, the longer the res-
idence time of cumene in the reactor and the lower the temper-
ature needed to obtain a given conversion.
BUSINESS CONFIDENTIAL8
5. WHY IS PURE OXYGEN OR ENRICHMENT NOT USED TODAY?
Despite these results shown in part 4, Bound Brook did not
convert their reactors to agitated, oxygenated reactors. One
reason for maintaining the reactors as mass transfer limiting
devices was to insure that most of the oxygen would dissolve
within the reactor, which would insure that the headspace oxygen
concentration would remain below the flammability limit of
cumene. In addition, they did not want to increase their gas
flow rate (which they believed was necessary in order to use
enrichment) because they were afraid that more cumene would be
lost with the larger air stream (at 109°C, the vapor pressure of
cunene is 0.27 atm). It was mentioned that the use of a more ef-
ficient sparger than the current slotted version should be
beneficial in terms of increased transfer efficiency. However,
the spargers were not replaced.
In 1973, an economic analysis was conducted to determine the
best way to increase the productivity of the oxidizers (14). At
that time, the productivity of the plant was 150 MM pounds of
phenol per year. The first option was to increase their blower
speed to capacity, increase the oxidizer pressure and increase
the CHP concentration. With this option, the plant could have
boost its capacity to 180 MM pounds of phenol per year for a
capital investment cost of $130,000. ‘The second case involved
either buying a second blower or using oxygen enrichment. The
option of buying a second blower was ruled out because it was
unattractive economically and in terms of ease of installation.
It was calculated that with oxygen enrichment, the plant's
capacity could have been boosted to 200 MM pounds of phenol per
year for a capital investment of $185,000 per year. The third
option involved a combination of both options 1 and 2 for produc-
tivity greater than 220 MM pounds per year and required a capital
investment of approximately $2.67 MM. The study considered the
impact of increased oxidizer productivity on all oxidizer and
downstream equipment and includes a rather complete list of all
of the major equipment in the phenol production plant.
As a result of this economic analysis, Linde was approached
in 1974 about the possibility of conducting tests with oxygen
enrichment (15). Supply options and oxygen pricing were dis-
cussed. Unfortunately, to this author's knowledge, these tests
were never conducted. ‘It should be noted that the study did
mention that increases in productivity would probably be ac-
companied by decreases in reaction efficiency.
Bound Brook was approached (by Linde) about the idea of
using oxygen enrichment in the production of CHP in 1979 (9). At
that time, it was felt that oxygen enrichment could not be
BUSINESS CONFIDENTIAL9
justified based on the anticipated improvement in yield. They
expected the yield to actually decrease with enrichment (although
the stated reasoning is highly suspect). They stated that the
use of oxygen could only be justified if an improvement in
capacity were realized (if such an improvement was needed).
Since they were operating under capacity at that time, oxygen
enrichment was only of academic interest to then.
6. ARE THERE OTHER PRODUCTS MADE FROM THE OXIDATION OF CUMENE
(BESIDES PHENOL)?
Another product made from cumene is acetophenone. It is a by-
product of the phenol process and is formed from the cunyloxy
radical. Acetophenone production is likely to be favored in the
cumene oxidation process when the temperature and hydroperoxide
concentrations are increased over those used in phenol production
(17). Acetophenone production is also enhanced through the use
of various metallic catalysts.
The rate of acetophenone production also increases slightly
with high dissolved oxygen levels. Therefore, the efficiency of
the phenol process decreases slightly (actually, negligibly) with
increased oxygen, as is stated above.
Acetophenone may also be made from CHP by thermal cleavage of
80% cumyl hydroperoxide at a temperature of 165°C (18).
7. REFERENCES
1. Pujado, P.R., Salazar, J.R. and Berger, C.V., "Cheapest
Route to Phenol", Hydrocarbon Processing, March 1976, pp.
91-96.
2. International Petroleum Encyclopedia, pp. 425-426.
3. Streitwieser and Heathock, Introduction to organic
Chemistry, 1976, pp. 1001-1002.
4, Kirk-othmer Encyclopedia of Chemical Technology, v. 17,
1983, pp. 374-377.
5. Shertok, J.T., Baez, E.E. and Chin, H.W., "Cumene oxidizer
Optimization: Results of optimization Plant Trials",
Bound Brook, NJ, 8/12/77.
6. Stueben, K.C., "Cumene Phenol Process Improvement Program I.
The Effect of Mass Transfer on Oxidation Rate", Bound Brook,
NJ, 1/20/71.
BUSINESS AL10.
ui.
12.
13.
14.
15.
16.
ng.
18.
19.
10
Letter, K.C. Stueben-C. Ho, 9/19/74.
U.S. Patent #3,523,977, "Process for Preparing Cumene Hydro-
peroxide", C. Reni and L. Lugo, 8/11/70.
trip Report, K. Kiyonaga, UCC, Bound Brook, NY, 7/19/79.
Brydia, L.E., "Analysis of Mixtures Containing Cumene and
Cunene oxidation Products by Liquid Chromatography", 2/71.
U.S. Patent #2,547,938, "Manufacture of Alkyl Benzene Per-
oxides", 1951.
U.S. Patent #2,632,772, "Manufacture of Peroxidic
Compounds", 1953.
Gregory, R.A., "The Design of Bubble Reactors with Applica-
tion to Cumene Oxidation", Bound Brook, NJ, 8/10/71.
Gregory, R.A., "Capacity Study of Bound Brook Phenol unit",
Bound Brook, NJ, 9/5/73-
Letter, K.C. Stueben-R.J. Cotter, 8/29/74.
stueben, K.C., "Ponce Phenol Plant Support Cumene oxidation
Program", Bound Brook, NJ, 10/30/73.
stueben, K.C., "Approaches to Increased Acetophenone Pro~
duction I: Cumene Oxidation Route", Bound Brook, NJ, 1/3/75.
Stueben, K.C., "Approaches to Increased Acetophenone Pro-
duction II: Controlled Decomposition of Cumene
Hydroperoxide", Bound Brook, NJ, 1/17/75.
shertok, J.T., "Cumene oxidizer Optimization: A status
Report", Bound Brook, NJ, 6/3/76.
ecgun!)
yh awe Jie emo
TJ. Bergnan ig
USINESS Ci TAL1.
TABLE 1
KINETIC CONSTANTS USED IN EQUATIONS 1-3
TEMPERATURE = 35°C ‘TEMPERATURE = 105°C
i 0.2 1/(mol-sec) 4.3 1/(mol-sec)
x, 2.0 x 10‘ 1/(mol-sec) 12.2 x 10° 1/(mol-sec)
x . 0.224 x 10° sec’!
PARAMETERS USED TO MODEL BOUND BROOK CUMENE OXIDATION REACTORS
Dy = 0.4-1.27 em
Dy 1s ft
H = 1,13 x 10° mm Hg 1/mole
k° = 3.06 x 10% - 3.51 x 107 cm/sec
Liquid Volume = 1.7 x 10° liters
D, = 9.3 x 10% cn*/sec
Gas Holdup = 0.1
BUSINESS CONFIDENTIALBovND BRooK ENOL PLANT
ESTIMBTED ‘OPERATING CONDITIONS
Str ain ro
AIM, HBX Oy,
fae
F308
és
$0, /20ppm
SRO
pECOMPOSER,
Varn R,
ra
v3035, f
rorcHr
s-30azat
DFHYDRATORS
Fron gph Onenr
_FRM T-Bloz ovo .
W30RR vero y- Zou
Cumens scAveaERE
CRUDE Tow aFRE erat .
Provecr CovitR, aaeaare 50, = mi Fig) * 1A
REKelm ye] 173eh
Aarayy
OF aunvis
ee ee
Valen — Saar maniatneny sheng
TOAD ae satis 2
. Sa wee
)
al? ¢ a
jee
efiter onxoquMN spToudex 2
our fuaraupia alana ura(00%) “UIAET/aHO STOW ‘2pey
NOMWINIONOD NIDAXO dO Loads
eaunoid |
4WILNadI4NOO (axe)
ssanisna LNONe
eeort Dot
08
wost = Sq
“suoLqeAdasqg aALyeyt Lent
og Alug asp “Lea uoayy
oupbou
tuo, ane
patterquop ATTeoTUTY ——— —
autbou
peTtoxqucp segsuer, ssew
= eK pe
Tame) Ww
sinqeredie, Sistoh syey UOTIep FO
v gunora
foqeu
ston
(.0TX)JOTUL FW UOTIDeTE aTOW UEbAXO
£ o s b es
T T I T
2,06 = 2
D9s0T = &
*suorzeasasag aAyqeyt Lend
4oy €lug as *[eaIga40ay{ :uoLgneg : aT
awy bor
peTTorgueg ATTeOTyeUTX —— ——
oupbex
PeTTOszUOD JezsuerTy ssey
4 73Mg
— 70 = dig Tor
woraseiry Stow wel
KO OTUT SMsTeA 07k UOPIEDTKO
§ minora
faqey
STOR
(07x)ot ort oot 06
I T =
wos? = %q S
set = Sq a
ast? = Sa
*suoyqensasqg alae tend
oj Alug asp *Leapyauoay, +
aurzbox
peTTor3zu0D ATTROTQOUTY
outbox
PeTTOTyUOD Toysuery, sey,
1429
g muno1e
suqut
eTOW ‘aap
(07x)