vyEPA
United States
Environmental Protection
Agency
Robert S Kerr Environmental Research
Laboratory
Ada OK 74820
EPA-600/2-80-064
April 1980
Research and Development
Solvent Extraction of
Wastewaters from
Acetic-Acid
Manufacture
-------
RESEARCH REPORTING SERIES
Research reports of the Office of Research and Development, U.S. Environmental
Protection Agency, have been grouped into nine series. These nine broad cate-
gories were established to facilitate further deveJopment and application of en-
vironmental technology. Elimination of traditional grouping was consciously
planned to foster technology transfer and a maximum interface in related fields.
The nine series are:
1. Environmental Health Effects Research
2. Environmental Protection Technology
3. Ecological Research
4. Environmental Monitoring
5. Socioeconomic Environmental Studies
6. Scientific and Technical Assessment Reports (STAR)
7. Interagency Energy-Environment Research and Development
8. "Special" Reports
9. Miscellaneous Reports
This report has been assigned to the ENVIRONMENTAL PROTECTION TECH-
NOLOGY series. This series describes research performed to develop and dem-
onstrate instrumentation, equipment, and methodology to repair or prevent en-
vironmental degradation from point and non-point sources of pollution. This work
provides the new or improved technology required for the control and treatment
of pollution sources to meet environmental quality standards.
This document is available to the public through the National Technical Informa-
tion Service. Springfield, Virginia 22161.
-------
EPA-600/2-80-064
April 1980
SOLVENT EXTRACTION OF WASTEWATERS FROM
ACETIC-ACID MANUFACTURE
by
N. Lawrence Ricker & C. Judson King
Department of Chemical Engineering
University of California
Berkeley, California 94720
EPA Grant No. R803773
Project Officer
John Matthews
Source Management Branch
Robert S. Kerr Environmental Research Laboratory
Ada, Oklahoma 74820
ROBERT S. KERR ENVIRONMENTAL RESEARCH LABORATORY
OFFICE OF RESEARCH AND DEVELOPMENT
U. S. ENVIRONMENTAL PROTECTION AGENCY
ADA, OKLAHOMA 74820
-------
DISCLAIMER
This report has been reviewed by the Robert S. Kerr Environ-
mental Research Laboratory, U.S. Environmental Protection Agency,
and approved for publication. Approval does not signify that the
contents necessarily reflect the views and policies of the U.S.
Environmental Protection Agency, nor does mention of trade names
or commercial products constitute endorsement or recommendation
for use.
-------
FOREWORD
The Environmental Protection Agency was established to coordinate admin-
istration of the major Federal programs designed to protect the quality of our
environment.
An important part of the Agency's effort involves the search for infor-
mation about environmental problems, management techniques and new technologies
through which optimum use of the nation's land and water resources can be
assured and the threat pollution poses to the welfare of the American people
can be minimized.
EPA's Office of Research and Development conducts this search through a
nationwide network of research facilities.
As one of these facilities, the Robert S. Kerr Environmental Research
Laboratory is responsible for the management of programs to: (a) investigate
the nature, transport, fate and management of pollutants in ground water;
(b) develop and demonstrate methods for treating wastewaters with soil and
other natural systems; (c) develop and demonstrate pollution control tech-
nologies for irrigation return flows; (d) develop and demonstrate pollution
control technologies for animal production wastes; (e) develop and demonstrate
technologies to prevent, control, or abate pollution from the petroleum re-
fining and petrochemical industries; and (f) develop and demonstrate technolo-
gies to manage pollution resulting from combinations of industrial wastewaters
or industrial/municipal wastewaters.
Industrial process wastewater streams frequently contain appreciable
concentrations of raw materials, side reaction, compounds, and process
products. Those compounds which are marketable represent an attractive area
for recovery/recycle/reuse research since their separation from a waste stream
reduces the waste treatment loading concurrently with offering a means of
capital investment recovery on separation equipment. Solvent extraction
technology has not been applied to waste stream cleanup as vigorously as it
has to manufacturing operations. This report contains the findings of study
applying solvent extraction technology in the removal of product losses from
acetic acid manufacturing wastewater streams. A cost estimate prepared from
bench scale data indicates a possible 244% per year return on investment
before taxes; this is an attractive alternative to wastewater treatment costs.
O. +~
W. C. Galegar
Director
Robert S. Kerr Environmental Research Laboratory
-------
ABSTRACT
Wastewater samples were obtained from two major processes
for the manufacture of acetic acid. The chemical oxygen demand
(COD) of these wastewaters varied from about 1000 parts per
million (ppm) to over 410,000 ppm. The acetic acid was generally
the major organic chemical in the samples; its concentration
varied from about 30 ppm to nearly 18 weight percent (wt:%).
Solvent extraction was evaluated as a potential treatment
method for such wastewaters. Several possible goals for an ex-
traction process were considered: the recovery of chemical
values, a large COD reduction, i.e., for pollution abatement or
as a pretreatment for a biological oxidation system, and/or the
removal of toxic or refractory chemicals. For the wastewater
samples studied, extraction appeared to be too expensive to be
practical unless recovery of a marketable chemical were possible.
Acetic acid was the only chemical in the wastewater having a
market potential.
Many organic chemicals were screened for use as extractants
for acetic acid. Long-chain, tertiary alkyl amines, dissolved
in organic diluents, appeared to be the most promising extract-
ants, except for certain wastewaters containing chlorinated
acetaldehydes. Amine extractants were studied extensively in
small-scale experiments to determine phase equilibria, extract-
ant regenerability, mass-transfer characteristics, and emulsifi-
cation tendencies. The choice of the organic diluent had a
large effect on behavior of the extractant; Cg ketones were
found to be attractive.
A cost estimate was prepared for an extraction process to
recover acetic acid from a 22,700 kilogram/hour (kg/h) (100 gals/
minute (GPM)) wastewater containing 5 wt. % acid. Estimated
direct-fixed-capital was $1,030,000, with an annual operating
cost of $253,000/year ($5.90/1000 gal), resulting in a return on
investment before taxes (ROIBT) of 244% per year. The ROIBT for
a 1 wt. % acetic acid wastewater was only about 30%; however,
this might be increased by further optimization of the amine/
diluent combination.
This report was submitted in fulfillment of Grant No. R803773
by the University of California under the sponsorship of the U.S.
Environmental Protection Agency and covers a period from 6-1-75
to 9-30-78, and work was completed as of 9-30-78.
iv
-------
CONTENTS
Foreword iii
Abstract iv
Figures vii
Tables viii
Acknowledgment x
1. Introduction 1
General classifications of Western
treatment processes. . 1
Steam-stripping processes 2
Solvent-extraction processes 5
Comparison of steam-stripping and solvent-
extraction processes 7
General characteristics of wastewaters from
acetic-acid manufacture 11
2. Conclusions and Recommendations 13
3. Quantitative Analysis of Wastewaters from the
Manufacture of Acetic Acid 15
Methods of wastewater analysis 17
Wastewater composition 27
4. General Process-design Strategy 28
Extraction for profit 28
Extraction for removal of hazardous chemicals. . 33
Extraction for waste COD reduction 34
Summary of treatment strategy—goals for
the extraction process 34
5. Solvents for Extraction of Acetic Acid and
Chlorinated Acetaldehydes 35
Desirable solvent properties 35
Weakly-interacting acetic-acid extractants ... 36
Organophosphorus extractants 37
Amine extractants 39
Liquid-liquid equilibria for amine extractants . 41
Effect of diluent type, amine concentration
and acid concentration of Kn 50
Extraction equilibria for Alamine 336 in
ketone diluents 52
Prediction of the acetic acid distribution
coefficient for an amine extractant 57
Extractants for chlorinated acetaldehyde.s. ... 60
Solvent-regeneration studies 63
Heat stability of amine extractants 65
-------
CONTENTS (continued)
6. Experimental Extraction Results 68
RDC extraction runs 68
Countercurrent separatory funnel system 68
7. Extraction Processes for Acetic Acid Recovery. . • • 109
Solvent volatility considerations 109
Solvent K- and selectivity 112
Solvents selected for detailed process-
design study 113
Cost estimates 114
Selection of a diluent for the Alamine 336
extractant 127
Optimization of the Alamine 336/DIBK
extraction system 128
Alternatives for regeneration of amine
extractant 129
Loss of solvent—cost and prevention 131
Solvent purification—removal of non-volatile
solutes 135
Summary of recommended extraction process
for recovery of acetic acid 136
8. Treatment of Wacker-process Wastewaters 141
Reactivity of Chlorinated Acetaldehydes 141
Process-design considerations for Wacker
wastewaters 142
Pre-treatment to remove chlorinated
acetaldehydes 143
Dual-solvent extraction 144
Recommendations for treatment of Wacker-
process wastewaters 146
9. Summary of Findings 148
References 152
VI
-------
FIGURES
Number Page
1 Recovery of butyl acetate from water by
steam-stripping 4
2 Schematic of phenosolvan process 6
3 Volatile-solvent extraction process 9
4 Dual-solvent extraction process 10
5 Extraction of acetic acid by di-tridecyl
amine in various diluents 47
6 Extraction of acetic acid by Alamine 336 in
various diluents 53
7 Extraction equilibria for 50 vol % Alamine 336
ketone diluents 54
8 Extraction equilibria for 50 vol. % Alamine 336 in
DIBK, and in 3 parts DIBK/2 parts polar additive. 56
9 Extraction of acetic acid from wastewater H by
Alamine 336 in DIBK 58
10 Water-content of organic phase for extraction
of wastewater H by Alamine 336/DIBK 59
11 Extraction of acetic acid with a lower-boiling
solvent 110
12 Extraction of acetic acid with a higher-boiling
solvent Ill
13 ROIBT for acetic acid recovery vs. concentration
of acid in wastewater feed 124
14 Chemical regeneration of amine extractant by
caustic-wash process 130
15 Flowsheet of recommended acetic-acid recovery
process 138
16 Conceptual design of dual-solvent extraction
process for wacker-process wastewaters 145
-------
TABLES
Number
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
Sources of Wastewater Samples
Composition of Wastewater Sample A
Composition of Wastewater Sample B
Composition of Wastewater Sample C
Composition of Wastewater Sample D
Composition of Wastewater Sample E
Composition of Wastewater Sample F
Composition of Wastewater Sample G
Composition of Wastewater Sample H
Chemical Value of Wastewater Constituents ....
Estimated ROIBT for Recovery of Acetic Acid
from Samples E-H
Amines Investigated in This Work
Results for Extraction of Acetic Acid by Tertiary
Amines Supplied by Ashland Chemical Company. .
Representative Results for Extraction of Acetic
Acid by Alamine 336
Extraction of Acetic Acid by Asymmetric Tertiary
Amines
Extraction of Acetic Acid by Several Secondary
Amines
Extraction of Chlorinated Acetaldehydes by
Several Common Solvents
Conditions and Results for Mini-plant Extraction
Run
Run
Run
Run
Run
1,
2,
3.
4.
5.
Run 6.
Run 7.
Run 8.
Run 9.
Run 10
Run 11
Run 12
Run 13
Run 14
Run 15
Page
. 16
, 19
. 20
. 21
. 22
. 23
. 24
. 25
. 26
. 29
. 32
. 42
. 43
. 44
. 45
. 46
. 61
. 69
. 70
. 72
. 74
. 76
. 77
. 79
. 81
. 83
. 84
. 87
. 90
. 93
. 96
. 99
Vlll
-------
TABLES (continued)
Number Page
33 Conditions and Results for Mini-plant Extraction
Run 16 101
34 Run 17 103
35 Run 18 105
36 Run 19 107
37 Key Solvent Properties Used in Design Study .... 117
38 Estimated Cost of Extraction of Acetic Acid by
Ethyl Acetate 119
39 Estimated Cost for Extraction of Acetic Acid by
Cyclohexanone 120
40 Estimated Cost for Extraction of Acetic Acid by
40% TOPO in 2-hrpysnone 121
41 Estimated Cost for Extraction of Acetic Acid by
50% Alamine 336 in DISK 122
42 Cost Estimate for DIBK/Alamine 336 Purification—
Batch System 137
IX
-------
ACKNOWLEDGMENT
This report was prepared by the two authors. However,
many others contributed in important ways to this project.
Gilbert W. Sanchez and Edward E. Pittman were research
assistants with the project and completed M.S. theses closely
related to it. Much of their work is represented in this
report. Professor John M. Prausnitz served as co-investi-
gator and provided insight into the thermodynamics of solvent
selection and performance. Professors F. Jensen, C. Heatcock
and J. Hildebrand of the University of California, Department
of Chemistry also advised on solvent possibilities. Nam-Sum
Wang provided considerable help as a laboratory assistant.
Professor F. Nakashio of Kyushu University, Japan, conducted
several studies associated with this project during a sab-
batical leave at Berkeley. Reiko Kubota and Marie Devlin
handled the typing of the report efficiently.
Messrs. Jack Hale, Frank Mayhue, John Matthews and
Leon Myers of the Robert S. Kerr Environmental Research
Laboratory, U.S. Environmental Progection Agency, provided
effective liaison and participated in several useful
discussions.
-------
SECTION 1
INTRODUCTION
It is now widely recognized that water pollution is a
serious industrial problem. Solvent extraction, a well-
established unit operation in the chemical industry, is be-
coming increasingly important as a treatment method for certain
types of industrial wastewaters, primarily those containing or-
ganic chemicals.
This report describes research on the use of solvent ex-
traction for the removal and recovery of organic chemicals from
wastewaters. Although the results of the study have consider-
able generality, the specific problem considered is the develop-
ment of an extraction process to treat acetic-acid-manufacturing
wastewaters.
GENERAL CLASSIFICATIONS OF WASTEWATER TREATMENT PROCESSES
Wastewater treatment processes can be divided into two
general classes, depending on what is eventually done with
the pollutant(s). Non-recovery processes, such as biological
oxidation, destroy the pollutant during treatment. Biological
oxidation is the major technique for treatment of industrial
wastewaters. It is an especially good method for dilute wastes
that contain a wide range of organic contaminants. However,
it is less effective for more concentrated industrial wastes,
or for a waste stream containing a pollutant that is either
slowly or non-biodegradable, or toxic to the bacteria in the
process.
Thermally-regenerated, activated-carbon adsorption is
another example of a non-recovery method suitable for dilute
wastewaters. Incineration is sometimes used for more concen-
trated wastewaters, and in some areas of the United States
wastewaters not amenable to other treatment methods are in-
jected into deep wells in stable geological formation.
Recovery methods, on the other hand, recover one or more
wastewater constituents. The recovered chemical value then
helps to offset the cost of treatment, or in some cases, can
even provide a net profit.
-------
Since the recovery methods are typically more complex and
require a higher capital investment, a wastewater must usually
contain significant, marketable chemical value to justify re-
covery. However, a recovery method might also be chosen when
non-recovery methods are unsuitable for a particular wastewater.
For example, a wastewater might contain a highly toxic or re-
fractory chemical, such as crotonaldehyde, that could be removed
most economically by a recovery method.
Many well-established separation techniques are used
industrially as recovery methods for wastewater treatment.
Activated carbon adsorption with a recovery-type carbon-
regeneration method, ion exchange, bubble flotation, steam
stripping, and solvent extraction are typical examples (Nathan,
1978; Fox 1973; Cooper, 1966). For a given treatment problem,
the physical and chemical characteristics of the pollutant and
its concentration level in the wastewater determine the most
appropriate recovery method. Activated-carbon adsorption and ion
exchange are used primarily in fixed-bed processes for dilute
wastewaters. These processes are usually too expensive to use
for concentrated wastewaters because the high bed loadings
involved in such cases would require either very large beds
(high capital investment) or frequent regeneration (high
operating costs).
On the other hand, stripping and extraction tend to be
relatively insensitive to pollutant concentration. In
extraction, for example, the solvent circulation rate, which
has a large influence on equipment size and operating costs, de-
pends mainly on the equilibrium distribution coefficient (Kn) of
the principal pollutants at equilibrium, with K being defined as
the weight fraction of pollutant in the solvent phase divided
by that in the aqueous phase. Since K_ is usually not a strong
function of pollutant concentration, at the concentrations
usually found in wastewaters, the process cost per volume of
wastewater is roughly constant.
STEAM-STRIPPING PROCESSES
Steam-stripping is probably the most widely used
recovery process for organic pollutants, and as such is
often the base case in a comparison of process alternatives
for a particular industrial waste-treatment problem. Although
solvent extraction treatment is the main concern
of this work, some understanding of steam-stripping is
important, both for proper perspective and because steam-
stripping is occasionally used as one of the steps in
solvent-extraction treatment.
Steam-stripping exploits the liquid-phase non-idealities
-------
found in aqueous solutions of organic chemicals. A typical
process schematic appears in Figure 1. This example
process recovers butyl acetate, a common industrial solvent,
which has a pure-component boiling point of 126.5°C at
atmospheric pressure. In dilute aqueous solution, however, butyl
acetate is actually more volatile than water (the relative
volatility is about 390) ; hence, butyl acetate is the
overhead product from the stripper. Energy consumption is, of
course, far lower than would be the case if water were
the overhead product. The water/butyl acetate system also
exhibits a heterogeneous azeotrope; even though
considerable water goes overhead with the butyl acetate,
upon condensation the overhead forms two liquid phases, and
simple phase-separation results in a 99 wt. % butyl
acetate product.
The wide appeal of steam-stripping is mainly due to
its simplicity and low capital investment relative to
other recovery methods. Design is straightforward and
usually does not require much experimental development. The
major design uncertainty is often the stage efficiency (or
its equivalent in a packed-tower design), which can be very
low, on the order of 20% for systems involving pollutants
with volatilities much different from that of water. Of
course, if the relative volatility is large, the stage
requirements are very small, and the stage efficiency does
not affect the economics greatly.
As with any distillation-type process, steam-stripping
is a high energy-user; this is probably its major
disadvantage. Energy is needed to raise the feed water to
the saturation temperature at the column pressure and then
to vaporize the pollutant (and often a considerable amount
of water) to form the overhead product. Feed-bottoms and
condensate heat-exchange can reduce energy costs at the
expense of increased complexity and capital investment.
An alternative is to operate the stripping column under
vacuum. Vacuum operations can reduce energy consumption and
yet avoid extra heat-exchange equipment (Rasquin, et al.,
1978). The resulting lower operating temperature reduces
the energy needed to raise the feed to the saturation
point and often allows the use of waste steam, rather
than more expensive high-pressure steam. Moreover, the
lower temperature usually increases the relative volatility
of the pollutant, which makes the separation easier, and it
tends to reduce the scaling and corrosion problems that can
occur with some wastewaters. The cost of the vacuum system
will offset these advantages to some extent.
-------
Condenser
• Waste water
Steam
1
BuAc
Decanter
Stripping
Column
Heat exchangers f
Aqueous,
Prod:
(Waste)
Figure 1. Recovery of butyl acetate from water by steam
stripping.
-------
Vacuum operation is especially attractive when the
concentration of the pollutant to be recovered is low, which
is usually the case for wastewaters. Then the vapor loading
is small, and the column diameter is not much larger than
would be the case for atmospheric pressure operation.
SOLVENT-EXTRACTION PROCESSES
As a recovery method for wastewater treatment, solvent-
extraction is not as widely used as steam-stripping, but
extraction does have important industrial applications.
One of the best known of these is the "Phenolsolvan" process
which was developed in Germany to treat coke-oven condensate
wastewaters, containing high concentrations of phenolics
(Wurm, 1968). The Phenosolvan process is worthy of some
discussion since it is typical of solvent-extraction
processes for wastewater treatment. It includes the usual
three major processing steps: extraction (and subsequent
phase separation); solvent recovery from the wastewater; and
solvent regeneration (Figure 2). In the extraction step,
a solvent with an affinity for the pollutant(s) to be
recovered contacts the wastewater in an extraction device.
The Phenolsolvan process uses di-isopropyl ether (DIPE) as
the solvent, and the extractor is a mixer-settler cascade.
There are many types of continuous-extraction devices
(Treybal, 1963). "Stagewise" contactors, such as the
mixer-settler, attempt to provide a single equilibrium
stage within each unit. These units can be connected in a
counter-current cascade if the separation requires more than
one stage. A mixer-settler consists of a mixing chamber,
usually agitated, followed by a phase-separation device.
Mixer-settlers are reliable, flexible and efficient, in that
they often approach the operation of an ideal, equilibrium
stage. Their main disadvantage is in separations requiring
many equilibrium stages, when complexity, floor space, and
capital investment requirements can be prohibitive.
"Differential" contactors are usually columns in which the
solvent and water phases flow counter-currently by gravity.
The columns may be packed, spray, sieve-tray, rotating disc
contactor (RDC) or pulsed. The most efficient differential
extractors are agitated in some manner to enhance inter-phase
mass transfer. These are advantageous for separations re-
quiring the equivalent of a large number of equilibrium stages,
Solvents for wastewater treatment often have an
appreciable water solubility (5,900 ppm for DIPE), and
economics and pollution-control considerations usually
dictate some form of solvent recovery. The Phenolsolvan
process solvent-recovery system uses a recirculating, inert
-------
Feed
=s
Water
fi.
A
MM
fr
A
Or
— i
l
1
rh
B
1
I
1
1
N
LT' f
1
f~
—
D
— if
s-U
i
T2
'J
•*- \
T^
A
p
Liin*.
-s
•IV «•
1
i 1
J" f =
i !
I i
! !
i t
==*•
_jj j
f
1 1 Miser- H !
Seffier
c
|| Purified
Watei
Equipment
A,B,C, &
Tl & T2 -
S - Steam
Flow Le
Recovered '
B* o^enol
Water
Isopropyl
Ether
Phenol
Stripping
Gas
Ficrure 2. Schematic of phenosolvan process.
-------
gas in a series of stripping and absorption columns (labelled
A, B, and C in Figure 2). Steam-stripping is an alternative
solvent-recovery method; another is a separate extraction
with a water-insoluble solvent to recover the primary solvents
(Earhart, et al., 1977).
For solvent regeneration, the Phenolsolvan process uses
distillation (column D). Regenerated DIPE is the overhead
product and recovered phenolics are the bottom product.
Distillation is the most common regeneration method,
although other methods are possible, e.g., back-extraction
of phenolics with concentrated aqueous sodium hydroxide (NaOH)
solution.
It is interesting to note that less energy would be
consumed in the regeneration step if a high-boiling solvent
were used and the pollutants were the overhead product.
The Phenolsolvan process uses DIPE rather than a high-
boiling solvent because coke-oven condensates contain
emulsified, non-volatile organics and tar-like substances
that would tend to build up in a high-boiling solvent and
would eventually degrade solvent performance. This is a
common problem in wastewater treatment and encourages the
use of volatile solvents.
COMPARISON OF STEAM-STRIPPING AND SOLVENT-EXTRACTION PROCESSES
The Phenolsolvan process is illustrative of both the
drawbacks and advantages of solvent extraction as a wastewater
treatment method. The major drawback is complexity. The
extraction step -is only a small part of a typical extraction
process. A good deal of the capital investment and nearly
all of the energy requirements are in the regeneration
step. This inherent complexity makes extraction relatively
difficult to design and operate, and it results in a
capital-intensive process.
Moreover, the fundamental phenomena involved in
extraction are not really well understood, at least when
compared with distillation, and extraction processes usually
require pilot-plant development. A critical step in this
development is the selection and testing of a solvent.
Solvent selection must usually be done individually for
each waste-treatment problem, i.e., there is no "universal sol-
vent", and the best solvent for a given problem is determined
by the pollutant(s). Solvents are often expensive, and the
process must be carefully designed to avoid any appreciable
loss of solvent.
These factors tend to complicate the use of extraction
-------
for wastewater treatment. One notable attempt to design a
simpler, generalized extraction process that could be used in
a wide range of wastewater treatment problems is the work of
Earhart, et al. (1977) . These investigations show that a
volatile C-4 hydrocarbon such as isobutane or isobutylene can
be a good extractant for many non-polar and slightly polar or-
ganic pollutants, and that the use of this solvent eliminates
the need for solvent recovery from the wastewater (Figure 3).
However, this "volatile-solvent extraction" (VSE)
process is not suitable for polar pollutants
like phenol. For this type of problem, Earhart, et al.
suggest a "dual-solvent extraction" process in which a
polar solvent, such as DIPE, extracts the polar pollutant
from the water, and then VSE recovers the polar solvent
from the water (Figure 4). While VSE may indeed be a good
solvent-recovery method, the dual-solvent process does not
reduce process complexity, and the choice of a polar solvent
is still a problem, in general.
Steam stripping, on the other hand, is simpler, and a
single conceptual design applies to most wastewater-
treatment problems. Consequently, extraction is not
competitive with stripping unless the separation factor for
the pollutant-water separation is poor for stripping and/or
very favorable for extraction. For many organic pollutants,
the separation factor for stripping, i.e., the pollutant-
water relative volatility, is high; and while the
corresponding K value is likely to be high as well, it is
usually not hign enough to justify the complexity and larger
capital investment of an extraction process.
Earhart, et al. (1976) present a detailed economic
analysis for the recovery of several representative organic
pollutants from water by extraction and by steam-stripping
at atmospheric pressure. For butyl-acetate recovery, for
example, they estimate a treatment cost of $0.16/cubic meter
(m3) ($0.61/1000 gal) wastewater with a capital investment of
$41,700 for stripping and $0.19/m ($0.71/1000 gal) and $78,600
capital for extraction. The operating costs are comparable,
but the capital investment for extraction is nearly double
that for stripping. A vacuum-steam-stripping process would
probably be even cheaper than the atmospheric-pressure
stripper. In any case, some kind of stripping process
would almost certainly be selected for recovery of butyl
acetate, and these relative costs are typical of the results
for many organic pollutants. While it is dangerous to
generalize, it seems likely that stripping will continue
to be the dominant recovery method for organic pollutants.
-------
Waste,
Water
o
w
O)
c
a>
o>
w
a:
Loaded
Isobutylene
Counter-
current
Extractor
Direct-
contact
Condenser
Settling
Tank
Side-stream
Re boiler
Recovered
Pollutants
Isobutylene Vapor
Reflux
Isobutylene
Vent
' Isobutylene
Holding
Tank
Make-up
Isobutylene
Aqueous
Effluent
Figure 3* Volatile-solvent extraction process,
-------
Waste
Water
Loaded P.S.
Polar
Solvent
Extractor
Volatile
Solvent
Extractor
P.S.-Pollutants
Splitter
Pollutants
Recycle P.S.
V.S. - P.S.
Splitter
Recycle V.S.
— —*" V.S. Vapor
Holding
Tank
.Purified
Water
Figure 4. Dual-solvent extraction process,
10
-------
However, there is a significant number of organic
pollutants for which stripping is not attractive. Phenol,
for example, forms a homogeneous azeotrope with water at
only 9.2 wt. % phenol, so a simple stripping process
cannot recover pure phenol from a dilute, i.e., below 9.2 wt. %,
wastewater. The Phenolsolvan process, on the other hand, is
viable because its extraction step circumvents the azeotrope,
and the added phenol-DIPE and DIPE-water separations are
not difficult.
Other potentially attractive applications for extraction
exist when a pollutant (in aqueous solution or as dispersed
droplets) is less volatile than water, or when the
separation factor for stripping is close to 1.0. Energy
and capital requirements for stripping are prohibitively
large in such cases, and if a good extractant can be found,
extraction should be favored.
GENERAL CHARACTERISTICS OF WASTEWATERS FROM ACETIC-ACID
MANUFACTURE
The wastewaters studied in this work were from the
manufacture of acetic acid by two different processes. One
of these is the liquid-phase oxidation of n-butane, which
accounted for 50% of U.S. production in 1974 (Lowry and
Aguilo, 1974). The other is a two-step process—the
oxidation of ethylene to acetaldehyde (Wacker process),
followed by the oxidation of acetaldehyde to acetic
acid—that accounted for another 35% of U.S. acetic-acid
production. The carbon monoxide/methanol process has a
smaller percentage of the market but appears to be the
process of the future; however, wastewater samples
were not available.
Samples representative of each process wastewater,
including both steps of the ethylene-oxidation route, were
obtained and analyzed by gas chromatography. The results
showed that these wastewaters contain at least five organic
chemicals for which recovery by extraction should be favored
over stripping. One is acetic acid, which appears in concen-
trations of less than 1000 ppm to over 5 wt. %, depending on
the water source. At concentrations of approximately 1 wt. %
and greater, the chemical value of the acetic acid is high
enough to encourage recovery of the acid. Since acetic acid is
less volatile than water and the acid-water relative volatility
is close to 1.0, acetic-acid recovery by a simple
distillation-type process is very costly, especially at
concentrations below 2 wt. % (Eaglesfield, et al., 1953; Brown,
1963). Fixed-bed, anion exchange has been suggested for
dilute solutions but has several drawbacks.
In industry, the separation is now usually
11
-------
accomplished by extraction or by extractive distillation
(Brown, 1963) . However, industrial extraction processes
have been designed mainly for recovery from high-concentration
aqueous solution, i.e., greater than 5 wt. % acetic acid. A
major goal of the present work has been the development of
a process that is economical at the lower concentrations
found in wastewaters.
Other chemicals of interest are mono-, di-, and tri-
chloroacetaldehyde (chloral) and crotonaldehyde. These are
present in the Wacker-process wastewaters at concentrations
up to about 8,000 ppm. The market for these chemicals is
probably not large enough to justify recovery. However,
chlorinated organics are potential carcinogens and are difficult
to degrade biologically (Nathan, 1978), and crotonaldehyde is a
notorious poison to biological oxidation. There is an incentive
to develop alternative treatment processes, such as extraction,
to remove such chemicals from water.
The chlorinated acetaldehydes form stable hydrates in
aqueous solution (Miller, 1969) so they are difficult to
recover by stripping. Extraction is more promising.
Earhart, et al. (1977), for example, report K values
of 50-120 for extraction of chloral by 2-ethyIhexanol.
Crotonaldehyde can be removed by stripping, but it could also
be co-extracted with the chlorinated acetaldehydes with little,
if any, additional cost. Thus, another major goal of the
present work has been to develop an extraction process to re-
move chlorinated acetaldehydes and chloral from the Wacker-
process wastewaters.
12
-------
SECTION 2
CONCLUSIONS AND RECOMMENDATIONS
Several wastewater streams from acetic-acid manu-
facturing plants contain sufficient acetic acid to make
recovery attractive. The principal role of solvent
extraction appears to be for recovery, rather than for broad-
brush reduction of COD and biochemical oxygen demand (BOD), or
for removal of toxic components.
Projected costs fo^ recovery of acetic acid by ex-
traction are about $4/m of wa^er for use of ethyl
acetate as solvent and $2.50/m for use of cyclohexanone
as solvent. Organic bases are attractive alternatives
to these physically interacting solvents. The projected
cost for a solvent composed of 40% trioctyl phosphine
oxide (TOPO) in 2-heptanone is about $2.80/m of water,
while that for 50% Alamine 336 (Cg-C.,Q tertiary amine)
in diisobutyl ketone (DIBK) is about 52.00/m of water.
The organic bases provide higher equilibrium distri-
bution coefficients (K ) than the physically interacting
solvents, with K being a maximum at some intermediate
composition of tne organic base extractant in a diluent. The
K_ values are highly sensitive to the diluent employed in
tne amine systems. Alcohols provide the best K values,
but were found to react with extracted acid through ester-
ification during regeneration by distillation. Ketones
were therefore selected. A Cg ketone appears to offer
the best compromise between high K and a volatility
sufficiently below that of acetic acid during solvent
regeneration.
Tertiary amine solvent systems should also be
attractive for acetic acid/water separations at higher
feed concentrations.
Secondary amine systems give higher values of KQ
than the tertiary amines, but there is solvent loss
from amide formation during regeneration. The tertiary
amines cannot form amides. Primary amines are too
miscible with water to be effective.
13
-------
Wacker process effluent waters contain chlorinated
acetaldehydes, which are corrosive, toxic and generally
difficult to handle. It appears that these chlorinated
acetaldehydes react irreversibly with amine solvents,
making extraction with amines unsuitable for these waters
unless the chlorinated acetaldehydes are somehow removed
first. Such waters are often disposed of by deep-well in-
jection or even incineration, at present.
Important questions for further exploration include
(1) the longevity of amine solvents under sustained usage,
(2) the reactivity of TOPO with chlorinated aldehydes,
(3) ways of overcoming entrainment problems with amine
solvents and/or TOPO, (4) the need for continual purge and
possible reprocessing of a fraction of the solvent system
to prevent build-up of heavy tars, and (5) the use of
additives to increase K in the Alamine 336/DIBK system
at low acetic acid concentrations.
14
-------
SECTION 3
QUANTITATIVE ANALYSIS OF WASTEWATERS FROM THE MANUFACTURE
OF ACETIC ACID
Several large acetic-acid^manufacturers provided a
total of eight, 55-gal (0.21-m ) wastewater samples.
The process source of each wastewater sample, designated
A to H, is given in Table 1. The samples represent wastewaters
from two major processes for the manufacture of acetic
acid. The samples are from six different plants; samples B and
G are from different parts of the same plant, and samples D and
H are from different parts of another plant.
In the initial stages of the project, the samples
were stored in a room maintained at about 4°C to reduce
the likelihood of degradation. Small portions of
the samples were also kept at ambient temperature in glass
containers. Repeated analysis over a period of several
months showed that the ambient-temperature samples were
as stable as the 4°C samples. However, the ester content
decreased significantly in both the chilled and the
ambient samples (Wardell, 1976). The mechanism is
apparently a slow hydrolysis to the corresponding alcohol
and carboxylic acid
ester + water —» carboxylic acid + alcohol
occurring over a period of about one month. This is an
equilibrium reaction; equilibrium constants are given by
Groggins (1958) and are on the order of 4.0 for n-acetates.
Hydrolysis of the esters is thermodynamically favored when
the ester is dilute and when the alcohol and/or acid
concentrations are small. The esters are completely
hydrolyzed in certain samples (A,B) and partially
hydrolyzed in the more concentrated samples (D,H).
The ester contents of the wastewaters were small, and
esters are relatively easy to extract; the observed changes
in concentration do not alter the essential characteristics
of the wastewaters. The concentration of acetic acid,
which was usually the primary solute in each sample, did
not appear to change once the esters had hydrolyzed completely.
Reported compositions are the final stable values.
15
-------
TABLE 1. SOURCES OF WASTEWATER SAMPLES
Sample Plant code
designation Process number
A
B
C
D
E
F
G
H
butane-oxidation
butane-oxidation
acetaldehyde-oxidation
acetaldehyde-oxidation
ethylene-oxidation (Wacker)
ethylene-oxidation (Wacker)
butane-oxidation
acetaldehyde-oxidation
1
2
3
4
5
6
2
4
Note: Ethylene-oxidation followed by acetaldehyde-
oxidation is a two-step method for manufacture of
acetic acid.
16
-------
METHODS OF WASTEWATER ANALYSIS
Three methods were used for wastewater analysis: gas
chromatography (GC), acid-base titration, and COD measurements.
The analytical techniques are summarized here; detailed dis-
cussions have been previously reported. (Wardell, 1976;
Michaels, 1977; Wardell and King, 1978; Pittman, 1979).
The GC analyses employed a Varian Model 600-D flame-
ionization chromatograph equipped with 3.175 millimeter outer
diameter (mm OD) , stainless steel columns. Poropak Q, QS, and
T column packings were used for wastewater analysis since these
greatly reduce the peak tailing that often occurs in GC
analysis of aqueous samples.
The wastewater samples were first analyzed
qualitatively to determine the identities of the wastewater
components. Their relative retention times on
chromatographic columns of differing polarity (Poropak Q vs
Poropak T), indicated the class (carboxylic acid, alcohol,
ketone) of each compound in the sample (Wardell, 1976). The
retention times also indicated the position of the compound
within the homologous series for its class. To confirm the
identity of a peak in the chromatogram, a small amount of the
pure, suspect compound was added to the sample. If this re-
sulted in an increase in the size of the appropriate peak,
confirmation was assumed.
This is not a completely fool-proof identification
method, since different compounds can have the same
retention times on a given chromatographic column.
However, the chances of two compounds having the same
retention times on both the Q and T columns are very
small, especially for the low-molecular-weight compounds
found in these wastewaters. One sample (G) was analyzed
by GC and then by GC-mass spectrophotometry, performed by
Dr. A. Newton of the Lawrence Berkeley Laboratory. The
mass spectrometer is accepted as a positive identification
method, and it confirmed the identifications hypothesized
in the original GC analysis. It also identified
acetonitrile, which had previously been an unknown peak
in the chromatogram for sample G.
The GC was also the primary method used for
quantitative analysis, as described by Wardell (1976) and
Michaels (1977) . A two-meter, Poropak Q column operating
at approximately 150°C provided adequate peak-resolution
in most cases. Samples were injected manually into a
heated injector port. The sample size was typically
3-5microliters (uD for dilute samples, e.g., sample C, and
0.5 uL for the more concentrated samples.
17
-------
Several standard analytical methods for the
determination of solute concentrations by GC were tested
throughout the course of this work. The internal-standard
method seemed to give the best results. The sample to be
injected is first spiked with a measured amount of a
standard compound, which is chosen such that it does not
interfere with any of the solute peaks. Methyl ethyl
ketone (MEK) or 2-pentanol were usually used as internal
standards. The sample is injected, and the solute peak
areas, measured by mechanical integration, are compared
to the area of the standard peak. The weight fraction of
each solute can then be calculated directly from these
relative peak areas and known relative response factors
for the flame ionization (FI) detector (Bonelli and McNair,
1968). This method does not require accurate injection volumes,
as does the external-standard method.
Samples were usually injected at least three times in
succession. Reproducibility was typically ±3%, except for
solutes present at less than about 100 ppm and for the
one-carbon oxygenated organics (formic acid, formaldehyde,
methanol) which respond poorly to the FI detection system.
The estimated accuracy of the GC measurements is indicated for
each sample in Tables 2-9.
COD measurements were performed by oxidation of organic
and inorganic solutes with potassium dichromate in a 50% (by
volume) sulfuric acid solution (EPA, 1971). At least two COD
measurements were made for each wastewater sample.
Reproducibility within 10% was considered to be acceptable.
A theoretical oxygen demand (TOD) was calculated from
the GC analysis as the amount of oxygen required to
oxidize the identified solutes completely to carbon
dioxide, water, chlorine and nitrogen dioxide. TOD and COD
values were then compared. Reasonable agreement was taken to
indicate that the GC analysis had correctly quantified all
major pollutants. The COD method does not oxidize some organic
solutes completely, so the ratio of COD to TOD was expected
to be less than 1.0 in most cases. The results are shown in
Tables 2-9.
Potentiometric titrations were used to determined the
total acid equivalent of each sample. The titrant was a
strong base, NaOH or potassium hydroxide (KOH), and the
titration curve was determined with a Corning Model 12 pH meter
equipped with standard pH and reference electrodes. Curves
contained a single,sharp inflection at about pH 5-8. No strong
mineral acids were detected. The inflection was assumed to be
due to the combined quantities of carboxylic acids in the
18
-------
TABLE 2. COMPOSITION OF WASTEWATER SAMPLE A*
Concentration
Compound (ppm)
Carboxylic Acids
formic
acetic
propionic
Alcohols
me than ol
ethanol
i-propanol
n-propanol
2-butanol
1-butanol
Ketones
acetone
methyl ethyl ketone
Aldehydes
i-butyraldehyde
Unidentified
Total
Measured COD: 5780 ppm
COD/TOD: 1.10
920
805
75
90
560
180
100
310
30
25
320
55
50
3520
Est . uncertainty TOD
(ppm) (ppm)
40 320
25 860
5 115
5 135
15 1165
5 430
5 240
10 805
4 80
4 55
10 780
4 130
125
5240
* Source: butane-oxidation, plant number 1
19
-------
TABLE 3. COMPOSITION OF WASTEWATER SAMPLE B*
Concentration
Compound (ppm)
Carboxylic Acids
formic 1340
acetic 385
Alcohols
methanol 1600
ethanol 350
Ke tones
acetone 125
methyl ethyl ketone 490
Aldehydes
formaldehyde 2700
Unidentified 220
Total 7390
Measured COD: 8800
COD/TOD: 0.99
Est. uncertainty TOD
(ppm) (ppm)
50 465
10 410
20 2395
10 730
5 275
15 1195
300 2875
550
8895
* Source: butane-oxidation, plant number 2
20
-------
TABLE 4. COMPOSITION OF WASTEWATER SAMPLE C*
Concentration Est. uncertainty TOD
Compound (ppm) (ppm) (ppm)
Carboxylic Acids
formic 175 15 60
acetic 30 5 30
Alcohols
ethanol 110 5 230
propanol 150 5 360
Aldehydes
formaldehyde 370 40 395
Total 835 - 1075
Measured COD: 1055
COD/TOD: 0.98
* Source: acetaldehyde-oxidation, plant number 3
21
-------
TABLE 5. COMPOSITION OF WASTEWATER SAMPLE D*
Concentration
Compound (ppm)
Carboxylic Acids
formic
acetic
propionic
Alcohols
methanol
ethanol
propanol
Ketones
acetone
Aldehydes
formaldehyde
acetaldehyde
chloroacetaldehyde
chloral
crotonaldehyde
Esters
methyl formate
methyl acetate
Unidentified
Total
Measured COD: 413300
COD/TOD: 0.91
89400
177100
1200
27000
2600
910
660
13000
45000
31000
290
3100
6500
26000
1350
424900
Est. uncertainty TOD
(ppm) (ppm)
3600
3400
500
40
10
10
1300
450
500
5
60
100
260
—
31100
188800
1815
40400
5420
2180
1455
13900
81700
28400
110
7085
6930
39300
3375
452QO.O
* Source: acetaldehyde-oxidation, plant number 4
22
-------
TABLE 6. COMPOSITION OF WASTEWATER SAMPLE E*
Concentration
Compound (ppm)
Carboxylic Acids
formic 1510
acetic 7560
Aldehydes
monochloroacetaldehyde 140
dichloroacetaldehyde 4900
chloral 1400
crotonaldehyde 220
Unidentified 40
Total 15900
Measured COD (ppm) : 12100
COD/TOD: 0.95
Est . uncertainty TOD
(ppm) (ppm)
150 525
120 8155
5 130
75 2775
30 530
10 500
100
12700
* Source: ethylene-oxidation, plant number 5
23
-------
Concentration
Compound (ppm)
Carboxylic Acids
formic
acetic
Alcohols
methanol
ethanol
Aldehydes
formaldehyde
acetaldehyde
chloroacetaldehyde
dichloroacetaldehyde
chloral
crotonaldehyde
Unidentified
Total
Measured COD (ppm) : 25600
COD/TOD: 0.91
3180
15500
55
190
200
95
6300
7200
140
330
150
32800
Est. uncertainty TOD
(ppm) (ppm)
350 1110
300 16500
4 80
5 395
20 215
5 175
80 5775
100 4075
5 55
8 755
375
29500
* Source: ethylene oxidation, plant number 6
24
-------
Concentration
Compound (ppm)
Carboxylic Acids
formic
acetic
Alcohols
methanol
ethanol
2-butanol
1-butanol
Aldehydes
acetaldehyde
Ke tones
acetone
methyl ethyl ketone +
n-propanol
Miscellaneous
benzene + unknown
acetonitrile
Total
Measured COD (ppm) : 57000
COD/TOD: 1.10
3760
49360
411
502
459
411
231
1620
389
242
747
58130
JMATJiK 5AMPJ.F. Pi*
Est. uncertainty TOD
(ppm) (ppm)
2550
2470
8
27
12
14
7
40
23
20
20
1310
52500
597
1050
1190
1060
419
3570
950
63600
* Source: butane-oxidation, plant number 2
25
-------
TABLE 9. COMPOSITION OF WASTEWATER SAMPLE H*
Concentration Est. uncertainty TOD
Compound (ppm) (ppm) (ppm)
Carboxylic Acids
formic 3760 (?)
acetic 66300
Alcohols
methanol 1890
ethanol 30
n-propanol 45
Aldehydes
formaldehyde 13800
acetaldehyde 7660
Ke tones
acetone 90
Esters
methyl formate 7
methyl acetate 4390
ethyl acetate 65
Total 94280
Measured COD (ppm) : 83300
COD/TOD: 0.75
2550 1310
4600 70675
28 2831
1 62
2 108
524 14711
76 13910
5 198
1 7
344 6638
3 118
109260
* Source: acetaldehyde-oxidation, plant number 4
26
-------
sample. The pK values of these acids are too similar to allow
separate inflections. pK values for formic, acetic, and
propionic acids are 3.75,a4.76, and 4.87, respectively
(Daniels and Alberty, 1966).
The concentrations of acetic and higher carboxylic
acids were also determined by GC. The GC measurements
predicted slightly lower acidities than those measured by
titration. The differences were attributed to formic acid,
which could not be measured with precision by GC due to
its very poor detector response.
WASTEWATER COMPOSITIONS
The results of wasterwater analyses are given in Tables
2-9. Samples A-P were analyzed by Wardell (1976) and samples
G and H were analyzed by Pittman (1979). Sample H is notable
for its poor COD/TOD agreement. The COD measurements procedure
consistently gave very low values. The reason for the poor
agreement was not found, although several hypotheses were in-
vestigated. In any case, the GC analysis is conservative.
27
-------
SECTION 4
GENERAL PROCESS-DESIGN STRATEGY
Once the sample wastewater compositions were
measured, it was possible to judge whether a given
wastewater was a likely candidate for treatment by
extraction. A sample was considered promising if (1) it
contained a marketable chemical that could be extracted
for profit, (2) it contained an environmentally hazardous
chemical that could not obviously be removed more cheaply
by other treatment methods, such as biological oxidation
or adsorption on activated carbon, or (3) extraction could
substantially reduce the total organic content of the
wastewater, thereby decreasing the loading on downstream
treatment facilities.
EXTRACTION FOR PROFIT
The possible economic incentives for extraction
treatment were examined in two steps. First, the potential
recovery value of each wastewater component was calculated
from its concentration and current market value as a
technical-grade chemical. The results are given in
Table 10, with the potential recovery value given in
terms of $/m wastewater. A value of zero indicates a
chemical for which there is no established market.
According to Earhart, et al. (1977), recovery of a
wastewater constitutent is economically promising if its
potential recovery value is $0.80/m , or more. Samples A-C
are thus poor prospects. With the exception of sample D, the
only chemicals with economic recovery potential are the acetic
acid in samples E-H and the formaldehyde and acetaldehyde in
sample H. The formic acid in samples F and G is marginal.
Sample D was eventually found by the supplier to have been
collected incorrectly, and it was eliminated from further
consideration. Sample H was subsequently received as a
replacement.
Formic and acetic acids and formaldehyde and
acetaldehyde are much more difficult to extract from
water than the solutes considered by Earhart, et al.
Because of this and because of recent increases in energy
28
-------
TABLE 10. CHEMICAL VALUE OF WASTEWATER CONSTITUENTS
Acids
formic
acetic
propionic
Alcohols
methyl
ethyl
i-propyl
n-propyl
2-butanol
1-butanol
Aldehydes
formaldehyde
acetaldehyde
Market value*
<«Ag)
47
40
45
14
36
32
59
51
46
32
44
Recovery value ($/m^
A B C D E F
0.43 0.63 0.08 41.90 0.71 1.49
0.32 0.15 0.01 70.31 3.04 6.15
0.03 0.54
0.01 0.23 3.37 0.01
0.20 0.13 0.04 0.95 0.07
0.06
0.06 0.09 0.54
0.16
0.02
0.86 0.12 4.16 0.06
19.85 0.04
G H
1.76
19.60 26.32
0.06 0.27
0.18 0.01
0.03
0.23
0.19
4.41
0.10 3.38
(continued)
-------
TABLE 10. (continued)
co
o
Market value*
(«Ag)
monochloroacetaldehyde
dichloroacetaldehyde
chloral
crotonaldehyde
i-butyraldehyde
Ketones
acetone
methyl ethyl ketone
Esters
methyl formate
methyl acetate
butyl acetate
ethyl acetate
-
-
-
-
54
40
46
62
62
64
53
Recovery value ($/m3)
A B C D E F
000
000
000
000
0.03
0.01 0.05 0.26
0.15 0.23
4.02
16.06
G H
0.64 0.04
0.18
0.01
2.69
0.03
(continued)
-------
TABLE 10. (continued)
Market value* Recovery value ($/m-^)
«=/kg) ABC DBF G H
Miscellaneous
benzene
acetonitrile
29 0.07
88 0.66
OJ
Chemical Marketing Reporter, 213(21), 1978
-------
and equipment costs, it was felt that the recovery value
of the acids and aldehydes would have to be substantially
higher than $0.80/m for recovery to be economically
promising.
A subsequent cost estimate for the recovery of acetic
acid by extraction with ethyl acetate, a solvent used
industrially for this purpose (Brown, 1963; Eaglesfield,
et al., 1953), indicated that the capital investment for
a 22.7 m /hr (100 GPM) capacity would probably §e about
$1.7 million, with an operating cost of $4.08/ra wastewater.
The operating cost does not include a credit for recovered
acetic acid. These costs were used to estimate the return on
investment before taxes ROIBT of an extraction process recover-
ing 99% of the acetic acid from wastewaters E-H. The results
are shown in Table 11.
TABLE 11. ESTIMATED ROIBT FOR RECOVERY OF ACETIC ACID
FROM SAMPLES E-H.
Wastewater sample Value of acetic acid ($/m ) ROIBT (%)
E
F
G
H
3.04
6.15
19.60
26.32
- 11.0
22.1
165.8
238.0
, , , . IL ,
Most companies would require an ROIBT of at least 30%
to justify such a project on a purely economic basis.
Samples E and F are thus not particularly attractive by
these standards, but recovery from samples G and H would
seem justifiable whether or not extraction would provide
other benefits. Furthermore, identification of a more
attractive solvent could still make recovery of acetic
acid from wastewaters E and F attractive.
ROIBT calculations also indicated that selective
extraction of acetaldehyde and formaldehyde from sample H
would probably not be economical. Formaldehyde is
particularly difficult to extract. However, co-extraction
of acetic acid and acetaldehyde from sample H is attractive,
since these solutes are easy to separate from one another.
The acetaldehyde could be recycled to make additional
acetic acid.
One might also consider co-extraction of formic acid
and acetic acid in sample G. However, while a significant
32
-------
amount of formic acid is produced in the butane-oxidation
process, on the order of 10% of the acetic acid production
rate (Lowery and Aguilo, 1974), formic acid does not appear
to be a desirable by-product. For example, in one
process, quite pure formic acid is burned rather than
marketed (Anon, 1973) . The market for formic acid is
apparently much more limited than that for acetic acid
(Lawler, 1977).
Finally, although it was not possible to determine a
market price for chlorinated acetaldehydes, several
authors have pointed out that recovery of chloroacetaldehyde
from the Wacker process is potentially attractive (Lowry
and Aguilo, 1974; Miller, 1969b). This very reactive
chemical can be used for production of insecticides,
fungicides, etc. (Miller, 1969a). The next section
considers this option in more detail.
EXTRACTION FOR REMOVAL OF HAZARDOUS CHEMICALS
Each wastewater sample was also examined for chemicals
that might be unsuitable for release to the environment
due to toxicity, carcinogenicity or non-biodegradability.
Samples A, B, C, G, and H contain no such chemicals.
Samples E and F, however, contain mono-, di- and tri-
chloroacetaldehyde (chloral) and crotonaldehyde. The
organic chlorides are potential carcinogens and are
difficult to degrade biologically (Nathan, 1978).
Crotonaldehyde is known to upset biological treatment
facilities. The chlorinated acetaldehydes form hydrates
with water and are difficult to strip from aqueous
solution. Activated-carbon adsorption would probably be
expensive due to the high solute concentrations.
A possible treatment strategy would be to extract
acetic acid, chlorinated acetaldehydes and crotonaldehyde
simultaneously. Acetic acid is generally more difficult
to extract than the other solutes, and costs for
simultaneous extraction should be comparable to those for
simple acetic-acid extraction. As noted previously,
chloroacetaldehyde may be marketable. Even if it is not,
the acetic acid could be purified and sold or reused to
offset the treatment costs.
A problem with the simultaneous extraction approach
is that the extremely reactive chloroacetaldehyde solute
reacts irreversibly with amines (Michaels, 1977) , which
appear to be the most attractive extractants for acetic-
acid recovery. The reaction is an alkylation to a
quaternary salt. For example, for the reaction of
33
-------
chloroacetaldehyde and a tertiary amine is:
C1CH2CHO + R3N—*(R3N-CH2CHO)+C1~
Another approach is then to extract the chlorinated
acetaldehydes selectively, followed by recovery of the
acetic acid with another solvent, e.g., an amine. This
also may avoid some or all of the need for subsequent
separation of the extracted chemicals, since the two
solvents would be regenerated separately. Treatment
options for the Wacker-process wastewaters are developed more
fully in a subsequent section.
EXTRACTION FOR WASTE COD REDUCTION
Finally, extraction was considered as a means to
reduce the organic loading in the wastewaters substantially,
i.e., a COD reduction to approximately 100 ppm. Due to
the presence of low-molecular-weight, oxygenated organics,
which are difficult to extract, this degree of COD
reduction would be very expensive to accomplish by
extraction. COD reduction alone did not appear to be
sufficient justification for treatment of any of the sample
wastewaters by extraction. However, some COD reduction
could be accomplished within other treatment strategies,
such as recovery of acetic acid for profit.
SUMMARY OF TREATMENT STRATEGY—GOALS FOR THE EXTRACTION
PROCESS
In the final treatment strategy, samples A, B, and C
were termed poor candidates for treatment by extraction.
Sample D was not considered for reasons mentioned
previously. However, co-extraction of the chlorinated
acetaldehydes, crotonaldehyde, and acetic acid from the
Wacker-process wastewaters (E and F) seemed worthwhile,
and samples G and H contained enough chemical value
(mainly acetic acid) to make extraction potentially
profitable. An additional benefit would be a large
reduction in COD for these wastewaters. Biological
treatment could then remove the remaining chemicals;
however, that aspect of the treatment strategy was not
studied.
34
-------
SECTION 5
SOLVENTS FOR EXTRACTION OF ACETIC ACID AND
CHLORINATED ACETALDEHYDES
The treatment strategy outlined in the previous
chapter identified acetic acid and chlorinated
acetaldehydes as the prime candidates for recovery by
extraction. The search for effective extractants for
these solutes was thus a critical aspect of this work.
DESIRABLE SOLVENT PROPERTIES
To be effective, a solvent should:
1. exhibit a large distribution coefficient, K , for the
solute(s) to be recovered;
2. not participate in irreversible reactions with any
of the wastewater components;
3. be easy to separate from the solutes after extraction
so that the solvent can be re-used;
4. have a low mutual solubility with water so as to
reduce the solvent loading in the aqueous raffinate,
which might otherwise result in costly solvent
losses, and to reduce the amount of water carried to
the regeneration step;
5. have a low K for solutes that are not to be recovered,
i.e., selectivity, to simplify purification of the
desired solutes;
6. have a density, viscosity, and interfacial tension
against water such that the extractor operates
effectively and the phases separate easily; and
7. be inexpensive, commercially available, non-toxic,
and biodegradable.
Ideally, the solvent selected would have an optimum
combination of the above properties, resulting in the
lowest-cost process. However, a quantitative optimization
of all solvent properties is a very difficult and time-
consuming procedure, since many process-design
alternatives must be considered for each solvent.
Consequently, the solvent-selection strategy used here
was to identify a group of solvents, based on qualitative
use of the above guidelines, that appeared to have the
most promise as extractants for acetic acid. Process
35
-------
cost calculations and small-scale experiments were then
used to evaluate those solvents in detail.
The K value for extraction of acetic acid was a
primary consideration in the selection procedure. A high
recovery of the solute to be extracted is only possible
when the extraction factor (E = K F /F , where F and F are
the mass flowrates of the solvent and water phases) is
greater than 1.0. If K is small, F must be large;
equipment size and energy costs are usually prohibitive
in such cases.
As noted by Earhart, et al. (1977), higher K values have
a strong, favorable effect on the process economics up
to a value of about 20.0. Higher K values are useful, but
do not reduce costs significantly. For extraction of
acetic acid, the KQ value is seldom much greater than 1.0. For
many common solvents, it is less than 0.1; such solvents
were immediately dropped from consideration, and attention
was concentrated on the much smaller group of solvents
giving K values on the order of 1.0, or greater.
Basically, these solvents fall into either of two
categories—those having a relatively weak "physical"
attraction for acetic acid, e.g., a hydrogen bonding
capability, and those having a stronger, specific
"chemical" attraction, e.g., strongly basic properties.
WEAKLY-INTERACTING ACETIC-ACID EXTRACTANTS
The physical class of solvents is the larger, by far, and
has received more attention in the past (Eaglesfield, et al.,
1953; Brown, 1963; Othmer, et al, 1941). Examples are
alcohols, ketone, esters, and ethers. Won (1974) recently
compiled K data for acetic acid and other aliphatic
carboxylic acids between water and various weakly-
interacting solvents. He shows that in highly-dilute
acetic-acid solutions, the ranges for the distribution
coefficient, K , are 0.63 to 0.14 for C. to Ca ethers,
0.89 to 0.17 for C. to C,0 acetates, 1.20 to 0.61 for C4
to Cg ketones, and 1.68 to 0.64 for C. to Cg alcohols.
Within a given molecular-weight range, K increases in
the following order: ether < ester < ketone < alcohol. Cyclic
alcohols and ketones have higher distribution coefficients
than their straight-chain counterparts. For example, the
KD for MIBK, a Cg ketone, is about 0.60, while that for
cyclohexanone is about 1.3. Hydrocarbons and aromatics
are generally much poorer extractants.
Eaglesfield, et al. (1953) give an extensive review
of weakly-interacting solvents and recommend specific solvents
for use as acetic acid extractants. The investigators argue
36
-------
against the use of alcohols, even though these have high
distribution coefficients, because they can esterify
with acetic acid. Potential esterification is indeed a
?a™?U* drawback for alcohols, as is reported elsewhere (Ricker,
1978) from the present work. The investigators recommend ethyl
acetate, with cyclohexanone and methyl cyclohexanone as
alternatives.
However, the extraction of acetic acid with these
solvents is not economically attractive if the
concentration of acid in the aqueous feed is less than
about 2,0 wt. % (Brown, 1963; Helsel, 1977). The main reason
is the relatively low K_; there is a strong incentive to find
and develop solvents giving K_ values higher than those typical
of the weakly-interacting extractants.
ORGANOPHOSPHORUS EXTRACTANTS
In his search for acetic acid extractants, Won (1974)
noted that tri-butyl phosphate (TBP) gave a K^ of 2.84
for acetic acid at high dilution. Wardell (.1576) inferred
from considerations of chemical structure that related
phosphorous compounds, notably the phosphine oxides, would
give even higher K^ values, due to the increased polarity
of the phosphorus-oxygen (P-O) bond. He then verified this
hypothesis experimentally.
One of the most promising organophosphorus compounds
studied by Wardell and King (1978), TOPO, was recently patented
as a novel acetic acid extractant (Grinstead, 1974). Helsel
(1977) and Kohn (1976) describe efforts by Hydroscience, Inc.,
to develop a TOPQ-extractant process for large-scale acetic
acid recovery, pointing out that the high K_ and other
favorable solvent properties should result in large
economic advantages over more conventional extractants
when the aqueous acid-feed concentration is less than
about 5,0 wt. %.
TOPO is a white, waxy solid that melts at about 56°C and
boils at 460°C at atmospheric pressure. It is thermally and
chemically stable and reportedly has a water^solubility of only
1 ppm (Helsel, 1977), It has a good selectivity for acidic
compounds (.Wardell and King, 1978). However, it must be
dissolved in an organic solvent, or "diluent", before it can be
used in an extraction process. It is quite soluble in many
organic solvents, typically up to 50-60 wt. %.
The choice of a diluent is a complicating factor in
the design of an extraction process using TOPO. Wardell
inferred from his measurements that the diluent had only
37
-------
a small effect on K , and thus could be chosen to enhance
other solvent properties. However, in a subsequent study
of three different diluents, Michaels (1977) found a
significant variation in K values, with K increasing
in the following order: 2-ethyl-l-hexanol < Chevron Solvent 25
(a solution consisting mainly of alkylated aromatics in the
Cg range) < 2-heptanone. The K value of 2.83 for TOPO
in 2-heptanone was more than twice that with the alcohol
diluent. The trend is not in order of diluent polarity
or hydrogen-bonding ability, as is true for the amine
extractants discussed in the next section.
Michaels hypothesized that the lower K for the
alcohol diluent was due to competition between the
alcohol and acetic acid; both are capable of hydrogen-
bonding with the phosphoryl group. Neither 2-heptanone
nor Chevron Solvent 25 can hydrogen bond with the phosphoryl.
The more polar ketone appears to be a better extractant
for the TOPO/acetic-acid complex, and thus gives a
higher K_ value than Chevron Solvent 25.
The Hydroscience process uses a proprietary,
aliphatic-hydrocarbon diluent (Kohn, 1976). The
resulting extractant mixture is reported to provide a low
water solubility, good phase-separation properties, and
a high boiling point, which facilitates recovery of
acetic acid by distillation.
The concentration of TOPO in the diluent also has an
appreciable effect on K . Wardell and King (1978), for
example, report a K variation from 0.30 to 3.5 for
TOPO concentrations of 0.0 to 58.6 wt.% in a heptane/
hexanol diluent, with a maximum K of 4.7 at 25.9 wt.%
TOPO. The TOPO concentration is thus likely
to be an important process-optimization parameter.
There are three basic disadvantages to the TOPO
extractant system. First, TOPO is relatively expensive.
Prices quoted by the two major U.S. producers, American
Cyanamid and Eastman Organic Chemicals, vary from
$15.42/kg to $44.05/kg, depending on purity and quantity
desired (Helsel, 1976) . A value of $22/kg was used in
this work as a representative figure; prices
could decrease if the TOPO market expands. Prices for
the more common weakly-interacting extractants of the
previous section average only $0.55/kg (Chem. Mark. Rep.,
1978) .
The cost of the solvent is an important consideration
because there is always some loss of solvent in an
industrial-scale extraction process. An example is
38
-------
entrainment of dispersed droplets of solvent in the
aqueous effluent from the extractor. This can be very
costly, especially when large volumes of wastewater are
to be processed. For extractants as expensive as TOPO/
great pains must be taken to keep solvent make-up costs
low.
The second disadvantage is its extremely high boiling
point. Removal of any soluble, non-volatile substances
from TOPO would be difficult to accomplish by distillation
Such substances might be extracted from the wastewater,
as in the Phenolsolvan process described in Section 1,
or the substances might be the result of solvent
degradation. Finally, TOPO has the potential for
solidification in process lines, which mandates expensive
steam-tracing or operation at a TOPO concentration well
below its solubility limit in the chosen diluent.
Extensive development of a TOPO-extractant process
did not seem to be appropriate to the present research.
However, sufficient development and evaluation were done
to allow a comparison between TOPO and alternative
solvents. Wardell and King (1978) and Michaels (1977)
measured K values for solutions of TOPO in batch
extractions of prepared aqueous acetic acid solutions and
wastewater samples, and Sanchez (1977) measured
extraction efficiencies of a TOPO extractant in a
continuous miniplant RDC extractor. These findings are dis-
cussed in later sections dealing with process design and cost
comparisons for alternative solvents.
AMINE EXTRACTANTS
Amines are strong organic bases. Aliphatic amines
are especially basic, more so than ammonia, for example
(Morrison and Boyd, 1966b), and as a result the amines interact
strongly with acidic solutes and can provide high K
values.
The extraction mechanism is an acid-base
neutralization. Acid in the aqueous phase transfers to
the solvent phase, where it reacts reversibly with the
amine. If the amine is sufficiently water-insoluble, the
reaction product, or complex, stays almost entirely in
the solvent phase:
-------
In the extraction of acetic acid, the viscosity of
the organic phase increases noticeably with increasing
complex formation (Kohler, et al., 1972). For example,
in one case studied in this work, the viscosity of the
feed solvent to an extraction experiment was 2.12 milli Pascal
seconds (mPa*s) at 30°C, while that of the extract phase was
3.04 mPa*s (Ricker, 1978). This suggests that the complex may
self-associate in the organic phase.
Page and Smith (1948) appear to have first suggested
extraction of acids as an application for amines. They
demonstrated that even strong mineral acids could be
completely removed from aqueous solution. In fact, K
tended to increase with increasing acid strength.
More recently, amines have been suggested as
extractants for large-scale recovery of specific acidic
solutes. Examples are hydrofluoric acid recovery
(Hardwick and Wace, 1965), pharmaceutical separations (Kunin,
1962), nitrate ion extraction (Coleman, et al., 1958), and
phenol recovery (Pollio, et al. 1967). However, with the
possible exception of the pharmaceutical applications,
these suggestions have not been implemented by industry,
probably because of the lack of a large enough economic
incentive to justify the use of this untried technology
in place of well-established methods.
The current major industrial application of such
amines is in the selective extraction of valuable metallic
and rare-earth ions from acidified aqueous solutions, as,
for example, in the recovery of uranium from ore-leach
liquors (Coleman, et al., 1958). The amine is used in
the salt form as a liquid anion-exchanger. The general
chemical-reaction representation of the extraction step is:
•*
[R-NHA] + [B~] , *' [R.NHB] + [A~]
1 3 org Jaq 3 org Jaq
where R,N represents the amine, in this case a tertiary_
amine, and A and B are anions. The extent to which_B
is transferred to the organic phase in exchange for A
depends on the relative affinities of these anions for
the amine, which can vary greatly.
Liquid ion-exchange can be considered to be the sum
of two neutralization reactions, and neutralization equilibria
can be used to predict ion-exchange behavior. A great deal of
amine neutralization data are available (Scibona, et al., 1966;
Kertes, 1965) but nearly all deal with strong mineral
acids or complex metal ions and are not directly useful
in the present research.
40
-------
Very little has been published on the extraction of
acetic acid by amines. Hogfeldt and Fredlund (1967)
studied extraction of acetic acid by tri-lauryl amine,
but their emphasis was on elucidation of the structure of
the amine-acid salt, and their data are not useful for
process design. More pertinent is the work of Sakai, et
al. (1969), which gives extraction equilibria for acetic
acid between aqueous solution and solutions of Amberlite
LA-2 (a Rohm and Haas secondary amine) in several organic
solvents. Additional measurements were made in the present work
and are reported by Wardell and King (1978), and Michaels (1977),
as well as later in this report. These measurements were made
to obtain a clearer picture of the extraction behavior and to
determine the best amine extractant system for an industrial-
scale acetic-acid-recovery process.
LIQUID-LIQUID EQUILIBRIA FOR AMINE EXTRACTANTS
As for TOPO, amines are usually dissolved in an
organic diluent to provide proper physical properties for
extraction and better values of KD« Amine extractants
are viscous liquids when pure. Extraction behavior varies
not only with the type of amine used, but with the type
of diluent and its concentration in the solvent phase, as
well.
Wardell and King (1978) measured KD values for
extraction of acetic acid from prepared aqueous solutions
by tri-octyl amine (TOA) and tri-isooctyl amine. The type of
diluent was shown to have a strong effect on K , with
the more polar diluents providing the higher K_ values.
The distribution coefficient was also a strong function
of the amine concentration in the solvent. For TOA in
chloroform, the combination that gave the highest K
value measured, K varied from 0.028 to 0.94 as the TOA
concentration increased from 0 to 100 wt.%, with a
maximum K of 9.9 at 18.6 wt.% TOA. Sakai, et al. (1969)
observed similar behavior, although their data are
mainly for amine concentrations below 50 volume %.
Additional measurements were performed in this work
to determine the effect of the type and structure of the
amine and to obtain more information on the effects of
diluent, amine concentration, and acid concentration.
Commercially available primary, secondary, and tertiary
amines were tested; a complete list is given in Table 12.
The experimental technique is described elsewhere (Ricker, 1978).
The experimental extraction equilibria are summarized
in Tables 13-16 and in Figures 5 and 6. More details are given
elsewhere (Ricker, 1978). No results are given for the primary
41
-------
TABLE 12, AMINES INVESTIGATED IN THIS WORK
Trade name
Class
Chemical make-up/structure
Manufacturer
to
Primine JMT
Amberlite LA-3
Adogen 283 D
Amberlite LA-2
Amberlite LA-1
Adogen 383
364
367 D
368
345 D
381
methylated
Adogen 283 D
Alamine 336
1°
1°
1°
2°
2°
3°
3°
3°
3°
3°
3°
3°
3°
highly-branched, 18-24 carbon atoms
branched-chain, ave. MWT=353
di-tridecyl amine, branched chains
highly-branched
it
tri-laurel, straight-chain
tri Cg to C,Q isomers, straight-chain
di-methyl Coco amine
tri Ca~cio~ci2 i-somers' straight-chain
di-methyl hydrogenated tallow amine
tri-iso-octyl amine
methyl di-tridecyl amine
tri C0 and C,n straight-chain
o JLU
Rohm & Haas
ti
Ashland Chem.
Rohm & Haas
it
Ashland Chem.
General Mills
*synthesized from Adogen 283 D by Eschweiler-Clarke modification of Leukart Reaction
(Moore, 1949)
-------
TABLE 13, RESULTS FOR EXTRACTION OF ACETIC ACID BY TERTIARY
AMINES SUPPLIED BY ASHLAND CHEMICAL COMPANY
Amine
Adogen 381
ii
ii
ii
Adogen 364
ii
ii
ii
Adogen 368
n
H
n
Adogen 363
n
n
n
Amine
vol. %
20
ii
30
80
20
n
30
80
20
ii
30
80
20
n
30
80
Equil . aqueous
Diluent acetic acid wt.
2-heptanone
n
chloroform
2-heptanone
2-heptanone
ii
chloroform
2-heptanone
n
ii
chloroform
2-heptanone
n
ii
chloroform
2-heptanone
3.23
0.198
0.0448
2.98
2.90
0.145
0.0381
2.96
3. -11
0.145
0.0447
3.06
3.44
0.145
0.0623
3.50
% KD
1.72
1.09
7.79
2.87
1.98
1.62
9.69
2.79
1.83
1.58
8.28
2.77
1.58
1.58
7.82
2.26
43
-------
TABLE 14. REPRESENTATIVE RESULTS FOR EXTRACTION OF ACETIC
ACID BY ALAMINE 336
Amine vol. %
10
20
50
75
75
30
"
"
"
20
"
100
II
II
II
Equil. aqueous
Diluent acetic acid wt.%
2-heptanone 3.01
2.99
1.07
0.98
2.77
chloroform 0.0465
0.648
2.98
4.79
octanol 0.664
decane 2.43
0.095
1.20
1.36
3.18
KD
1.48
2.24
2.50
1.68
3.29
9.68
5.60
1.87
1.39
3.73
0.22
0.75
1.05
0.934
0.97
44
-------
TABLE 15, EXTRACTION OF ACETIC ACID BY ASYMMETRIC TERTIARY AMINES
Equil. aqueous
Amine Amine vol. % Diluent acetic acid wt.T
methyl di-tridecyl 50 methyl iso-amyl ketone
„
„
0.988
2.11
2.91
KD
1.77
2.58
3.54
Adogen 367 D
Adogen 345 D
single phase
-------
TABLE 16. EXTRACTION OF ACETIC ACID BY SEVERAL SECONDARY AMINES
(Ti
Amine
Amber lite LA-1
H
ii
ii
Amber lite LA- 2
H
ii
n
Adogen 283 D
n
ii
n
n
Amine vol. %
50
"
30
100
50
If
30
100
50
II
II
30
100
Diluent
Chevron 25
ii
chloroform
none
Chevron 25
n
chloroform
none
Chevron 25
n
n
chloroform
none
Equil . aqueous
acetic acid wt.%
1.16
3.78
0.0469
2.25
0.53
2.19
0.0218
1.83
0.983
2.63
3.85
0.0133
0.383
KD
1.27
2.26
4.48
4.22
3.82
4.48
9.86
6.49
9.65
4.55
3.46
32.11
33.4
-------
0
100
c
-------
amines because these co-extracted a large amount of water with
the acetic acid. At high acid loadings,, e.g., an aqueous feed
of 10 wt.%, extraction resulted in a single liquid phase.
The same behavior was observed for the asymmetric tertiary
amines, Adogen 345D and Adogen 367D, which are essentially
primary amines with methyl groups substituted for the
hydrogens on the nitrogen (N) atom.
Secondary Amines
The symmetric secondary amines, Adogen 283 D, and
Amberlite LA 1 and LA 2, gave the highest KD values
measured in this work (Table 15). Adogen 283 D, which
is made up of two branched C,- chains attached to the N
atom, appears to be the more powerful extractant for
acetic acid, as shown by a comparison of the experimental
results for identical extractions with the amines in a
chloroform diluent and with the pure amines. Amberlite
LA 2 is slightly more effective than LA 1. It is
interesting to note that Adogen 283 D supplies fewer basic
equivalents per weight of amine than Amberlite LA 1 or LA 2
(Ricker, 1978). Their relative strengths as extractants must
be due, primarily, to structural effects. However, the
structures are poorly characterized, and no structural
comparisons are possible.
Selected results for Adogen 283 D are plotted in
Figure 5 to show the strong effects of amine
concentration and acetic acid concentration.
KD values of as high as 160 were measured under
certain conditions. This is about two orders of magnitude
higher than the K values for conventional solvents. The
decrease in K with increasing acid concentration is
characteristic of chemically-interacting extractants, and
is discussed in detail in the next section. Figure 5
shows that pure Adogen 283 D is an effective extractant,
in contrast to the behavior of the tertiary amines (Figure
6). However, the pure amine becomes quite viscous when
loaded with acetic acid and could not be easily used in
an industrial extractor.
The results obtained by Sakai, et al. (1969) for
extraction of acetic acid by Amberlite LA 2 in methyl
isobutyl ketone (MIBK) are qualitatively similar to those
obtained for the Adogen 283/2-heptanone system in the
present work (Figure 5). The LA 2/MIBK extractant gives
K_ values approximately 20% lower than the Adogen 283/2-heptanone
system for similar extraction conditions. Again, this in-
dicates that Adogen 283 is the more powerful extractant for
acetic acid.
Unfortunately, Adogen 283 D cannot be regenerated
48
-------
by distillation because it reacts with acetic acid to form an
amide, as follows:
0
11
CH3C-OH
(acetic acid) (secondary amine) (amide)
This reaction forms the basis for a test to determine
the amount of secondary amine in a mixture of secondary
and tertiary amines (Ashland Chem. Co.), and was assumed
to be the reason for disappearance of acid in regeneration
experiments performed in this work (see below) . The
reaction is encouraged by the high temperature and
dehydrating conditions that occur in distillation. The
resulting loss of amine would be too costly for such a
process to be economical.
Distillation has a substantial cost advantage over
other regeneration methods for the extraction applications
considered in this work. Amide formation was thus a serious
disadvantage for secondary amines, and they were not investiga-
ted further.
Tertiary Amines
Tertiary amines do not form amides and are thus more promis-
ing for general use. The K values of the amines are not as high
as those for Adogen 283 D, But with the better diluents are
often much higher than those for TOPO. Values as high as
22.0 were measured for Alamine 336 (Figure 6).
The symmetrical tertiary amines are all quite similar
as extractants for acetic acid — there is not much
variation in K with amine structure, probably because the
commercial amines are isomeric mixtures of quite similar
composition (Ricker, 1978) . In general, as molecular weight
increases, the water-solubility of the amine and its acid
complex decrease, as would be expected, due to the
increasing hydrocarbon-like character of the amine.
Tri-butyl amine is a relatively poor extractant because
the reaction product is too water soluble (Wardell and
King, 1978) . An independent, GC measurement showed that the
solubility of tri-butyl amine in water is 3.1 wt %, in the
presence of 1% acetic acid (Pittman, 1979). Adogen 381, tri-
iso-octyl amine, also seemed to be somewhat soluble in aqueous
solutions of acetic acid, although its solubility was
not determined accurately. The higher-molecular-weight
amines are essentially insoluble (Ricker, 1978) , and K values
49
-------
for similar extractions tend to be slightly higher than
for Adogen 381, up to a point. As molecular weight continues
to increase, the equivalent weight of base decreases.
The optimum seems to be in the tri-C8-C,Q range (Adogen
364, Alamine 336)'; tri-C,, amine (Adogen 363) appears
to be beyond the optimum molecular weight (Table 13) .
A tertiary amine with two symmetrical alkyl chains,
but with a methyl group as the third alkyl group was
synthesized from the secondary amine, Adogen 283 D. Its
structure is thus more like that of Adogen 283 D than the
more symmetrical tertiary amines. However, it performed
much like Alamine 336 in extraction experiments, indicating
that secondary amines are fundamentally more powerful
extractants than tertiary amines of similar
structure.
Alamine 336 was the major extractant investigated in
this work. However, the results obtained with Alamine 336
are probably representative of other commercial tertiary
amines in the same molecular weight range, e.g., Adogen
364.
Water-solubility of Tertiary Amines
The water-solubility of the major components of
Alamine 336, i.e., the Cg-C,- tertiary amines, is reported
to be less than 5 ppm (General Mills). Chromatographic
analyses by Pittman (1979) and Ricker (1978) showed that
even in the presence of 10 wt % acetic acid, the combined
solubility of these amines was less than 10 ppm. However,
as pointed out in the product literature of several amine
manufacturers (General Mills, Rohm & Haas, Ashland Chemical
Co.), commercial grades of these extractants contain
on the order of 2-5 wt.% impurities, which can be primary
and secondary amines, lower-molecular-weight tertiary
amines, and starting materials for the amine synthesis
reaction. Some of these may be quite water-soluble.
In several cases, Pittman (1979) noted the presence of
unidentified soluble impurities in the raffinate phase
following extraction with Alamine 336. Tests indicated
that the impurities might be amines of some type. Such
impurities would gradually extract into the aqueous phase
in an operating extraction process and would represent
a one-time loss of solvent, albeit a relatively small
loss.
EFFECT OF DILUENT TYPE, AMINE CONCENTRATION AND ACID CONCENTRA-
TION ON KD
The acetic acid K value in amine extractant systems
50
-------
was a strong function of several factors. Alamine 336 was
tested thoroughly to see which combinations of variables
resulted in high K values. Representative results are
shown in Figure 6.
As previously mentioned, for chemically-interacting
extractants, K generally decreases with increasing acetic
acid concentration, especially in the concentration range
shown in Figure 6. This effect is due to the chemical
equilibrium between the acid and the amine in the organic
phase. Qualitatively, as the acid concentration increases,
the amine complexes with the acid more and more completely,
which decreases the capacity of the amine to extract
additional acid. The degree of complexing at any point
depends on the equilibrium constant of the acid-base
complexing reaction and the activities of the reactants
and products. When the amine is highly complexed, the
extraction behavior becomes like that of a non-chemical
extractant,and K_ is no longer a strong function of acid
concentration.
One would then expect an increase in the amine concentra-
tion to give higher distribution coefficients, and this
does happen for low-to-moderate amine concentrations,
as shown in Figure 6. However, if the amine concentration
increases beyond a certain point, which depends on the
diluent, K begins to decrease. For example, the curve
for 100% amine in Figure 6 is lower than that for a solution
of 5% amine in 2-ethyl hexanol.
Such behavior cannot be explained by a simple chemical
equilibrium theory; it appears to be due to a combination
of chemical and physical interactions in the solvent
phase. Pure Alamine 336 is a relatively non-polar
compound, since the N atom is surrounded by long
hydrocarbon chains. The acid-amine complex, on the other
hand, is much more polar; its formation appears to be
inhibited in non-polar solutions. A more polar solvent
such as 2-ethyl hexanol allows a greater degree of complex
formation at a given aqueous-phase acid activity.
As previously mentioned, the pure secondary amines
are much more effective extractants than the pure tertiary
amines. Apparently, the secondary amines are sufficiently
polar to solvate the complex to a high degree, whereas
the tertiary amines are not.
Similar considerations explain the variation of K with
the type of diluent. The more polar diluents give higher
K values, in general (Wardell and King, 1978), and the
51
-------
order of effectiveness appears to be the same as for
extraction of acetic acid with the pure -diluent (i.e.,
alcohols ketones esters ethers). Hydrocarbon-like
solvents are very poor diluents (Table 14), as would be ex-
pected.
The effect of the diluent can be seen most clearly
in the dilute-acid region, below about 1 wt. %. Here,
complex-diluent interactions predominate. For dilute
acid in the presence of an excess of amine, a "good"
diluent, such as 2-ethyl hexanol allows a high degree of
complex formation and thus gives high K values. In ^
other words, the activity coefficient of the acid-amine
complex is reduced by a good diluent.
The primary alcohols, as a class, give the highest
KD values measured, much higher than ketones (Figure 6).
A possible explanation is that an alcohol has an acidic
hydrogen available for hydrogen bonding, whereas a ketone
does not. The acidic hydrogen on the acid molecule is
probably tied up in the acid-amine complex, leaving only
the carbony1 oxygen for hydrogen bonding, i.e., an
acidic hydrogen would be required for such a bond.
It is interesting to note that chloroform appears to
be a good diluent at low acid concentrations, giving K
values on the order of 10.0 (Wardell and King, 1978) .
Chloroform also has a slightly acidic hydrogen available
for bonding with a carbonyl oxygen.
Unfortunately, alcohols are probably a poor choice
for extraction of acetic acid because of the potential for
esterification (see below). Chloroform is quite volatile
and is a suspected carcinogen. Aldehydes might hydrogen-
bond with the acid-amine complex; however, aldehydes are
relatively reactive organic chemicals, and they were not
investigated experimentally.
EXTRACTION EQUILIBRIA FOR ALAMINE 336 IN KETONE DILUENTS
Ketone diluents were studied extensively, since they
give relatively high K values and yet are not as prone to
chemical reaction as are alcohols. Extraction equilibria for
50 volume % Alamine 336 dissolved in 2-heptanone, DIBK, and a
combination of 2 parts by volume of cyclohexanone plus
3 parts of DIBK are compared in Figure 7. More detailed data
are given elsewhere (Ricker, 1978; Ricker et al, 1979a).
One would probably expect the Alamine 336/DIBK
combination to give lower K values than those for Alamine
336/2-heptanone. The data in Figure 7 indicate that
52
-------
10.0
0)
o
o
c-
o
s j-°
o
<
u
-------
10
8
O)
(ft
£ ?
CL
O
g 6
?
O
.E 5
TJ
"o
< 4
o
o _
1 1 1—
Diluents:
0 2-Heptanone
o DIBK
A 3parts DIBK, 2 parts Cyclohexanone
I
~0 I 2 3
Wt % Acetic Acid in Aqueous Phase
Figure 7. Extraction equilibria for 50 vol.
Alamine 336 in ketone diluents.
54
-------
this is the case. However, there is an important
difference between the two equilibrium curves: the curve
for the 2-heptanone diluent is concave-downward at the
lower acetic acid concentrations, as for the 2-ethyl
hexanol diluent (Figure 6), whereas that for DIBK is
concave-upward. The concave-upward shape is a
disadvantage because it makes it difficult to achieve a
high degree of acid removal in a multi-stage extractor,
i.e., the operating and equilibrium lines "pinch" at the
dilute end of the extractor cascade.
For the 2-heptanone diluent, the KQ for acetic acid
is highest in the dilute-acid region, and gradually
decreases with increasing acid concentration (Figure 6),
due to the disappearance of un-complexed amine. For DIBK,
the KD is quite low in the dilute region, but increases
with increasing acid concentration to at least 5 wt.%
aqueous acid. The K_. probably begins to decrease again
at some higher acid concentration; such measurements were
not performed in the present work.
The increase in Kn at intermediate acid concentrations
is probably due to complex-complex or complex-acid
interactions, i.e., the pure DIBK seems to inhibit complex
formation, but once enough polar species are present in
the extractant phase, the complex forms more readily.
Possible support of this hypothesis is given by two
experimental results. First, in initial experiments using
technical grade DIBK as received from the supplier, the
extraction equilibria for DIBK was very much like that of
2-heptanone, i.e., it did not exhibit a concave-upward
equilibrium curve. Unfortunately, this supply
of DIBK was not analyzed carefully for impurities, and the
reason for this behavior is not clear. Second, the acetic
acid K value is usually higher for Alamine 336/DIBK
extractions of wastewaters than for extractions of
synthetic aqueous acetic acid solutions. This may be due
to the co-extraction of polar impurities, although these
are usually dilute. For example, the result for 50%
Alamine 336 in distilled DIBK which has a KD value of
1.00, is about 20% below the curve shown in Figure 7
Ricker, 1978).
The performance of pure DIBK diluent might then be
enhanced by the addition of an appropriate polar species.
Figure 7 shows that an Alamine 336/DIBK/cyclohexanone
combination gives higher K values than Alamine 336/DIBK.
The use of acetophenone or 3-methyl cyclohexanone as
additives results in even higher K values, as shown in
Figure 8.
55
-------
10
8
6 i
c
73
I 4
O
Additive
o None, i.e., Pure DIBK
A Cyclohexonone
D Methylcyclohexcmone
x Acetophenone
I
0123
Wt % Acetic Acid in Aqueous Phase
Figure 8. Extraction equilibria for 50 vol. %
Alamine 336 in DIBK, and in 3 parts
DIBK/2 parts polar additive.
56
-------
The results in Figure 7 also suggest that a C-
methyl-ketone, such as 2-nonanone, might be more effective
than DIBK. Ashland Chemical Company literature advertises
2-nonanone as well as 3-octanone, another interesting
possibility. These diluents would be much higher-boiling than
2-heptanonef an advantage for the regeneration step in the
extraction process.
TOPO was also found to be an effective additive for the
Alamine 336/DIBK system. A solvent mixture of 8% TOPO, 50%
DIBK and 42% Alamine 336 gave K_ = 1.9 and K = 1.75 for 0.44%
and 0.32% respectively, acetic acid in water (Pittman, 1979).
The extraction equilibria for Alamine 336/DIBK are
shown in Figure 9 as a function of Alamine concentration.
The curves for the various amine concentrations are quite
similar in the dilute region, but diverge at the higher
acid concentrations because of acid-base stoichiometry.
The corresponding equilibrium water-content of the ex-
tractant phase is shown in Figure 10 (Ricker, et al, 1979b).
Hardwick and Wace (1965) report a roughly 1:1
correspondence between moles of acid-base complex and moles
of water in the extractant phase at equilibrium for the
extraction of HF. If the acid in the organic phase is
assumed to be mainly in the complexed form, this would
seem to be a reasonable approximation for the data in
Figure 10 as well, although the data are scattered.
PREDICTION OF THE ACETIC-ACID DISTRIBUTION COEFFICIENT FOR
AN AMINE EXTRACTANT
Some consideration was given to the development of a
thermodynamic model of the extraction of acetic acid by
an amine extractant. An effective model would allow the
prediction of extraction equilibria from limited
experimental data.
Sakai, et al. (1969) studied the extraction of acetic
acid by Amberlite LA-2, a secondary amine, dissolved in
four organic diluents—carbon tetrachloride, chloroform,
n-hexane and methyl isobutyl ketone (MIBK). They proposed
a simple chemical-equilibrium model of the organic phase
in which the species present were acid, diluent and amine
monomers, an acid dimer, and complexes of one amine molecule
plus 1-4 acid molecules. The resulting equations fit the
data quite well, but six adjustable parameters were
required. The model does not account for physical interactions
in the organic phase and would not, for example, predict
the decrease in K at high amine concentrations observed
experimentally in the present work.
57
-------
15
u>
o
10
o
1*
o
<
o
4)
O
Vol. % Alomine 336
in DIBK
& 20 %
Q 3 5 %
O 50 %
0 65 %
012345
Wt % Acetic Acid, Aqueous Phase
Figure 9. Extraction of acetic acid from
wastewater H by Alamine 336 in
DIBK.
58
-------
2.5
o
.c
£L
o
0>
6
.1.5
o
0.5
o>
0>
0*0
-••—01
Volume % Alamine 336
in DIBK
« 1.0
£
C3D
0
O A 20 %
GJ CD 35 %
O 50 %
O 65%
0 5 10 I5L
Weight Percent Acetic Acid in the Organic Phase
Figure 10. Water-content of organic phase for extraction
of wastewater H by Alamine 336/DIBK.
59
-------
Inoue and Baba (1978) attempted to use the same model
to correlate the extraction of acetic acid by Amberlite
LA-2 diluted with cyclohexanone, benzene and chlorobenzene.
Physical interactions apparently caused a poor fit to the
data for the aromatic diluents, even though six adjustable
parameters were used.
In the present work, a study of the Alamine 336/2-ethyl
hexanol/acetic acid/water system indicated that a simple
chemical equilibrium theory would be inadequate for. amine
concentrations above about 5 wt.% (1.5 mole %) in
the organic phase. A better model might be developed
along the lines of the model of Harris and Prausnitz (1969),
which accounts for both chemical and physical interactions
and requires only two adjustable parameters. However, the
activity-coefficient equations used by Harris and
Prausnitz (1969) do not seem to give a good representation
of the data obtained in the present work; a more
sophisticated model of the physical interactions is
probably required, e.g., as in the UNIQUAC theory (Abrams
and Prausnitz, 1975). Development of such a model would
be quite difficult and was felt to be beyond the scope of
the present work.
EXTRACTANTS FOR CHLORINATED ACETALDEHYDES
The extraction of chlorinated acetaldehydes from
water has received very little attention, especially in
comparison to the voluminous literature on extraction of
acetic acid. Some mention is made of the use of ether for
extraction of chloroacetaldehyde (Miller, 1969a), but no
details are given.
The work of Earhart, et al. (1977), and Michaels
(1977) indicates that the most effective extractants for
chlorinated acetaldehydes are primary alcohols of six to
eight carbon atoms. A comparison between the primary
alcohol, 2-ethyl-l-hexanol, and several other solvents is
shown in Table 17. Distribution coefficients for
chloral are not given; these are always higher than for
monochloro (MCA) and dichloro (DCA) acetaldehyde. For
example, Earhart, et al. (1977) reported a K value of
about 50.0 for extraction of chloral by 2-etHyl-l-hexanol.
As shown in Table 17, K_ values for MCA are always
substantially lower than for the other chlorinated
acetaldehydes. Since MCA is present in each of the
wastewater samples that contain chlorinated acetaldehydes,
MCA is the limiting solute for extraction.
The primary alcohols give relatively high K values
because of a specific chemical interaction between the
60
-------
TABLE 17. EXTRACTION OF CHLORINATED ACETALDEHYDES BY
SEVERAL COMMON SOLVENTS
Solvent CKD)MCA
2-ethyl-l-hexanol 8.34
2-octanol 2.04
3-methyl-3-pentanol 0.64
2-heptanone 0.30
cyclohexanone 1.24
ethyl ether 1.61
butyl acetate 0.28
methyl isobutyl ketone 0.44
MCA = monochloroacetaldehyde
DCA = dichloroacetaldehyde
(VDCA
14.3
3.58
1.65
1.23
5.60
-
1.11
2.22
Reference
Michaels
ii
H
ii
n
this work
ii
n
All results are for equilibrium values, i.e., long contact time
61
-------
aldehyde and the alcohol (Morrison and Boyd, 1966a):
o
C1CH2CH + ROH ^ C1CH2C-OH
I
H
(MCA) (primary alcohol) (hemi-acetal)
t
Distribution of the hemiacetal favors the alcohol phase/
while un-complexed MCA is hydrated in aqueous solution
and does not favor the organic phase.
Earhart, et al. (1977) and Michaels (1977) found that
the extraction performance of 2-ethyl-l-hexanol varied
significantly with the interphase contact time in the
extraction step. Distribution coefficients increased
slowly with time, finally attaining an equilibrium value.
The time constants for extraction of MCA and chloral were
about 30 and 6 minutes, respectively. Michaels (1977)
shows that the extraction of MCA is probably limited by
the rate of hemiacetal formation. Earhart, et al. (1977)
found that the rate of chloral extraction could be
accelerated somewhat by the addition of carboxylic acids.
However, the effective K in an operating extractor is
likely to be lower than the equilibrium value.
Another disadvantage of the primary alcohols is that
they cannot be regenerated by distillation. Esterification
is a problem because acetic acid co-extracts with the other
solutes. Moreover, the chlorinated acetaldehydes are reactive
chemicals; the hemiacetal formed in the extraction step can
react with another alcohol molecule to yield an acetal
(Miller, 1969):
OR H
l I
C1CH.C-OH + ROH^CICH-C-OR + H00
l 2, 2
H OR
Other undesirable side reactions are also possible. For
this and other reasons discussed subsequently, treatment
of the Wacker-process wastewaters by extraction is likely
to be costly.
62
-------
SOLVENT-REGENERATION STUDIES
The regenerability of acetic-acid extractants was tested
qualitatively in bench-scale batch distillations. A
small quantity of the extractant, typically 200 ml, was
used to extract solutes from one of the wastewater
samples. Alternatively, pure acetic acid was added to
the pure extractant. The extractant was then distilled
in a 500-ml boiling flask equipped with a small packed
column. Refluxing occurred due to natural heat losses
and resulting condensation. The vapor leaving the column
was condensed and collected in a receiver. The temperature
of the overhead vapor was measured with a thermometer.
The condensate and residue were analyzed by GC, and
mass balances were performed for all the components. The
very volatile components, such as acetaldehyde and
chlorinated acetaldehydes, always gave poor mass balance
closures. Typically 20-40% of the volatile components
disappeared during the distillation, probably due to
incomplete condensation.
Similar behavior for the less volatile components
was taken to indicate a chemical reaction between that
solute and another solute or the extractant. For example,
large losses of acetic acid were observed in distillation
with alcohols due to esterification, and in distillations with
secondary amines (e.g., Adogen 283 D) due to amide formation.
For the secondary amine, molar losses of amine and
acid were nearly 1:1, which is the correct stoichiometry
for amide formation. Nearly 40% of the amine was
converted to amide in a 3-hour distillation. On the other
hand, the mass balances for the tertiary amine case were
within experiment error, indicating no reaction.
Tertiary amines are incapable of amide formation, so good
closures would be expected.
An attempt was made to inhibit amide formation by
distilling at lower temperature, i.e., under vacuum.
Otherwise, the experimental conditions were the same as
before. About 6% amine loss still occurred, and
low-temperature operation decreases the acid/diluent
relative volatility, making regeneration more difficult
Although Adogen 283 D gave the highest KD values observed in
extraction experiments, the regeneration experiments showed
that secondary amines are probably a poor choice for an ex-
traction process involving acetic acid unless an alternative
regeneration method is used.
The batch distillation experiments also provided dramatic
63
-------
evidence of the incompatability of chlorinated acetaldehydes
and amines. When an attempt was made to distill a batch °*
monochloroacetaldehyde and Alamine 336, as soon as the temper
ature of the mixture increased appreciably, the contents of tne
flask turned dark 'and became viscous. A titration of *&«
residue showed that the amine content had been greatly reduced.
Studies of regeneration of Alamine 336 extractant by con-
tinuous and semi-batch distillation are reported elsewhere
(Ricker, 1978). This evidenced no difficulties.
Regeneration of Solvents Containing Chlorinated Ac^^ldehydes
By Sodium Bisulfite Wash,
A quick, screening experiment was performed to test the
regeneration of an extractant for Wacker -process wastewaters by
a sodium bisulfite (NaHSCX,) wash. The basic concept was to re-
move chlorinated acetaldenyde by the following reaction
(Morrison and Boyd, 1966c) :
- C - + Na+HSO~ < *:. -OSO~Na+
ll J i J
0 OH
Caldehyde or ketone) (sodium bisulfite) (addition product)-
Addition involves nucleophylic attack by bisulfite ion on car-
bony 1 carbon, followed by attachment of a hydrogen ion to the
carbonyl oxygen. The reaction is reversible; addition of acid
or base destroys the bisulfite ion in equilibrium with the
addition product and regenerates the carbonyl compound:
_ . ^H+ S00 + H_0
C-SO-Na .-»• -C- + HSO~ ^ *
1 "
OH 0 OH" S0~ 4- H20
The addition product is usually only partially soluble in
aqueous bisulfite ion solution, so that it might conceivably be
removed as a solid product by filtration for eventual disposal.
In the experiment, 450 ml of wastewater sample F was con-
tacted with 50 ml of 2-ethyl-l-hexanol in a separatory funnel.
The extract phase was decanted, and 0.323 kg was subsequently
contacted with 0.020 kg H20 plus 0.00504 kg NaHSO,, . A strong
odor of sulfur dioxide CSO,) was noted but gas bubbles were not
released. The 2-phase mixture was agitated by hand for 5 min-
utes, then allowed to separate for several hours. The phases
were analyzed by GC for MCA and DCA (mono- and dichloroacetalde-
hyde) content, using the chroma tographic techniques developed by
Michaels (.1977). The feed extract solution was also analyzed;
64
-------
a rough mass balance gave the following results:
% removal from apparent % apparent K
solvent phase reacted
MCA 63 49 0.45
DCA 44 32 0.96
The per cent reacted was determined from the non-closure of the
mass balance. Absolute concentrations were not measured; the
results are relative to the extract feed concentration, which
could be estimated from the K values of Michaels (1977), if
desired.
The apparent K given above was based on a mass balance on
the organic phase: apparent K = C(l-x) • (wt. of aqueous phase)/
[x-(wt. of organic phase)], where x is the fractional removal
from the feed organic phase. In the presence of NaHSO-, DCA
only slightly favors the aqueous phase. Thus a large amount of
wash solution would be required for complete regeneration of
the extract, even if the wash step were multi-stage counter-
current. The cost of the NaHSO^ would be prohibitive. The un-
favorable equilibrium may have Been due to the presence of
acetic acid and/or other solutes, such as acetone and
acetaldehyde.
Another problem is that the addition product was soluble,
and a substantial amount of unreacted MCA and DCA remained in
the aqueous wash phase. Disposal of MCA and DCA in this form
would probably be as serious a problem as disposal of the
original wastewater. For these reasons, the idea was not
pursued.
HEAT STABILITY OF AMINE EXTRACTANTS
Two experiments were performed to check the stability
of Alamine 336 at the maximum temperature likely to be
encountered in an extraction process, i.e., in the
reboiler of the regeneration column. Hardwick and Wace
(1965) report small but measureable degradation of
Alamine 336 at 280°C in the presence of hydrogen fluoride. The
maximum operating temperature contemplated in the present work
is considerably less severe, about 180°C, and it was
anticipated that, unless degradation were to be caused by
a reaction with one of the wastewater solutes, degradation
should be negligible compared to other mechanisms for
amine loss, e.g., by entrainment.
In the first experiment to test for degradation,
100 ml of DIBK, 100 ml of Alamine 336, and 2 ml of acetic acid
were heated in a 250-ml boiling flask equipped with a refluxing
65
-------
condenser. A small flow of nitrogen gas (N^) was continuously
fed into the top of the refluxing condenser via a glass tube,
the outlet of which was about 5 cm above the active condensing
region in the condenser. The purpose of the N2 was to prevent
diffusion of air into the boiling flask.
The temperature of the boiling liquid rose slowly
throughout the experiment from 173°C to a maximum of 184°C,
indicating a gradual loss of volatiles—probably due to
incomplete condensation. The boiling liquid remained a
clear, pale, yellow color throughout; it did not turn
dark brown as in the regeneration experiments.
Samples were taken at 72 and 193 hours, at which time
the experiment was terminated. A sample of the original
feed was also taken. The samples were analyzed for amine
content as described elsewhere (Ricker, 1978; Ricker et al,
1979a). Peak areas for amine peaks were totalled and relative
areas were determined. The relative areas were very
reproducible; the standard deviation was less than 1%.
The relative areas were essentially the same for all these
samples and compared well with the values determined for the
pure amines.
The total peak area for the chromatogram was also determined,
including the peaks preceding the first amine peak (DISK,
acetic acid, and volatile impurities). The total area of the
amine peaks increased slightly with respect to that of the
lighter peaks as the experiment progressed. In the feed sample,
the total amine peak area was 58.2% of the total area, while it
was 61.5% and 60.5% at 72 and 193 hours, respectively. Again,
this seemed to indicate a loss of volatiles rather than
degradation, which should have caused a reduction in the
relative amount of amine. No new peaks were found in the
chromatogram, and no other signs of a degradation reaction
were apparent.
In the second experiment, the extract phase from an
extraction of wastewater sample G was used in place of a
synthetic mixture of DIBK, Alamine 336, and acetic acid.
The extract was from Pittman's (1979) run #16, containing
about 5 wt.% acetic acid and lesser amounts of the
other solutes found in wastewater G.
The experimental procedure and method of analysis
were the same as for the first experiment. Again, there
was a slight increase in the relative amine concentration
with time, and there were no measureable signs of araine-
degradation. The feed solution was dark brown, having
been regenerated many times. There may have been some
additional coloring during the experiment; the final
66
-------
liquid was noticeably more turbid than the feed, but a
2-cm-thick sample was still quite transparent.
In summation, thermal degradation of the amine does
not appear to be a significant problem. Hardwick and
Wace (1965) suggest that a branched-chain amine should be
even more stable than the straight-chain amines typified
by Alamine 336.
67
-------
SECTION 6
EXPERIMENTAL EXTRACTION RESULTS
Experimental measurements of countercurrent extraction with
actual wastewaters and candidate solvents were made using a mini-
plant rotating-disc contactor (RDC) extractor. These served to
demonstrate performance attainable with regard to mass transfer,
axial mixing and solvent entrainment. A few runs were also made
with a system of separatory funnels, simulating a countercurrent
mixer-settler system.
RDC EXTRACTION RUNS
The RDC extraction runs were made with the miniplant system
described by Earhart, et al (1976, 1977). They are described in
more detail by Sanchez (1977) and Pittman (1979). Mass balance
non-closures are not reported in these tables. They were deter-
mined and are available elsewhere (Sanchez, 1977; Pittman, 1979).
Mass transfer analyses of these runs are reported by Ricker,
et al (1979c), and show that the performance of the RDC was
heavily dominated by axial mixing.
Results of these runs are reported in Tables 18 through 36.
COUNTERCURRENT SEPARATORY FUNNEL SYSTEM
Results of a countercurrent separatory funnel experiment
for extraction of acetic acid by Alamine 336/DIBK are given by
Ricker, et al (1979b). Results of scattered other experiments
of this sort are given by Ricker (1978).
68
-------
TABLE 18. CONDITIONS AND RESULTS FOR MINI-PLANT EXTRACTION
RUN 1
wastewater = synthetic mixture (continuous phase)
solvent = n-butyl acetate (dispersed phase)
water flow rate = 9.46 L/h =9.48 kg/h
solvent flow rate = 3.56 L/h = 3.18 kg/h
solvent-to-water ratio =0.33 kg/kg =0.38 L/L
density of solvent = 0.876 kg/L (25°C)
shaft rotation speed = 1200 rpm
estimated droplet size = 0.75 mm diameter
column pressure (top) = 390 kPa
average column temperature = 20.5°C
rotor disc diameter = 3.81 cm
stator hole diameter = 5.72 cm
Feed
Water
(ppm)
850
300
180
130
Component
acetic acid
acetaldehyde
2-butanol
methyl isobutyl ketone
acetone
n-butyl acetate
TOD
COD
COD/TOD
Steady-State
Raffinate
(ppm)
785
250
60
0
0
2300
2798
1.22
1910
7320
21800
18950
0.
87
Percent
Removal
8
17
67
97
69
-------
TABLE 19. CONDITIONS AND RESULTS FOR MINI-PLANT EXTRACTION
RUN 2
Conditions
wastewater = A
solvent = n-amyl alcohol
water flow rate = 4.54 kg/h = 4.54 L/h (continuous phase)
solvent flow rate =6.76 kg/h = 8.33 L/h (dispersed phase)
solvent-to-water ratio = 1.50 kg/kg = 1.83 L/L
density of solvent = 0.812 kg/L (25°C)
shaft rotation speed = 865 rpm
estimated droplet size = 0.2 mm diameter
column pressure (top) = 400 kPa
average column temperature = 20.6°C
rotor disc diameter = 3.81 cm
stator hole diameter = 5.72 cm
a. Gas Chromatograph (concentrations in ppm, by weight, unless
otherwise noted)
Component
formic acid
acetic acid
propionic acid
methanol
ethanol
isopropanol
n-propanol
2-butanol
n-butanol
methyl ethyl ketone
isobutyraldehyde
acetone
n-amyl alcohol
Water
Feed
870
560
78
50
539
130
41
226
70
309
129
23
0
Steady
State
Raffinate
174
128
23
19
36
p
5
27
26
21
16
943
1.78
Steady
State
Extract
p
(260)
p
25
292
?
34
133
p
154
95
389
wt.%
Percent
Removal
80
77
71
62
93
-
88
88
63
92
88
-
_
(continued)
70
-------
TABLE 19. (continued)
b. Acid/Base Titration
water feed: 0.0295 g equivalents acid/L
steady state raffinate: 0.0063 g equivalents acid/L
steady state removal of acid = 79%
c. COD Analysis (ppm)
TOD COD COD/TOC
water feed 4,570 4,760 1.04
steady state
raffinate 51,170 47,800 0.93
71
-------
TABLE 20.
Conditions
CONDITIONS AND RESULTS FOR MINI-PLANT EXTRACTION
RUN
wastewater = A
solvent = n-amyl alcohol
water flow rate =1.89 L/h =1.89 kg/h
solvent flow rate =3.48 L/h =2.84 kg/h
solvent-to-water flow ratio = 1.50 kg/kg = 1.84 L/L
solvent density = 0.812 kg/L (25°C)
shaft rotation speed = 750 rpm
estimated droplet size = 0.5 mm (approx.)
column pressure (top) = 400 kPa
average column temperature = 18°C (approx.)
rotor disc diameter = 3.81 cm
stator hole diameter = 5.72 cm
a. Gas Chromatograph (ppm, unless otherwise indicated)
Steady State Samples
Component
methanol
formic acid
ethanol
acetone
isopropanol
n-propanol +
isobutyraldehyde
acetic acid
methyl ethyl ketone
2-butanol
propionic acid
amyl alcohol
Water
Feed
85
940
594
39
197
225
744
318
298
139
0
Raffinate
31
230
73
7
3
4
113
2
4
-
17,300
Extract
27
158
330
19
115
152
634
174
169
-
_
Percent
Removal
63
75
88
83
99
98
85
99
99
-
_
b- Acid/Base Titration
water feed = 0.0346 g equiv. acid/L
steady state raffinate = 0.0069 equiv. acid/L
72
(continued)
-------
TABLE 20. (continued)
steady state extract = 0.014 g equiv. acid/L
solvent feed = 0.0004 g equiv. acid/L
steady state removal of acid = 80%
c. COD Analysis (ppm)
COD TOD TOD
2
water feed 6350 5339 5339 1.19
steady state raffinate 49700 49200 444 1.01
whole-run raffinate 44500 48100 1643 0.93
TOD: with arayl alcohol included
73
-------
TABLE 21. CONDITIONS AND RESULTS FOR MINI-PLANT EXTRACTION
RUN 4 -
wastewater = F
solvent = 2-ethyl hexanol (95%+)
water flow rate =1.74 L/h =1.74 kg/h
solvent flow rate =6.21 L/h =5.18 kg/h
solvent-to-water ratio = 3.00 kg/kg = 3.57 L/L
solvent density = 0.829 kg/L (25°C)
shaft rotational speed = 1090 rpm
estimated droplet size = 1 mm diameter (approx.)
column pressure (top) = kPa
average column temperature = 24°C
rotor disc diameter = 3.81 cm
stator hole diameter = 5.72 cm
a. Gas Chromatograph (ppm)
Component
methanol
acetaldehyde
ethanol
MCA
acetic acid
formic acid
DCA
chloral
crotonaldehyde
2-ethyl hexano.1
Water
Feed
76
150
174
7040
16500
1880
Solvent
Feed
8.5
~0
67
0
0
0
Steady
State
Raf finate
67
4
143
291
2680
863
Steady
State
Extract
12
62
77
3130
5740
463
Percent
Removal
12
98
18
96
84
54
6880
188
3010
97
0.14%
b. Acid/Base Titration
water feed: 0.316 g equiv. acid/L
solvent feed: 0.0004 g equiv. acid/L
steady state raffinate: 0.063 g equiv. acid/L
steady state extract: 0.016 g equiv. acid/L
steady state removal of total acid = 80%
(continued)
74
-------
TABLE 21. (continued)
c. COD Analysis (ppm)
TOD2
(without solvent) (with solvent) COD COD/TOD2
water feed 36700 36800 34500 0.94
steady state raffinate 4140 8260 7360 0.89
percent removal 89% 78% 79%
75
-------
TABLE 22. CONDITIONS AND RESULTS FOR MINI-PLANT EXTRACTION
RUN 5 —
wastewater = F
solvent = 50:50 (by volume) cyclohexanone and 2-ethyl hexanol
water flow rate =1.89 L/h = 1.89 kg/h
solvent flow rate =3.54 L/h =3.15 kg/h
solvent-to-water ratio = 1.65 kg/kg = 1.87 L/I
solvent density = 0.89 kg/L
shaft rotational speed = 638 rpm
estimated droplet size = 0.5 mm (approx.)
column pressure (top) = 370 kPa
average column temperature = 21°C
rotor disc diameter = 3.81 cm
stator hole diameter = 5.72 cm
run time = 2 hr, 32 min
a. Gas Chromatograph (ppm)
Component Steady State Removal
acetic acid 80%
acetaldehyde 76%
ethanol 43%
monochloroacetaldehyde 71%
dichloroacetaldehyde
chloral
very high
crotonaldehyde
b. Acid/Base Titration
water feed= 0.318 g. equiv. acid/L
solvent feed = 0.003 g equiv. acid/L
steady state raffinate = 0.063 g equiv. acid/L
steady state extract = 0.178 g equiv. acid/L
steady state removal of total acid = 80%
c. COD Analysis
water feed = 26000 ppm ;—£ =0.90
steady state raffinate = 120,500 ppm
76
-------
TABLE 23. CONDITIONS AND RESULTS FOR MINI-PLANT EXTRACTION
RUN 6
Conditions
wastewater = F
solvent = 50:50 (by volume) cyclohexanone and Alamine 336
water flow rate = 1.89 L/h =1.89 kg/h
solvent flow rate =2.20 L/h = 1.90 kg/h
solvent-to-water =1.00 kg/kg = 1.16 kg/h
solvent density = 0.864 kg/L
shaft rotational speed:
(first 10 minutes)—413 rpm (minimum)
(next 10 minutes)— 525 rpm
(final 50 minutes)—413 rpm
estimated droplet size:
(bottom of column)—0.7 mm (approx.)
(top of column)—1.0-1.5 mm
column pressure (top) = 200 kPa
average column temperature = 26°C
stator hole diameter = 5.72 cm
run time =70 minutes
a. Gas Chromatograph
Compound
acetic acid
acetaldehyde
ethanol
monochloroacetaldehyde
Water
Feed
20,500
401
154
5,870
Steady
State
Raffinate
2980
217
88
2750
Steady
State
Extract
19800
66
81
3140
Percent
Removal
85
46
43
53
b. Acid/Base Titration
Steady State:
water feed: 0.358 g equiv. acid/L
solvent feed (blank): 0.00864 g equiv.
raffinate: 0.041 g equiv. acid/L
extract: 0.444 g equiv. acid/L
steady-state removal of acid =89%
77
acid/L
(continued)
-------
TABLE 23. (continued)
c. COD Analysis
water feed = 32,300 ppm (COD/TOD = 1.09)
steady state raffinate = 58,200 ppm
pure water equilibrated with solvent = 189,500 ppm (centrifuged,
but remained cloudy with a slight yellowish tint)
78
-------
TABLE 24. CONDITIONS AND RESULTS FOR MINI-PLANT EXTRACTION
RUN 7
wastewater = F
solvent = 75:25 (by volume) Alamine 336/2-heptanone
water flow rate = 1.89 L/h =1.89 kg/h
solvent flow rate =2.38 L/h = 1.89 kg/h
solvent-to-water ratio = 1.00 kg/kg = 1.26 L/L
density of solvent = 0.812 kg/L
shaft rotational speed:
(first 1-3/4 hours): 413 rpm (minimum)
(next 2 hours): 600 rpm
estimated droplet size:
(first 1-3/4 hours):
top of column: ~1 mm (wide distribution)
bottom of column: 1-4 mm (wide distribution)
13 cm below bottom stator: very tiny
(next 2 hours):
top of column: ~1.0 mm
bottom of column: ~0.5 mm
column pressure (top) = 200 kPa
average column temperature = 25°C
rotor disc diameter = 3.81 cm
stator hole diameter = 5.72 cm
total time for run = ~4 hours
a. Gas Chromatograph (concentration in ppm)
Compound
acetic acid
methanol
acetaldehyde
ethanol
monochloroacetaldehyde
dichloroacetaldehyde
chloral
crotonaldehyde
79
Water
Feed
15000
106
1320
227
e 4830
5510
Steady
State
Raffinate
6600
99
1110
205
4280
2980
Steady
State
Extract
10100
10
31
37
459
1760
/ .
Percent
Removal
60
6
16
9
12
46
.6
.7
_ _ n
(continued)
-------
TABLE 24. (continued)
b. Acid/Base Titration
water feed: 0.357 g equivalents of acid/L
steady-state raffinate: 0.156 g equivalents of acid/L
steady-state extract: 0.185 g equivalents of acid/L
steady-state removal of total acid = 56.3%
c. COD Analysis
water feed: 27,500 ppm
steady-state raffinate: ~30,800 ppm
whole-run raffinate: -26,000 ppm
pure water equilibrated with solvent: 18,100 ppm
80
-------
TABLE 25. CONDITIONS AND RESULTS FOR MINI-PLANT EXTRACTION
RUN 8
wastewater = F
solvent = 70:30 (by weight) Chevron Solvent 25/trioctyl phosphine
oxide (TOPO)
water flow rate =1.89 L/h =1.89 kg/h
solvent flow rate =2.38 L/h =2.06 kg/h
solvent-to-water ratio =1.09 kg/kg =1.26 L/L
solvent density = .0.8627 kg/L
shaft rotation speed:
(first 10 minutes): 860 rpm
(final 122 minutes): 1090 rpm
estimated droplet size = 0.75 mm diameter (approx.)
column pressure (top) = 200 kPa
average column temperature = 23°C
rotor disc diameter = 3.81 cm
stator hole diameter = 5.72 cm
run time = 2 hours, 12 minutes
a. Results
Compound
acetic acid
methanol
acetaldehyde
ethanol
acetone
monochloroacetaldehyde
Water
Feed
15000
96
167
214
Steady
State
Raf finate
3700
74
96
157
Steady
State
Extract
13700
32
44
54
Percent
Removal
75
23
43
26
5 -
4990 3390 1590
DCA
chloral 5660 1280 5040
crotonaldehyde
b. Acid/Base Titration
water feed = 0.330 g equivalents acid/L
solvent feed = 0.037 g equivalents acid/L
steady state raffinate = 0.094 g equivalents acid/L
steady state extract = 0.293 g equivalents acid/L
81
32
77
(continued)
-------
TABLE 25. (continued)
pure water (blank) equilibrated with solvent = 0.037 g equivalents
acid/L
steady state removal of total acid = 71%
82
-------
TABLE 26. CONDITIONS AND RESULTS FOR MINI-PLANT EXTRACTION
RUN 9
Conditions
solvent: 50:50 (by volume) Alamine 336 and 2-ethyl hexanol
wastewater: G
water flow rate: 1.9 L/h =1.9 kg/h (continuous phase)
solvent flow rate: 1.3 L/h =1.0 kg/h (dispersed phase)
solvent-to-water ratio: 0.53 kg/kg
solvent density: 0.816 kg/L
extraction factor (acetic acid):
shaft speed: 1000 rpm
column pressure (top): 170 kPa
rotor disc diameter: 3.81 cm
stator hole diameter: 5.72 cm
run time: 5 h
a. Gas Chromatograph
KDS/W = 2
Component
formic acid
methanol
acetaldehyde
ethanol
acetonitrile
acetone
acetic acid
MEK
2-butanol
1-butanol
benzene
b. Acid/Base
water feed:
steady state
steady state
steady state
Water
Peed
2,450
409
213
506
721
1,538
51,300
470
512
561
Titration
Steady
State
Raffinate
0
359
146
413
650
750
4556
296
127
8
Steady
State
Extract
8,230
77
74
175
136
3,417
65,800
220
502
733
235
Percent
Removal
100
3.4
31.5
18.4
9.8
51.2
91.1
37.0
75.2
98.6
0.906 g. equivalents/L
raf finate :
extract: 1
removal of
0.049 g. equivalent s/L
.040 g. equivalents/L
acid = 96%
83
-------
TABLE 27. CONDITIONS AND RESULTS FOR MINIPLANT EXTRACTION
RUN 1Q
Conditions
solvent: 50:50 (by volume) Alamine 336 and 5-methyl-2-hexanone
wastewater: G
Flow rates: First steady state
water feed 1.47 L/h = 1.47 kg/h (continuous phase]
solvent feed 1.11 L/h =0.92 kg/h (dispersed phase)
Second steady state
water feed 2;69 L/h =2.69 kg/h (continuous phase;
solvent feed 2.41 L/h =2.0 kg/h (dispersed phase)
solvent-to-water ratio:
First steady state =0.63 kg/kg
Second steady state = 0.74 kg/kg
Extraction factor (acetic acid): KpS/W
First steady stats = 1.6
Second steady state = 1.9
shaft speed: 1000 rpm
column pressure: 101 kPa
rotor disc diameter: 3.81 cm
stator hole diameter: 5.72 cm
run time: 7 h
a. Gas Chromatograph (ppm) - 1st Steady State
Steady Steady
Water State State Percent
Component Feed Raffinate Extract Removal
formic acid 3,763 184 3,657 95.2
methanol 411 412 79 1.8
acetaldehyde 231 156 131 33.9
ethanol 502 426 172 16.9
acetonitrile 747 503 428 34.0
acetone 1,620 1,374 980 17.4
acetic acid 49,360 9,186 64,500 81.4
n propanol + MEK 389 170 393 57.2
(continued)
84
-------
TABLE 27. (continued)
oueetuy oteauy
Water State State Percent
Component Feed Raffiriate Extract Removal
2-butanol 459 122 551 74.0
1-butanol 411 49 562 88.3
Benzene + unknown 242 ~0 484 100.0
b. Acid/Base Titration - 1st Steady State
steady state water feed: 0.904 g«equivalents/liter
steady state raffinate: 0.157 g*equivalents/liter
steady state extract: 0.958 g-equivalents/liter
steady state removal of acid = 82.6%
c. COD Analysis - 1st Steady State
COD TOD TOD/COD
water feed 57000 63600 1.10
raffinate (centrifuged) 22800 15900
% COD removal = 60%
a. Gas Chromatograph (ppm) - 2nd Steady State
Steady Steady
Water State State Percent
Component Feed Raffinate Extract Removal
formic acid 3,763
methanol 411 384 25 6.6
acetaldehyde 231 122 63 47.2
ethanol 502 396 111 21.1
acetonitrile 747 406 128 45.6
acetone 1,620 1,225 24.5
acetic acid 49,360 5,814 65,060 88.0
n prop + MEK 389 145 204 62.7
2-butanol 459 116 364 74.7
1-butanol 411 22 435 94.6
benzene + unknown 242 ~0 211 100.0
(continued)
85
-------
TABLE 27. (continued)
b. Acid/Base Titration - 2nd steady state
steady state water feed: 0.904 g«equivalents/L
steady state raffinate: 0.114 g-equivalents/L
steady state extract 0.798 g-equivalents/L
steady state removal of acid = 87.4
86
-------
TABLE 28. CONDITIONS AND RESULTS FOR MINIPLANT
RUN 11
Conditions
Solvent: 50:50 (by volume) Alamine 336 and Diisobutyl
ketone (DIBK)
Wastewater: G
Flow rates: water feed: 1.25 liter/h = 1.25 kg/h
solvent feed: 1.22 liter/h =1.00 kg/h
raffinate: 1.21 liter/h =1.21 kg/h
extract: 1.25 liter/h =1.02 kg/h
Solvent-to-water ratio: 0.80 kg/kg
Solvent density: 0.818 kg/L
Extraction Factor
(Acetic Acid): Kn Fs/Fw = 2.64, K-. = 3.3 (at 5% wt%
D u Acetic Acid)
Shaft Speed: 1000 RPM
Column pressure (top): Atmospheric
Rotor disc diameter: 3.81 cm
Stator hole diameter: 5.72 cm
Estimated solvent hold-up: 0.022
Height of mass transfer zone: 78.5 cm
Run time: 5h
Results of_ Analysis
a. Gas Chromatograph
steady steady
Inter state state Removal
face feed raffinate extract %
Formic
Acid - ppm 3760 ppm - ppm - ppm ^100.%
(continued)
87
-------
TABLE 28. (continued)
Results of Analysis (continued)
Methanol
Acetal-
dehyde
Ethanol
Acetoni-
trile
Acetone
Acetic
Acid 23
n-prop
+MEK
2-butanol
1-butanol
unknown
(Benzene
+ other)
Inter-
face
407
145
464
604
1721
,222 49
253
216
121
0
feed
411
231
502
747
1620
,360
389
459
411
242
steady
state
raf f inate
400
154
450
556
1413
13,560 47
246
188
68
steady
state
extract
64
50
123
259
711
,760
312
410
401
386
Removal
%
5.6
35.5
13.2
29.0
15.6
73.4
38.8
60.4
84.0
100.0
steady
state non-
closure %
-6.9
+17.8
"" O • 8
-0.3
-20.2
-5.5
-26.7
-12.5
+4.4
-30.1
b. Acid/Base Titration
Steady State:
water feed:
solvent feed:
raffinate:
extract:
interface:
0.904 g. equivalents/liter
0.0 g. equivalents/liter
0.204 g. equivalents/liter
0.715 g. equivalents/liter
0.312 g. equivalents/liter
(continued)
88
-------
TABLE 28. (continued)
Run 11 (continued)
Steady state removal of acid: 78.2%
Steady state acid non-closure: -0.7%
C ~ C i
c. Jump ratio: ^ interface = Q>846
CF ~ ^outlet
d. Entrainment:
DIBK = 360 ppm
Alamine 336 = 275 ppm
89
-------
TABLE 29. RESULTS FOR MINIPLANT RUN 12
Conditions
Solvent: 50:50 (by volume) Alamine 336 and DIBK
Wastewater: G
Flow rates: water feed: 1.26 liter/hr =1.26 kg/hr
solvent feed: 1.26 liter/hr =1.00 kg/hr
raffinate: 1.21 liter/hr =1.21 kg/hr
extract: 1.21 liter/hr =0.99 kg/hr
Solvent-to-water ratio: .79 kg/kg
Solvent density: .816 kg/L
Extraction Factor (Acetic Acid): K_ Fs/Fw = 2.64
Shaft Speed: 580 RPM
Column Pressure: Atmospheric
Rotor disc diameter: 3.81 cm
Stator hole diameter: 5.72 cm
Estimated Solvent hold-up: 0.011
Height of mass transfer zone: 81.0 cm
Run time: 5h
Results of Analysis
a. Gas Chromatograph
steady steady
state state Inter- Removal
feed raffinate extract face %
Formic
Acid 3760 ppm - ppm - ppm - ppm
Methanol 411 409 58 407 4.4
(Continued)
90
-------
TABLE 29. (continued)
Results of Analysis (continued)
Acetal-
dehyde
Ethanol
Acetoni-
trile
Acetone
Acetic
Acid
n-prop
MEK
2-butanol
1-butanol
feed
231
502
747
1620
49,360
389
•%
459
411
steady
state
raffinate
162
448
581
1288
20,770
234
194
73
steady
state
extract
61
109
261
708
45,176
233
351
432
Inter-
face
187
459
614
1430
27,600
309
280
159
Removal
%
32.5
14.3
25.3
23.6
59.6
42.2
59.4
82.9
unknown
(Benzene
+ other) 242
410
100.
b. Acid/Base Titrations
steady state:
water feed: 0.904 g. equivalent/liter
solvent feed: 0.078 g. equivalent/liter
0.304 g. equivalent/liter
0.671 g. equivalent/liter
raffinate:
extract:
interface:
0.450 g. equivalent/liter
steady state acid removal:
67.7%
(Continued)
91
-------
TABLE 29. (continued)
Run 12 (continued)
c. Jump ratio: ^F " ^interface
CF " goutlet
d. Entrainment:
DISK = 9 ppm
Alamine 336 =38 ppm
92
-------
TABLE 30. CONDITIONS AND RESULTS FOR MINIPLANT RUN 13
Conditions
Solvent: 50:50 (by volume) Alamine 336 and DISK
Wastewater: G
Flowrates: water feed: 1.26 liter/hr = 1.26 kg/hr
solvent feed: 1.23 liter/hr = 1.00 kg/hr
raffinate: 1.23 liter/hr =1.23 kg/hr
extract: 1.30 liter/hr = 1.06 kg/hr
Solvent-to-water ratio: .79 kg/kg
Solvent density: 0.816 kg/liter
Extraction Factor (Acetic Acid): KD Fs/Fw =2.64
Shaft speed: 720 ppm
Column pressure: Atmospheric
Rotor disc diameter: 3.81 cm
Stator hole diameter: 5.72 cm
Estimated solvent hold-up: 0.011
Height of mass transfer zone: 75.5
Run time: 3h
Results of_ Analysis;
a. Gas Chromatograph
steady steady
state state Inter- Removal
feed raffinate extract face %
Formic
Acid 3760 ppm - ppm - ppm - ppm
Methanol 411 378 49 424 10.2
(Continued)
93
-------
TABLE 30. (continued)
Results of Analysis (continued)
Acetal-
dehyde
Ethanol
Acetoni-
trile
Acetone
Acetic
Acid 49
n-prop
+MEK
2-butanol
1-butanol
unknown
(Benzene
+ other)
feed
231
502
747
1620
,360
389
459
411
242
steady
state
raffinate
151
428
551
1193
21,770
134
173
64
0
steady
state
extract
63
120
251
680
42,510 31
275
441
397
388
Inter-
face
186
481
639
1377
,557
318
291
145
0
Removal
%
36.2
16.8
27.9
28.1
56.9
66.3
63.2
84.8
100.0
b. Acid/Base Titration
steady state:
water feed: .904 g. equivalents/liter
solvent feed: 0.0 g. equivalents/liter
raffinate: 0.285 g. equivalents/liter
extract: 0.659 g. equivalents/liter
interface: 0.441 g. equivalents/liter
steady state acid removal:
69.2%
(continued)
94
-------
TABLE 30. (continued)
Run 13 (continued)
C - C
F interface _ n
c. Jump ratio: g—_ g - "•
F outlet
d. Entrainment:
DIBK = 0 ppm
95
-------
TABLE 31. MINIPLANT RUN 14
Conditions
Solvent: 50:50 (by volume) Alamine 336 and DIBK
Wastewater: G
Flowrates: water feed: 1.21 liter/hr =1.21 kg/hr
solvent feed: 1.22 liter/hr =1.00 kg/hr
raffinate: 1.15 liter/hr =1.15 kg/hr
extract: 1.26 liter/hr =1.03 kg/hr
Solvent-to-water ratio: .83 kg/kg
Solvent density: 816 kg/L
Extraction Factor (Acetic Acid) KQ Fs/Fw =2.74
Shaft speed: 850 RPM
Column pressure: Atmospheric
Rotor disc diameter: 3.81 cm
Stator hold diameter: 5.72 cm
Estimated solvent hold-up: .008
Height of mass transfer zone: 74.8 cm
Run time: 4h
Results of Analysis
a. Gas Chromatograph
steady steady
state state Inter- Removal
feed raffinate extract face %
Formic
Acid - ppm - ppm - ppm - ppm
Methanol 376 391 43 391 1.2
(continued)
96
-------
TABLE 31. (continued)
Results of Analysis (continued)
Acetal-
dehyde
Ethanol
Acetoni-
trile
Acetone
Acetic
Acid 47
n-prop
+MEK
2-butanol
1-butanol
unknown
(Benzene
+ other)
feed
175
523
675
1610
,850
424
497
389
243
steady
state
raffinate
151
414
510
1428
14,675
191
192
95
0
steady
state
extract
75
167
255
635
45,700
298
386
401
273
Inter-
face
180
419
579
1269
21,415
227
243
142
0
Removal
%
18.0
24.8
28.2
15.7
70.9
57.2
63.3
76.8
100.
steady
state non-
closure %
-18.5
-2.4
-4.0
-17.9
-10.5
-2.6
-2.8
-11.0
-4.4
b. Acid/Base Titration
steady state:
water feed:
solvent feed:
extract:
raffinate:
interface:
.898 g. equivalents/liter
.016 g. equivalents/liter
.662 g. equivalents/liter
.241 g. equivalents/liter
.377 g. equivalents/liter
steady state acid removal: 74.5%
steady state acid non-closure: +0.4%
(continued)
97
-------
TABLE 31. (continued)
Run 14 (continued)
C - C
-,„„ ts,4.,'« F interface
c. Jump Ratio: _ j,
F ~ outlet
d. Entrainment:
DIBK =131 ppm
98
-------
TABLE 32. CONDITIONS AND RESULTS FOR MINIPLANT
RUN 15
Conditions
Solvent: 50:50 (by volume) Alamine 336 and DIBK
Wastewater: 6
Flowrates: water feed: 1.25 liter/h = 1.25 kg/h
solvent feed: 1.26 liter/h =1.00 kg/h
raffinate: 1.18 liter/h = 1.18 kg/h
extract: 1.28 liter/h =1.06 kg/h
Solvent-to-water ratio: 0.80
Solvent density: .818 kg/L
Extraction Factor: K_ Fs/Fw =2.64
Shaft speed: 530 RPM
Column pressure: Atmospheric
Rotor disc diameter: 4.45 cm
Stator hole diameter: 5.08 cm
Estimated solvent hold-up: .028
Height of mass transfer zone: 74.8 cm
Run time: 4.5h
Results of Analysis
a. Acid/Base Titration
steady state:
water feed: .935 g. equivalents/liter
solvent feed: .010 g. equivalents/liter
(continued)
99
-------
TABLE 32. (continued)
Results of Analysis
steady state (continued)
raffinate: .214 g. equivalents/liter
extract: .691 g. equivalents/liter
interface: .396 g. equivalents/liter
steady state acid removal: 78.4%
steady state acid non-closure: +2.4%
C - C
K -r,,™rx o=4-^« F interface -,0
b. Jump Ratio: « „ = .748
F " outlet
c. Entrainment:
DIBK= 108 ppm
100
-------
TABLE 33. CONDITIONS AND RESULTS FOR MINIPLANT
RUN 16
Conditions
Solvent: 50:50 (by volume) Alamine 336 and DIBK
Wastewater: G
Flowrates: water feed: 3.0 liter/h =3.0 kg/h
solvent feed: 2.9 liter/h = 2.4 kg/h
raffinate: 2.8 liter/h =2.8 kg/h
extract: 3.1 liter/h =2.6 kg/h
Solvent-to-water ratio: 0.80 kg/kg
Solvent density: .818 kg/L
Extraction Factor: K_ Fs/Fw =2.64
Shaft speed: 530 rpm
Column pressure: Atmospheric
Rotor disc diameter: 4.45 cm
Stator hole diameter: 5.08 cm
Estimated solvent hold-up: .029
Height of mass transfer zone: 74.8 cm
Run time: 3h
Results of_ Analysis
a. Acid/Base Titrations
steady state:
water feed: .935 g. equivalents/liter
solvent feed: .010 g. equivalents/liter
(continued)
101
-------
TABLE 33. (continued)
Results of Analysis
steady state (continued)
raffinate: .191 g. equivalents/liter
extract: .715 g. equivalents/liter
interface: .419 g. equivalents/liter
steady state acid removal: 80.9%
steady state acid non-closuret +0.7%
b. Jump Ratio: !? " /"terface = .694
CF outlet
c. Entrainment:
DIBK - 265 ppm
102
-------
TABLE 34. CONDITIONS AND RESULTS FOR MINIPLANT
.RUN 17
Conditions
Solvent: 50:50 (by volume) Alamine 336 and DIBK
Wastewater: G
Flowrates: water feed: 3.53 liter/h =3.53 kg/h
solvent feed: 3.45 liter/h = 2.82 kg/h
raffinate: 3.35 liter/h =3.35 kg/h
extract: 3.58 liter/h =2.93 kg/h
Solvent-to-water ratio: 0.80 kg/kg
Solvent density: 0.818 kg/L
Extraction Factor (Acetic Acid): KD Fs/Fw =2.64
Shaft speed: 530 RPM
Column pressure: Atmospheric
Rotor disc diameter: 4.45 cm
Stator hole diameter: 5.08 cm
Estimated solvent hold-upi 0.020
Height of mass transfer zone: 74.8 cm
Run time: 4h
Results of_ Analysis
a. Acid/Base Titrations
steady state:
water feed: 0.937 g. equivalents/liter
solvent feed: 0.010 g. equivalents/liter
(continued)
103
-------
TABLE 34. (continued)
Results of Analysis
steady state (continued)
raffinate: 0.220 g. equivalent/liter
extract: 0.705 g. equivalent/liter
interface: 0.459 g. equivalent/liter
steady state acid removal: 77.7%
steady state acid non-closure: -1.4%
b. Jump ratio: = 0.67%
CF outlet
c. Entrainment:
DIBK = 427 ppm
104
-------
TABLE 35. CONDITIONS AND RESULTS FOR MINIPLANT
RUM 18
Conditions ~~~ "—~
Solvent: 50:50 (by volume) Alamine 336 and DIBK
Wastewater: G
Flowrates: water feed: 2.93 liter/h = 2.93 kg/hr
solvent feed: 2.87 liter/h =2.34 kg/hr
raffinate: 2.81 liter/h =2.81 kg/hr
extract: 2.99 liter/h =2.45 kg/hr
Solvent-to-water ratio: 0.80 kg/kg
Solvent density: 0.816 kg/L
Extraction Factor (Acetic Acid): K- Fs/Fw =2.64
Shaft speed: 460 RPM
Column pressure: Atmospheric
Rotor disc diameter: 4.45 cm
Stator hole diameter: 5.08 cm
Estimated solvent hold-up: 0.026
Height of mass transfer zone: 74.8 cm
Run time: 3.5h
Results o_f Analysis
a. Acid/Base Titrations
steady state:
water feed: .942 g. equivalents/liter
solvent feed: .009 g. equivalents/liter
(continued)
105
-------
TABLE 35. (continued)
Results of Analysis
steady state (continued)
raffinate: .253 g. equivalents/liter
extract: .666 g. equivalents/liter
interface: .495 g. equivalents/liter
steady state acid removal: 74.2%
steady state acid non-closure; +2.1%
b. Jump Ratio: gF " ^interface .
F outlet
c. Entrainment:
DIBK = 186 ppm
106
-------
TABLE 36. CONDITIONS AND RESULTS FOR MINIPLANT
RUN 19
Conditions
Solvent: 50:50 (by volume) Alamine 336 and DIBK
Wastewater: G
Flowrates: water feed: 3.0 liter/h =3.0 kg/h
solvent feed: 2.96 liter/h =2.4 kg/h
raffinate: 2.84 liter/h =2.84 kg/h
extract: 2.98 liter/h =2.44 kg/h
Solvent-to-water ratio: 0.80 kg/kg
Solvent density: .818 kg/L
Extraction Factor (Acetic Acid): KD Fs/Fw =2.64
Shaft speed: 530 RPM
Column pressure: Atmospheric
Rotor disc diameter: 4.45 cm
Stator hole diameter: 5.08 cm
Estimated solvent hold-up: 0.026
Height of mass tranfer zone: 74.8 cm
Run time: 4.5h
Results of_ Analysis
a. Acid/Base Titration
steady state:
water feed: .934 g. equivalents/liter
solvent feed: .009 g. equivalents/liter
(continued)
107
-------
TABLE 36. (continued)
Results of Analysis
steady state (continued)
raffinate: .224 g. equivalents/liter
extract: .739 g. equivalents/liter
Interface: .472 g. equivalents/liter
steady state acid removal: 77.3%
steady state acid non-closure: 0.0%
b. Jump Ratio: ^ " ^interface .
HP outlet
c. Entrainment:
DIBK = 178 ppm
Alamine 336 - 145 ppm
108
-------
SECTION 7
EXTRACTION PROCESSES FOR ACETIC ACID RECOVERY
This Section examines in detail the impact of a number of
different factors on the process design and costs. Emphasis is
on extraction processes for selective recovery of acetic acid
from wastewaters not containing chlorinated acetaldehydes. The
presence of chlorinated acetaldehydes in the Wacker-process
wastewaters results in some special problems which are discussed
in Section 8.
SOLVENT VOLATILITY CONSIDERATIONS
As noted earlier, solvent-regeneration is often the most
costly step in an extraction process. Distillation is the
most common regeneration method. If the, solvent is to be
distilled, its volatility is a major design consideration,
affecting both the overall process configuration and the
process energy consumption.
In the past, lower-boiling solvents such as ethyl acetate
[Normal Boiling Point (NBP) = 77°C] have been favored for acetic
acid recovery (Brown, 1963; Eaglesfield, et al., 1953). A typi-
cal process configuration is shown in Figure 11. Note that
water, solvent, and any volatile solvents co-extracted with
acetic acid are separated from the acetic acid in a single
distillation step (the dehydration column).
Figure 12 shows the typical configuration for a
process using a higher-boiling solvent. A two-step
regeneration is necessary. The first step removes water
and any volatile solutes that would otherwise contaminate
the acetic acid product. The second separates acetic acid
from the solvent.
The'lower-boiling solvents have the following general
advantages: (1) Since the solvent is taken overhead, it
is continuously separated from soluble, non-volatile
contaminants. (2) The distillation system operates at low
temperatures, so reboiler energy can be supplied by low-
pressure steam or possibly by low-level heat from another
process stream. The maximum temperature is 118°C at
109
-------
Make-up Solvent
Solvent Recycle
Aq.
Acetic Acid
Feed jfr f
Extractor
t
cicl
:id
>duct
)~
-* I
^5
Dehydr
Colu
«-==a
=:
Decanter
Steam
r
Water to
Waste
Solvent
Stripping
Column
Steam
k-CD
Figure 11. Extraction of acetic acid with a lower-boiling solvent.
-------
Solvent +
Acetic Acid
Aq.
Acetic Acid
Feed J--J
Extractor
Make-up
Solvent
Oeconter
4 -_
Aq.
Phose
Regeneration
Column
Sfeom
kCD
Solvent*
Acetic Acid
Decanter
Solvent
Recycle
-> Glociel Acetic
Acid Product
Fracfionotor
H
Wofor
to
Waste
Sfeom
KID
Solvent
Stripper
Steam
KZD
Figure 12. Extraction of acetic acid with a higher-boiling solvent.
-------
atmospheric pressure if the bottom product is mainly
acetic acid. (3) The process is relatively simple,
requiring one less regeneration step than that for higher-
boiling solvents. However, if non-volatile solutes are
co-extracted with acetic acid, the acid product shown in
Figure 11 may require additional purification, depending
on its intended end-use. In this work, it was assumed
that the extraction process need not produce high-purity
acid, because the recovered acid could be recycled to the
main separation system in the acetic-acid production
facility for final purification.
The higher-boiling solvents have advantages as well.
One possible advantage is a higher K , as discussed in the
next section. Other advantages of higher-boiling solvents,
independent of K considerations, are: (1) the process
shown in Figure 12 can produce a high-purity acid product
(not important in this work), and (2) the solvent is the
bottom product in the distillation. In a wastewater-
extraction process, the solvent is the major component
(typically about 90 wt.%) in the feed to the regeneration
system. If the solvent is the overhead product, as for
lower-boiling solvents, larger condensers and reboiler
duties and a larger column diameter are required.
The relative volatility of the solvent to the solute
is another important consideration. If possible, the
solvent should be either much more or much less volatile
than the solute to be considered; this reduces the number
of stages, the tower diameter, and the energy required
to effect the distillation. The relative volatility
should be at least 2.0 and preferably much higher. Ethyl
acetate is a good choice in this respect, since its
volatility relative to acetic acid is approximately 3.5
at atmospheric pressure. Higher-boiling solvents
can provide similar separation factors with, however, the dis-
advantage of higher operating temperatures or vacuum operation.
The separation factors for regeneration of amine extractants
are determined mainly by the choice of the diluent. For process
design, separation factors had to be measured experimentally
since no data were available and the systems are too non-ideal to
allow prediction by standard correlations. Results of such
measurements are given elsewhere (Ricker, 1979; Ricker et al,
1979b) .
SOLVENT KD AND SELECTIVITY
Many solvents have acceptable volatilities for the
regeneration step in an acetic-acid extraction process,
112
-------
but very few exhibit attractive K_ values in the extraction
step. Large K_ values are advantageous, up to K values
of about 20.0 ferhart, et al., 1977). Unfortunately?
KD values for acetic-acid extraction are usually on the
order of 1.0, or less.
Low KD values are especially typical of the lower-
boiling solvents. As discussed in Section 6, higher K
values are usually the result of a specific, chemical D
interaction between the extractant and the solute. If the
extractant is volatile, i.e., its molecular weight is low,
the extractant-acetic acid complex is almost certain to be
too water-soluble for practical applications (Wardell and
King, 1978) . Consequently, only the weakly-interacting
solvents in the lower-boiling class (e.g., ethyl acetate)
can be considered).
Higher-molecular-weight organics, on the other hand,
can be insoluble in water and yet have a specific strong
attraction for acetic acid. Examples are certain amines
and organophosphorus compounds. These provide higher K_ values
and good selectivity for acidic compounds, which simplifies
purification of the extracted solute.
SOLVENTS SELECTED FOR DETAILED PROCESS-DESIGN STUDY
Based on the above factors and on the data presented in
Section 6 the following specific solvents were chosen for a
detailed process-design study:
1. ethyl acetate, to represent the class of lower-boiling
solvents;
2. cyclohexanone, a weakly-interacting, higher-boiling
solvent;
3. Alamine 336 (tertiary Cg-C10 amine) diluted with DIBK, a
novel extractant in the strongly-interacting, higher-
boiling class; and
4. TOPO a semi-commercialized, strongly-interacting, higher-
boiling extractant.
The TOPO diluent was assumed to be 2-heptanone. Reasons
for selection of amine and TOPO diluents are discussed
in a subsequent section.
The purpose of the design study was to quantify the
effects of various solvent properties and process-design
alternatives. Approximate design methods were
used because data were scarce and rapid evaluation of
many alternatives was considered more important than
rigorous design. Accuracy of equipment costs is estimated
to be ± 30%.
113
-------
COST ESTIMATES
Estimated Direct Fixed Capital
Plant capital-investment estimates made extensive
use of the cost curves and the "process module" technique
proposed by Guthrie (1969) . Cost data for special types
of heat exchangers and other miscellaneous equipment items
came from other sources, such as Woods, et al. (1975) and
Peters and Timmerhaus (1968) . Costs were adjusted to
January, 1978, using a Marshall & Stevens cost index of.520.
Expected accuracy is ± 30%.
It was necessary to make a number of assumptions in
order to establish process enthalpy and material balances
and to size the equipment. For example, determination
of the extractor size and solvent flowrate for a given
wastewater required specification of the wastewater
flowrate, the percent recovery of pollutant desired and
an assumption regarding the optimum trade-off between
running the extractor near the minimum solvent rate to
reduce energy costs vs. a high rate to reduce the extractor
cost. Heuristics were usually used rather than full cost
optimizations in such cases, due to the large uncertainties
in the costs. In sizing extraction equipment, for example,
the design solvent rate was 1.5 times the minimum solvent
rate as recommended by Happel and Jordan (1975), whose
recommendations for equipment sizing for preliminary cost
estimates were very useful for other capital items as well.
After sizing, cost diagrams from Guthrie (1969) and
other sources were used to determine costs of major
equipment items. Multiplication by the module factors
recommended by Guthrie gave the total installed cost.
These were increased by 18% to account for contingency
plus contractor's fee. Costs for off-site development,
buildings, working capital, incremental investment for
steam, power, and cooling capacity, were assumed
negligible. The total of the above equipment costs is
the "direct fixed capital" (DFC).
Operating Costs
The plant design stream time is 8000 hours/year.
114
-------
The following items make up the annual operating costs:
1. Cost of Chemicals—Solvent make up (for losses to the
product water due to upsets, degradation reactions,
and to incomplete recovery in the regeneration step)
was charged at the market price for the solvent.
2. Utilities—as estimated by Petterson and Wells (1977)
for the year 1979-80:
utility
fuel
high pressure steam
(1500 psig)
low pressure steam
(50 psig)
cooling water, 30°F rise
electrical power
cost
$2.86/GJ ($3.00/10^ BTU)
$3.17/GJ ($3.33/10 BTU)
$1.52/GJ ($1.60/106 BTU)
3 BTU)
3
4
$0.56/GJ ($0.48/10
$0.03/KWH
Maintenance and Repairs—annual cost of 6% of DFC.
Operating Supplies—annual cost of 15% of maintenance
and repairs.
5. Depreciation—annual cost of 8% of DFC.
6. Insurance and Taxes—annual cost of 3% of DFC.
The above costs are summarized in the following
equation: annual operating costs = chemicals +
utilities + 0.18 DFC. Other operating costs (e.g.,
operating labor and laboratory charges) could have been
included but were not because an accurate estimation
method was not apparent, as discussed by Earhart, et al (1976)
They points out that these could amount to an additional
charge to the process of as much as $0.16/m wastewater
processed, which is not negligible. However, these costs
depend primarily on the wastewater flowrate and would not
change appreciably from one process alternative to the
next, so their exclusion should not affect the relative
evaluation of alternatives.
A detailed example of cost estimation and design for
one case is given elsewhere (Ricker, 1979).
Assumptions for Cost Estimates
The major assumptions used in the cost estimates are
the following:
1. The design capacity was 22,700 kg/h (100 GPM) of
butane-oxidation wastewater G. This seemed to be an
appropriate capacity, based on published data giving
the amount of wastewaters produced per unit of acetic
115
-------
acid production (Sittig, 1974); such a capacity would
be typical of large-scale treatment facility.
2. The extraction system was designed to recover 99% of
the acetic acid from the feed stream. This recovery
value was selected as a practical upper limit so as
to provide a large reduction in the COD due to the
acetic acid. As will be shown later, a lower recovery,
say 95%, would probably be selected if recovery of the
acid for profit were the only motive for installation of
the process.
3. In general, constant separation factors were assumed
in the extractors and distillation equipment. For
separations in which a large variation in the separation
factor was expected within the equipment, as in the
azeotropic separation of ethyl acetate and water from
acetic acid, a limiting value of the separation factor
was used to calculate the energy requirements, and the
size of the equipment was adjusted to reflect the
higher values that applied to certain sections of the
separation device. However, in the extractor
calculations for the Alamine 336/DIBK system, the
change in K_ with acetic acid concentration was
accounted for in a rigorous manner.
Limiting relative volatility and K_ values are given
in Table 37; these are from publisned or measured
data, if available. The data for the ethyl acetate
and cyclohexanone systems are well-established, and
accuracy is probably on the order of 5%. The data
for the amine and TOPO systems are less reliable,
since K_, solubilities, and vapor-liquid equilibria
can vary greatly with changes in process variables.
Large inaccuracies, on the order of 50% are conceivable.
4. The extractor is a sieve-tray column, chosen because
it can be designed more reliably than other types of
differential contactors (Treybal, 1963). An agitated-
column extractor, such as an RDC, might be preferable
in a full-scale process, because the extraction is
relatively difficult and requires approximately 10
equivalent equilibrium stages for a 99% recovery. A
mixer-settler cascade might also be considered.
5. To keep solvent loss to a minimum, the acetic-acid
product from the regeneration column was to contain
no more than 0.1 wt.% solvent (or diluent in the case
of the amine and TOPO systems). The regenerator bottom
product was to contain acetic acid equivalent to one-half
116
-------
TABLE 37. KEY SOLVENT PROPERTIES IISF.D TN DRSTCTJ STITHY
Solvent
ethyl acetate (1'3)
cyclohexanone '
50% Alamine 336
in DIBK
40% TOPO in^3*
2-heptanone
KD
1.05
1.45
1.0*
1.5
Solubility
in H20
7.1
2.4
>10 ppm
0.06
>1 ppm
0.10
(wt. %)
H20 in
3.5
7.0
0.6
1.0
a
3.9
0.33
0.30
0.29
Price
(SAg)
0.53
0.66
2.42
0.66
22.0
a is the limiting solvent/acetic acid relative volatility
References
1. Eaglesfield, et al. (1953)
2. Garner, et al (1954)
3. Michaels (1977)
* This is the value at the intermediate pinch-point (see Ricker,
1978, and Ricker, et al, 1979b).
117
-------
that in equilibrium with the acid in the extractor
raffinate, so as to provide an adequate mass-transfer
driving force at the lean end of the extractor.
6. The extractor raffinate was assumed to be saturated
with solvent and the extract saturated with water.
Solubilities are given in Table 37. They were
assumed independent of wastewater solute concentration,
since the solutes were quite dilute.
7. Steam stripping of the raffinate was assumed to produce
an aqueous effluent containing less than 50 ppm
dissolved solvent. For the TOPO and amine systems,
only the diluent was stripped; solubilities of TOPO and
amines were assumed to be negligible. The ethyl
acetate and cyclohexanone systems use atmospheric
strippers, and the remaining two are vacuum-stripping
columns.
8. The aqueous effluent was assumed to contain an average
of 20 ppm entrained solvent droplets. This is only a
rough assumption. Experimental results indicated that
the TOPO, amine, and cyclohexanone systems would
require more than primary settling by gravity to
achieve this level of entrainment. Estimated costs for a
coalescer are included in Tables 38-41.
9. The higher-boiling solvents require a purge system to
remove soluble non-volatiles from the solvent, as
discussed later. The net loss of extractant in the
purge system was assumed to be 1 kg/h. The estimated
installed cost of the system is included in Tables 38-41.
10. The dehydration system is operated such that the final
acetic-acid product contains less than 4.0 wt.% water
in all cases.
11. Stainless steel is required for equipment in contact
with hot, concentrated acetic acid (Brown, 1963) .
12. Steam was assumed to cost $1.52/GJ ($1.60/10 BTU).
Results of the Design Study
The capital investment, operating cost, and
profitability of each alternative are summarized in Tables
38 through 41. Operating costs are given both as a
charge per cubic meter of wastewater processed and as an
annual cost; depreciation, maintenance, etc., are included.
The direct fixed capital (DFC) includes a charge for a one-hour
solvent inventory.
118
-------
TABLE 38. ESTIMATED COST OF EXTRACTION OF ACETIC ACID BY
ETHYL ACETATE
Direct-fixed capital (DFC) k$ % of DFC
extractor (16.2 m x 2.21 m) 185 10.7
solvent-recovery column (15.2mxO.692m) 48 2.8
solvent-regeneration column (27.1 m x 2.56 m) 695 40.1
heat exchange 458 26.4
pumps, drums, tanks 20 1.1
contingency & contractor's fee 308 17.8
solvent inventory 18 1.0
1732
$/m3 % of
Operating costs k$/yr wastewater operating
steam (low-pressure) 300 1.65 42
cooling water 90 0.50 13
make-up solvent 10 0.05 1
depreciation, maintenance, etc. 311 1.71 44
711 3.91
value of acetic acid recovered = 2846 k$/yr
ROIBT = 123%
119
-------
TABLE 39. ESTIMATED COST FOR EXTRACTION OF ACETIC ACID
BY CYCLOHEXANONE
Direct fixed capital
extractor (16.2 m x 1.99 m)
sol vent -recovery column (13.7 m x 1.22 m)
solvent-dehydration column (6.1 m x 1.16 m)
solvent-regeneration column (32.6 m x 1.29
heat exchange
pumps , tanks , drums
coalescer
batch system for solvent purification
contingency & contractor's fee
solvent inventory
k$
160
72
96
m) 392
455
18
20
10
268
15
1506
% of
10.
4.
6.
26.
30.
1.
1.
0.
17.
1.
DFC
6
8
4
0
2
2
3
7
8
0
Operating costs k$/yr
steam (low-pressure) 118
cooling water 35
make-up solvent 17
depreciation, maintenance, etc. 271
441
$/m3
wastewater
0.65
0.19
0.09
1.50
2.43
%
27
8
4
61
value of acetic acid recovered = 2846 k$/yr
ROIBT = 160%
120
-------
TABLE 40. ESTIMATED COST FOR EXTRACTION OF ACETIC ACID
BY 40% TOPO IN 2-HEPTANONE
Direct fixed capital
extractor (16.2 m x 2.04 m)
solvent recovery column (12.2 m x 0.55 m)
solvent -dehydration column (6.1 m x 0.48 m)
solvent-regeneration column (28.7 m x 1.10 m)
heat exchanger
pumps , tanks , drums
coalescer
batch system for solvent purification
contingency and contractor's fee
solvent inventory
k$ %
169
29
66
316
300
18
20
20
198
207
1342
Of DFC
12.6
2.2
4.9
23.5
22.4
1.3
1.5
1.5
14.8
15.4
Operating cost k$/yr
steam (lovr pressure) 32
cooling water 9
solvent make-up 221
depreciation, maintenance, etc. 242
504
$/m3
wastewater
0.18
0.05
1.22
1.33
2.78
%
6
2
44
48
value of acetic acid recovered = 2846 k$/yr
ROIBT = 175%
121
-------
TABLE 41. ESTIMATED COST FOR EXTRACTION OF ACETIC ACID
BY 50% ALAMINE 336 in DIBE
Direct fixed capital
extractor (12.2 m x 2.32 m)
solvent-recovery column (12.2 m x 0.52 m)
solvent-dehydration column (6.1 m x 0.71 m)
solvent-regeneration column (26.8 m x 1.08 m)
heat exchanger
pumps, tanks, drums
coalescer
batch system for solvent purification
contingency and contractor ' s fee
solvent inventory
k$ %
160
29
56
308
376
18
20
16
215
17
1215
of DFC
13
2
5
25
31
2
2
1
18
1
Operating costs k$/yr
steam (low pressure) 42
cooling water 13
make-up solvent 34
depreciation, maintenance, etc. 219
308
$/m3
wastewater
0.26
0.08
0.21
1.35
1.90
%
14
4
11
71
value of acetic acid recovered = 2846 k$/yr
ROIBT = 220%
122
-------
The return on investment before taxes (ROIBT) is the annual
profit before taxes, i.e., the value of recovered acetic acid
less operating costs, divided by the DFC. Acetic acid is
valued at 80% of the current market price of $0.40/kg
(Chem. Market. Rep., 1978) for glacial acetic acid, since
the product from the process is only about 95% acetic acid
and must be recycled for further purification before it can
be marketed.
The height and diameter of the major process vessels
are in parentheses following the name of the item. Since
approximate design methods were used to size the equipment,
these dimensions should be taken as only rough estimates
of the values that would be used in a final design.
Costs for extraction by ethyl acetate are based on
the process configuration of Figure 11. The regeneration
system and steam costs are especially high for ethyl
acetate because the solvent is the overhead product in the
regeneration step. Even so, a high ROIBT is predicted,
due to the relatively high acetic acid concentration in
the feed wastewater (approximately 5 wt.%).
Effect of Feed Concentration and Feed Rate
At lower feed concentrations, the ROIBT for the
ethyl acetate process is less attractive. Below about
10 wt.%, the operating cost per cubic meter of wastewater
and the direct fixed costs (DFC) are nearly independent of
the acid feed concentration. Given this approximation, for a
certain plant investment and base operating cost, ROIBT varies
directly with the feed concentration.
Figure 13 shows the effect of acid feed concentration
on ROIBT for base operating costs ranging from $1.0/ijf3 to
$10.0/m3 of wastewater. The figure assumes a 22.7 m /hr (100
GPM) wastewater feed rate and a DFC of $1,730,000. The ethyl
acetate process breaks even at a feed concentration of about
1.3 wt.%; installation of the process is not economically
attractive until the feed concentration exceeds about 2.2 wt.%
(i.e., where ROIBT is greater than 30%).
Figure 13 can be adjusted for other DFC values by
ROIBT = (ROIBT) . 13.(1.73 x 10D/DFC) . Thus, the break-even
point for the Alamine 336-DIBK process is at a feed concentra-
tion of about 0.5 wt.%, and the ROIBT exceeds 30% at about 1.2
wt.%, i.e., at about one-half the concentration required for
the ethyl-acetate process.
The process costs are also a function of the assumed
feed rate. Direct-fixed-capital varies with approximately the
123
-------
150
100
CD
O
50
0 12345
WT % ACETIC ACID IN WASTEWATER
Figure 13. ROIBT for acetic acid recovery
vs. concentration of acid in
wastewater feed.
124
-------
0.7 power of plant capacity (Guthrie, 1969), and utilities and
solvent make-up vary with capacity. Thus, base DFC and operating
o?SJS 5" Tables 38-41 ca» be adjusted for capacities other than
22.7 m /h by:
DFC = (DFC)base(Q/22.7)0'7
OP = t(OP)base.182,000 - 0.18(DFC)base](Q/22.7) +
0.18(DFC)base(Q/22.7)°'7 /(8000 Q)
where DFC and (DFC). are in dollars, OP and (OP)
are the adjusted anS Else operating3costs in $/m waffewater,
and Q is the adjusted capacity in m /h. The factor 8,000
is the number of operating hours per year and the factor
0.18 is the fraction of the DFC taken for annual depreciation,
maintenance, etc.
If the wastewater feed ratg to the ethyl-acetate-
extraction process were 2.27 m /h, the adjusted DFC would
be $346jOOO and the adjusted operating cost would be
$5.63/m . To break even would then require a feed
concentration of about 1.8 wt.%, and a 30% ROIBT would
require a 3.6 wt.% feed. In general, as the wastewater
flowrate decreases, it becomes increasingly difficult to
justify acetic acid recovery on an economic basis.
Costs for the remaining solvents in the design study,
which are all less volatile than acetic acid, are based
on the process configuration of Figure 15, which appears later.
Estimated costs are lower than for the ethyl acetate process in
each case. '
Costs for Extraction with Cyclohexanone
Extraction with cyclohexanone, for example, is
estimated to be 40% less expensive than with ethyl acetate.
However, in pilot-plant tests, Eaglesfield, et al. (1953),
noted that the specific gravity of water-saturated
cyclohexanone (0.953) is quite close to that of water and
becomes even closer as acetic acid is extracted. This
results in low flooding velocities in the extractor and
long phase-separation times'. Coalescence is also slow ••
due to low interfacial tension, and fairly stable
water/solvent emulsions are easily produced under
agitation; the extent of these problems is difficult to \
predict without more extensive testing. The calculation
method used to size the extractor (Happel and Jordan,
1975) is more accurate for solvents having specific
gravities very different from that of water, and the
125
-------
estimated extractor cost in Table 39 may be too low.
The properties of cyclohexanone might be improved by
the addition of another organic solvent. For example, 50%
cyclohexanone in 2-ethyl hexanol gives a K of 1.20, i.e.,
only 18% less than for pure cyclohexanone TMichaels, 1976).
The specific gravity of this mixture is about 0.89, and
its mutual solubility with water, though not measured by
Michaels, should be much lower than for pure cyclohexanone.
Sanchez (1977) attained good acetic-acid removals and
encountered no special operating problems with this
mixture in an extraction experiment with wastewater F in
a miniplant RDC, as shown in Section 6- However, an alcohol
is unsuitable for a large-scale process because of its tendency
to esterify with acetic acid.
Other solvents might be combined with cyclohexanone
to form a mixture with favorable solvent properties.
Also, methyl cyclohexanone may be an attractive alternative,
as suggested by Eaglesfield, et al. (1953), although it does not
seem to be as readily available as cyclohexanone (Lawler, 1977).
However, the amine extractant system seemed to
offer greater promise, and the use of cyclohexanone and
methyl cyclohexanone was not investigated further.
Costs for Extraction with TOPO
It is interesting to compare the cost estimate for
TOPO with that for cyclohexanone. The K values for
acetic-acid extraction are very similar and the estimated
acetic acid/solvent relative volatilities are comparable.
The TOPO system co-extracts much less water, but this
advantage is more than offset by the very high cost of
TOPO, as evidenced by the solvent inventory and make-up
charges. It should be noted that the make-up charges are
very uncertain.
2-Heptanone was selected as the diluent for TOPO
because Michaels (1976) found that this combination
resulted in higher K values than for other common
diluents. Alternatively, an aliphatic hydrocarbon, such
as that used in the Hydroscience process (Helsel, 1976) ,
would reduce the water content of the extract and might allow
elimination of the solvent-recovery stripper. K would also
decrease, but the overall effect might be to reauce the process
costs. A lower concentration of TOPO might also be advanta-
geous, especially since the cost of the solvent would be
reduced. However, since the amine extractants seemed to have
some significant advantages over TOPO, little attempt was made
to optimize the TOPO solvent system.
126
-------
Comparison of Costs for TOPO and Amine Extractants
Under the assumptions used to develop Tables 38-41,
the Alamine 336/DIBK system is about 40% less
costly to operate than the TOPO system. The largest factor
differentiating the systems is the cost of the extractants.
The estimated cost for solvent make-up, for example, is
lower for the amine system by a factor of approximately 5.
It should be noted that the cost of TOPO may decrease
in the future due to increased production for use in
uranium recovery (anon., 1978) and other applications,
in which case the operating costs for the TOPO system
would compare more favorably with those of the amine system.
However, the costs for the Alamine 336 system could also be
reduced significantly.
SELECTION OF A DILUENT FOR THE ALAMINE 336 EXTRACTANT
Selection of the diluent for the Alamine 336
extractant is a critical design decision. The acetic acid
distribution coefficient and the extractant/water mutual
solubility are strong functions of the diluent.
Alcohol diluents provide very favorable extraction
equilibria, but are prone to esterification. Ketones, such
as 2-heptanone and DIBK, appear to offer the best compromise
between extraction equilibria, regenerability, and other
important factors.
However, selection of the optimum ketone diluent is
not a trivial task. For example, vapor-liquid equilibrium
measurements indicate that a C_ ketone, such as 2-heptanone,
would be too volatile for regeneration by distillation at
reasonable cost. The presence of the amine in the liquid
phase reduces the volatility of acetic acid, so that the
diluent must be less volatile than would be necessary if the
amine were not present.
The volatility of DIBK is more favorable than that
of 2-heptanone. However, the extraction equilibria are less
favorable for DIBK, especially for recovery of acetic acid from
dilute solutions (i.e., less than 1-2 wt.%). Of the two,
DIBK appears to be the better choice, but it is not an
ideal one.
Unfortunately, no other potential ketone diluents
were tested as extensively as were DIBK and 2-heptanone.
There are some interesting possibilities, including
cyclohexanone, methyl cyclohexanone, 3-octanone (ethyl amyl
ketone, Ashland Chem. Co.), and 2-nonanone (methyl heptyl
127
-------
ketone, Ashland Chem. Co.). Another approach would be the
addition of a third component to the Alamine 336/DIBK
system to enhance the extraction equilibria, as discussed in
Section 5.
In any case, if a diluent or mixture of diluents
could be found that gave extraction equilibria like those
shown in Figure 7 for 2-heptanone and yet was as
regenerable as DIBK, the DFC of the extraction process
of Table 41 would decrease by 33%, the operating cost by
31%, and the profitability, as measured by the ROIBT,
would be 326% rather than 220%. The major benefit would
be a reduction in the required solvent rate. The search
for such a diluent would seem to be an important area for
further research.
OPTIMIZATION OF THE ALAMINE 336/DIBK EXTRACTANT SYSTEM
Another way to reduce the cost of the process is to
optimize the operating conditions for a given amine/diluent
combination. The Alamine 336/DIBK system was optimized
with respect to the concentration of amine to be used in
the extractant phase, and the recovery of acetic acid to
be achieved in the extraction step. Details of this study
are given elsewhere (Ricker, 1978; Ricker et al, 1979b).
The optimum amine concentration appeared to be
between 30% and 50% by volume. The process costs were
found to be a weak function of amine concentration in
this range. One would probably choose a concentration at
the lower end of the range to minimize the solvent cost.
The optimum amine concentration was independent of the
concentration of acetic acid in the feed wastewater in
the range of 0-5 wt.% acid.
The costs were more sensitive to the design recovery;
the optimization procedure showed that from an economic
standpoint, the recovery should be on the order of 95-97%,
rather than 99%, for a wastewater containing 5 wt. % acetic
acid. This would reduce the DFC by about 11% and increase
the ROIBT by 5%, compared to the process of Table 41.
For a wastewater containing 1 wt.% acetic acid, the
optimum recovery of acid was 86%. Even at the optimum
conditions, the ROIBT was only about 29.3% for this
wastewater.
A sensitivity analysis was also performed for the 1 wt. %
wastewater case to see which solvent parameters had the
greatest effect on process costs, (Ricker, 1978; Ricker et al,
1979b). The results showed that the water content of the
128
-------
extract or the entrainraent of extractant could be doubled
without changing the costs by more than 6%. However, a
change of only 10% in either the limiting K for extraction
or the acetic acid/diluent relative volatility would have
a 6% effect on the costs. This again illustrates the
potential for improvement that could be achieved by a
better amine/diluent combination. It also suggests that
the values of these key parameters should be well
established before the design of a full-scale process is
attempted.
ALTERNATIVES FOR REGENERATION OF AMINE EXTRACTANT
Since regeneration of the araine extractant by
distillation is the most expensive step in the acetic-acid-
recovery process, it is worthwhile to consider
alternatives to distillation. One possiblity is to back-
extract the acetic acid into an aqueous solution of strong
base, such as NaOH or sodium carbonate (Na2C03l, as shown in
Figure 14. Since these are much stronger bases than the free
amine, an excess of the aqueous base should strip
virtually all of the acid from the organic phase. The
operating principle is similar to that used in liquid-ion-
exchange processes (Coleman, et al., 1958).
There are several disadvantages to this approach.
First, the cost of the base is excessive. For NaOH, the
cost would be $12.9/m wastewater, assuming 5 wt.% acid
in the wastewater feed and a cost of $0.19/kg for 50%
aqueous NaOH (Chem. Market. Rep., 1978). This is much
higher than the total cost of the extraction process with
regeneration by distillation (Table 41).
Second, the process produces an aqueous sodium
acetate solution. The solubility of sodium acetate at 20°
is only about 32 wt.% (Perry, 1963). It is unlikely that
this solution could be utilized directly, since the market
for sodium acetate is much smaller than that for acetic
acid (Lawler, 1977) . One could consider "springing" the
acetic acid by addition of a strong acid, such as sulfuric acid
(H,SOJ, followed by distillation, but this would be even more
costly and would produce an undesirable, salty waste
stream.
It is interesting to note that the same
considerations apply to the recovery of acetic acid from
aqueous solution by fixed-bed, anion-exchange processes.
After the bed has been loaded with acetic acid, it must
be regenerated by stripping with an aqueous base that is
stronger than the anion-exchange resin, or by washing
with strong acid followed by strong base. Both options
129
-------
Loaded Solvent
U)
o
Wastewater^
Fe«d
Acetic
Acid
Extractor
A
Regenerated
^
Aqueous
NaOH
Caustic
Wash
Aqueous
Effluent
Aqueous
NaOAc
Figure 14. Chemical regeneration of amine extractant by caustic-
wash process.
-------
involve chemical consumption and produce either a
concentrated aqueous acetate solution or an aqueous acetic
acid solution plus an aqueous salt solution, rather than
anhydrous acetic-acid. One must also contend with the
additional problems inherent in fixed-bed processes, e.g.,
poorer mass-transfer rates, incomplete bed utilization
due to solute breakthrough, etc. For some resins, thermal
regeneration may be an alternative.
For the amine-extraction process, methods other than
chemical regeneration have been considered, e.g., a
stripper-absorber system, but even when these are
feasible, the methods do not seem to be competitive with
distillation. Distillation is relatively simple and does
not require the addition of chemicals or produce an
additional waste stream; also, its design procedures are
well-understood.
LOSS OF SOLVENT—COST AND PREVENTION
Loss of solvent can have a very detrimental effect on
the economics of an extraction process, especially if the
solvent is costly. Effective solvent selection can
prevent certain types of solvent loss. For example, one
would avoid a solvent known to react irreversibly with a
solute in the extractor feed stream or a solvent known to
form an emulsion with water. Other sources of loss are
common to all extraction processes, regardless of the
solvents and solutes involved. Examples are incomplete
phase separation following extraction and incomplete
solvent-solute separation in the regeneration step. .These
problems cannot be circumvented completely, but their
effects can usually be minimized by appropriate process-
design measures.
Loss of Solvent by Dissolution and Entrainment in Raffinate
Loss of solvent to the extractor raffinate is always
an important consideration. Solvent can be lost both in
dissolved and in dispersed-droplet form (entrainment).
Solubility losses are relatively predictable. The
raffinate is typically very diluent (less than 99 mole%
H50) , so that the solvent solubility can usually be taken
as its solubility in pure water, although this is often a
poor approximation when surface-active agents are present,
even if they are quite dilute, due to micelle formation
(Adamson, 1967). High solubilities can also result when
a solute forms a water-soluble complex with the
extractant, e.g., acetic acid/low-molecular-weight amine.
Miniplant extractions of wastewater G with DIBK Alamine 336
mixtures gave solubilities for both the amine and the
131
-------
DIBK that were quite close to those in pure water (Pittman,
1979).
Entrainment, on the other hand, is quite difficult
to predict with any certainty. It depends strongly on
the properties of the solvent, e.g., the specific gravity,
interfacial tension against water, emulsification
tendencies, etc., as well as on the type of extractor and
the manner in which it is operated. One must also consider
the possibility of a process upset or operator error.
For ethyl acetate, solubility .,alone would result in
a loss of 71,000 ppm, or $37.63/m raffinate. Fortunately,
ethyl acetate is quite volatile and is easily recovered
by a steam-stripping process operated at atmospheric
pressure or under vacuum.
The cost of such a solvent-recovery system is only a
small fraction of the overall cost of the extraction
process. For ethyl acetate (Table 38), raffinate-
stripping at atmospheric pressure represents only 1/10 of
the DFC and an operating cost of $0.05/m raffinate,
including both utilities and amortized capital items.
Since the ethyl acetate is more volatile than water,
entrained ethyl acetate droplets could probably be
recovered in the stripper along with the dissolved
solvent. However, dispersions of the non-volatile TOPO
and amine solvents and their typical diluents cannot be
recovered by stripping. As explained by Rasquin, et al.
(1978), steam-stripping is essentially a liquid-phase-
mass-transfer-controlled process, especially in the case
of operation under vacuum. Consequently, the primary
driving force for stripping of non-volatile, dispersed,
extractant droplets is the solubility of the extractant
in water. This is usually very low, e.g., 600 ppm for
DIBK, and less than 10 ppm for Alamine 336 and TOPO.
In one study, pilot-plant-scale extraction of
uranium by amine solvents resulted in at least 0.02 volume %
entrainment in the raffinate (Grouse and Brown, 1955).
This level of entrainment would require extractant
make-up at a cost of $0.20/m raffinate for an amine
system and $1.83/m for TOPO, if the extractant were 50 wt. %
of the solvent phase in each case. Entrainment levels re-
ported in Section 6 are even higher.
Control of Entrainment Losses
There are many possible ways to control the cost of
entrainment, including the following: (1) avoid the use
132
-------
of very expensive extractants such as TOPO, or use them
in dilute solution in an inexpensive diluent. The latter
would reduce K , of course. (2) Run the extractor
in a manner that produces large, easily-settled
droplets. This reduces the mass-transfer efficiency of
the extractor, and it is difficult to produce a uniform
dispersion. However, drop-size optimization can be
beneficial. (3) Use an extractor that limits entrainment
by the nature of its operation. The Merichem extractor
(anon., 1974), for example, contains tightly-packed fibers
that are wet preferentially by one of the phases. This
phase adheres to the fibers and can be made to flow
countercurrently to the other phase with very little droplet
formation. Columns containing "static mixers" operate in a
similar manner (Rosenzweig, 1977). Other possibilities (Hanson,
1968) include the Podbielniak extractor, which increases
settling rates by the use of centrifugal force. The
Podbielniak is, however, relatively expensive for a
large-scale application. (4) Use some type of
secondary-recovery device. Treybal (1963) describes
many alternatives, including centrifuging, special
separating membranes, electrostatic coalescers, etc.
(5) Utilize a dual-extraction process, as.suggested by Earhart,
et al. (1977); this possibility is evaluated by Ricker (1978).
(6) Use a flotation-skimming process—not studied in this work.
A secondary-recovery device is certainly a reasonable
possibility. For example, Earhart (1978) reports that a
Mapco #1500 coalescer (Mapco, Inc.), a series of fibrous,
fixed-bed separators similar to cartridge filters, can
virtually eliminate entrainment of a variety of organic
chemicals from wastewater. Effluent levels of 10-15 ppm
are reported, which is the basis of the 20 ppm used in
the cost estimates (Tables 38 through 41). Installed
cost is estimated to be $20,000 for a 22.7 m /hr capacity
(Gale, 1978) . This means an operating cost of about
$0.02/m raffinate. The device is simple and requires
minimal maintenance.
The Use of Alternative Extractor Designs
Serious consideration should also be given to the
type of extraction device to be used for the primary
extractor. Unfortunately, in this work, the only
experimental extractor available was an RDC. Many
experimental runs were made with the DIBK-amine
extractant in the RDC and a wide range of raffinate ex-
trainment was observed, depending primarily on the agitator
speed and the size of the discs and stators used. The RDC
seemed to produce a large variation in drop size rather than a
narrow range, and the smaller droplets were easily entrained.
133
-------
This is probably typical of other extractors with similar
agitation systems.
Treybal (1963) and Hanson (1968) discuss alternative
designs, some of which have already been mentioned. One
must keep in mind that entrainment problems may be reduced
in an alternate extractor, but possibly at the expense of
increased complexity, lower mass-transfer efficiency,
and/or higher overall cost. The state-of-the-art is such
that these factors cannot be optimized a priori; each
alternative must be tested with the wastewater and solvent
under consideration, and at nearly full-scale, before a
quantitative comparison can be made. Obviously, the
engineer will usually rely on his judgement or on the
experience of a contractor, rather than resort to such
expensive tests.
In this work, it seemed reasonable and sufficiently
conservative to assume that significant entrainment from
the primary extraction step would be inevitable, and to
allow for the cost of a recovery device, as well as
significant (20 ppm) residual entrainment. Still,
entrainment is an important uncertainty in the process
design and in the cost analysis, especially for the amine
and TOPO extractants; experimental resolution is necessary
before these extractants can be widely used.
Loss Due to Solvent Degradation
The other source of solvent loss common for most
extraction processes is the regeneration step. The
discussion here is limited to regeneration by distillation,
since this is the most common method.
Other than irreversible solute-solvent reactions,
discussed previously, there are three main possibilities
for loss in the regeneration step. The first is
incomplete solvent-solute separation, i.e., solvent lost
to the acetic-acid product. This can be easily avoided
by proper column design. The second is thermal
degradation, including oxidation with traces of dissolved
air and cracking reactions.
Surprisingly, there are some quantitative data
available on the thermal degradation of Alamine 336
(Hardwick and Wace, 1965), indicating that heat
stability should not be a problem at the conditions
contemplated for the reboiler in the regeneration system,
as long as residence times in the hot sections of the
process, especially the reboilers for the dehydrator and
the regenerator, are on the order of 5-10 minutes or less.
134
-------
Tri-octyl phosphine oxide is reported to be stable (Helsel,
1977); no stability tests were performed on TOPO
extractants in the present work.
SOLVENT PURIFICATION—REMOVAL OF NON-VOLATILE SOLUTES
The third and potentially most serious source of
solvent loss is that due to the possible build-up of
non-volatile solutes in the higher-boiling solvents. The
obvious way to prevent the accumulation of non-volatiles
would be to take a purge stream from the regenerated
solvent. The purge stream rate would depend on the rate
of build-up and on the maximum allowable concentration
of non-volatiles in the solvent. If, for example, the
wastewater feed rate were 22,700 kg/h, the solvent
circulation rate 10,000 kg/h, the build-up rate 1 kg/h,
and the maximum allowable concentration 5.0 wt.%, the
required purge rate would be 20.0 kg/h. For a 50 gt. % Alamine
336 in DIBK solventj this purge would cost $1.36/m
wastewater ($9.96/m for 50% TOPO), or more than the total
of all other costs in Tables 40 and 41. The build-up
rate may be higher than 1/10,000, and 5 wt. % non-volatiles
may be too high; costs would be even higher'in such
cases.
Obviously, if the required purge rate is on the order
of 1 kg/h or more, the extractant must be recovered from
the non-volatile solutes. One approach would be to
evaporate the extractant, leaving a non-volatile residue
for disposal. However, evaporation is probably not
practical for TOPO, due to its very high boiling point.
An alternative for TOPO might be to evaporate most of the
diluent, and then partially crystallize the TOPO from the
residue. The non-volatiles might solidify simultaneously,
however. Another alternative would be to wash the purge with a
liquid that would dissolve either the non-volatile
solutes or the TOPO preferentially. Both of these
alternatives are likely to result in large losses of TOPO,
due to poor separation factors in the crystallization and
wash steps. Neither method was tested experimentally in
this work; recovery of TOPO from a solvent purge stream
remains an open question. The estimated cost of the
purification system given in Table 40 is only a very
rough estimate.
For the amine extractant, it might be possible to
evaporate the diluent and part of the amine at atmospheric
pressure, then cool the residue and separate out the non-
volatiles as a second liquid phase or in solid form.
This would avoid the cost of a vacuum system and reduce
135
-------
the energy consumption for the purification process.
However, the concentration/cooling process is rather
speculative, and only simple evaporation was considered
for the cost estimates.
Table 42 gives estimated costs for a batch still
to be used in conjunction with an extraction process with
a capacity of 22.7 m /h wastewater. The following
assumptions were used in the cost estimate: (1) A
20.0 kg/h purge rate; (2) storage capacity for a 2-week
accumulation of purge; (3) size of still such that a
2-week accumulation could be evaporated in one day (24
hours); (4) loss of 5% of the amine to the residue; and
(5) use of high-pressure steam at $3.17/GJ ($3.33/10 BTU) .
The major cost item, by far, is the loss of amine to
the residue. The assumption of a 5% loss is rather
arbitrary, as is the 20 kg/h purge rate. It was not
possible to determine these figures more accurately in
this work. To do so would require operation of an
extraction pilot-plant for a long-enough period of time
to measure the build-up rate of non-volatiles and to
accumulate enough non-volatiles for a batch distillation
experiment. No build-up of non-volatiles was observed
in miniplant extraction experiments, even
when the solvent was regenerated and recycled several
times. This suggests that the acetic-acid-manufacturing
wastewaters contain little, if any, extractable non-
volatiles.
SUMMARY OF THE RECOMMENDED EXTRACTION PROCESS FOR RECOVERY
OF ACETIC ACID
The complete flowsheet of the recommended extraction
process is given in Figure 15 / The discussion to follow
assumes the use of 30-50 volume % Alamine 336, or a
comparable tertiary alkyl amine, dissolved in DIBK. However,
DIBK is not an ideal diluent, and the development of a better
diluent could reduce the cost of the process by 30% or more.
Alamine 336 should not be used with wastewaters containing
chlorinated acetaldehydes.
The first step in the process is the extraction of
acetic acid by the amine/DIBK solvent. The recovery of
acetic acid to be achieved in the extractor is a key
design specification. For a 5 wt. % wastewater feed, a 97%
recovery requires 10 equilibrium stages and an F /F of
1.10, which is about the maximum that can be jus€if¥ed
economically.+ As demonstrated experimentally by Ricker, et al
(1979b), a 93 % recovery can be achieved in six mixer-settler
stages with an extractant containing 50% Alamine 336 and
an F /F of 0.81.
s w
136
-------
TABLE 42. COST ESTIMATE FOR DIBK/ALAMINE 336
PURIFICATION—BATCH SYSTEM
Direct fixed capital
still
heat exchanger
vacuum system
pumps/ drums, etc.
contractor's fee, contingency, etc.
M$
12.0
2.0
1.0
1.0
4.._P_
20.0
% Of DFC
60.0
10.0
5.0
5.0
20.0
Operating cost $/yr
steam and cooling water 200
loss of amine 19,400
depr ec iat ion , e tc . 3,600
23,200
$/m3
wastewater
0.001
0*11
0.02
0.13
%
0.8
82.3
16.9
137
-------
Decanter
u>
oo
Aqueous
Purge Aqueous
Acetic
Acid
Coalescer
Solvent-
Recovery
Stripper
Solvent -Regeneration
Column
Solvent Makeup from
Storage and Batch
Distillation
Aqueous
Effluent
Regenerated Solvent
Steam
Purge to Botch
Distillation
Figure 15. Flowsheet of recommended acetic-acid recovery process.
-------
Operation at low acid concentrations is relatively
expensive, because the Alamine 336/DIBK equilibrium curve
is non-linear; KD values are less than 1.0 below approximately
0.5 wt.% aqueous acid. Consequently, in the treatment of a
1 wt. % wastewater feed, the maximum practical acid recovery
is about 86%.
Since the extraction requires the equivalent of a
large number of equilibrium stages, a non-agitated column
is probably a poor choice for the extraction device. A
mixer-settler cascade or an agitated column, such as an
RDC, are more likely candidates. Solvent-entrainment
problems may influence extractor selection, although the
Alamine 336/DIBK system appears to have good phase-
separation characteristics.
The aqueous raffinate passes through a coalescer
to remove dispersed solvent droplets. It then enters a
vacuum steam-stripper, operating at 7320 Pa (55 mm Hg) ,
where approximately 92% of the dissolved DIBK is
recovered (the vacuum system is not shown in Figure 15).
The final aqueous effluent contains 1500 ppm acetic
acid, assuming a 97% recovery from a 5% wastewater feed,
50 ppm dissolved DIBK, and an average of 20 ppm entrained
DIBK-amine mixture.
The effluent will also contain other solutes,
depending on the nature of the feed wastewater. Solutes
in the feed that are more volatile than acetic acid will
leave the process in the aqueous effluent, for the most
part. Solutes that have roughly the same or a somewhat
higher volatility than acetic acid, and are extracted at
least as easily as acetic acid, will appear in the
acetic acid product.
The COD removal from wastewaters containing acetic
acid as the primary solute will be on the order of 80% or
more. Subsequent treatment by biological oxidation would
result in a more complete COD reduction. The loading on
such a biological-treatment facility will be much less
than if extraction were not used as a pre-treatment.
The extract phase from the extraction step contains
about 1 wt.% water and 5 wt.% acetic acid, depending on the
feed concentration, and small amounts of other solutes
extracted from the wastewater. Water has a high activity
coefficient in the extract phase and is easily separated
from acetic acid and solvent in a reboiled stripping
column. The water/acetic-acid relative volatility is
estimated to be 15.0 for operation at atmospheric
pressure.
139
-------
The overhead from this column i§ at about 90°C.
Since the formation of the amine-acetic acid complex is
exothermic (Kohler, et al., 1972), the complex is more
stable at lower temperatures. At 40°C, the water/acetic
acid relative volatility is estimated to be 31.0, based
on liquid-liquid equilibria, and in this respect, there
is some advantage to vacuum operation for the^
dehydration column. However, cost estimates indicate
that the energy savings resulting from vacuum operation
would be offset by increased capital costs for a vacuum
system and for a larger distillation column.
The condensed vapor from the dehydrator is two-phase,
since a small amount of DIBK is taken overhead. The
decanted organic phase recycles to the dehydrator, and
the aqueous phase, which contains acetic acid and volatile
pollutants, recycles to the extractor inlet. A small
aqueous purge can be used to rid the system of volatile
solutes, if desired. The same decanter is used for the
separation of the two-phase overhead from the solvent-
recovery stripper. The combination of solvent-recovery
and solvent-dehydration columns is similar, in principle,
to the well-known system for the azeotropic separation of
water and 1-butanol (Kirk and Othmer, 1963).
Finally, the acetic acid and the Alamine 336/DIBK
extractant are separated in a plate-type distillation
column operating at atmospheric pressure. The stripping
section of this column separates acetic acid from DIBK
and amine. The rectifying section separates DIBK and
acetic acid. A total condenser returns some of the
overhead acid for reflux, and the rest is product. The
reflux ratio (external reflux to distillate product) is
about 1.2. The overhead product is estimated to be less
than 95 wt.% acetic acid, the impurities being residual
water and wastewater solutes not removed in the
dehydrator. Higher purities could be achieved, if necessary,
by an increase in the dehydrator size and stripping-vapor
rate.
The hot solvent bottoms product, at about 175°C,
provides reboiler energy for the solvent-recovery stripper
and pre-heats the dehydrator feed. A small purge stream,
perhaps 10-20 kg/h for a solvent-circulation rate of
approximately 25,000 kg/h, removes non-volatile solutes
from the solvent. The solvent portion is recovered by
batch distillation, as discussed previously.
140
-------
SECTION 8
TREATMENT OF WACKER-PROCESS WASTEWATERS
The presence of chlorinated acetaldehydes makes the
wastewaters from the ethylene-oxidation process (Wacker
process) difficult to treat by extraction. This Section
summarizes the experimental problems encountered in
extractions of the Wacker wastewaters and discusses process
design considerations for a treatment system based on
extraction.
REACTIVITY OF CHLORINATED ACETALDEHYDES
The chlorinated acetaldehydes are very reactive
chemicals; MCA is especially reactive, behaving as both a
carbonyl compound and a chlorinated hydrocarbon (Miller, 1969a).
In an extraction process, the reactivity of these compounds can
cause problems during the extraction step and in regeneration
of the solvent by distillation.
Michaels (1977) found that the chlorinated
acetaldehydes react irreversibly with amine extractants
during extraction. He also found experimental indications of
a reaction with TOPO—a consistent disappearance of MCA in
TOPO extractions, whereas mass-balance closures in extractions
with other solvents, such as 2-ethyl-l-hexanol, were within
experimental error. However, he was not able to show con-
clusively that the disappearance of MCA was due to a reaction
with TOPO.
In the present work, an additional experiment was
performed to check for MCA-TOPO reactivity. An amount of 0.210
milliliters(ml) of a solution of 30 wt.% TOPO in Chevron
Solvent 25 was placed in a 5jOO-mi flask, maintained at 96°C by
a temperature controller and equipped with a reflux condenser.
Thirty wt.% aqueous MCA was added, and samples were withdrawn at
intervals. The TOPO content of each sample was measured
by GC. The GC column was 0.5 m of stainless steel tubing
packed with 5% OV-17 (supplied by Varian, Inc.) on 80-100
mesh, acid-washed Chromosorb W. The column was maintained
at 275°C and the injector at 350°C.
The unknown TOPO peak response (TC detector) was
141
-------
compared to that of a standard solution of TOPO. No
decrease in the TOPO peak was measured in six samples taken
over a period of 24 hours; the standard deviation of the
data was 3.7%. Approximately 0.030 kg more MCA solution
was added to the contents of the flask, and the experiment
was continued for another 24 hours. Again, there was no
apparent decrease in the TOPO concentration.
This experiment suggests that MCA and TOPO do not
react, in contrast with the conclusions of Michaels. The
disappearance of MCA observed in his experiments may have
been due to other factors, such as a reaction with an
impurity in the TOPO extractant.
Another pertinent reaction of the chlorinated
acetaldehydes is the addition of alcohol molecules to form
hemiacetals and acetals (Morrison and Boyd, 1966a;
Miller, 1969a). Hemiacetal formation is easily reversible
and is actually beneficial in that it results in large
distribution coefficients during extraction with alcohol
solvents. However, acetal formation, which is encouraged by
conditions in a regeneration by distillation, is much less
readily reversible. The problem is similar to that posed by
esterification.
Still another consequence of the reactivity of
chlorinated acetaldehydes is corrosiveness. An industrial
source has reported serious corrosion problems in
distillation equipment in which chlorinated acetaldehydes
were present. The use of special resistant metal alloys
in such equipment did not prevent corrosion. The
chlorinated acetaldehydes seemed to cause deposits of
polymeric materials on metal surfaces, possibly with the
release of hydrochloric acid (HCl) by hydrolysis. These
deposits appeared to be corrosion sites.
PROCESS-DESIGN CONSIDERATIONS FOR WACKER WASTEWATERS
Many of the design requirements for an extraction
process for the Wacker wastewaters are the same as for
the basic acetic-acid-recovery processes. However, the
reactivity of the chlorinated acetaldehydes results in some
additional constraints. First, the acetic acid cannot be
recovered from a Wacker wastewater by extraction with an amine
extractant unless the chlorinated acetaldehydes are first re-
moved in a pre-treatment step. Virtually complete removal
would be required; stoichiometrically, 1 part by weight MCA
reacts with 6.7 parts Alamine 336, so that a residual
concentration of only^lO ppm MCA could result in a cost
of as much as $0.16/m wastewater for amine make-up.
Moreover, the resulting quaternary-ammonium compound is
likely to be surface active and might thus promote
142
-------
emulsion formation, as experienced by Sanchez (1977) in
miniplant RDC extractions of sample F with amine
extractants.
Second, the primary alcohols, which are the best
known extractants for chlorinated acetaldehydes
cannot be regenerated by distillation because of
esterification and acetal formation. Secondary and
tertiary alcohols are probably much more resistant to
esterification (Morrison and Boyd, 1966d) and may be less
prone to acetal formation due to steric hindrance.
However, those alcohols are also significantly less effective
than primary alcohols as extractants for chlorinated
acetaldehydes.
Other extractants give relatively poor distribution
coefficients, on the order of 1.0 and less (Michaels,
1977), and would be expensive to use for the removal of
chlorinated acetaldehydes by extraction.
Third, corrosion problems would tend to discourage
distillation of any solvent containing chlorinated
acetaldehydes.
PRE-TREATMENT TO REMOVE CHLORINATED ACETALDEHYDES
A number of alternative methods of pre-treatment for
removal of chlorinated acetaldehydes were considered in
the present work. The conclusions as to their potential
utility are as follows:
(1) The hydrated form of MCA is less volatile than water;
steam stripping is thus prohibitively expensive.
(2) In an activated-carbon-adsorption process, the polarity
of MCA and its low molecular weight would probably
result in low equilibrium carbon loadings (Nathan,
1978) . This, combined with the relatively high MCA
concentration, would require large bed volumes and
rapid carbon turn-over, i.e., the process would be
expensive.
(3) The chlorinated acetaldehydes might be converted
chemically to a less troublesome form. For example,
the acetaldehydes might be chemically oxidized or converted
to insoluble products that could be removed by filtration.
As noted by Fox (1973), the cost of the chemical agent is
often prohibitive, especially when the solute is
concentrated.
(4) Wet air-oxidation is an alternative to chemical
conversion (Pradt, 1972; Hirai, 1975). For example,
the chlorinated acetaldehyde might be oxidized to
chloro-acetic acids; these strong acids could then be
recovered from the aqueous phase by an amine extractant
143
-------
or by other means. Acetic acid is resistant to
oxidation under fairly extreme wet air-oxidation
conditions (Day, et al., 1973; Williams, et al., 1975)
and could probably be co-extracted from the oxidation-
reactor effluent with the chloro-acetic acids.
Acetaldehyde and other wastewater solutes would
probably be oxidized to form additional acetic
acid. A key question would be the compatibility of
amine extractants with and the corrosiveness of the
chloro-acetic acids, i.e., are they more desirable
than chlorinated acetaldehydes? Air-oxidation pre-
treatment appears to be an interesting alternative,
but it was not investigated experimentally in the
present work.
(5) One could consider a dual-solvent process with a
preliminary selective extraction step to remove the
chlorinated acetaldehydes, e.g., with 2-ethyl-l-hexanol
as the extractant. As mentioned previously, the
2-ethyl-l-hexanol would have to be regenerated by some
means other than distillation.
DUAL-SOLVENT EXTRACTION
The conceptual design of a dual-solvent process is
shown in Figure 16. Solvent A is for pre-treatment and
solvent B is for acetic-acid recovery; the solvent B
loop is conceptually the same as in the processes discussed
earlier.
A detailed cost analysis of the dual-solvent process
was not performed because of the uncertainty of the
regeneration step. However, an order-of-magnitude estimate
can be derived from the costs of the processes in Section 7,
(1) Since the solvent cycles are independent and the
pre-extraction is nearly as difficult as the acetic-acid
extraction, the capital cost is assumed to be about twice
that of the DIBK-Alamine 336 process (Table 41), or
$2 million. (2) Since the major operating-cost items are
equipment related plus solvent make-up, the operating cost
is also assumed to double. (3) The assumed recovery value
of the3acetic acid in sample F is $882,000/yr, for a
22.7 m /h (100 GPM) wastewater. Given these assumptions,
the ROIBT for recovery of acetic acid from sample F is
only 19%, i.e., too low to be justified economically,
especially considering the risk and the complexity of the
process.
A detailed study is likely to be even less promising.
For example, one possible regeneration method for the
chlorinated-acetaldehyde extractant is a back-wash with
sodium bisulfite (Morrison and Boyd, 1966c):
144
-------
Solvent A +
Wastewater
Feed
Pre-Treatment
Extractor
Chlorinated
Acetaldehydes
Solvent A
Regeneration
Regenerated
Chlorinated
Acetaldehydes
Solvent A
Acetic Acid
Extraction
Water Effluent
Extract
B
•> Acetic Acid
Regenerated
Solvent B
(to Solvent B
Recovery)
Figure 16. Conceptual design of dual-solvent extraction process
for wacker-process wastewaters.
-------
o
II + _ +
-C - + Na HSCU > -C-SO-Na
3 , 3
OH
(aldehyde or ketone) (sodium bisulfite) (addition product)
Under certain conditions, the addition product is an insoluble
crystalline compound that could conceivably be removed by
filtration.
The method was tested experimentally in the present
work the presence of co-extracted acetic acid and other solutes
in the solvent A extract (Figure 16) resulted in a prohibitively
large bisulfite ion consumption. Moreover/ the addition product
was completely soluble at the concentration tested; thus an
aqueous waste stream would be produced.
Other potential drawbacks of a dual-extraction
process include the following: (1) If an amine were used
as the acetic-acid extractant, a process upset in the
solvent-A loop might result in the release of chlorinated
acetaldehydes to the solvent-B loop/ which would then
attack the amine. (2) Sufficiently complete removal of
chlorinated acetaldehydes would be difficult. For sample
F, reduction of the MCA content to 10 ppm would require a
99.8% removal, a very stringent separation specification.
(3) Any solvent A entrained into the solvent B loop would
not be recoverable without an extra separation step. If
solvent A were an alcohol, it would probably esterify to
a large extent before it could be recovered. (4) If the
solvent A loop cannot be regenerated by distillation, the
disposition of the extracted chlorinated acetaldehydes
might be more of a problem than would have been the case
with the original wastewater.
RECOMMENDATIONS FOR TREATMENT OF WACKER-PROCESS WASTEWATERS
Extraction is much less attractive for treatment of
the Wacker-process wastewaters than for the butane-
oxidation wastewaters. A better approach would seem to be a
process modification to eliminate the chlorinated acetladehydes
from the wastewater.
If this is not feasible, some form of destructive
terminal treatment, such as chemical or biological
oxidation, would probably be more appropriate than
extraction. For example, Hirai (1975) reports a wet air-
oxidation process to treat wastewaters containing
chlorinated acetaldehydes; the process is non-catalytic
and operates at 100°-200° and moderate pressures.
146
-------
However, such terminal treatment methods were not
investigated in the present work.
Another treatment alternative is deep-well disposal,
which is the current disposal technique for similar
wastewaters in areas of the country in which there are
suitable underground geological formations. However,
while this practice may be permitted at present, it is
certain to come under increasing scrutiny and may be
outlawed in the future due to the danger of an unexpected
release of toxic chemicals.
147
-------
SECTION 9
SUMMARY OF FINDINGS
The results of the present work suggest that there
should be a strong economic incentive to recover acetic acid
from large-volume (e.g., 100 GPM) wastewaters containing on the
order of 1.3 wt. % or more acetic acid. An exception would be a
wastewater containing other solutes that complicate recovery
and purification of the acetic acid, as in the case of the
two wastewater samples studied in this work which contained
chlorinated acetaldehydes.
Solvent extraction is probably the least expensive
recovery process for dilute wastewaters, i.e., those
containing less than 10 wt. % acetic acid. Of the
extractants considered, an amine, such as Alamine 336
(tri Cg-Cjn amine), dissolved in an appropriate organic
diluent, such as a Cg ketone, seems to be the most
promising for acetic-acid recovery. Such amines are
powerful extractants for acetic acid and are commercially
available at reasonable cost (approximately $2.40/kg).
A design study was done to determine the best process
configuration for acetic-acid recovery with an amine
extractant. The most economical configuration appeared to
be extraction in an agitated, differential-type contactor,
followed by a two-step distillation to regenerate the
extractant. The first regeneration step would remove water
from the extracted acetic acid, and the second step would
recover acetic acid from the extractant. The acetic-acid
product was designed to be more than 95% pure, with water
as the main contaminant. Glacial acetic acid could be
produced at additional cost. Steam-stripping seemed to be the
best way to recover dissolved amine diluent from the aqueous
extractor raffinate.
The proportion of amine to diluent in the extractant
phase was found to be an important process variable. Studies
of extraction equilibria showed that there was a maximum in
the function of K versus wt. % amine in the extractant. The
position of the maximum varied with the amount of acetic
acid in the aqueous feed and with the type of diluent. The
more polar diluents gave the higher K values. For a 5 wt. %
148
-------
acetic acid feed, and with DISK (di-isobutyl ketone) as the
diluent, the maximum K value appeared to be at more than
75 wt. % amine in the extractant.
However, other factors influence diluent selection and
the optimum extractant concentration. The diluent has a
large effect on the co-extraction of water by the extractant,
i.e., its selectivity for acetic acid, and on regeneration
by distillation. The diluent must have a high boiling point,
because the presence of the amine depresses the volatility
of acetic acid. The diluent must also be inexpensive and
resistant to degradation by oxidation and other chemical
reactions. The use of DIBK appeared to be a good compromise
between high KD values and these competing factors.
The optimum amine concentration in the diluent is
somewhat lower than that at the maximum K_ value, in
general. The higher amine concentrations result in an
increase in viscosity, which lowers extraction rates,
hinders phase-separation, and can lead to emulsification in
some cases. Also, the extractant becomes more difficult to
regenerate as KD increases, i.e., the volatility of acetic
acid decreases, as mentioned previously. Finally, the amine
is 3-4 times more expensive than common diluents, such as
DIBK, so that higher amine concentrations require a greater
investment for solvent inventory, and solvent losses, e.g.,
by entrainment, are more costly.
Quantitative process-optimization was not
performed in the present work; more extensive data, including
data from continuous, pilot-plant studies, would be required.
However, based on the information available, preliminary
cost analyses indicated that the optimum extractant
concentration for a wastewater feed containing 5 wt. % acetic
acid should be about 30-50 wt. % amine, and that the
optimum design recovery of acid would be about 95-97% for
a 22.8 m /h (100 GPM) wastewater. For these conditions,
the capital investment was estimated to be $1,030,000, with
an operating cost, including utilities, solvent make-up,
depreciation, etc., of $253,000/yr ($5.90/1000 gal
wastewater). Recovered acetic acid is valued at
$2,700,000/yr, for a ROIBT of 244%.
Two of the eight wastewater samples received from
acetic acid manufacturers contained on the order of 5%
acetic acid and were thus attractive candidates for an
acetic-acid extraction process. Two other samples contained
about 1 wt. % acetic acid but also contained chlorinated
acetaldehydes. These were found to attack amine solvents;
for this and other reasons, these samples were termed poor can-
didates for acetic-acid recovery. Three of the remaining
149
-------
samples were too dilute for economical acid recovery, and the
final sample was found to have been collected incorrectly by
the manufacturer and was not considered further.
The acetic-acid-recovery process was tested
experimentally on a miniplant scale with the two more
attractive wastewaters (Section 6) . The extractor was a
1.22 m x 7.65 cm rotating disc contactor (RDC). Typical acetic-
acid removals were 70-80%; the performance of the extractor
seemed to be dominated by back-mixing in the continuous
(aqueous) phase, and higher removals could not be obtained at
realistic values of F /F . However, a 93% removal was achieved
for a 6-stage mixer-settler cascade at an F /F of 0.808 with
50% Alamine 336 in DIBK, and a 99.8% recovery was obtained in a
similar experiment with 2-ethyl-l-hexanol diluent at
VFw = °-4'
The only notable problem encountered in the miniplant
RDC extractions was the formation of a dispersion of very
small droplets in the aqueous phase. These droplets were
entrained into the aqueous effluent. Since the extractant
is valuable, a large amount of entrainment would be a
serious problem unless it were recovered by some means.
Gravity-settling is not likely to be effective; samples of
the aqueous effluent from the RDC often required several
days of standing before the effluent becomes clear.
In some cases, it was possible to run the extractor at
low agitator speeds so as to reduce entrainment, and still
obtain acceptable acetic-acid removals. Entrainment in
such runs was less than 150 ppm amine, equivalent to a cost
of $0.36/m of wastewater. Other process costs, excluding
solvent make-up, are typically about $1.50/m .
It is difficult to predict the level of entrainment
that would occur in a large-scale extraction process. It
might be necessary to use a type of extractor less prone to
small-droplet formation. Special entrainment-removal devices,
such as the Mapco coalescer described in Section l, should also
be investigated.
Another important uncertainty in the process design
concerns the possible accumulation of non-volatile solutes
in the solvent phase. Such solutes might be extracted from
the wastewater or might be the result of a solvent-degradation
reaction and would have to be purged from the system by some
means. This would result in a net loss of solvent and could be
a serious problem if the rate of accumulation were large. No '
such accumulation was noted in this work; however, the solvent
phase was not recycled enough times to provide a definitive
measurement of the rate of accumulation. A meaningful test
150
-------
would require a long-term pilot-plant run with continuous
recycling of the solvent and with the actual wastewater to
be processed in a full-scale plant. Such a study should
also determine fouling rates in heat exchange equipment,
possible corrosion problems, and other factors that are
often a problem in wastewater treatment processes (Lang, et al.f
1975) but could not be studied in this work.
It would also probably be necessary to obtain
additional fundamental data, e.g., extraction and vapor-
liquid equilibria. The data obtained in this work seemed
to be adequate for a preliminary cost estimate and
feasibility study, but are probably not adequate for a
detailed design. Moreover, the chemical systems involved are
too highly non-ideal to allow the necessary data to be predicted
by common correlations. A thermodynamic model would reduce
the amount of experimental data required in that it would
allow accurate interpolation of limited data. One possible
framework for such a model was discussed.
Extraction was also considered as a treatment method
to reduce the COD in the eight wastewater samples, without
regard to chemical recovery, i.e., as a potential pre-
treatment for biological oxidation or some other terminal
treatment method. However, it was concluded that due to the
presence of significant amounts of low-molecular-weight, highly-
polar organic chemicals, which are difficult to extract as a
rule, extraction would be a very costly pre-treatment. The
removal of possibly toxic or refractory chemicals, such as
the chlorinated acetaldehydes, was considered another possible
application for extraction, but did not appear to be practical
for the wastewaters studied in this work.
151
-------
REFERENCES
Abrams, D.S., and J.M. Prausnitz, AIChE J., 11(1), 116 (1975).
Adarason, A.W., Physical Chemistry of Surfaces, 2nd ed., Inter-
science Publishers, New York, 492 (1967).
Anonymous, Hydro. Proc., 5£ (11) , 93 (1973).
, Chem. Eng., 81, Sept. 30, 54 (1974).
, Chem. Eng. News, 31 July, 20 (1978).
, Chem. Eng., £5(21), 71 (1978a).
Ashland Chemical Company, Bulletin #1181-6.
Barrow, G.M., and E. A. Yerger, J. American Chem. Soc., 76, 5211
(1954); I!.' 4474 (1955).
Biribauer, F.A., H.T. Oakley, C.E. Porter, J.H. Staib, and J.
Stewart, Ind. Eng. Chem., j49_(10), 1673 (1957).
Bonnelli, E.J., and H.M. McNair, Basic Gas Chromatography,
Consolidated Printers, Berkeley, California (1968).
Brown, W.V., Chem. Eng. Prog. , J59 (10) , 65 (1963).
Chemical Marketing Reporter, 213 (21), (1978).
Coleman, C.F., K.B. Brown, J.G. Moore, and D.J. Grouse, Ind. Eng.
Chem. , J50(12), 1756 (1958).
Cooper, J.C. , and D.G. Harper, Chem. Eng. Prog. , 62^(10) ,85
(1966).
Grouse, D.J., and K.B. Brown, AEC Report, ORNL-1959, Sept. 30,
(1955).
Danesi, P.R., R. Chiarizia, and M. Muhammed, J. Inorg. Nuel. Chem.,
submitted for publication (1977).
Daniels, F., and R.A. Alberty, Physical Chemistry, 3rd ed.
Wiley, 218 (1966).
152
-------
and P-L- Silveston, Can, J. chem. Eng.,
J-L i J J3 \.j.y / j ) , —
Eaglesfield P. B K Kelly, and J.F. Short, Ind. Chem., 29, 147
(1953) .
Earhart, J.P., Ph.D. Thesis, University of California, Berkeley,
California (.1975). , J
Earhart, J.P., and C.J. King, J. Food Sci., £1, 1247 (1976).
Earhart, J.P., K.W- Wong, C.J. King, and J.M. Prausnitz, U.S.EPA-
600/2-76-220, 1976.
Earhart, J.P., K.W. Won, H.Y. Wong, J.M. Prausnitz, and C.J.
King, Cheiu. Eng. Prog., 73_(5), 67 (.1977).
Earhart, J.P., Stauffer Chem. Co., Personal Communication,
July 1978.
EPA, U.S., "Methods for Chemical Analysis of Water and Wastes,
1971," Storet No. 00340, 17 (1971)
Fox, R.D., Chem. Eng., _80_(18), 72 (1973).
Gale, P.C., D.L. Bracken Co., Personal Communication, July 1978.
Garner, F.H., S.R.M. Ellis, and C.J. Pearce, Cheiu. Eng. Sci.,
3_, 48 (1954).
General Mills Co., "Product Notes," Alamine 336.
Grinstead, R.R. , U.S. Patent #3,816,524, June 11, 1974.
Groggins, P.H., Unit Processes in Organic Synthesis, 5th ed.,
McGraw-Hill, New York, 697 (.1958).
Guthrie, K.M. , Chem. Eng., 76(6), 114 (1969).
Hanson, C. , Chem. Eng., 75(18) , 76 (1968).
Happel, J. , D.G. Jordan, Chemical Process Economics, 2nd ed.,
Appendix C, Marcel Decker, Inc., New York (1975) .
Hardwick, W.H., and P.F. Wace, Chem. and Proc. Eng., £6(6),
283 CL965).
Harris, H.G., and J.M. Prausnitz, Ind. Eng. Chem. Fund., £,
180 (1969).
153
-------
Helsel, R.W. , Private Communication to C.J. King, August 17, 1976,
, Chem. Eng. Prog., 7_3_(5) , 55 (1977).
Hibbert, F., and D.N.P. Satchell, J. Chem. Soc. (B) , j>,
573 (1968).
Hirai, T., Japan. Kokai, 75, 13 262 (Cl. 13(7) All, 16B531,
15B31), Feb. 12, 1975.
Hogfeldt, E. , and F. Fredlund, Solvent Extraction Chemistry,
J. Rydberg, ed., North Holland Pub. Co., Amsterdam, 383 (1967).
Inoe, K., and Y. Baba, Kagaku Kogaku Ronbunshu, £(1), 95 (1978).
Kawano, Y., and F. Nakashio, Kagaku Kogaku Ronbunshu,
3_(5), 440 (1977).
Kertes, S., J. Inorg. Nucl. Chem., 27, 209 (1965).
King, C.J., Separation Processes, McGraw-Hill, New York
447 (1971).
Kirk, R.E., and D.F. Othmer, Encyclopedia of Chemical Technology,
2nd ed. , 2_, 839 (1963) .
, 12, 136 (1967).
Kohler, F. , E. Leiebermann, G- Miksch, and C. Kaing,
J. Phys. Chem., 7j>d9)> 2764 (1972).
Kohn, P.M., Chem. Engn., £3(6), 58 (1976).
Kunin, R. , Angew. Chem., Int. Ed., _1(3), 149 (1962).
Lang, W.C., F.P. Dace, and K.H. Pearson, Proceedings of the
2nd National Conference on Complete Water Re-use, AIChE,
New York, 1975.
Lawler, G.M. , Chemical Origins and Markets, SRI, Inc., Palo
Alto, California (1977)
Lowry, R.P., and A. Aguilo, Hydro. Proc. , J53 (11), 103 (1974).
Mattison, P., General Mills, Inc., Private communication to
C. J. King Jan. 31, 1977.
Meissner, H.P., and B. Chertow, Ind. Eng. Chem., 3_8_(8) , 856
(1946) .
154
-------
Mellan, I., Industrial Solvents Handbook. Noyes Data Corp.,
New Jersey, 337 (1970) . ~
Meyers, H.S., Ind. Eng. Chem., 50, 1671 (1958).
Michaels, J N., M.S. Thesis, University of California, Berkeley,
California (1977).
Miller, S.A., Ethylene and Its Industrial Derivatives,
Ernest Benn Ltd., London, 656 (1969a); 682 (1969b).
Moore, M.L., Organic Reactions, R. Adams, ed., 5, Wiley,
New York, 301 (1949).
Morrison, R.T., and R.N. Boyd, Organic Chemistry, Allyn and
Bacon, Inc., Boston, 642 (1966a); 718 (1966b); 639 (1966c);
671 (1966d).
Nathan, N.F. , Chem. Eng., _85_(3) , 93 (1978).
Nakashio, F., Kyushu University, Japan, Private Communication
(1978) .
Othmer, D.F., R.E. White, and E. Trueger, Ind. Eng. Chem., 33
(10), 1240 (1941).
, Ind. Eng. Chem., 3J>(5), 614 (1943).
, Anal. Chem., 20(8), 763 (1948).
, R. Gilmont, and J.J. Conti, Ind. Eng. Chem., 32, 625
(1960).
Page, J.E., and E.L. Smith, J. Soc. Chem. Ind. (London), 67,
48 (1948).
Perry, R.H., C.H. Chilton, and S.D. Kirkpatrick, Chemical
Engineers' Handbook, 4th ed., McGraw-Hill, New York, 1963.
Peters, M.S., and K.D. Timmerhaus, Plant Design and Economics
for Chemical Engineers, 2nd ed., McGraw-Hill, New York, 1968.
Petterson, W.C., and T.A. Wells, Chem. Eng., £4(20), 78 (1977).
Pittman, E., M.S. Thesis, University of California, Berkeley,
California (1979).
Pollio, F.X. , R. Kunin, and A.F. Preuss, Envir. Sci. Techno 1. , I
(6), 495 (1967).
Pradt, L.A., Chem. Eng. Progr., 6i8_(12) , 72 (1972).
155
-------
Prausnitz, J.M., Molecular Thermodynamics of Fluid-Phase
Equilibria, Prentice-Hall, New Jersey, 314 (1969).
Rasquin, E.A., S. Lynn, and D.N. Hanson, Ind. Eng. Chem. Fund.,
17. (3) , 170 (1978) .
Ricker, N.L., Ph.D. Dissertation in Chemical Engineering,
University of California, Berkeley, 1978.
Ricker, N.L., J.N. Michaels and C.J. King, J. Separ. Process
Technol., 1, No. 1, (1979a).
Ricker, N.L., E.F. Pittman and C.J. King, J. Separ. Process
Technol., 1, No. 1, (1979b)
Ricker, N.L., F. Nakashio and C.J. King, paper presented at
Am. Inst. Chem. Eng. Meeting, Boston, MA, August (1979c).
Rosenzweig, M.D., Chem. Eng., £4(10), 95 (1977).
Sakai, W., F. Nakashio, and K. Inoue, Kagaku Kogaku, 33,
1243 (1969).
Sanchez, G.W., M.S. Thesis, University of California, Berkeley,
California (1977).
Scibona, G., F. Orlandini, and P.R. Danesi, J. Inorg. Nucl.
Chem. , 218, 1701 (1966).
Schmidt, V.S., Amine Extraction, Israel Program for Scientific
Translations, Ltd.(1971).
Sittig, M., Pollution Control in the Organic Chemical Industry,
Noyes Data Corp., London(1974).
Treybal, R.E., Liquid Extraction, 2nd. ed., Chapters 10 and 11,
McGraw-Hill, 1963.
Wardell, J.M., M.S. Thesis, University of California, Berkeley,
California (1976).
, and C.J. King, J. Chem. Eng. Data, 23(2), 144 (1978).
Weast, R.C., Handbook of Chemistry and Physics, 51st Ed.,
CRC, Cleveland, Ohio (1970).
Williams, P.E.L., P.L. Silveston, and R.R. Hudgins, Can. J.
Chem. Eng., 53, 354 (1975).
Won, K.W., Ph.D. Thesis, University of California, Berkeley,
California (1974).
156
-------
, and J. M. Prausnitz, AIChE J., ,20(6), 1187 (1974).
Woods, D.R., S.J. Anderson, and S.L. Norman, "Evaluation of
Capital Cost Data: Heat Exchangers," Department of Chemical
Engineering, McMaster University, Hamilton, Ontario, Canada
(1975).
WUrm, H.J., Glxikauf, 104(12), 517 (1968). Reprints available
from American Lurgi Corp., 235 E. 42nd Street., New York,
NY 10017.
Yerger, E.A., and G.M. Barrow, J. Amer. Chem. Soc., 7_7_ 6206
(1955).
157
-------
TECHNICAL REPORT DATA
(Please read Instructions on the reverse before completing)
1. REPORT NO.
EPA-600/2-80-064
3. RECIPIENT'S ACCESSION-NO.
4. TITLE AND SUBTITLE
Solvent Extraction of Wastewaters from Acetic-acid
Manufacture
5. REPORT DATE
April 1980 issuing date
6. PERFORMING ORGANIZATION CODE
w. Lawrence Ricker and C. Judson King
8. PERFORMING ORGANIZATION REPORT NO
9. PERFORMING ORGANIZATION NAME AND ADDRESS
Department of Chemical Engineering
University of California
Berkeley, California 94720
10. PROGRAM ELEMENT NO.
C33B1B
11. CONTRACT/GRANT NO.
Grant No. R803773
12. SPONSORING AGENCY NAME AND ADDRESS
Robert S. Kerr Environmental Research Laboratory
Office of Research and Development
U.S. Environmental Protection Agency
Ada, OK 74820
13. TYPE OF REPORT AND PERIOD COVERED
Final/ 6-1-75-to 9-30-78
14. SPONSORING AGENCY CODE
EPA/600/15
15. SUPPLEMENTARY NOTES
16. ABSTRACT Solvent extraction was evaluated as a potential treatment method for
wastewaters generated during the manufacture of acetic acid. Possible goals for an
extraction process were considered. For the wastewater samples studied, extraction
appeared to be too expensive to be practical unless recovery of a marketable chemi-
cal were possible.
Long-chain, tertiary alkyl amines, dissolved in organic diluents, appeared
to be the most promising extractants, except for certain wastewaters containing
chlorinated acetaldehydes. Amine extractants were studied extensively in small-scale
experiments to determine phase equilibria, extractant regenerability, mass-transfer
characteristics, and emulsification tendencies.
A cost estimate was prepared for an extraction process to recover acetic
acid from a 22,700-kg/h (100-gpm) wastewater containing 5 wt.% acid. Estimated
direct-fixed-capital was $1,030,000, with an annual operating cost of $253,000/year
($5.90/1000 gal), resulting in a return on investment before taxes (ROIBT) of 244%
per year. The ROIBT, for a 1 wt.% acetic acid wastewater was only about 30%; how-
ever, this might be increased by further optimization of the amine/diluent combina-
tion.
7.
KEY WORDS AND DOCUMENT ANALYSIS
DESCRIPTORS
b.lDENTIFIERS/OPEN ENDED TERMS JC. COSATI Field/Group
Water Pollution
Solvent Extraction
Industrial Wastewater
Acetic Acid Manufacture
Organic Solvents
Cost Estimates
Wacker-Process
07A
8. DISTRIBUTION STATEMENT
Release to Public
19. SECURITY CLASS (ThisReport)
Unclassified
21. NO. OF PAGES
168
20. SECURITY CLASS (Thispage)
Unclassified
22. PRICE
EPA Form 2220-1 (9-73)
158
> U S. GOVERNMENT PRINTING OFFICE: 1980-657-146/5658
------- |