Production of

Calcium Magnesium Acetate (CMA)

from Dilute Aqueous Solutions
of Acetic Acid
by

Daniel B. Leineweber

A thesis submitted in partial fulfillment of

the requirements for the degree of

Master of Science
(Chemical Engineering)

at the

University of Wisconsin-Madison
December 25, 2002
 Abstract

Calcium Magnesium Acetate (CMA) can be used as a biodegradable non-

corrosive deicing salt, and has already been selected by the U.S. Federal
Highway Administration as the road salt to replace sodium chloride. Another
potential use for CMA is as an additive to coal combustion, where it acts as a
catalyst and sulfur remover. Hence CMA is likely to become a bulk chemical
used in multi-million ton quantities in the future.
In order to reduce the high production cost associated with conventional
processes, fermentation has been proposed as an advantageous route for CMA
production. The bioconversion of low-grade biomass yields acetic acid which
is then reacted with dolime or dolomite to make CMA. This can be done
directly by addition of dolime or dolomite to the fermentor, neutralizing the
acid and thus helping to avoid product inhibition of the organisms. However,
producing a stoichiometric solution of CMA in this manner is complicated by
the different reactivities and solubilities of the various corresponding calcium
and magnesium compounds, particularly in dilute acetic acid solutions such
as those present in the context of fermentation.
The main objective of this study is to gain a better understanding of the
thermodynamics and kinetics of the various reactions involved in the pro-
duction of CMA and to find ways to control the composition of the product.
Equilibrium models are examined both theoretically and experimentally to
obtain some information about thermodynamic limitations, and a dynamic
model is developed for a hypothetic continuous fermentation process to give
at least a qualitative idea of the kinetic features. A more practical approach
is made to specifically address the problem of composition control—it is
shown that a stationary solids phase allows producing a stoichiometric so-

i
lution of CMA, and, more generally, that it is possible to achieve a wide
range of other product compositions if desired. Feasibility is confirmed by an
experimental continuous dissolution process, and the performance of three
different neutralizer materials is compared: type S dolime and selectively
calcined dolomite are both found to be adequate choices for future applica-
tions; however, the use of uncalcined dolomite is not recommended due to
its very low reactivity. Selectively calcined dolomite provides the option of
producing a high-magnesium CMA with a calcium to magnesium mole ra-
tio of 2:3, which might be suitable for crystallization as a double salt; this
finding makes the almost unknown selectively calcined material particularly
interesting. The preliminary design of a process for the production of CMA
from wood hydrolyzate (waste sugars) and selectively calcined dolomite is
presented, and an economic analysis shows that this new process compares
well to previously proposed ones.

ii
Acknowledgements

Many individuals and institutions have contributed either directly or indi-

rectly to this study; I should like to mention some of them by name:
Professor D. C. Cameron deserves credit for his idea to investigate the
problem of composition control in the context of CMA fermentation processes
and to look at CMA production from the viewpoint of aquatic chemistry—
nobody had ever done this before. By asking the right questions he helped
me to determine the direction of my research; as my major professor he also
read and commented on several draft versions of my thesis.
Professor R. E. Swaney, who has worked with me on this project as a
co-advisor, spent hours discussing problems related to process design and
economics, and he has given me many useful suggestions. It also was a
valuable experience for me to be a teaching assistant in his ChE 450 Process
Design course—as a matter of fact, this course even inspired my idea of
examining CMA as a possible new byproduct of the pulp and paper industry.
The Department of Chemical Engineering here at UW-Madison has al-
ways provided a friendly atmosphere and a great environment for doing re-
search. They have also supported me financially during summer and fall 1992
by offering me a combined teaching/research assistantship, when it turned
out that it was not possible to get funding from other sources. All this help

iii
and support is greatly appreciated.
Several Undergraduate Students have worked with me on experiments
during their ChE 424 Summer Lab, and D. Kotowsky has helped me with
the preparation of selectively calcined dolomite during the fall semester.
Last but not least I also want to thank my parents in Germany for their
continuous support and my girl friend Elizabeth for her patience and love.

iv
Contents

1 CMA and Its Applications 1

1.1 Historical Background . . . . . . . . . . . . . . . . . . . . . . 1
1.2 Properties of CMA . . . . . . . . . . . . . . . . . . . . . . . . 5

2 The Production of CMA 10

2.1 Raw Materials and Reactions . . . . . . . . . . . . . . . . . . 10
2.2 Production Routes and Costs . . . . . . . . . . . . . . . . . . 15
2.3 In-Fermentor Production of CMA . . . . . . . . . . . . . . . . 19
2.3.1 Process and Organism . . . . . . . . . . . . . . . . . . 19
2.3.2 Encountered Problems . . . . . . . . . . . . . . . . . . 29

3 Equilibrium Models 32
3.1 Nonideal Effects in Aquatic Solutions . . . . . . . . . . . . . . 33
3.2 The Hydroxide System . . . . . . . . . . . . . . . . . . . . . . 37
3.3 The Carbonate System . . . . . . . . . . . . . . . . . . . . . . 49

4 The Dynamic Approach 59

4.1 Heterogenous Reactions . . . . . . . . . . . . . . . . . . . . . 61
4.2 Kinetics of Dissolution . . . . . . . . . . . . . . . . . . . . . . 66

v
 4.2.1 Carbonates: Calcite, Magnesite, Dolomite . . . . . . . 67
4.2.2 Oxide Minerals: CaO, MgO . . . . . . . . . . . . . . . 83
4.3 The Dynamic Model and Its Limitations . . . . . . . . . . . . 89
4.3.1 Outline of the Process Model . . . . . . . . . . . . . . 90
4.3.2 Model Uncertainties and Conclusions . . . . . . . . . . 105

5 Process Design Considerations 107

5.1 How to Solve the Mole Ratio Problem . . . . . . . . . . . . . 109
5.1.1 Stationary Solids Phase . . . . . . . . . . . . . . . . . 109
5.1.2 Experimental Realization . . . . . . . . . . . . . . . . . 115
5.2 Preliminary Plant Design . . . . . . . . . . . . . . . . . . . . . 129
5.2.1 Process Description and Flowsheet . . . . . . . . . . . 130
5.2.2 Economic Analysis and Discussion . . . . . . . . . . . 140

6 Analysis of Results 144

6.1 Results and Conclusions . . . . . . . . . . . . . . . . . . . . . 144
6.2 Future Work and Recommendations . . . . . . . . . . . . . . . 149

A Original ICP Data 151

A.1 Equilibrium Experiments (Chapter 3) . . . . . . . . . . . . . 151
A.2 CMA Production (Section 5.1.2) . . . . . . . . . . . . . . . . 153

B Lime and Limestone Analyses 157

B.1 Type S Dolomitic Hydrated Lime . . . . . . . . . . . . . . . . 157
B.2 Dolomitic Limestone . . . . . . . . . . . . . . . . . . . . . . . 159

C Equipment List (Section 5.1.2) 160

vi
List of Tables

1.1 Properties of CaAc2 and MgAc2 . . . . . . . . . . . . . . . . . 6

2.1 Inhibition of C. thermoaceticum by CMA . . . . . . . . . . . . 25

2.2 Dependence of product inhibition on pH . . . . . . . . . . . . 27

3.1 Tableau of the hydroxide system . . . . . . . . . . . . . . . . . 40

3.2 Equilibrium speciation for the hydroxide system . . . . . . . . 42
3.3 Tableau of the carbonate system . . . . . . . . . . . . . . . . . 51
3.4 Equilibrium speciation for the carbonate system . . . . . . . . 53

4.1 Rate constants for calcite and magnesite . . . . . . . . . . . . 70

4.2 Rate constants for dolomite . . . . . . . . . . . . . . . . . . . 79
4.3 Summary of carbonate and oxide dissolution rates . . . . . . . 89
4.4 Tableau of the liquid phase equilibrium model . . . . . . . . . 96

5.1 Capital requirement for CMA plant . . . . . . . . . . . . . . . 141

5.2 Annual operating cost for CMA plant . . . . . . . . . . . . . . 142

A.1 ICP results for equilibrium experiments . . . . . . . . . . . . . 152

A.2 ICP results for CMA from type S dolime . . . . . . . . . . . . 154
A.3 ICP results for CMA from selectively calcined dolomite . . . . 155

vii
A.4 Concentrations of impurities . . . . . . . . . . . . . . . . . . . 156

B.1 Sieve analysis of type S dolime . . . . . . . . . . . . . . . . . . 158

B.2 Chemical analysis of type S dolime . . . . . . . . . . . . . . . 158

viii
List of Figures

1.1 Phase diagram for CaAc2 – H2 O . . . . . . . . . . . . . . . . . 8

1.2 Phase diagram for MgAc2 – H2 O . . . . . . . . . . . . . . . . 9

2.1 In-fermentor production of CMA (batch process) . . . . . . . . 28

3.1 Titration of hydroxides—solid species . . . . . . . . . . . . . . 46

3.2 Titration of hydroxides—soluble species . . . . . . . . . . . . . 47
3.3 Titration of carbonates—solid species . . . . . . . . . . . . . . 55
3.4 Titration of carbonates—soluble species . . . . . . . . . . . . 56

4.1 Concentration profiles for heterogenous reactions . . . . . . . . 62

4.2 Surface reaction and mass transfer . . . . . . . . . . . . . . . 65
4.3 Calcite dissolution—dominance of reaction mechanisms . . . . 72
4.4 Carbonate dissolution rates vs. pH . . . . . . . . . . . . . . . 78
4.5 Magnesium oxide dissolution rates vs. pH . . . . . . . . . . . . 88
4.6 Overall structure of model system . . . . . . . . . . . . . . . . 91

5.1 Setup of experimental system . . . . . . . . . . . . . . . . . . 116

5.2 Production of CMA from type S dolime . . . . . . . . . . . . . 120
5.3 Production of CMA from selectively calcined dolomite . . . . 123
5.4 Ideal mole ratio curves . . . . . . . . . . . . . . . . . . . . . . 127

ix
5.5 Block flow diagram of acetic acid pulping . . . . . . . . . . . . 131
5.6 Mass balance for production of 1:1 CMA . . . . . . . . . . . . 134
5.7 Mass balance for production of 2:3 CMA . . . . . . . . . . . . 136
5.8 Process flow diagram for CMA production . . . . . . . . . . . 137

x
Notation

Dimensions are given in terms of mass (M ), length (L), time (t), temperature
(T ), and amount of substance (N ).

Latin Letters
A area, L2
A constant in Debye-Hückel formula, L3/2 N −1/2
A constant in Davies’ formula, dimensionless
a adjustable size parameter in Debye-Hückel formula, L
Ai molar activity of species i, N L−3
Āi effective surface area for species i, L2
B constant in Debye-Hückel formula, L1/2 N −1/2
b constant in Davies’ formula, dimensionless
C molar concentration in bulk solution, N L−3
C (s) molar concentration at surface layer, N L−3
Ceq equilibrium concentration, N L−3
(s)
Ceq equilibrium concentration at surface layer, N L−3
Ci molar concentration of species i, N L−3
D coefficient of diffusion, L2 t−1

xi
D dilution rate, t−1
G Gibbs free energy, M L2 t−2
∆G molar free energy change of reaction, M L2 t−2 N −1
∆G0 standard free energy change of reaction, M L2 t−2 N −1
F volumetric flow rate, L3 t−1
I ionic strength, N L−3
J molar flux, N t−1 L−2
JC molar flux due to surface reaction, N t−1 L−2
JT molar flux due to transport process, N t−1 L−2
K equilibrium constant, dimension varies
Kc concentration equilibrium constant, dimension varies
Ksp solubility product, dimension varies
ki rate constant of chemical reaction i, dimension varies
kC rate constant of heterogenous chemical reaction, dimension varies
kT mass transport coefficient, Lt−1
M molar mean molecular weight, M N −1
Mi molecular weight of species i, M N −1
N amount of substance, N
Ni amount of species i, N
Nx amount of biomass, N
n empirical reaction order, dimensionless
P pressure, M L−1 t−2
Pi partial pressure of component i, M L−1 t−2
Q reaction quotient, dimension varies
R gas constant, M L2 t−2 T −1 N −1
R net rate of heterogenous reaction, N t−1 L−2

xii
Rf forward rate of heterogenous reaction, N t−1 L−2
Rb backward rate of heterogenous reaction, N t−1 L−2
Si symbol for species i, dimensionless
T absolute temperature, T
TOT total (analytical) concentration, N L−3
t time, t
V volume, L3
xi mole fraction of species i, dimensionless
z rectangular coordinate, L
zi charge number of species i, dimensionless

Greek Letters
αP growth related constant for product formation, dimensionless
αS growth related constant for substrate uptake, dimensionless
βP non-growth related constant for product formation, t−1
βS non-growth related constant for substrate uptake, t−1
δ thickness of diffusion layer (Nernst theory), L
γi activity coefficient of species i, dimensionless
µ growth rate, t−1
µi chemical potential of species i, M L2 t−2 N −1
µ0i standard free energy of species i, M L−2 t−2 N −1
νi stoichiometric coefficient of species i, dimensionless
ξi mass fraction of species i, dimensionless

xiii
Subscripts and Superscrips
(aq) aqueous
b backward
C chemical reaction
dol dolomite
eq equilibrium
f feed
f forward
(g) gaseous
P product
S substrate
(s) solid
(s) surface layer
sp solubility product
T transport process
T total (analytical)
x biomass

xiv
Chapter 1

CMA and Its Applications

1.1 Historical Background

Sodium chloride (rock salt) has been used as a deicing agent on roads and
highways in the United States since the 1930s. Because of its effectiveness
over a wide range of temperatures, easy application, and low cost it quickly
became the most common road deicer. The annual consumption of salt for
highway deicing in the U.S. increased from approximately 4 million tons in
the early 1960s to a peak value of 12 million tons by the late 1970s. Cur-
rently about 10 million tons are spread on the roadways every year [Cho91].
Only at very low temperatures calcium chloride—which is significantly more
expensive—is used instead (about 0.3 million tons per year) because it has a
lower eutectic temperature than sodium chloride.
Dumping these huge amounts of salt on the roadways caused serious
corrosion problems, and also a lot of environmental damage. The major
drawbacks are

1
 2

• bridge-deck deterioration caused by corroding reinforcing bars,

• corrosion of vehicle chassis,

• pollution of soil, groundwater, and aquatic habitats by sodium and

chloride ions in runoff,

• damage to roadside vegetation.

There is also concern about possible effects on human health caused by high
levels of sodium in drinking water. It has been estimated that the total annual
damage resulting from the use of sodium chloride as road deicer is at least
15 times the cost for purchase and application of the salt [Cho91]. However,
these economic losses must be related to the benefits of road salting like
reduced number of accidents, or reduced business losses due to transportation
delays—economically the benefits still outweigh the losses. Hence it pays to
keep highways snow- and ice-free in winter, but it would pay even more if
this could be done without the adverse effects.
In the early 1970s an extensive research program was initiated by the
Federal Highway Administration (FHWA) in order to investigate the effects
of road salt and find ways to minimize the damage. As a result many im-
provements were made to bridge construction, e.g. sealed concrete bridge
decks, epoxy-coated reinforcing bars, and cathodic protection, which are ex-
pected to slow down bridge deterioration considerably. Measures to reduce
the amount of salt needed for the desired melting effect were also successfully
implemented, mainly through improved application techniques that made it
possible to get the salt to the proper location in the proper amount and
to keep it there. But all of this could not really solve the problems associ-
 3

ated with the use of sodium chloride, thus a search for alternative deicing
chemicals was begun.
In 1976 the FHWA awarded a contract for the study of alternative de-
icers to Bjorksten Research Laboratories, Inc. in Madison, WI. This study
was conducted by Dunn and Schenk who became the “inventors” of CMA
[Dun80]. Starting from the periodic table they first eliminated all chemical
elements that were radioactive, toxic or expensive, and by considering deicing
properties, corrosivity, etc. of the compounds that could be made from the
remaining ones they were able to eliminate several more candidates. Only
nine elements were left as possible constituents of new deicing chemicals:
H, C, N, O, Na, Mg, P, K, and Ca. Compounds containing these elements
were grouped in organic compounds, inorganic salts, and mixed salts (con-
sisting of an inorganic and an organic component), and solubility, eutectic
temperature, solution pH, and estimated cost were evaluated for each com-
pound. From the continued elimination process finally two candidate deicers
remained, methanol and calcium magnesium acetate. Both were studied
extensively in lab and field experiments to assess their deicing properties,
potential corrosive effects, and toxicity. Later methanol was also eliminated
because of its flammability, solvent nature, and poor persistency on road
surfaces, thus leaving only CMA for further consideration.
At this point the FHWA and the state highway agencies directed their
efforts towards a comprehensive project for the complete evaluation of CMA.
Numerous institutions were involved in those studies1 made during the 1980s.
General characteristics and deicing properties of CMA were determined, its
practical application was tested in several field trials, production methods
1
For a list of references, see [Cho91].
 4

were developed, and potential environmental effects were evaluated. It was

found that compared to sodium chloride CMA is

• almost as effective as an deicer and can be applied using the same

equipment,2

• less corrosive to steel and other highway-related metals,

• less damaging to concrete (scaling occurred only for CMA solutions

below pH 7 containing traces of free acetic acid),

• less harmful to soil, groundwater, human health, plant and animal life,

• but unfortunately much more expensive (currently CMA costs 10–

20 times as much as rock salt).

CMA is expected to be environmentally safe: the acetate part is biodegradable

and calcium and magnesium ions are biocompatible, since they are already
present in large amounts in most natural soils. Oxidation of calcium mag-
nesium acetate by soil microorganisms would generate calcium magnesium
carbonates which are non-toxic and could even improve soil quality in some
cases. One minor concern is a possible increase in biological oxygen demand
(BOD) when CMA is flushed into natural waters, but this is not likely to
cause damage because the run-off occurs when temperatures are low and
biological systems function slowly.
Hence this new deicer is in fact quite promising—it has almost ideal
characteristics, the only remaining problem being its high production cost.
The FHWA ended their involvement in CMA research in 1989, but industry,
2
The amount of CMA required for the same deicing effect is slightly higher, and there
may be problems with clumping due to the hydroscopic nature of the material.
 5

universities, and states are continuing to search for economically feasible ways
of manufacturing CMA, and to further improve the deicing and handling
properties of the material.
Still another potential application for CMA has been proposed recently
[Wis91, Chapters 9–14]: it can be used as an additive to coal combustion,
where it catalyzes the combustion and at the same time acts as a “sulfur
grabber”, forming solid calcium sulfate and thus substantially reducing sulfur
dioxide in the stack gases. The coal merely has to be impregnated with a
CMA solution before combustion. Used in this manner, CMA would help to
prevent acid rain, and would also enhance the efficiency of coal-fired boilers.3
In conclusion, CMA is a very interesting new chemical for environmental
applications—both of its proposed major uses can be expected to have a
large positive impact on the environment.

1.2 Properties of CMA

Usually dry CMA is considered to be a physical mixture of calcium acetate
CaAc2 ·H2 O and magnesium acetate MgAc2 ·4H2 O, both in their stable hy-
drated form as the monohydrate and tetrahydrate, respectively.4 Therefore
it should be possible to produce CMA with an arbitrary mole ratio of Ca:Mg.
Experiments have shown that under conditions of slow crystallization CaAc2
and MgAc2 in fact crystallize separately, while fast crystallization may pro-
duce a double salt with a low Ca:Mg ratio of about 1:3 [Mar85]. A process
for the manufacture of a double salt with empirical formula Cax Mgy Ac2(x+y)
3
It might be preferable to use calcium acetate instead of CMA in this application, since
the magnesium does not react.
4
Ac− is widely used as a shorthand notation for the acetate ion CH3 COO− .
 6

Table 1.1: Properties of CaAc2 and MgAc2 ; solubilities from [Lin65], all
other properties from [Mer83].

Calcium Acetate Magnesium Acetate

formula Ca(CH3 COO)2 Mg(CH3 COO)2

molecualar weight 158.17g/mol 142.40g/mol
mass fractions O 40.46% O 44.94%
C 30.37% C 33.73%
Ca 25.34% Mg 17.08%
H 3.82% H 4.25%
stable form (25◦ C) monohydrate tetrahydrate
density 1.50g/cm3 (anhydrate) 1.45g/cm3 (tetrahydrate)
solubility (25◦ C) 34.2g CaAc2 /100g H2 O 65.6g MgAc2 /100g H2 O
(55◦ C) 32.8g CaAc2 /100g H2 O 97.9g MgAc2 /100g H2 O
melting point (decomposes at 160◦ C) about 80◦ C
prior uses in leather manufacture none reported
in lubricants
as food stabilizer
as corrosion inhibitor

where x = 3 to 4 and y = 7 to 6 has been patented [Tod90], and is currently

used by Chevron Chemical Co. for commercial production of CMA.
Table 1.1 shows some properties of the individual acetate salts. Both
CaAc2 and MgAc2 have a high solubility in water—a necessary condition for
effective freezing point depression and good deicing performance. Note that
the solubility of CaAc2 decreases with increasing temperature, while MgAc2
has a positive temperature coefficient of solubility. The eutectic tempera-
 7

tures with water as reported by [Dun80] are −15◦ C for CaAc2 and −30◦ C
for MgAc2 compared to −21.1◦ C for NaCl; the corresponding phase diagrams
are shown in Figures 1.1 and 1.2. From this data one should expect bet-
ter deicing properties for high-magnesium CMA or even consider the use of
pure magnesium acetate. In fact the evaluation of eutectic data for CMA
solutions of various Ca:Mg mole ratios revealed that optimum freezing point
depression was obtained for a Ca:Mg ratio of 3:7 over a wide range of CMA
concentrations. However, under field conditions this 3:7 CMA tended to be
much slower in its ice melting rate than the originally proposed 1:1 CMA, so
finally the 1:1 ratio was recommended again as the optimum deicer composi-
tion [Sch91]. This last remark may not apply to the double salt mentioned
above—[Tod90] claim for their CMA unique properties that are not com-
parable to those of a physical mixture with the same composition.
Probably as important as the chemical composition is the physical form
of the deicer. As Gancy points out [Gan84a, Gan84b], dense and course
particles are required for effective ice removal, because it is sufficient to just
break up the ice layer by “drilling holes” down to the pavement and weak-
ening the pavement-ice interface. The ice layer then fractures and is finally
cleared away by the ongoing traffic. Hence it would be an enormous waste
of deicer material to use a fine powder or a brine in an attempt to melt all
the ice on the road surface. The most preferred physical form for a CMA
deicer is the hard, coarse, and non-friable flake which can be produced from
wet CMA using a drum pelletizer or similar equipment.
 8

10
CaAc2 – H2 O liquidus
measured data 3
CaAc2 – H2 O eutectic line
I
03 measured data 2
NaCl – H2 O eutectic +
3
3
3
3
3 3
−10 II
temp. 2 3
(◦ C) 2 2 2 2
2
−20 +

III
−30

−40
0 10 20 30 40 50
CaAc2 conc. (wt%)

Figure 1.1: Phase diagram for CaAc2 – H2 O. In region I there is only liquid
CaAc2 solution present, in region II there is water ice and liquid CaAc2 solu-
tion, and in region III there are two solid phases, water ice and CaAc2 ·2H2 O.
Eutectic of NaCl – H2 O included for comparison. Adapted from [Dun80].
 9

10
MgAc2 – H2 O liquidus
measured data 3
I MgAc2 – H2 O eutectic line
03 measured data 2
NaCl – H2 O eutectic +
3

3
−10
temp. 3
(◦ C) 3
II
−20 +
3

2
−30 2 2 2
2 III

−40
0 10 20 30 40 50
MgAc2 conc. (wt%)

Figure 1.2: Phase diagram for MgAc2 – H2 O. In region I there is only

liquid MgAc2 solution present, in region II there is water ice and liquid
MgAc2 solution, and in region III there are two solid phases, water ice and
MgAc2 ·4H2 O. Eutectic of NaCl – H2 O included for comparison. Adapted
from [Dun80].
Chapter 2

The Production of CMA

2.1 Raw Materials and Reactions

CMA is produced by neutralization of acetic acid with either dolomitic lime-
stone, mainly consisting of calcium and magnesium carbonates, or with its
burned and hydrated form called dolomitic lime, which is mainly calcium
and magnesium hydroxide. Some important properties of these raw materi-
als shall be discussed at this point.

Acetic acid, formula CH3 COOH, is a flammable, colorless liquid with a

characteristic sharp odor; it solidifies at 16.7◦ C, thus its pure form is
called glacial acetic acid. In aquatic solutions it is weakly dissociated
(pKa = 4.76); at concentrations of 1.0 M, 0.1 M, and 0.01 M the result-
ing pH is 2.4, 2.9, and 3.4, respectively [Mer83]. Many metals, as well
as their oxides and carbonates, dissolve in aqueous solutions of acetic
acid to give simple salts. The reactions are considerably slower than
those of hydrochloric acid or sulfuric acid, but the rate is still higher

10
 11

than with most other organic acids [Wag78]. Vinegar, a dilute aque-
ous solution of acetic acid (approximately 5%), has been used as a food
acidulant and preservative for thousands of years. Today acetic acid
is also widely used in the chemical industry as a solvent and as a raw
material for many organic syntheses, e.g. the manufacture of vinyl ac-
etate and cellulose acetate. The acetic acid production capacity in the
United States was 1.6 million tons in 1990, while only about 1.1 million
tons were actually consumed [Bus90]. At the present time the main
production routes for acetic acid are liquid-phase oxidation of n-butane,
and methanol carbonylation, both with feedstocks derived from natu-
ral gas or petroleum [Bus90, Wag78]. Alternative processes include
the destructive distillation of wood, and the fermentation of ethanol or
sugars, but except for the production of vinegar which is still made by
fermentation, these routes were considered no longer competitive when
cheap feedstocks based on natural gas and petroleum were introduced
following World War II. However, more recently there is again a grow-
ing interest in producing acetic acid from renewable biomass sources
like corn or even organic wastes, and much research has been done in
this field [Bus90, Gho85, Sch82b, Wan78]. The price1 for synthetic
glacial acetic acid was between $ 500 and $ 600 per ton in 1990.

Dolomitic limestone is generally understood to contain at least 20 wt%

and up to 44 wt% of MgCO3 , it may consist of the carbonate minerals
dolomite, calcite, aragonite, and magnesite in varying relative amounts
along with impurities like silica or alumina. Dolomite is the double
carbonate CaMg(CO3 )2 , calcite and aragonite are two different crys-
1
All product prices and costs are given per (US) ton; 1 ton = 2000 lb = 907.18 kg.
 12

talline forms of CaCO3 , and magnesite refers to MgCO3 . Normally a

magnesium-rich dolomitic limestone (40–44 wt% of MgCO3 ) is more or
less pure dolomite and therefore has a Ca:Mg mole ratio of approxi-
mately 1:1. Dolomitic stones with 20–40 wt% MgCO3 contain also a
considerable amount of calcite. Since calcite is the main constituent
of high-calcium limestone, calcite and dolomite are by far the most
abundant carbonates of calcium and magnesium. Both aragonite and
magnesite occur in small amounts together with the other carbonate
minerals, but their pure forms are rare and expensive. For the produc-
tion of a 1:1 CMA, dolomite seems to be most suitable, because it has
the required mole ratio. Other dolomitic limestones could be used as
well if a higher calcium content in the product is acceptable. Dolomitic
limestone currently sells for about $ 16 per ton.

Dolomitic lime is produced from dolomitic limestone by calcining, i.e.,

heating the stone to a temperature between 900 and 1100◦ C in a kiln.
The carbonates are transformed into oxides, and CO2 is released. Lime-
stone particle size, temperature, and calcination time are important
factors determining the quality of the product. Ideally, the particles do
not change in size and form during calcination; they just loose weight
and become more and more porous. A lime particle is said to be un-
derburned, when it still has an unreacted core of carbonate, and it is
overburned, when it has lost its porous structure through sintering.
The limestone fed into the kiln should be of uniform size, since par-
ticles smaller than average tend to be overburned, while larger lumps
are often underburned. The reactivity of the lime with water decreases
with increasing degree of sintering. For most applications, a highly
 13

reactive product is desired, and hence overburning of the lime has to

be avoided. However, the so-called dead burned dolomite (sintered at
very high temperatures) is used as a basic refractory. The thermal de-
composition of dolomite is assumed to be a two step process: at low
temperatures (600–725◦ C) magnesium oxide and calcite are formed,
and only above 800–900◦ C the calcite is decomposed to calcium oxide.
This has two implications:

1. The MgO in fully calcined dolomitic lime is often hardburned and

unreactive, because the sintering of MgO starts at lower temper-
atures than that of CaO.

2. It is possible to produce selectively calcined dolomite CaCO3 ·MgO

by calcining at low temperature. The MgO produced has a very
low degree of sintering; therefore, it is chemically more reactive.
Although this product seems to have some interesting properties
(and could actually be produced with less energy than required
for fully calcined lime) it has never been produced on a large scale
[Bol22, Boy80, Sha22, Sta53].

In most applications the lime is reacted with water in a process called

slaking before use. The reaction of CaO with water is highly exother-
mic and proceeds very fast as long as the lime is not hardburned; CaO
is almost completely transformed into Ca(OH)2 . However, under nor-
mal conditions the MgO does not hydrate equally well, since its lower
reactivity tends to be further decreased by overburning. Thus normal
hydrated lime (type N) is comprised of mainly Ca(OH)2 · MgO, and
only small amounts of Mg(OH)2 , but it is possible to obtain a fully
 14

hydrated product by applying high pressure and temperature during

slaking. This specially treated lime is called type S dolime. Depending
on the amount of water used in the process, either a dry hydrate, a
putty, a slurry, or a liquid (“milk of lime”) result. In the context of
CMA production mainly the use of type N dolomitic lime (obtained
from dolomitic quicklime by “normal” slaking) has been proposed, for
instance by [Mar85], but type S dolomitic lime is more suitable,2 and
selectively calcined dolomite could also have some potential. The mar-
ket price for type S dolime is about $ 67 per ton, for unslaked dolomitic
quicklime about $ 57 per ton. If produced on a large scale, selectively
calcined dolomite could actually be cheaper than quicklime.

Limestone as well as lime reacts with aqueous solutions of acetic acid to

form the desired acetate salts. The governing neutralization reactions for
dolomite, calcite and magnesite are

CaMg(CO3 )2 + 4HAc = CaAc2 + MgAc2 + 2CO2 + 2H2 O (2.1)

CaCO3 + 2HAc = CaAc2 + CO2 + H2 O (2.2)
MgCO3 + 2HAc = MgAc2 + CO2 + H2 O, (2.3)

and for the hydroxides

Ca(OH)2 + 2HAc = CaAc2 + 2H2 O (2.4)

Mg(OH)2 + 2HAc = MgAc2 + 2H2 O. (2.5)

As will be seen later, equilibria and kinetics of these reactions are highly
dependent on pH and tend to be quite different for calcium and magnesium,
2
According to a personal communication with Dunn, the type N material suffers from
the low reactivity of the hard-burned magnesium component.
 15

respectively. Generally the hydroxides react much faster with acetic acid
than the carbonates.

2.2 Production Routes and Costs

Various kinds of processes have been proposed for the manufacture of CMA,
but only some of them have been actually tested at a pilot plant level. CMA
has not yet been produced on a large scale, and almost all patented and
commercially used processes start from purchased acetic acid as raw material.
Depending on the source of acetic acid, the processes can be grouped into
three basic categories:

Conventional Processes use glacial or concentrated acetic acid which may

be derived from either petroleum or natural gas. Acetic acid produced
in this manner is quite expensive although large scale processes are well
established, because big amounts of valuable raw materials are utilized.
Since acetic acid cost is the key factor determining the cost of the final
product (the weight fraction of acetate in CMA is almost 80%), little
margin is left for improvements that would substantially lower the price
at which CMA can be produced.

The ways of reacting acetic acid with dolime or dolomite to make CMA,
and the subsequent steps necessary to get a dry product in the form
of flakes or pellets have been extensively studied, and many of them
are patented [Gan83, Gan84a, Gan84b, Gan86, Gan87b, Tod90,
Rip86]. Chevron Chemical Co. began commercial production of their
CMA-deicer ICE-B-GONTM in 1985, and they are now one of the main
producers of CMA. The selling price for ICE-B-GONTM was $ 657
 16

per ton in 1989; it has been estimated that on a large production scale
a price between $ 400 and $ 450 per ton could possibly be reached—still
too much to really consider complete replacement of sodium chloride
by CMA, since rock salt is available for between $ 25 and $ 50 per ton.
In order to obtain an affordable deicing product it has been suggested
to produce CMA-coated sand or substitute part of the acetate by cheap
chloride [Gan87a].

Fermentation Processes use a feedstock of glucose, corn, or low-grade

biomass (e.g. organic municipal waste) which is converted to acetic acid
by microorganisms like Clostridium thermoaceticum. Depending on
the raw material, a pretreatment may be required before fermentation.
Basically three different routes can be followed for CMA production:

• The acetic acid is extracted from the fermentation broth using

a liquid-ion exchanger and then reacted with dolime [Tra90,
Wis88, Yan92]. The advantages of this approach are that it
would allow continuous operation and would yield a CMA so-
lution at relatively high concentration; potential problems are
the high cost of exchanger-materials and the fact that efficient
extraction requires an acidic pH (only the undissociated acid is
extracted)—therefore, either the fermentation must be done with-
out pH control3 which greatly reduces the production rate and
attainable concentration of acetic acid, or the broth must be acid-
ified before extraction, thus adding another costly step to the pro-
cess and generating a salt solution which has to be disposed of as
3
A strain of C. thermoaceticum capable of growth and acetic acid production at pH 4.5
has actually been isolated, but productivity and growth rate were low [Sch82a].
 17

a waste stream. It has been proposed to integrate the fermen-

tation and the extraction process to remove the acid from the
reactor while it is being produced in order to avoid or at least re-
duce product inhibition of the organisms, but much more work is
needed before it is possible to assess the viability of such a process.

• CMA is produced directly in the fermentor by controlled addition

of dolime or dolomite to the broth, so that the pH remains in a
favourable range for the organisms during fermentation [Hud88,
Mar85, Wie91]. Batch fermentation as well as continuous pro-
cesses can be used; however, to date significantly higher final
product concentrations were obtained through batch fermenta-
tion. The advantages of this route are higher acetic acid produc-
tion rates (because of pH control) and probably lower capital in-
vestment, since smaller equipment can be used. Among the prob-
lems are the difficulties to produce a stoichiometric CMA solution
from dilute acetic acid at a near-neutral pH, and the inhibition of
the organisms at higher CMA concentrations. The latter restricts
CMA concentrations in the broth to about 5 wt% with currently
available organisms, hence energy requirements for product recov-
ery are relatively high.

• Another interesting route has been described in an early U.S. Pat-

ent dating back to 1932 [Car32]—decades before anyone was talk-
ing about CMA and its potential use as a deicer chemical: During
fermentation the pH is controlled using ammonia, thus producing
a dilute aqueous solution of ammonium acetate. The broth is fil-
tered, and lime (or another nonvolatile base) is added which gives
 18

calcium acetate (or some other acetate salt) and easily decom-
posable ammonium compounds like ammonium hydroxide. This
solution is first sent into a scrubbing column to recover the ex-
pensive ammonia and then through a multi-effect evaporator and
dryer, where the water is boiled away and a dry acetate salt is
obtained as final product. In the scrubbing tower, the broth is
contacted with the vapor from the first effect of the evaporator;
the vapor leaving the top of the scrubbing tower serves as the
heating medium in the second effect. It contains about 96% of
the ammonia which can be directly recycled back to the fermen-
tor for reuse as neutralizer.

Cost estimates for fermentation processes range from about $ 260 to

$ 500 per ton, depending on the feedstock, organism, and process scale.
There is still much research under way to find better organisms [Hud88]
or improve existing ones [Lju85, Par90, Wis91] in order to achieve
higher production rates and higher product concentrations.

Alkaline Fusion Processes involve heating cellulosic waste materials in an

excess of alkali to a temperature above 200◦ C; an exothermic reaction
then converts the cellulose to acetate (up to 30%), methanol, acetone,
carbonate, and oxalate. Processes of this kind have been already de-
scribed 100 years ago [Cro92], and lab experiments have shown that
they can be easily adapted for CMA production [Dun80], but it has
to be mentioned that the acetate yield is significantly lower for earth
alkaline metals like calcium and magnesium than for alkaline metals
like sodium. There is a patent on an alkaline fusion process for CMA
 19

manufacture [San84], but recently there seems to be little interest in

developing this route further, although it might be very promising—
[Dun80] estimated the cost being as low as $ 100 per ton of unpurified
CMA if solid organic wastes were used as raw material.

Throughout the subsequent part of this work, focus will be on fermentation

processes only, and particularly in-fermentor production of CMA will be
considered. Hence a more detailed description of this specific route and of
some of the problems encountered in this context is required, which will be
given in the next section.

2.3 In-Fermentor Production of CMA

2.3.1 Process and Organism

It has been estimated that in-fermentor production of CMA is advantageous
compared to conventional processes, even when a high-cost corn feedstock is
used [Mar85]. The potential use of low-grade biomass like municipal wastes,
agricultural and forest residues, etc. instead of corn would result in further
significant cost reductions, and a selling price of about $ 260 per ton CMA
seems to be attainable [Hud88].
None of the proposed CMA fermentation processes uses the commercially
available technology for vinegar production, a two step process involving
two different organisms, where glucose is first converted anaerobically to
ethanol by Saccharomyces cerevisiae and then further metabolized to acetic
acid under aerobic conditions by Acetobacter aceti. A one step fermentation
with thermophilic homoacetogenic bacteria like Clostridium thermoaceticum
 20

[Wan78] is preferred for several reasons:

1. The actual yields of acetic acid are often more than 85% (g acetic acid
per g glucose) for homoacetate fermentations, compared to only 50%
for the vinegar process.

2. Homoacetogenic bacteria are able to produce acetic acid not only from
C6 sugars like glucose, but also from some C5 sugars like xylose that
would represent a significant portion of the sugars produced by hydrol-
ysis of lignocellulosic feedstocks. In the vinegar process only glucose is
metabolized.

3. The homoacetate process is anaerobic, hence simpler equipment can

be used (no aeration, less agitation), and the total energy requirements
including product recovery are lower than in the vinegar process, if the
acetate concentration in the broth is at least 2 wt% [Gho85].

4. Since homoacetate fermentations are carried out at a fairly high tem-

perature level (about 60◦ C), the growth of unwanted other organisms
is effectively suppressed. If sterilization is necessary at all, the related
cooling costs are reduced, and there may be other process advantages
like a positive effect on dissolution kinetics of the lime or limestone
introduced for neutralization.

For instance, glucose is converted to acetic acid by C. thermoaceticum with

a theoretical yield of 100% according to the reactions [Wie91]

C6 H12 O6 + 2H2 O −→ 2CH3 COOH + 2CO2 + 8H+ + 8e− (2.6)

2CO2 + 8H+ + 8e− −→ CH3 COOH + 2H2 O, (2.7)
 21

or in sum,
C6 H12 O6 −→ 3CH3 COOH. (2.8)

It is remarkable that this organism is able to synthesize acetic acid by fixation

of CO2 . A similar overall reaction can be written for the metabolization of
xylose
2C5 H10 O5 −→ 5CH3 COOH. (2.9)

Since acetic acid is the only fermentation product, a relatively “clean” solu-
tion of CMA could be obtained without need for expensive separation tech-
niques. If organic wastes containing cellulose and hemicellulose are to be used
as a feedstock, these materials must be transformed into a mixture of soluble
sugars by hydrolysis under mild conditions before they can be fermented.
Wang et al. found that C. thermoaceticum produces acetic acid not only
during growth, but also when cell growth ceases [Wan78]. They used the fol-
lowing model to describe the growth-related and non-growth-related aspects
of product formation,
dNP
= (αP µ + βP )Nx , (2.10)
dt
where NP denotes the produced amount of acetic acid in moles, µ the spe-
cific growth rate, and Nx the total biomass in moles, based on the average
elemental composition of cells.4 Since for exponential growth
dNx
= µNx , (2.11)
dt
equation (2.10) can be rewritten as
dNP dNx
= αP + β P Nx , (2.12)
dt dt
4
Often the composition CH1.8 O0.5 N0.16 S0.0045 P0.0055 is used for biomass, corresponding
to a molecular weight of 24.4 g mol−1 [Rie91].
 22

i.e., the total acetic acid production rate depends linearly on both the biomass
growth rate and the amount of biomass already present; the corresponding
terms in equation (2.10) are often referred to as growth term and maintenance
term. The linear relationship was found to be valid for specific growth rates in
the range of 0 to 0.15 h−1 with αP = 0.8 (mole acetic acid per mole cells) and
βP = 0.065 h−1 (mole acetic acid per mole cells per hour) at pH 7.0 [Wan78].
In a very similar manner, the substrate consumption can be modeled by the
expression
dNS
= (αS µ + βS )Nx ,
− (2.13)
dt
where NS denotes the amount of substrate consumed in moles, and the overall
yield of product on substrate (“actual” yield) is then given by
ov dNP /dt
YPS =− . (2.14)
dNS /dt
ov
A typical value for YPS on a mole basis is 2.55 (mole acetic acid per mole glu-
cose), or, equivalently, 0.85 on a weight basis (g acetic acid per g glucose).5
From these results it can be concluded that growth is not absolutely re-
quired for acetic acid production—a high productivity could also be achieved
through high cell density at minimal growth using immobilized cells or cell
recycling.
Direct anaerobic digestion of cellulosic waste by a mixed culture of acid-
formers is another process that has been considered for CMA production
[Tra91, Tra90, Wis88]. The organisms are found in stable manure, pond
mud, and sewage sludge; some of them—called cellulolytic bacteria—are able
to hydrolyze cellulose or hemicellulose, while others depend on the solu-
ble sugars and alcohols provided by the metabolic activity of the former.
5
This shows that it is very important to indicate on which basis the yield has been
calculated, although it is a dimensionless quantity.
 23

The thermophilic homoacetogenic bacteria discussed above belong to the

second group of organisms and are therefore not capable of using cellulose
as a substrate. Acetic acid is only one of the organic acids produced by
the acid-formers, other products are formic, butyric, propionic, lactic, suc-
cinic, and isobutyric acids [Gho85]. Under normal conditions, these organic
acids are further metabolized to methane by another group of bacteria called
methanogens. Their growth must be suppressed in order to allow for accu-
mulation of acetic acid. Sewage sludge, woody biomass, and in principle, any
kind of organic waste can be directly used in this process, and since these
feedstocks are available in large amounts at a very low cost, or, in some
cases even a credit can be taken for the disposal of such waste materials,
anaerobic digestion seems to be economically quite attractive. On the other
hand, the deicer produced is not pure CMA, but rather a complex mixture
comprising the salts of different organic acids, and hence is likely to be less
effective. In addition, the acetic acid production rate is extremely low, and
the acetate concentration that has been obtained in experiments is also very
low, only about 0.8 wt%, although theoretically 3 wt% should be possible
[Tra91]. This makes product recovery by conventional methods like multi-
effect evaporation too expensive, and an extraction process as described in
the last section must be used. Therefore, anaerobic digestion does not seem
appropriate for in-fermentor production of CMA.
At the moment, batch fermentation appears the most economical, be-
cause it yields the highest final product concentration, and hence minimizes
subsequent drying requirements; however, it is possible to make consider-
able improvements to continuous processes with cell recycling or immobi-
lized cells—these continuous fermentations, when developed, will ultimately
 24

be the preferred processes [Wie91]. Contiuous operation per se is advan-

tageous in many respects, but the main advantage is that the continuous
fermentation allows a much higher production rate per unit volume than the
batch process. Acetic acid production rates of up to 8 g l−1 h−1 were re-
ported for continuous processes, compared to less than 1 g l−1 h−1 for batch
fermentations [Bus90]. According to Schwartz and Keller, a productiv-
ity of 5 g l−1 h−1 should be achieved in commercial fermentations for acetic
acid production [Sch82b]. Hence in our context, a CMA production rate of
5 g l−1 h−1 shall be assumed; this corresponds to an acetic acid productivity
of about 4 g l−1 h−1 which should be attainable using a continuous process,
but probably can not be achieved with batch fermentation. A reasonably
concentrated solution of CMA would have 7 to 10 wt% or 75 to 111 g l−1
[Hud88, Mar85]. This target concentration requires microorganisms that
are acclimated to fairly high levels of calcium, magnesium, and acetate. “Nor-
mal” strains of C. thermoaceticum are strongly inhibited by their product
even at near-neutral pH and are also negatively affected by higher concentra-
tions of calcium and magnesium; this led Ljungdahl et al. to the conclusion
that the “properties of the wild type strains (. . . ) do not suggest success-
ful applications of these strains for industrial production of CMA or acetic
acid” [Lju85]. More recently, Parekh and Cheryan were able to isolate
an “improved” strain of C. thermoaceticum that could grow in the presence
of 70 g l−1 sodium acetate, 50 g l−1 magnesium chloride, and 20 g l−1 calcium
chloride at pH 6.8 [Par90]. From this data, the approximate tolerance levels
for acetate, magnesium, and calcium ions are 50 g l−1 , 12 g l−1 , and 7 g l−1 ,
respectively (under the conservative assumption that high concentrations of
sodium and chloride do not cause inhibition, i.e., without these additional
 25

Table 2.1: Concentrations of calcium, magnesium, and acetate ions in differ-

ent CMA solutions compared to the tolerance levels of C. thermoaceticum at
pH 6.8. A 1:1 molar ratio of Ca:Mg was assumed for the three fictive CMA
solutions with total concentrations 50, 60, and 70 g l−1 . Also included are
experimental results for a 1:3 CMA obtained by [Wie91] in batch fermen-
tation at pH 6–6.8. The tolerance levels were calculated from the data of
[Par90]. All concentrations are given in units g l−1 .

Concentration of
2+
Ca Mg2+ Ac− Total CMA

4.8 wt% CMA (1:1) 6.7 4.0 39.3 50.0

5.7 wt% CMA (1:1) 8.0 4.9 47.1 60.0
6.5 wt% CMA (1:1) 9.3 5.7 55.0 70.0

3.1 wt% CMA (1:3) 2.4 4.1 25.8 32.3

(experimental)

tolerance levels 7 12 50

ions even higher levels of acetate, magnesium, and calcium might be toler-
ated). Calcium has a relatively high toxicity and would definitely be one of
the factors limiting the attainable concentration of CMA. As shown in Ta-
ble 2.1, it should be theoretically possible to obtain a CMA concentration of
at least 50 g l−1 using this improved strain; at 60 g l−1 the calcium tolerance
would be slightly exceeded, and at 70 g l−1 both calcium and acetate would
be limiting. For these calculations, a 1:1 molar ratio of Ca:Mg was assumed.
Experimentally, 32 g l−1 of a 1:3 CMA were obtained by Wiegel et al. in
their pH controlled batch fermentation [Wie91]. Thus it still is somewhat
optimistic—but not too unrealistic—to assume that it will be possible to at-
 26

tain a CMA concentration of about 50 g l−1 (4.8 wt%) with currently available
strains of C. thermoaceticum. This assumption shall be used throughout the
rest of this study, although it falls short of the target concentration stated
above. Possibly, an organism might be genetically engineered that possesses
a higher tolerance to calcium and acetate in order to meet the target. It
has been also proposed to use halophilic organisms from the Dead Sea that
can naturally tolerate the conditions of highly concentrated ionic solutions,
in particular high levels of calcium ion; actually such organisms have been
successfully weaned from sodium chloride and encouraged to produce acetic
acid [Hud88]. But keeping in mind that it is quite difficult to obtain high
product concentrations with continuous processes, 7–10 wt% does not seem
to be a reasonable goal at the moment.
Different studies have shown that C. thermoaceticum is far more sensi-
tive to the undissociated acetic acid than the acetate ion [Sch82b, Wan84].
While acetate ion can be tolerated at concentrations of up to 48 g l−1 , total
growth inhibition by undissociated acetic acid already occurs when its con-
centration is as low as 2.8 g l−1 [Wan84]. It is quite instructive to calculate
the maximum attainable total acetic acid concentration (undissociated acetic
acid plus acetate ion) as a function of pH by the Henderson-Hasselbalch
equation
[Ac− ]
log = pH − pKa (2.15)
[HAc]
and the mole balance

[HAc]T = [HAc] + [Ac− ], (2.16)

assuming that neither of the two respective inhibitory concentrations for ac-
etate ion and undissociated acid may be exceeded. Then the maximum total
 27

Table 2.2: Calculation of undissociated acetic acid concentration, acetate ion

concentration, and total acetic acid concentration at different values of pH;
all concentrations are given in units g l−1 .

pH Concentration of
HAc Ac− HAc plus Ac−

5.0 2.8 4.8 7.6

5.2 2.8 7.6 10.4
5.4 2.8 12.0 14.8
5.6 2.8 19.0 21.8
5.8 2.8 30.2 33.0
6.0 2.8 48.0 50.8
6.2 1.8 48.0 49.8
6.4 1.1 48.0 49.1
6.6 0.71 48.0 48.7
6.8 0.45 48.0 48.5
7.0 0.28 48.0 48.3

concentration at different values of pH is limited by the inhibitory concen-

tration which is reached first. The results presented in Table 2.2 show that
below pH 6, undissociated acetic acid is responsible for growth inhibition
since the tolerance level of 2.8 g l−1 is reached before attaining an acetate
ion concentration of 48 g l−1 . Above pH 6, acetate ion is the limiting factor,
because now the inhibitory concentration of undissociated acetic acid is not
exceeded before acetate ion becomes inhibiting. From the data it also be-
comes obvious that only very low total concentrations of acetic acid can be
obtained at pH < 6. Therefore, the pH of the broth should be maintained
in the range of pH 6–7 during fermentation by controlled addition of lime or
limestone, and perhaps bubbling with CO2 , in order to avoid serious inhibi-
 28

substrate CO2
6
?

¾
dolime
slurry
fermentor 6
(pH 6–6.8)

CO2 - - pH control
unit

Figure 2.1: Fed batch fermentation process for CMA production, pH is con-
trolled by addition of a dolime slurry; process as used by [Wie91].

tion of HAc production. A simple pH-controlled fed batch process for CMA
production as the one used by Wiegel et al. [Wie91] is shown in Figure 2.1,
a continuous process could be controlled in a similar way.
After completion of the fermentation, the pH of the broth must be ad-
justed to pH 8–9, because traces of free acetic acid in the product would
be damaging to concrete. This pH adjustment could be done by adding an
extra amount of lime, but the sluggish reaction makes it difficult to obtain
a uniform and stable product. For this reason, it has been suggested to use
a small amount of potassium hydroxide instead of lime for the “fine tuning”
[Gan84a]. At the same time, the desired 1:1 mole ratio of Ca:Mg should
be achieved in the solution. If there remain undissolved solids, they may or
may not be removed before drying the product. Remaining particles could
actually improve deicing properties, since they would enhance friction.
 29

2.3.2 Encountered Problems

As already mentioned, some difficulties arise from the different solubilities
and reactivities of the various corresponding calcium and magnesium com-
pounds, particularly at near-neutral pH. Hence process control seems to be
quite complicated. Marynowski et al. describe these problems in the fol-
lowing way [Mar85, page 459–460]:

Dolomite is essentially insoluble in neutral or alkaline solu-

tions, and its rate of dissolution in well-agitated acid solutions is
a negative exponential function of the pH (i.e., each decrease of
one pH unit results in about a tenfold increase in the dissolution
rate for a given particle size of dolomite). We have found that
even finely pulverized (< 100 mesh) dolomite requires a pH < 6
before any CO2 evolution is observable and that a pH < 4 is
required for complete dissolution within approximately 1 hour.
Thus, if use of dolomite were to be considered for pH control in
a fermentation medium, we estimate that the fermentation or-
ganism would have to be able to survive and produce acetic acid
efficiently at a pH of about 5 or less. (. . . ).
Light-burned dolime is much more reactive than dolomite, and
we estimate that it may be useful for fermentation pH control at
pH values up to about 6. At higher pH, however, the reactivities
of CaO and MgO differ greatly; CaO continues to dissolve rapidly
up to pH > 12, whereas MgO becomes essentially inert at pH > 6.
Therefore, any attempt to use dolime for fermentation pH control
at pH > 6 would produce a solution of calcium acetate only, not
one of CMA.
 30

These observations were obviously made on the basis of simple model sys-
tems and equilibrium calculations—“real” fermentation experiments showed
remarkably different results. As part of their CMA-project Wiegel et al.
produced 150 lb of CMA by fermentation [Wie91]. They grew C. ther-
moaceticum in fed batch fermentation on hydrolyzed corn starch with bub-
bling of CO2 through the medium; the temperature was maintained at about
60◦ C; the pH was kept in the range of pH 6–6.8 by addition of a dolime
slurry during fermentation, and it was finally adjusted to about pH 8 using
an extra amount of dolime; the duration was 5–6 days. In contrast to the
results cited above they found [Wie91, page 384]:

One of the most interesting observations is that magnesium of

the dolime was preferentially solubilized over calcium. Thus, the
CMA produced was more a magnesium acetate than calcium ac-
etate. The ratio between the Mg/Ca was from 1.8 to as high
as 3.

For instance, they reported final soluble concentrations of 58.8 mM for cal-
cium and 168.0 mM for magnesium, while the total concentrations for these
elements were 142.09 mM and 179.25 mM, respectively. They did not check
the nature of the precipitates (e.g. hydroxide, carbonate, phosphate), and
they did not offer an explanation. Among the possible reasons for these dif-
ferences are temperature effects, the precipitation of calcium carbonate (due
to CO2 bubbling) or less significantly, the precipitation of calcium phosphate
(because the medium contained a certain amount of phosphate). It should
be also pointed out that the fermentation is necessarily a non-equilibrium
process and that ultimately a dynamic approach must be taken in order to
explain the process behaviour.
 31

Although inconclusive regarding the details, the available literature clear-

ly indicates that one has to expect some problems with producing a stoichio-
metric solution of CMA in the context of fermentation. These problems are
less important if the neutralization reactions can be carried out using concen-
trated acetic acid at pH < 5, but still then the final pH adjustment remains
complicated [Gan84a]. As a basis to address these difficulties, a better un-
derstanding of the thermodynamics and kinetics of the various dissolution
reactions is required. In the next chapter, several simple equilibrium models
will be examined both theoretically and experimentally to find the thermody-
namic limitations. Then an attempt will be made to develop a more realistic
dynamic model that will allow process simulation and the design of a control
strategy.
Chapter 3

Equilibrium Models

The equilibrium of a closed chemical system is determined by its lowest

possible Gibbs free energy G. In this case, the free energy change ∆G of all
chemical reactions that can take place in the system has to be zero. If the
“recipe” of the system—i.e., its initial composition—is known, its equilibrium
composition at given temperature and pressure can be calculated; however,
this kind of calculation does not answer the question how fast the equilibrium
will be approached. Many reactions proceed very slowly, so that systems can
appear virtually stable although far from equilibrium.
There are two basic methods to calculate chemical equilibria: The first
technique directly minimizes the total Gibbs free energy as a function of
composition under the constraints of mass conservation for each independent
component; the second one consists of solving a set of nonlinear equations
provided by the mole balances and mass law equations, which is equivalent
to the condition of free energy minimum. For both methods, the complete
set of species present at equilibrium must be known. In addition, the min-
imization approach requires knowledge of the standard Gibbs free energy

32
 33

for each species, while the so-called equilibrium constant method depends on
information about the independent reactions that take place in the system
and their equilibrium constants. Since stability constants are readily avail-
able for most complexes and solids of interest in aquatic systems [Mar89],
the second technique is used more frequently in this context.
“Small” chemical equilibrium problems can be solved by hand using a
methodology that allows to make significant simplifications to the original
set of nonlinear equations and also gives some insight on how the system
would behave under altered conditions [Mor83]. More complex problems
involving many different species and reactions must be solved numerically on
a computer; appropriate software for the study of aquatic systems has been
developed. Most of these programs use the Newton-Raphson algorithm
to solve the set of nonlinear equations for equilibrium composition. The
sophisticated ones like MINEQL [Mor72] can handle nonideality corrections
necessary for concentrated systems and can even find the correct set of solids
in cases where it is unknown which solid phases are present at equilibrium.
Since nonideal effects must be considered for the systems to be discussed, a
brief introduction to nonideality corrections is in place.

3.1 Nonideal Effects in Aquatic Solutions

In an ideal thermodynamic system the partial molar free energy or chemical
potential µi of a species i depends exclusively on the mole fraction xi of that
species
µi = µ0i + RT ln xi , (3.1)
 34

the presence of other species has no effect on µi . The composition of aquatic

solutions is expressed in molar concentrations Ci (number of moles solute
per liter of solution) rather than on the mole fraction scale. It is possible to
rewrite (3.1) for molar concentrations

µi = µ0i + RT ln Ci (3.2)

by incorporating an approximately constant conversion term in the standard

free energy µ0i . Thus standard free energies are strictly dependent on the
chosen concentration scale, each of which corresponds to a different standard
state. In real aquatic solutions equation (3.2) holds only for infinite dilution—
otherwise interactions of solute molecules or ions occur that influence the
chemical potential of a given soluble species. In order to account for these
nonideal effects, a correction factor γi , the so-called activity coefficient, is
introduced in (3.2), so that

µi = µ0i + RT ln γi Ci . (3.3)

The product Ai = γi Ci is called the activity of species i. It remains to

determine the activity coefficients γi , an important and extensively studied
problem in applied thermodynamics.
For dilute ionic solutions the Debye-Hückel theory provides the nec-
essary basis. In this theory only interactions due to long-range electrostatic
forces among ions are considered: attraction of ions of opposite charge and
repulsion of ions of like charge. As a result of these interactions the distribu-
tion of ions in solution is not uniform—in the vicinity of a positive ion there is
a greater probability to find negative ions than positive ions, and vice versa.
This separation of charges leads to local variations in the electrical potential
which effectively decrease the partial molar free energy of each ionic species.
 35

By evaluation of the relevant electrostatic energy it is possible to determine

the activity coefficients, typically γi < 1 for ionic species. Uncharged solutes
are assumed to behave ideally, i.e. for neutral species γi = 1.
As a general measure for ionic concentration, the ionic strength I of the
solution is defined by
1X 2
I= z Ci , (3.4)
2 i i
where zi is the charge number and Ci the molar concentration of species i.
Then according to Debye-Hückel theory the activity coefficients are given
by [Mor83]
I 1/2
ln γi = −Azi2 (3.5)
1 + BaI 1/2
in which the constants A and B depend on absolute temperature T and the
dielectric constant of the system, and a is an adjustable size parameter that
corresponds roughly to the radius of the hydrated ion. The dimensions of
A and B are chosen in a way that ln γi is dimensionless (I has dimensions
mole per liter). This formula can be used for ionic strengths up to 0.1 M; at
higher concentrations other nonideal effects come into play, and some of the
approximations on which (3.5) is based are no longer valid.
There have been several attemps to extend the applicability of the classic
Debye-Hückel formula (3.5) to higher ionic strength systems, and a variety
of empirical and semiempirical expressions have been proposed. One of the
simplest and most widely used expressions of that kind is Davies’ formula
given by
I 1/2
à !
ln γi = −Azi2 − bI (3.6)
1 + I 1/2
where (for water at 25◦ C) A = 1.17, b = 0.3, and I has to be “stripped”
of its dimensions mole per liter [Mor83, Dav62]. This equation is fairly
 36

accurate in the range I = 0 to 0.7 M. For higher concentrated systems a

more sophisticated approach becomes necessary that accounts for specific
as well as unspecific ion interactions and also predicts the large increase in
activity coefficients beyond I = 0.7 M.
With the knowledge of activity coefficients, the effect of nonideality on
equilibrium is easily established. For a chemical reaction symbolized by

0 = ν 1 S1 + ν 2 S2 + ν 3 S3 + · · · (3.7)

where Si denotes species i and νi , the corresponding stoichiometric coefficient

(positive for products, negative for reactants), the molar free energy change
∆G is given by
X
∆G = ν i µi , (3.8)
i

or after substitution of equation (3.3) for µi

à !
νi µ0i + RT ln γiνi Ciνi
X Y
∆G =
i i
0
= ∆G + RT ln Q. (3.9)

The constant term ∆G0 is called the standard free energy change of the reac-
tion, and Q is the reaction quotient. At equilibrium, ∆G = 0 and therefore

∆G0
à !
Q = exp − =K (3.10)
RT

in which K denotes the equilibrium constant of reaction (3.7). In terms of

activities Ai = γi Ci , this mass law equation can be written

Aνi i = K,
Y
(3.11)
i
 37

but in most practical applications it is more convenient to deal with con-

centrations rather than activities. This leads to a second form of (3.11),
namely
Ciνi = K γi−νi ,
Y Y
(3.12)
i i

where the activity coefficients have been brought to the other side. The term
on the right side of (3.12)
γi−νi
Y
Kc = K (3.13)
i

is called the concentration equilibrium constant of reaction (3.7). Hence mass

laws can be written in terms of concentrations even for concentrated systems
where the approximation that activities and concentrations are the same is
no longer valid, if the standard equilibrium constants K are corrected using
equation (3.13).

3.2 The Hydroxide System

As a very simplified model for CMA production, the neutralization of HAc
with Ca(OH)2 and Mg(OH)2 was examined. It was assumed that there is
no CO2 present in the system. First the equilibrium speciation for a closed
system containing HAc and an excess of both Ca(OH)2 and Mg(OH)2 was
calculated, i.e., neither of the two solids could dissolve completely. The initial
composition of the system was given by
 38

Recipe 1:

• 0.5 M HAc

• 0.3 M Ca(OH)2 (s)

• 0.3 M Mg(OH)2 (s).

The problem was solved “almost” by hand using only a programmable pocket
calculator; essentially the same solution was obtained from the TITRATOR
software package which is available on MS-DOS machines [Cab87]. The
species considered to be present at equilibrium were H2 O, H+ , OH− , HAc,
Ac− , Ca2+ , CaOH+ , Ca(OH)2 (s), CaAc+ , Mg2+ , MgOH+ , Mg(OH)2 (s), and
MgAc+ . Among these species, the following independent reactions can be
written (equilibrium constants from [Mar89], at zero ionic strength and
25◦ C):
H2 O = H+ + OH− , log K = −14.0
HAc = H+ + Ac− , log K = −4.76
CaOH+ = Ca2+ + OH− , log K = −1.15
Ca(OH)2 (s) = Ca2+ + 2OH− , log K = −5.19
(3.14)
CaAc+ = Ca2+ + Ac− , log K = −1.2
MgOH+ = Mg2+ + OH− , log K = −2.6
Mg(OH)2 (s) = Mg2+ + 2OH− , log K = −11.1
MgAc+ = Mg2+ + Ac− , log K = −1.3
As principal components, the species H2 O, H+ , Ac− , Ca(OH)2 , and Mg(OH)2
were chosen. The number of components (five) is in fact equal to the number
of species (thirteen) minus the number of independent reactions (eight) since
each reaction defines a stoichiometric relationship among species which allows
to express one of them as a formula of the others, i.e. for each species it is
 39

possible to write a mass law involving only principal components by simple

rearrangement of the given independent reactions. The linear operations
necessary to obtain the desired form of mass laws may include adding two or
more reactions (their log K values have to be added, too) and changing the
direction of a reaction (what simply changes the sign of the corresponding
log K value).
All the information regarding the setup of the equilibrium problem can
be organized conveniently in form of a tableau which is shown in Table 3.1.
At the top of this tableau, the stoichiometric formulae of the species as a
function of the components are given along with the corresponding log K
(and also log Kc ) values, and at the bottom, stoichiometric coefficients and
total amounts are included for the chemicals used in the recipe. Hence each
row in the upper part corresponds to a mass law, and each column stands for
a mole balance. Water is completely omitted from the tableau although it is
chosen as a component, since its presence is understood in aquatic systems.
From the columns, the mole balances for the components are ([. . .] denoting
concentrations)

TOT H = [H+ ] − [OH− ] + [HAc] + 2[Ca2+ ] + [CaOH+ ]

+ 2[CaAc+ ] + 2[Mg2+ ] + [MgOH+ ] + 2[MgAc+ ]
= [HAc]T = 0.5 M (3.15)
TOT Ac = [HAc] + [Ac− ] + [CaAc+ ] + [MgAc+ ]
= [HAc]T = 0.5 M (3.16)
TOT Ca(OH)2 = [Ca2+ ] + [CaOH+ ] + [Ca(OH)2 (s)] + [CaAc+ ]
= [Ca(OH)2 (s)]T = 0.3 M (3.17)
TOT Mg(OH)2 = [Mg2+ ] + [MgOH+ ] + [Mg(OH)2 (s)] + [MgAc+ ]
 40

Table 3.1: Tableau-representation of the hydroxide system (Recipe 1). Equi-

librium constants at 25◦ C; log K at zero ionic strength, log Kc at I = 0.35 M.

Species H+ Ac− Ca(OH)2 (s) Mg(OH)2 (s) log K log Kc

H+ 1
OH− −1 −14.0 −13.7
HAc 1 1 +4.76 +4.46
Ac− 1
Ca2+ 2 1 +22.81 +23.12
CaOH+ 1 1 +9.96 +9.96
Ca(OH)2 (s) 1
CaAc+ 2 1 1 +24.01 +23.71
Mg2+ 2 1 +16.9 +17.2
MgOH+ 1 1 +5.5 +5.5
Mg(OH)2 (s) 1
MgAc+ 2 1 1 +18.2 +17.9

Recipe TOT X

HAc 1 1 0.5 M
Ca(OH)2 (s) 1 0.3 M
Mg(OH)2 (s) 1 0.3 M
 41

= [Mg(OH)2 (s)]T = 0.3 M (3.18)

and from the rows, the mass laws for the species are (assuming zero ionic
strength)

[OH− ] = 10−14.0 [H+ ]−1 (3.19)

[HAc] = 10+4.76 [H+ ][Ac− ] (3.20)
[Ca2+ ] = 10+22.81 [H+ ]2 (3.21)
[CaOH+ ] = 10+9.96 [H+ ] (3.22)
[CaAc+ ] = 10+24.01 [H+ ]2 [Ac− ] (3.23)
[Mg2+ ] = 10+16.9 [H+ ]2 (3.24)
[MgOH+ ] = 10+5.5 [H+ ] (3.25)
[MgAc+ ] = 10+18.2 [H+ ]2 [Ac− ] (3.26)

In general, these mass laws are valid for activities rather than concentrations,
but it is possible to retain the given convenient form at ionic strength greater
than zero by simply using the concentration equilibrium constants Kc defined
in equation (3.13) instead of the standard constants K. The activities of solid
species are always fixed to one, hence they do not appear in (3.19)–(3.26).
To obtain a first rough estimation for the equilibrium composition, the
uncorrected mass laws (3.19)–(3.26) were substituted into the mole bal-
ances (3.15)–(3.18), resulting in four nonlinear equations for the component
concentrations [H+ ], [Ac− ], [Ca(OH)2 (s)], and [Mg(OH)2 (s)]. It was possi-
ble to further eliminate [Ac− ] from the TOT H balance (3.15) using equa-
tion (3.16); thus only one equation of one unknown had to be solved numeri-
cally, which could be easily done with the built-in equation solver of a pocket
calculator. The result is shown in Table 3.2 (first column). From this data
 42

Table 3.2: Equilibrium speciation for the hydroxide system (Recipe 1). pH
corresponds to negative log of hydrogen ion activity rather than concentra-
tion.

species Equilibrium Concentrations (M)

I=0 I = 0.35 experimental

H+ 8.32 · 10−13 9.07 · 10−13

pH 12.08 12.19 12.51
OH− 1.20 · 10−2 2.20 · 10−2
Ac− 0.293 0.352
HAc 1.40 · 10−8 9.20 · 10−9

Ca2+ 4.47 · 10−2 0.108

CaOH+ 7.59 · 10−3 8.27 · 10−3
CaAc+ 0.207 0.148
total soluble Ca 0.260 0.265 0.288
Ca(OH)2 (s) 4.02 · 10−2 3.49 · 10−2

Mg2+ 5.51 · 10−8 1.30 · 10−7

MgOH+ 2.63 · 10−7 2.87 · 10−7
MgAc+ 3.21 · 10−7 2.30 · 10−7
total soluble Mg 6.39 · 10−7 6.47 · 10−7 7.67 · 10−5
Mg(OH)2 (s) 0.3 0.3
 43

the ionic strength I could be calculated: the value found was I = 0.35 M,
indicating a fairly concentrated solution.
For a more accurate result, ionic strength corrections were required. Ap-
proximated values for the activity coefficients γi at I = 0.35 M were obtained
from Davies’ formula (3.6), which gave γi = 0.70 and γi = 0.24 for ions hav-
ing single and double charge, respectively. Then the concentration equilib-
rium constants Kc could be determined according to (3.13); their log values
are given in Table 3.1. Using the corrected constants, a slightly different
equilibrium composition was found (Table 3.2, second column). The differ-
ences were smaller than expected, in particular for the total soluble calcium
and magnesium concentrations, both of which did not change significantly.
The resulting new ionic strength was I = 0.48 M, its increase was mainly
due to the increase of [Ca2+ ] (double charge!) at the expense of [CaAc+ ].
It was not necessary to make another iteration using I = 0.48 M, because
the activity coefficients have almost constant values in the range I = 0.3
to 0.7 M, so that no further significant changes were expected, taking into
account the restricted accuracy of such calculations.
In order to confirm the validity of these calculations, the equilibrium
composition for Recipe 1 was also determined experimentally. The system
was prepared from deionized water and reagent-grade HAc, Ca(OH)2 , and
Mg(OH)2 . Stirring and sparging with nitrogen were provided during equi-
libration; the pH of the solution was monitored. After approximately half
an hour the pH had stabilized, and it was assumed that the system had
reached equilibrium. The undissolved solids were removed by filtration, and
the sample solution was analyzed for total soluble calcium and magnesium by
inductively coupled plasma emission spectroscopy (ICP). It was found that
 44

pH and total soluble calcium had been predicted with an error of less than
10%, but the concentration of magnesium had been grossly underestimated
by almost two orders of magnitude (see Table 3.2, third column1 ). One pos-
sible explanation for this significant deviation might be that the system had
not yet reached “true” equilibrium, i.e., the solution was still oversaturated
with respect to magnesium species.
Very similar values were obtained for industrial-grade type S dolime,
supplied by The Western Lime & Cement Co. in West Bend, Wisconsin:
pH 12.47, 0.258 M total soluble calcium, and 4.03 · 10−5 M total soluble
magnesium. Obviously the impurities2 did not have a strong influence on
the equilibrium speciation.
The theoretical as well as the experimental results show that in ac-
cordance with the observations by [Mar85], only a negligable amount of
Mg(OH)2 (s) dissolved because of the high pH, and essentially a solution of
calcium acetate was produced, not one of CMA. However, it must be pointed
out that in the context of the proposed fermentation process a pH > 12 will
never be reached. This leads to the more important question what happens,
if there is no excess of Ca(OH)2 (s) in the system, and the pH is allowed to
drop far enough for dissolution of Mg(OH)2 (s).
An answer to this question was found by simulating a titration of calcium
and magnesium hydroxide with acetic acid, i.e., by determining the equilib-
rium composition depending on the total amount of acetic acid added to
the system. These calculations were made using the TITRATOR program.
Unfortunately it turned out that the available option of “automatic” ionic
1
The original data on the ppm scale can be found in Appendix A.1.
2
See Appendix B.1 for a chemical analysis of the lime.
 45

strength corrections did not work properly, so zero ionic strength had to be
assumed. In order to restrict the error introduced at higher ionic strength,
only half the hydroxide concentrations of Recipe 1 were chosen according to

Recipe 2:

• 0–0.7 M HAc (titrant)

• 0.15 M Ca(OH)2 (s)

• 0.15 M Mg(OH)2 (s).

The results are visualized separately for solid species and total soluble con-
centrations in Figures 3.1 and 3.2; the equilibrium concentration of H+ has
been included in both diagrams. There are two interesting points, where
some of the concentrations change dramatically: one at TOT HAc = 0.3 M,
where just all Ca(OH)2 (s) is dissolved, and another at TOT HAc = 0.6 M,
where also the last bit of Mg(OH)2 (s) has disappeared. As long as there is
Ca(OH)2 (s) present, a pH > 12 is maintained, and the solution phase con-
tains only traces of magnesium species. After Ca(OH)2 (s) has disappeared,
the pH drops to a value of about pH 9–10 and again remains almost constant
while the dissolution of Mg(OH)2 (s) takes place; finally it drops further to
the region of pH 4–6 (the pKa of HAc is 4.76, i.e. a pH of about this value
would be reached when the concentrations of acetate ion and undissociated
acetic acid become equal).
Several real titration experiments were performed with the individual
hydroxides and the mixed system according to Recipe 2. After each addition
of HAc, the pH was allowed to stabilize before a reading was taken and the
titration was continued. The time required for a stable pH reading was less
 46

0
×××××
×××
×
−2 ×

−4

3
33
−6 3
log C

−8

3333333
333
−10 3

−12 33333
333333 Ca(OH)2 (s) ×
Mg(OH)2 (s)
H+ 3
−14
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7
TOT HAc

Figure 3.1: Titration of the hydroxide system (Recipe 2)—variation of solid

species concentrations with TOT HAc.
 47

×× ××××××××××××××××××××
× ××
−2 ×

−4

3
33
−6 3
log C

−8

3333333
333
−10 3

−12 33333
333333 total soluble Ca ×
total soluble Mg
H+ 3
−14
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7
TOT HAc

Figure 3.2: Titration of the hydroxide system (Recipe 2)—variation of total

soluble calcium and magnesium concentrations with TOT HAc.
 48

than 20 sec for Ca(OH)2 , but as long as 10 min for Mg(OH)2 , indicating that
Mg(OH)2 reacted significantly slower than Ca(OH)2 . A good agreement was
found between the obtained experimental curves of pH versus TOT HAc and
the simulation results stated above: Ca(OH)2 (s) dissolved at pH > 12 and
Mg(OH)2 (s) at pH 9–10.
Therefore, no thermodynamic limitation should be expected for the dis-
solution reaction of Mg(OH)2 (s) at any pH < 9, and the experiment showed
that although Mg(OH)2 (s) dissolved slowly compared to Ca(OH)2 (s) the dis-
solution rate was still considerable. This result is in contrast to the observa-
tion reported by [Mar85] that “MgO becomes essentially inert at pH > 6”.
Since [Mar85] used unslaked dolomitic lime CaO · MgO rather than highly
hydrated (type S) dolime Ca(OH)2 · Mg(OH)2 , their problem was most likely
due to hard-burned (sintered) MgO with low reactivity, hence the exergonic
reaction
MgO(s) + H2 O −→ Mg(OH)2 (s)

proceeded very slowy at pH > 6. Apart from using type S dolime which is
more expensive than quicklime and normal hydrated lime, two other possi-
bilities to overcome this problem would be

• to dissolve MgO at pH < 6 as proposed by [Mar85], or

• to use selectively calcined dolomite (containing light-burned MgO with

high reactivity) which could actually be produced at lower cost than
quicklime and might have other advantages.

The last case represents an interesting combined carbonate-hydroxide system

and will be considered again in the context of a dynamic model after the
carbonate system has been examined.
 49

3.3 The Carbonate System

Regardless wheter dolime, dolomite, or selectively calcined dolomite is used
for acetic acid neutralization in a real fermentation processs, carbonates
will always be present in the system since CO2 is an intermediate prod-
uct in homoacetate fermentations (and occasionally also provided by gas
bubbling). Thermodynamically the solids calcium hydroxide and calcium
carbonate cannot coexist in water at ambient partial pressure of CO2 which
is PCO2 = 10−3.5 bar, because the exergonic reaction

CO2 (g) + Ca(OH)2 (s) −→ CaCO3 (s) + H2 O (3.27)

must proceed to the right until Ca(OH)2 (s) is exhausted if PCO2 is fixed
[Mor83]. The same is true for the corresponding magnesium compounds;
again the reaction

CO2 (g) + Mg(OH)2 (s) −→ MgCO3 (s) + H2 O (3.28)

is exergonic at PCO2 = 10−3.5 bar. Therefore, the hydroxide system discussed

in the previous section would be ultimately transformed into the carbonate
system if there is enough CO2 present. Both reactions (3.27) and (3.28) can
be applied also to the dry compounds—the hydroxides absorb CO2 from the
air and are gradually transformed to carbonates (a possible source of error
in experiments, when “old” chemicals are used).
The carbonate system was examined in the same way as the hydroxide
system; hence only a brief discussion is necessary. Again, the closed system
was defined by
 50

Recipe 3:

• 0.5 M HAc

• 0.3 M CaCO3 (s) (calcite)

• 0.3 M MgCO3 (s) (magnesite)

• CO2 (g) at fixed partial pressure PCO2 = 10−3.5 bar,

having an excess of both calcium and magnesium carbonate. Now 21 species

had to be considered, 15 independent reactions among them could be written,
and hence 6 principal components had to be chosen. Species and components
can be found in Table 3.3, the independent reactions are

H2 O = H+ + OH− , log K = −14.0

HAc = H+ + Ac− , log K = −4.76
CO2 (g) + H2 O = H2 CO∗3 , log K = −1.5
H2 CO∗3 = H+ + HCO−
3, log K = −6.3
HCO−
3 = H+ + CO2−
3 , log K = −10.3
CaOH+ = Ca2+ + OH− , log K = −1.15
CaAc+ = Ca2+ + Ac− , log K = −1.2
CaHCO+
3 = Ca2+ + HCO−
3, log K = −1.26 (3.29)
CaCO3 = Ca2+ + CO2−
3 , log K = −3.2
CaCO3 (s) = Ca2+ + CO2−
3 , log K = −8.35
MgOH+ = Mg2+ + OH− , log K = −2.6
MgAc+ = Mg2+ + Ac− , log K = −1.3
MgHCO+
3 = Mg2+ + HCO−
3, log K = −1.01
MgCO3 = Mg2+ + CO2−
3 , log K = −3.4
MgCO3 (s) = Mg2+ + CO2−
3 , log K = −7.46,
 51

Table 3.3: Tableau-representation of the carbonate system (Recipe 3). Equi-

librium constants at 25◦ C; log K at zero ionic strength, log Kc at I = 0.33 M.

Species H+ Ac− CaCO3 (s) MgCO3 (s) CO2 log K log Kc

H+ 1
OH− −1 −14.0 −13.7
HAc 1 1 +4.76 +4.46
Ac− 1
CO2 (g) 1
H2 CO∗3 1 −1.5 −1.5
HCO− 3 −1 1 −7.8 −7.5
CO2−
3 −2 1 −18.1 −17.2
Ca2+ 2 1 −1 +9.75 +10.06
CaOH+ 1 1 −1 −3.1 −3.1
CaAc+ 2 1 1 −1 +10.95 +10.65
CaHCO+ 3 1 1 +3.21 +3.21
CaCO3 1 −5.15 −5.15
CaCO3 (s) 1
Mg2+ 2 1 −1 +10.64 +10.95
MgOH+ 1 1 −1 −0.76 −0.76
MgAc+ 2 1 1 −1 +11.94 +11.64
MgHCO+ 3 1 1 +3.85 +3.85
MgCO3 1 −4.06 −4.06
MgCO3 (s) 1

HAc 1 1 0.5 M
CaCO3 (s) 1 0.3 M
MgCO3 (s) 1 0.3 M
CO2 (g) 1 ?
 52

where H2 CO∗3 represents by convention H2 CO3 + CO2 (aq). The equilibrium

speciation was calculated using TITRATOR, first at zero ionic strength, then
at the resulting value I = 0.33 M (with activity coefficents γi = 0.7 for single
charged ions and γi = 0.25 for double charged ions); both compositions are
given in Table 3.4 together with the experimental result obtained for a system
that was prepared from deionized water and reagent-grade HAc, CaCO3 , and
MgCO3 .3 Again there were no great differences between the uncorrected and
the corrected total soluble concentrations of calcium and magnesium. This
time the equilibration of the real system took several hours (compared to half
an hour for the hydroxide system), indicating that the carbonates reacted
significantly slower than the hydroxides. The total concentration of soluble
magnesium species and the measured pH agree well with the predicted values,
but the total soluble calcium concentration was overestimated by a factor of
ten. There was no convincing explanation found for this deviation; maybe,
the powdered carbonates did not have the same thermodynamic properties
as calcite and magnesite.
In contrast to the hydroxides a considerable amount of both carbonates
dissolved, but now magnesium was preferentially solubilized over calcium
(about ten to hundred times as much magnesium as calcium), and the equi-
librium pH was much lower.
The simulated titration at zero ionic strength was carried out for the
system defined by
3
The carbonates used in this experiment were in form of very fine powders, possibly
produced by precipitation, and hence their crystal structure was not necessarily that of
calcite and magnesite, respectively.
 53

Table 3.4: Equilibrium speciation for the carbonate system (Recipe 3). pH
corresponds to hydrogen ion activity rather than concentration.

Species Equilibrium Concentrations (M)

I=0 I = 0.33 experimental

H+ 1.56 · 10−8 1.72 · 10−8

pH 7.81 7.91 8.22
OH− 6.42 · 10−7 1.16 · 10−6
Ac− 0.288 0.344
HAc 2.58 · 10−4 1.71 · 10−4
H2 CO∗3 1.00 · 10−5 1.00 · 10−5
HCO− 3 3.22 · 10−4 5.80 · 10−4
CO2−3 1.03 · 10−6 6.72 · 10−6

Ca2+ 4.32 · 10−3 1.08 · 10−2

CaOH+ 3.91 · 10−8 4.33 · 10−8
CaAc+ 1.97 · 10−2 1.45 · 10−2
CaHCO+ 3 2.53 · 10−5 2.80 · 10−5
CaCO3 7.08 · 10−6 7.08 · 10−6
total soluble Ca 2.40 · 10−2 2.53 · 10−2 3.60 · 10−3
CaCO3 (s) 0.276 0.275

Mg2+ 3.35 · 10−2 8.37 · 10−2

MgOH+ 8.57 · 10−6 9.47 · 10−6
MgAc+ 0.192 0.141
MgHCO+ 3 1.10 · 10−4 1.22 · 10−4
MgCO3 8.71 · 10−5 8.71 · 10−5
total soluble Mg 0.226 0.225 0.219
MgCO3 (s) 7.39 · 10−2 7.49 · 10−2
 54

Recipe 4:

• 0–0.7 M HAc (titrant)

• 0.15 M CaCO3 (s) (calcite)

• 0.15 M MgCO3 (s) (magnesite)

• CO2 (g) at fixed partial pressure PCO2 = 10−3.5 bar

using TITRATOR; Figures 3.3 and 3.4 show the resulting titration curves.
Initially the pH has a value of about pH 8, and it only slightly decreases while
MgCO3 (s) and then CaCO3 (s) dissolve. After both solids have disappeared, it
drops to the region of pH 4–6. There are no big concentration “jumps” at the
point TOT HAc = 0.3 M as was the case for the hydroxides—the reactivities
of MgCO3 (s) and CaCO3 (s) are different, but not very much. Because of the
slow reaction, no real titration experiment was performed; this would have
required an automated titrator.
An experiment using the double carbonate dolomite CaMg(CO3 )2 instead
of a mixture of CaCO3 and MgCO3 showed quite different results: the reac-
tivity was much lower than for the system defined by Recipe 3 (the pH was
still below pH 6 after several hours of equilibration), and very surprisingly,
more calcium than magnesium was found in the filtered sample taken after
24 hours. The concentrations were 0.121 M and 0.0854 M for calcium and
magnesium, respectively. Again, an excess of base was used (about 0.6 M
dolomite), and the material was finely pulverized. It is difficult to find an
explanation for this strange behaviour; first of all, the system probably did
not reach true equilibrium, and perhaps the limestone contained not only
dolomite, but also some calcite which could possibly dissolve much faster
 55

0
××××××××××××××××
×××
×
×
−2 ×

−4

3
33
−6 3
log C
33
333333 3333333
−8 333333
3

−10

−12
CaCO3 (s) ×
MgCO3 (s)
H+ 3
−14
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7
TOT HAc

Figure 3.3: Titration of the carbonate system (Recipe 4)—variation of solid

species concentrations with TOT HAc.
 56

0
××××
×× ×××××
××
×
−2 × ×××××
× ×××
×

×
−4

3
33
−6 3
log C
33
333333 3333333
−8 333333
3

−10

−12
total soluble Ca ×
total soluble Mg
H+ 3
−14
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7
TOT HAc

Figure 3.4: Titration of the carbonate system (Recipe 4)—variation of total

soluble calcium and magnesium concentrations with TOT HAc.
 57

than dolomite. However, an analysis of the limestone itself showed that it

actually had just slightly more calcium than would be theoretically expected
for pure dolomite,4 and it is not sure whether or not the given explanation
is correct. If the use of dolomite as a neutralizer is considered in spite of its
low reactivity, this problem definitely has to be addressed again.
Another experiment was made in order to examine the influence of CO2
bubbling: a hydroxide system prepared according to Recipe 1 was bubbled
with CO2 instead of nitrogen. After about three hours the pH had stabi-
lized at a value of pH 6.48. The sample was filtered and stored for several
days in a closed tube before it could be analyzed. During this time a small
amount of precipitate formed, so that the solution had to be filtered again.
Unfortunately, no second pH reading was taken at the time of analysis. The
total soluble concentrations for calcium and magnesium were 4.65 · 10−3 M
and 0.278 M, respectively. These values agree quite well with the ones found
for the carbonate system, indicating that the hydroxides had been actually
transformed to carbonates. Both concentrations are significantly higher than
in the carbonate case discussed above; this is probably due to a higher con-
centration of CO2 and therefore a lower equilibrium pH, because the sample
was not equilibrated with air.
In the light of these results, the following explanation seems to be plau-
sible for the observation by [Wie91] that magnesium was preferentially sol-
ubilized during their CMA-fermentation: although they did not directly use
carbonates, it is very likely that carbonates were formed according to reac-
tions (3.27) and (3.28) or just through simple precipitation,5 thus leading to
4
The results of this analysis can be found in Appendix B.2.
5
As a matter of fact, precipitated CaCO3 is normally produced by bubbling CO2
 58

a combined hydroxide-carbonate system with somewhat intermediate char-

acteristics. However, it is not possible to apply the results of the equilibrium
case directly to the real fermentation system, since during the whole pro-
cess the solid calcium and magnesium phases are not in equilibrium with
the liquid phase. Therefore, the equilibrium models discussed so far must
be considered as “limiting cases” that can only give information about basic
thermodynamic properties (e.g. the equilibrium pH values of the dissolution
reactions), but not about the behaviour of real non-equilibrium processes. A
dynamic approach is necessary to obtain more realistic models; this approach
will be discussed in the following chapter.

through an aqueous suspension of Ca(OH)2 (milk of lime).

Chapter 4

The Dynamic Approach

Although lime and limestone neutralization is widely used in waste water

treatment, it is difficult to find conclusive models for the dissolution kinetics
in literature. Obviously in most applications to date a detailed model was not
required—the only thing one had to know was how many minutes or hours
it takes to neutralize a given acid solution using a certain type and amount
of neutralization agent. From this empirical data it can be concluded that in
general the calcium minerals react faster than the magnesium minerals, and
that the hydroxides have a higher reactivity than the carbonates [Boy80].
In the case of CMA production it definitely is necessary to study the
kinetics of the neutralization reactions, since not only “neutralizing capacity”
and “reaction time”, but also the composition of the resulting solution are
important. Therefore, an extensive literature search was performed using
computerized versions of Engineering Index and Chemical Abstracts. About
25 papers related to the following fields were found:

1. Kinetics of acid neutralization with lime or limestone; most of these

59
 60

“engineering” papers dealt with the neutralization of sulfuric acid in

the context of acidic waste treatment (coal mine drainage, pickling
liquors) [Bar76, Bog85, Hoa45, Par65, Pea75].

2. Kinetics of heterogenous reactions in general [Bir52].

3. Calcite dissolution kinetics; many publications of detailed studies that

were made by geologists in order to understand the natural dissolu-
tion of carbonate rocks [Cho89, Com90a, Com90b, Plu76, Plu78,
Plu79, Ric83, Sjö76, Sjö84, Sua84, Ber74].

4. Dolomite and Magnesite dissolution kinetics [Bus82, Cho89, Sua84].

5. Dissolution kinetics of CaO and MgO [Cas91, Gor84, Mac71,

Seg78, Seg88, Ver69].

There was just one single publication found that specifically dealt with the
dissolution of limestone in acetic acid [Sua84], and nothing about the reac-
tion of acetic acid with lime. Probably some of the more general models could
be applied to the production of CMA, but difficulties are to be expected,
since these models were found using CO2 -water systems in combination with
strong mineral acids like HCl or H2 SO4 . Hence it would be necessary to per-
form experiments in order to verify model structure and parameters. Most
likely the dynamic model will turn out to be a system of differential algebraic
equations (DAEs)—some differential equations describing the heterogenous
surface reactions combined with a set of algebraic equations (mole balances
and mass laws) which determine the speciation in the bulk fluid.
Another question is whether the design of the process can be based on
such a dynamic model which will necessarily involve large uncertainties (liter-
 61

ature data on the dissolution rate of one specific compound at a given pH and
temperature may differ by an order of magnitude or more among different
sources). It may be possible to overcome this problem by a “clever” process
design that does not require exact knowledge of the dissolution kinetics.

4.1 Heterogenous Reactions

The dissolution of lime or limestone particles in an aquatic solution of acetic
acid is a heterogenous reaction—it occurs at the interface between two phases,
a solid and a solution. The rate of a heterogenous reaction is proportional to
the available surface area of the solid. In order to find the factors which may
control the rate it is helpful to decompose the whole process into a series of
five primary steps [Bir52]:

1. Transport of the reactants from the bulk solution to the interface.

2. Adsorption at the interface.

3. Reaction at the interface.

4. Desorption of the products.

5. Transport of the products from the interface to the bulk solution.

Of these, steps 1 and 5 are tranport processes, steps 2–4 are chemical pro-
cesses. Depending on which step is the slowest and therefore rate determin-
ing, the overall rate may be either

• transport controlled, if the rate of mass transfer to or from the solid

surface by diffusion and convection is very much slower than the rate
of chemical reaction at the interface,
 62

boundary bulk
solid layer solution
6

Ceq a

b
C

- z

Figure 4.1: Concentration profiles near the interface for (a) transport con-
trolled, (b) chemically controlled, and (c) intermediate-type heterogenous
reactions.

• chemically controlled, if the surface reaction is much slower than either

of the transport processes, or

• of intermediate type (general case), if both rates are of the same order
of magnitude and hence the observed rate is determined by a function
of the two.

Figure 4.1 shows the corresponding concentration profiles near the interface
for a dissolution reaction involving only one solute and the solvent. In case
of a chemically controlled reaction the bulk concentration C of the solute
extends right to the surface, and there is no boundary layer. In both other
cases the concentration at the interface is not the same as in the bulk so-
lution; there must exist a boundary layer with a concentration gradient. If
the observed rate is transport controlled then equilibrium is established at
 63

the interface, i.e., the surface concentration is equal to the equilibrium con-
centration Ceq . In the general case the surface concentration has some value
between C and Ceq .
The classical theory of heterogenous reactions was formulated by Nernst
at the beginning of the century. He assumed that the chemical processes are
always much faster than the transport processes, that there is no concen-
tration gradient within the bulk solution in well-stirred systems, but the
concentration varies linearly with distance (measured normal to the surface)
in a thin boundary layer adhering to the solid surface, and that the thickness
of this layer is only a function of the stirring rate and geometry of the system,
but does not depend on the coefficient of diffusion of the solute, the viscos-
ity, and the temperature [Bir52]. In general, none of these assumptions is
justified, but the theory provided a useful approximation at least for many
transport controlled reactions. Consider a solid of surface area A being in
contact with a volume V of solution. Then it can be shown by Fick’s law
and a mass balance that
dC DA dC
=− , (4.1)
dt V dz
where dC/dt is the rate of change of concentration in the bulk solution, dC/dz
is the concentration gradient normal to the surface and D is the coefficient of
diffusion of the solute. Assuming equilibrium at the interface and linearity,
the gradient can be written
dC C − Ceq
= , (4.2)
dz δ
where Ceq is the equilibrium concentration and δ the thickness of the diffusion
layer; by substitution in (4.1),
dC DA
= (Ceq − C). (4.3)
dt Vδ
 64

This is a very simple first order rate equation, but reasonable agreement was
found for many heterogenous reactions. Careful analyses have shown that
the values obtained for δ are too large to be interpreted as above, and that
the linearity assumption for the gradient is far from reality. As more and
more chemically controlled reactions were discovered, the need to generalize
from the transport-limited case was realized.
In order to extend the Nernst theory to reactions that are at least partly
controlled by surface reaction, the equilibrium assumption must be given up,
and an additional rate equation for the chemical reaction is required [Plu76].
The molar flux JT from the interface to the bulk solution can be expressed
as
JT = kT (C (s) − C), (4.4)

where C is again the bulk fluid concentration and C (s) is the concentration
at the outer edge of a very thin surface layer right at the interface, see Fig-
ure 4.2. The constant kT is the mass transport coefficient for the solute and
is determined by fluid properties and surface geometry. The concentration
difference C (s) − C acts as a driving force for the mass transport from the
interface to the bulk solution through the diffusion boundary layer. Equa-
tion (4.4) is a good approximation for relatively small fluxes and is valid only
if no homogenous reactions involving the solute occur within the boundary
layer. On the other hand, it is often possible to express the rate of surface
reaction by an empirical relation of the form

(s)
JC = kC (Ceq − C (s) )n , (4.5)

where JC is the molar flux into the surface layer that results from the chemical
(s)
reaction characterized by rate constant kC and empirical order n, and Ceq
 65

diffusion
boundary bulk
solid layer solution

C (s) C

- -
JC JT

surface
layer

Figure 4.2: Molar fluxes due to surface reaction and mass transfer for an
intermediate-type heterogenous reaction.

denotes the equilibrium concentration at the interface. Since only a small

amount of mass can be accumulated in the thin surface layer, a steady state
is established almost instantaneously; the mass balance for the surface layer
forces JC = JT . Hence equations (4.4) and (4.5) may be combined as

J = kT (C (s) − C) = kC (Ceq
(s)
− C (s) )n , (4.6)

where the subscripts of J have been dropped. Equation (4.6) applies to

the general intermediate case and allows to eliminate the unknown concen-
tration C (s) . For a first order chemical reaction, n = 1, and C (s) can be
expressed as
kT kC
C (s) = C+ C (s) , (4.7)
kT + k C kT + kC eq
thus leading to the simple relation
kT kC
J= (C (s) − C). (4.8)
kT + kC eq
 66

For kT À kC equation (4.8) reduces to

(s)
J = kC (Ceq − C), (4.9)

the limiting case of reaction control. Similarly, the transport controlled case
is obtained for kC À kT . The rate of change of concentration in the bulk
solution for dissolution from a surface of area A into a fluid of volume V is
related to the flux J by
dC A
= J (4.10)
dt V
as long as no homogenous reactions with the solute occur in the bulk fluid.
This allows to write the rate equation
dC A kT kC
= (C (s) − C) (4.11)
dt V kT + kC eq
for first order, intermediate-type reactions. In some cases, C may denote
not the bulk concentration of the dissolving species itself, but of some other
species in the system which reacts with the solid and is depleted, e.g., H+
for the dissolution of a metal in acid. However, it must be pointed out
that the equations discussed here can be applied only to very simple disso-
lution processes involving just one significant heterogenous reaction and no
homogenous reactions in the boundary layer or bulk solution. As will be seen
in the next section, surface processes involving more than one species, paral-
lel heterogenous reactions, and accompanying homogenous reactions actually
occur; the corresponding models are of course more complicated.

4.2 Kinetics of Dissolution

Various models for the dissolution of carbonates and oxides in acids can
be found in the literature; some of them shall be discussed briefly at this
 67

point because of their relevance for CMA production. From these models,
approximate dissolution rates will be calculated for the range of pH 5–7 in
order to give an idea of the orders of magnitude to be expected. The two
classes of potential raw materials, carbonate minerals and oxide minerals,
are discussed separately. More detailed information about the dissolution of
carbonates is available than about that of oxides, especially in the higher pH
range.

4.2.1 Carbonates: Calcite, Magnesite, Dolomite

The dissolution and precipitation of carbonate minerals plays an important
role in natural systems, e.g. for the weathering of carbonate rocks, or the
composition of natural waters. Hence the study of mechanisms and kinetics of
these processes has been an area of interest particularly for geochemists, geol-
ogists, and soil scientists, and most of the publications are found in geological
journals. Calcite has been the most extensively and rigorously studied, but
it has been shown that other simple carbonates like magnesite have a similar
behaviour. Only the dissolution of dolomite as a two-component carbonate
seems to follow a different mechanism, but still the same methodology could
be applied to its study. First the findings for calcite will be discussed, and
then directly compared with the corresponding results for magnesite; finally
the different behaviour of dolomite will be addressed.

Studies of Calcite Dissolution

Soon it was realized that there exist different mechanisms for calcite disso-
lution depending on the pH regime, and the partial pressure of CO2 . The
saturation of the bulk solution with respect to Ca2+ and CO2−
3 has also an
 68

important influence, since far from equilibrium the overall dissolution rate is
determined solely by the forward rate (the backward rate is neglegibly small),
while near equilibrium the forward and backward rate are of the same order
of magnitude. In some cases, experimental conditions may also play a role,
especially hydrodynamics and surface geometry of the system. For instance,
it has been shown only recently that the dissolution of calcite at pH < 4
is not necessarily transport controlled as supposed by most authors, but
rather limited by the rate of surface reaction under conditions of high mass
transport [Com90a].
The first comprehensive model of the dissolution process was the one de-
veloped by Plummer et al. [Plu78, Plu79]. In their experiments, they
used two different fractions of crushed and sized Iceland spar (estimated
surface area of the particles 44.5 cm2 /g and 96.5 cm2 /g, respectively) in a
well-stirred system at fixed PCO2 (provided by bubbling with CO2 , N2 , or a
mixture of both) and constant temperature. A “pH-stat” method was used
to study dissolution far from equilibrium at low pH, while “free drift” experi-
ments were made to examine systems nearer equilibrium. In the overlapping
region of the two methods, the obtained results were in agreement.
During the pH-stat runs, a constant specified pH was maintained through
controlled addition of a standard HCl solution using an automated titrator;
the rate of addition of standard solution dVHCl /dt was recorded. Then the
rate of calcite dissolution R could be estimated from
CHCl dVHCl
R = 0.5 , (4.12)
A dt
where CHCl is the molarity of the standard HCl solution and A the surface
area of the solid particles. This relation is justified, because at constant
2−
pH and PCO2 the activities of H2 CO∗3 , HCO−
3 , and CO3 are constant in
 69

solution. Hence for each mole of CaCO3 dissolved, two moles of hydrogen
ions are consumed, and one mole of CO2 leaves the system with the bubbled
gas phase; i.e., the rate of calcite dissolution is essentially half the rate of
“hydrogen ion addition”.
In case of the free drift runs, the pH was allowed to vary as the reaction
proceeded to near equilibrium at constant PCO2 and temperature. Measure-
ment of the pH as a function of time provided a means to follow the reac-
tion, since the composition of the bulk solution at each point of time could
be calculated from pH and PCO2 using an iterative algorithm (similar to the
algorithms discussed in Chapter 3). The rate of calcite dissolution had to be
computed stepwise from this data by means of a central difference formula.
By analysis of their pH-stat data, Plummer et al. found the following
empirical expression for the forward rate of dissolution

Rf = k1 AH+ + k2 AH2 CO∗3 + k3 AH2 O , (4.13)

where Ai is the activity of species i in the bulk fluid, and ki are rate con-
stants dependent on temperature. The dimensions used in [Plu78] are
M = mol l−1 = mmol cm−3 for the activities Ai , cm s−1 for the rate con-
stants ki , and hence mmol cm−2 s−1 for the rate Rf itself. Throughout this
section, another convention shall be adopted in order to make the models of
different authors better comparable and avoid confusion: all rates are mea-
sured in mol cm−2 s−1 , activities have units of M = mol l−1 , and first order
rate constants are assigned dimensions of mol cm−2 s−1 M−1 = l cm−2 s−1 ,
i.e., their numerical values differ from [Plu78] simply by a factor of 1000.
The temperature dependence of the rate constants at temperatures between
 70

Table 4.1: Summary of empirical rate constants for the dissolution of cal-
cite and magnesite; the corresponding rate expressions are given by equa-
tions (4.19) and (4.29). First order constants k1 , k2 , k3 and second order con-
stants k−3 , k4 have dimensions of mol cm−2 s−1 M−1 and mol cm−2 s−1 M−2 ,
respectively.

Calcite Magnesite

[Cho89] [Plu78] [Cho89]

25◦ C 25◦ C 48◦ C 25◦ C

k1 = 8.9 · 10−5 k1 = 5.1 · 10−5 k1 = 6.5 · 10−5 k1 = 2.5 · 10−9

k2 = 5.0 · 10−8 k2 = 3.5 · 10−8 k2 = 1.2 · 10−7 k2 = 6.0 · 10−10
k3 = 6.5 · 10−11 k3 = 1.2 · 10−10 k3 = 3.1 · 10−10 k3 = 4.5 · 10−14
k−3 = 1.9 · 10−2 k4 = 2.6 · 10−5 k4 = 1.3 · 10−4 k−3 = 4.5 · 10−6

5◦ C and 48◦ C is given by the empirical expressions

log k1 = −2.802 − 444/T (4.14)

log k2 = −0.16 − 2177/T (4.15)
log k3 = −8.86 − 317/T (5◦ C to 25◦ C) (4.16)
log k3 = −4.10 − 1737/T (25◦ C to 48◦ C), (4.17)

where T is temperature in K and ki have been transformed to new units.

Values for ki at 25◦ C and 48◦ C are shown in Table 4.1.
Three regimes of pH and PCO2 can be distinguished, where one of the
terms in equation (4.13) dominates both others:

1. At low pH, the rate is almost independent of PCO2 and approximately

proportional to the bulk fluid activity of H+ .
 71

2. At moderate to high pH and high PCO2 , the rate shows a linear depen-
dence on the bulk concentration of dissolved CO2 .

3. At high pH and in near absence of CO2 , the rate becomes independent

of both pH and PCO2 and is approximately constant.

These regions in the pH,PCO2 plane, as well as an area where the forward rate
depends significantly on more than one mechanism, are shown in Figure 4.3.
In addition, Rf was found to be dependent on the rate of stirring in regime 1,
but not in the other two regions. This led to the conclusion that the forward
rate is controlled by H+ transport at pH < 5, but essentially controlled by
surface reaction at high pH.
Data from the free drift experiments showed that the backward rate Rb
depends linearly on the activity product of Ca2+ and HCO−
3 (Rb was calcu-

lated as the difference between the forward rate Rf and the observed rate R)
and hence could be described by

Rb = k4 ACa2+ AHCO−3 , (4.18)

where k4 (dimensions mol cm−2 s−1 M−2 ) was found to be a function not
only of temperature, but also of PCO2 . From equations (4.13) and (4.18) it
follows that the net rate R is given by

R = Rf − Rb = k1 AH+ + k2 AH2 CO∗3 + k3 AH2 O − k4 ACa2+ AHCO−3 , (4.19)

which describes the whole range of experimental data in terms of individual

ion activities in the bulk fluid.
In order to derive a theoretical expression for k4 , Plummer et al. devel-
oped a reaction mechanism model for calcite dissolution. They concluded
 72

1
k1 AH+ = k2 AH2 CO∗3 + k3 AH2 O
k2 AH2 CO∗3 = k1 AH+ + k3 AH2 O
k3 AH2 O = k1 AH+ + k2 AH2 CO∗3

0.8

0.6 H2 CO∗3

PCO2
H+
(atm)

0.4

0.2

H2 O
0
3 4 5 6 7 8 9 10
pH

Figure 4.3: The rate of forward reaction is dominated by one of the terms
in equation (4.13) in the three regions labeled H+ , H2 CO∗3 , and H2 O. Along
the lines, one term just balances the other two, and in the area enclosed the
forward rate depends significantly on more than one mechanism. Adapted
from [Plu78].
 73

that three parallel reactions occur on the surface,

CaCO3 + H+ = Ca2+ + HCO−

3 (4.20)
CaCO3 + H2 CO3 = Ca2+ + 2HCO−
3 (4.21)
CaCO3 + H2 O = Ca2+ + HCO− −
3 + OH , (4.22)

and further assumed that reaction (4.20) is fast and therefore limited by
mass transport, while reactions (4.21) and (4.22) are much slower, and their
rates are chemically controlled. On basis of these assumptions, the surface
layer activities of H2 CO3 and H2 O are equal to the corresponding bulk fluid
activities, and the activities of all other species in the surface layer (including
H+ ) are determined by calcite and carbonate species equilibria. For instance,
(s) (s)
AH+ is given by the calcite saturation value at AH2 CO3 = AH2 CO3 and may
be calculated using an equilibrium model. A formal kinetic analysis of the
reactions (4.20)–(4.22) leads to a rate expression that is equivalent to the
empirical equation (4.19), and defines the rate constant k4 as

(s) (s)
k4 = k40 + k400 AHCO− + k4000 AOH− , (4.23)
3

(s)
where Ai are the surface layer activities of species i, and the new constants
k40 , k400 , and k4000 can be expressed in terms of the forward rate constants and
some equilibrium constants [Plu78, Plu79]. The agreement between the-
oretical and observed values of k4 was satisfactory over some range of CO2
partial pressures.
Plummer et al. had to make the crucial assumption that the partial
pressure of CO2 at the surface has the same value as in the bulk fluid (this is
(s)
equivalent to AH2 CO3 = AH2 CO3 since AH2 CO3 is proportional to AH2 CO∗3 which
in term depends linearly on PCO2 by Henry’s law)—an assumption that
 74

may not hold under certain conditions. In an attempt to verify their model
by comparison with several other studies in the literature, Plummer et al.
themselves found discrepancies of more than one order of magnitude, espe-
cially under conditions of low PCO2 [Plu79]. More recently, the model has
been criticized by Compton and co-workers as conveying only limited mech-
anistic information, because the empirical rate expression (4.19) is based on
bulk fluid activities instead of surface layer activities, and the experimental
conditions do not allow a well-defined, calculable and controllable transport
of the reactants to the interface [Com90a, Com90b]. Although the model
seems to agree quite well with the original set of data, its applicability to
data obtained under different experimental conditions is strictly limited, in-
dicating that there must be a significant dependence upon specific properties
of the experimental method applied.
A completely different experimental approach was used in the study by
Compton et al. which has already been cited [Com90a, Com90b]. They
proposed a new strategy for studying surface reactions and applied their
sophisticated method to calcite dissolution: a calcite crystal forms part of
one wall of a rectangular flow cell, through which a standard HCl solution
is pumped under laminar flow conditions. Immediately downstream of the
crystal, an appropriate detector electrode is placed in the wall, measuring
either directly the concentration of Ca2+ or the remaining level of the reactant
H+ as a function of flow rate. This technique has several advantages, for
instance

• defined surface area and topography of the crystal,

• very high rates of mass transport possible,

 75

• transport of the reactants to the interface and the products to the

detection system calculable and controllable because of defined flow,

• measurement of surface concentrations instead of bulk values,

• steady-state flow system, hence no chemostatic control necessary.

Two different pH regimes were examined: At pH < 4, the data was described
best by first-order heterogenous kinetics according to the rate law

R = k1 [H+ ](s) , (4.24)

where R is the dissolution rate with units mol cm−2 s−1 , [H+ ](s) denotes the
surface concentration of H+ in M, and the rate constant was found to be
k1 = (4.3 ± 1.5) · 10−5 mol cm−2 s−1 M−1 . Compton et al. were able to
avoid transport control during all their experiments, down to very low values
of pH. Under high pH conditions, i.e., at pH > 7, and in the near absence of
CO2 , they chose the following rate law as that which best fits the observed
data
R = k − k 0 [Ca2+ ](s) [CO2− (s)
3 ] , (4.25)

where R has the same units as above and k 0 = k/Ksp with the solubility
product of calcium carbonate Ksp . Again, [Ca2+ ](s) and [CO2−
3 ]
(s)
are sur-
face concentrations, and the zero order rate constant k is dependent upon
surface morphology (especially roughness). The value found for a surface of
Iceland spar, polished with a succession of grit sizes down to 0.25 µm, was
k = 9.5 · 10−11 mol cm−2 s−1 . It can be shown that in the absence of CO2
and at pH > 7 Plummer’s rate expression (4.19) reduces to the same ki-
netic form as (4.25) under the assumption that dissolution ceases when the
calcite solubility product is reached [Com90b]. However, equation (4.25)
 76

refers to surface concentrations rather than bulk phase quantities, and one
may conclude that it is more physically significant, i.e., closer related to the
“real” mechanism of surface reaction. On the other hand, model equations
like (4.19) are of great practical value, since in most practical cases the sur-
face concentrations can neither be measured nor calculated, and it is not
possible to eliminate the influence of transport processes.
Still another problem arises from the effect of impurities: it is known
that trace amounts of inhibitors (such as phosphate) can cause large changes
in rate without significantly altering the thermodynamic properties of the
system [Ber74, Plu79]. Unfortunately, there is no simple way of predicting
the total effect that an impurity or combination of impurities may have on the
rate of reaction, since theoretical models are not available to date. This shows
the difficulty in developing a useful general model of calcite dissolution—
very often a specific empirical relation for the system in question might be
preferred.

Comparative Study of Carbonate Dissolution

Another interesting publication that shall be discussed here is the compara-

tive study by Chou et al., who investigated the dissolution of various carbon-
ates (including calcite, magnesite, and dolomite) in standard HCl solutions
at 25◦ C [Cho89]. They used a continuous fluidized bed reactor and samples
of relatively coarse particle size (0.3–0.4 mm for calcite, 0.05–0.1 mm for
magnesite, and 0.1–0.2 mm for dolomite). Surface area and composition of
the solid phase remained approximately unchanged during each dissolution
experiment. The reactor was operated at a constant flow rate of standard
input solution, and the composition of the output solution quickly reached a
 77

steady state. The output solution was collected and analyzed for dissolved
species, and also the pH was measured. This allowed to determine the disso-
lution rate by multiplying the concentration of cations with the flow rate of
the output solution and dividing by the estimated surface area of the solid.
The partial pressure of CO2 could not be fixed at some arbitrary value like
in the experiments above, but it could be calculated from the measured data
(bulk fluid concentrations and pH). The experiments showed that both calcite
and magnesite have the same general rate dependence on pH, but the rate
of dissolution of magnesite is about three to four orders of magnitude lower
than that of calcite, see Figure 4.4. The authors proposed a reaction model
similar to that of Plummer et al. to describe their data on the dissolution
of one-component carbonates,

MCO3 + H+ = M2+ + HCO−

3 (4.26)
MCO3 + H2 CO∗3 = M2+ + 2HCO−
3 (4.27)
MCO3 = M2+ + CO2−
3 , (4.28)

where M represents either Ca or Mg. From these three reaction steps they
derived the rate expression

R = k1 AH+ + k2 AH2 CO∗3 + k3 − k−3 AM2+ ACO2−

3
, (4.29)

where the contribution of the first two reactions to the backward rate has
already been omitted since it is not significant. The rate constants ki were
determined from the experimental results; their numerical values are given
in Table 4.1 for both calcite and magnesite.
For dolomite as a two-component carbonate, Chou et al. found the fol-
lowing empirical rate equation,

Rf = k1 AnH+ + k2 AnH2 CO∗3 + k3 , (4.30)

 78

−7
calcite ×
magnesite
dolomite 3
−8
×
3
××
×
−9 3
×
×
3
××
−10 ××
×× ×
3 ×
××
×
log R−11 ×
33
3 ××
3
33 3
3 33
33
−12 3

−13

−14

−15
2 4 6 8 10
pH

Figure 4.4: Dissolution rates vs. pH for calcite, dolomite, and magnesite;
experimental data obtained with continuous fluidized bed reactor at 25◦ C.
Adapted from [Cho89].
 79

Table 4.2: Summary of empirical rate constants for the dissolution of

dolomite; the corresponding expressions for forward and backward rate are
given by equations (4.30) and (4.31), respectively. Forward rate constants
k1 , k2 , k3 have dimensions of mol cm−2 s−1 M−n , backward rate constant k4
has dimensions of mol cm−2 s−1 M−1 .

Dolomite

Ca0.498 Mg0.502 CO3 Ca0.509 Mg0.489 Fe0.002 CO3

[Cho89] [Bus82]

25◦ C 25◦ C 55◦ C

n = 0.75 n = 0.5 n = 0.6

k1 = 2.6 · 10−7 k1 = 4.1 · 10−8 k1 = 2.5 · 10−7

k2 = 1.0 · 10−8 k2 = 1.3 · 10−9 k2 = 2.0 · 10−9
k3 = 2.2 · 10−12 k3 = 5.0 · 10−9 k3 = 2.5 · 10−11
k4 = 5.6 · 10−8 k4 = 2.0 · 10−7

which only describes the forward rate. After a very short initial stage of
dissolution, calcium and magnesium are released stoichiometrically, i.e., in
a 1:1 mole ratio.1 An important difference in behaviour compared to the
other carbonates is that the reaction order n is a fractional number less than
one. The best fit of experimental data was obtained for n = 0.75 and the rate
constants given in Table 4.2. There is not yet a really convincing explanation
for the fractional order of reaction—Chou et al. assume that the formation
1
Since initially Ca is released faster than Mg, the dolomite surface acquires a Ca:Mg
mole ratio that is smaller than that of the bulk solid. This affects only a few atomic layers
on the surface [Bus82].
 80

of positively charged surface complexes due to H+ uptake causes electrostatic

interactions which in term may lead to the fact that the degree of protonation
of the surface increases only slightly with decreasing pH, i.e., the observed
reaction order is less than one. As shown in Figure 4.4, the dissolution rate
of dolomite is about one order of magnitude lower than the calcite rate, but
still two orders of magnitude higher than the magnesite rate.
Busenberg and Plummer used the same equation (4.30) to describe
their data on dolomite dissolution, but they found a reaction order of n = 0.5
at temperatures below 45◦ C which increases at higher temperatures [Bus82].
They also identified one backward reaction with empirical rate

Rb = k4 AHCO−3 , (4.31)

and they were able to show that this backward rate is significant even far
from equilibrium. This is one important reason for the very low net rate
of dissolution, particularly at pH > 6 as H+ attack becomes less dominant.
The dissolution reaction essentially stops quite far from equilibrium, and
the amount of time necessary to reach “true” equilibrium might be so large,
that this will never be observed in laboratory or nature. Rate constants for
the temperatures 25◦ C and 55◦ C can be found in Table 4.2, but it must
be pointed out that these constants are highly dependent on the source of
dolomite. Busenberg and Plummer studied eight different dolomite speci-
mens of both sedimentary and hydrothermal origin and found that the latter
dissolved significantly slower at least in the low temperature regime. The
rate constants given here refer to a microcrystalline sedimentary dolomite
from Wisconsin. There is also a significant increase in dissolution rate with
increasing temperature: dolomite dissolves five to ten times faster at 45–60◦ C
than at 25◦ C.
 81

Acid Neutralization with Limestone

The use of crushed or powdered limestone to neutralize acidic wastes like coal
mine drainage and pickling liquors has been investigated by several workers
since limestone is the least expensive neutralizing agent available [Bar76,
Hoa45, Par65, Pea75]. There have been some difficulties in the practical
application of limestone neutralization, because

• the reaction rates of limestone are relatively low,

• the active limestone surface may be fouled by oil, grease, or sulfate and
phosphate precipitates, thereby disrupting dissolution,

• the equilibrium pH is too low to precipitate certain metal ions which

have to be removed from the waste water.

Out of these reasons, lime is more commonly used for the treatment of acidic
wastes. Nevertheless, limestone neutralization has been successfully applied,
especially in cases where the waste water does not contain too much sulfuric
acid. Parsons described design and installation of an upflow limestone
neutralization plant; after some simple modifications like increased flowrates
in the limestone bed and removal of fouling substances from the input stream,
satisfactory operation of the system was achieved [Par65].
Hoak et al. proposed a combined limestone-lime treatment of spent pick-
ling liquors: pulverized high-calcium limestone is used to raise the pH to
about pH 6 and precipitate part of the metals, then the treatment is com-
pleted with lime. Dolomitic limestones are practically useless for this purpose
due to their very low reactivity. As a rule of thumb, the rate of reaction with
pickle liquor was found to be inversely proportional to the percentage of mag-
 82

nesium carbonate in the limestone, e.g., a limestone with 5 wt% MgCO3 has
only about 1/5 of the normal reaction rate [Hoa45].
Barton and Vatanatham presented a simple model for the kinetics
of limestone neutralization [Bar76]. They examined the dissolution rate of
calcite particles in 0.01 N sulfuric acid; the pH was allowed to drift freely.
Particles with almost uniform initial size were used, and a shrinking particle
model was applied to describe the change in surface area. All the reactions
in the bulk phase were assumed to be at equilibrium, and for the dissolution
rate the following expression—valid in the range between pH 2 and 6—was
found which is first order with respect to H+ concentration2
dN
− = Ak ([H+ ] − [H+ ]eq )
dt
1/3
6kM N0
= N 2/3 ([H+ ] − [H+ ]eq ) . (4.32)
ρD0
In this equation,

N = moles of solid CaCO3 at time t,

A = surface area of CaCO3 (s) at time t,
k = rate constant,
[H ] = molar H+ concentration in bulk solution at time t,
+

[H+ ]eq = molar H+ concentration in bulk solution at equilibrium,

M = molecular weight of CaCO3 (s),
N0 = moles of initial CaCO3 (s),
ρ = density of CaCO3 (s), and
D0 = initial particle diameter.
From their data at different temperatures, the authors concluded that the
dissolution rate was controlled by H+ diffusion rather than surface reaction
2
The model equation is misprinted in the original paper.
 83

in the given pH range. Since their dynamic model is not valid at pH > 6, it
cannot be applied to the production of CMA at near-neutral pH. The same
is true for the model proposed by Pearson et al., who examined the use
of crushed limestone for the treatment of coal mine drainage and derived
practical rules for the design of such neutralization processes [Pea75].
The only study that investigated the dissolution of calcite and dolomite
in acetic acid was done by Suarez and Wood in the context of soil analysis
[Sua84]. They measured the buildup of CO2 -pressure in a sealed reaction
vessel in order to determine calcite surface area and content in soil sam-
ples (surface area and total calcite were determined from maximum slope
of pressure vs. time and final pressure, respectively). A buffered solution
containing sodium acetate and acetic acid was used in these experiments;
therefore, the pH remained nearly constant at about pH 4.3 during the reac-
tion. Under these conditions, calcite dissolves approximately 70 times faster
than dolomite. This is but the only useful result since the publication con-
tains neither absolute rate data nor a theoretical model.

4.2.2 Oxide Minerals: CaO, MgO

Kinetics and mechanisms of the dissolution of metal oxides in aqueous me-
dia have been studied for several reasons—the reactivity of oxide surfaces in
solution is an interesting field of study and also has some practical implica-
tions, e.g. for the etching of metals, the extraction of ores, and the inhibition
of corrosion. Although quite a lot has been published in this field since the
1960s, it is difficult to find rate data or kinetic models for the higher pH
regime above pH 4. Therefore, it will be necessary to rely on rather crude
extrapolations, or to collect additional data.
 84

An excellent review of oxide dissolution is provided by Segall et al.

[Seg88]. They use an approach based on solid state science, thus emphasiz-
ing the role of the solid and including solution properties as necessary. There
are three basic types of oxides,

• ionic oxides like CaO, MgO

• semiconducting oxides like CoO (p-type), ZnO (n-type)

• covalent oxides like Al2 O3 , SiO2 ,

with different mechanisms of dissolution. Other important factors deter-

mining reactivity are pretreatment (effect on surface and bulk defect sites),
surface structure, and surface alteration in solution. Only the ionic oxides—
particularly CaO and MgO—shall be discussed in further detail now. Com-
pared to the other types, these oxides dissolve very fast: the dissolution
rates at pH 1 and 30◦ C (after the initial stage of acidic solution attack)
are 1.2 · 10−7 and 1.2 · 10−8 mol cm−2 s−1 for CaO and MgO, respectively
[Seg88]. It is believed that the dissolution process comprises the following
steps [Seg88]:

1. Restructuring of ionic surface upon initial contact with aqueous solu-

tion (produces a roughened surface structure with many defect sites).

2. Alteration of surface charge.

3. M2+ transfer to solution.

4. H+ transfer to O2− ion.

5. OH− (or H2 O) transfer to solution.

 85

Steps 1 and 2 are responsible for a significant alteration of reactivity that

occurs at the very beginning of the dissolution process (less than 5% of the
solid dissolved). After completion of this initial stage, the surface structure
remains nearly unchanged until about 75% of the solid are dissolved, and dis-
solution proceeds according to steps 3–5 only (“advanced stage”). However,
it should be pointed out that the reaction steps 3–5 given above are not the
only ones that have been proposed: for instance, Vermilyea assumed that
MgO first reacts with water to form a Mg(OH)2 layer on the surface and the
rate is controlled by dissolution of the Mg(OH)2 via steps 3 and 5 [Ver69].
MacDonald et al. supported the same view in their study [Mac71].
Depending on pH and saturation conditions, the observed dissolution rate
is controlled by either transport, surface reaction, or both. Three different ki-
netic regimes can be distinguished, which, in practice, often overlap [Seg88].

Regime I: At high product concentration approaching saturation, it is as-

sumed that the surface layer is in equilibrium with the solid, and the
rate is controlled by transport of either M2+ or OH− from the surface
layer to the bulk solution. The kinetics can be described by Nernst
theory.

Regime II: Transport control is found also in the extreme cases where the
dissolution rate is very high, or the concentration of a reactant is very
low. The first case occurs at rates greater than 5 · 10−8 mol cm−2 s−1 ,
e.g., with CaO dissolving at pH 1–2, and the diffusion of dissolved
species away from the surface is rate limiting. The second case can be
observed with ionic oxides at pH > 6; then the rate is controlled by
arrival of H+ .
 86

Regime III: In the intermediate pH regime (pH 2–6), the dissolution rate
is mainly reaction controlled, and it typically ranges between 10−9 and
10−12 mol cm−2 s−1 . The activation energy in this regime is usually
greater than 65 kJ mol−1 (compared to only 15–20 kJ mol−1 for dif-
fusion control). However, the results of MacDonald et al., who in-
vestigated MgO dissolution in dilute sulfuric acid at pH 3–5 using a
rotating disc technique, demonstrate the existence of mixed kinetic
control, with transport control playing a more and more dominant role
at higher temperatures [Mac71].

These different kinetic regimes combined with the necessary distinction be-
tween initial and advanced stages of dissolution clearly complicated the in-
terpretation of experimental results, and, in fact, the available literature is
not very conclusive. There are several conflicting results reported for the dis-
solution of MgO, and it is hard to find anything at all about CaO dissolution
kinetics, since in this case experiments are further complicated by the very
high reactivity of CaO (which is about one order of magnitude higher than
that of MgO).
Briefly, the main results can be stated as follows: During the initial stage
of dissolution, both CaO and MgO show a significant rise in rate despite
decreasing pH under free drift conditions; this initial rise is steeper for CaO
than for MgO. The effect has been attributed to surface roughening due to
initial solution attack [Seg88, Seg78], but is still not very well understood
and shall not be discussed in further detail here. For the advanced dissolu-
tion regime of MgO, most authors found a linear relationship between the
logarithm of rate R and pH, but both the slopes (ranging from −0.5 to −1)
and the absolute values differ greatly [Seg88]. Figure 4.5 gives a good im-
 87

pression of the variation in published rates for MgO at 20–25◦ C for pH 2–6.
The advanced kinetics of CaO are reported to show similar features to MgO
but at rates about ten times greater [Seg88].
At the moment there seems to be no way to explain these apparent exper-
imental contradictions. Some of the difficulties may arise from uncertainties
in estimating the effective surface area (by gas absorption or geometrical
considerations), changes in particle size distribution during dissolution (es-
pecially in the case of powders), and the use of background electrolytes in
some of the experiments. Vermilyea found that certain substances act as
inhibitors (e.g. periodate) or accelerators (e.g. phosphate, acetate3 ) in the dis-
solution of magnesium hydroxide [Ver69]. Another possible effect that may
occur in technical applications of calcium hydroxide as neutralizing agent is
the precipitation of calcium sulfate (gypsum) on the surface of the calcium
hydroxide particles, if sulfuric acid is present in the solution. Even a very
thin film of gypsum has been shown to significantly decrease the dissolution
rate [Bog85].
From these results one must conclude that it is even more difficult to
develop a kinetic model for the dissolution of CaO and MgO than it was
for the carbonates. As already stated at the beginning of this section, it is
therefore necessary to rely on rather crude estimates for the dissolution rates
in the interesting pH range, and to verify these estimates through subsequent
experiments under “real” process conditions. Reasonable extrapolated values
for the rates at pH 6 would be 1 · 10−9 and 1 · 10−10 mol cm−2 s−1 for CaO
and MgO, respectively. These values may be compared to similar estimates
3
This is particularly interesting, since both acetate and phosphate would be present in
a CMA fermentation.
 88

−7
MgO smoke crystals at 25◦ C
sized, sieved powder at 20◦ C
rotating disk at 20◦ C
rotating disk at 20◦ C

−8

−9

log R

−10

−11

−12
1 2 3 4 5 6 7
pH

Figure 4.5: Summary of published dissolution rates vs. pH for MgO, in units
mol cm−2 s−1 . The data was collected from various sources by Segall et
al.; diagram adapted from [Seg88].
 89

Table 4.3: Summary of dissolution rates at 20–25◦ C for carbonate minerals

and oxide minerals. The carbonate rates are based on the experimental data
of [Cho89], the values for CaO and MgO were chosen according to the data
presented in [Seg88], assuming a slope of −0.6 for log R vs. pH between
pH 5 and 7. Rates are given in units of mol cm−2 s−1 .

Mineral Dissolution Rate

pH 5 pH 6 pH 7

calcite 2 · 10−9 2 · 10−10 8 · 10−11

dolomite 9 · 10−11 2 · 10−11 1 · 10−11
magnesite 2 · 10−13 1 · 10−13 6 · 10−14
CaO 4 · 10−9 1 · 10−9 3 · 10−10
MgO 4 · 10−10 1 · 10−10 3 · 10−11

for the carbonate minerals. Table 4.3 gives a summary of dissolution rates at
pH 5, 6, and 7 for all the minerals discussed. The rates cover a range of almost
four orders of magnitude from the slowest dissolving mineral (magnesite) to
the fastest (CaO). It might be interesting to observe that the rates of calcite
and MgO are relatively high and of the same order of magnitude. Since such
a mixture can be easily prepared from dolomite by selective calcination, it
could have some potential as neutralizing agent in CMA production.

4.3 The Dynamic Model and Its Limitations

At this point, it is time to “put everything together”—most of the results
obtained so far will be used to develop a dynamic model of CMA production
 90

in a continuous fermentor. A brief discussion of the model and its struc-

ture forms the main part of this section. Simplifying assumptions had to be
made where certain aspects of the process were unknown or quantitative de-
scriptions were not available. Other significant uncertainties were introduced
by the underlying models used to describe acetic acid production, dissolu-
tion kinetics, and liquid phase equilibrium speciation. Altogether, this led
to the conclusion that quantitative results obtained from this model must
be expected to carry only very limited information about the real process.
Therefore, the original plan to examine process dynamics and control by
simulation has been abandoned in favour of a more practical approach to
be presented in the next chapter. Nevertheless, the theoretical model allows
to gain some insight into the dynamic structure of the process, it helps to
identify important factors determining the process behaviour, it puts the pre-
viously obtained results into a new perspective, and it allows to draw some
qualitative conclusions for process design and control.

4.3.1 Outline of the Process Model

Overall System Structure

The continuous reactor containing fermentation broth (with dissolved sub-

strate and product), cells, and undissolved lime or limestone solids is modeled
as an open system with lumped parameters. Four phases shall be distin-
guished as shown in Figure 4.6:

1. The liquid phase comprises water and all dissolved species, but not
the suspended cells and particles. It may be characterized by the vol-
ume V and the total concentrations of dissolved substrate, calcium,
 91

gas discharge
µ

gas phase
(PCO2 )
6
CO2

liquid phase product stream

feed stream - (V , CST , - (F , CST ,
(F f , CSf ) T T T T T T
CCa , CMg , CAc ) CCa , CMg , CAc )

µ I
HAc S Ca, Mg, CO2
ª R
cells solids phase
(µ, Nx ) (Ni , Āi )
µ
R R
cell loss solids feed solids discharge

Figure 4.6: Overall structure of the model system with definition of phases,
streams, and relevant dynamic variables.
 92

magnesium, and acetate, which are denoted by CST , CCa

T T
, CMg T
, and CAc ,
respectively. A feed stream with defined flow rate F f and substrate con-
centration CSf (all other concentrations zero) enters the liquid phase,
T
and a product stream characterized by F , CST , CCa T
, CMg T
, and CAc is
withdrawn. For our purposes, it shall be assumed that at any time
F = F f , i.e., the liquid volume V is a constant. The system is well
stirred, so that all concentrations are functions of time only.

2. The gas phase contains water and carbon dioxide. By maintaining a

constant pressure, the partial pressure of carbon dioxide, PCO2 , can be
fixed (water at fermentation temperature of 60◦ C has a vapor pressure
of about 0.2 bar). The exsolution of CO2 from the liquid phase is
assumed to be fast since the reactor is highly turbulent. Therefore,
H2 CO∗3 in the liquid phase is in equilibrium with the gas phase at fixed
PCO2 . Evolved gases are discharged. Bubbling with CO2 should be
avoided if possible, because a high PCO2 decreases carbonate dissolution
rates and may even transform hydroxides into carbonates. Instead, an
inert purge gas like nitrogen may be used to further facilitate exsolution
of CO2 .

3. The cells transform the substrate S, which may be glucose or xylose,

into acetic acid. For simplification, the substrate is assumed to be pure
glucose, and other components needed by the organism are neglected.4
The biomass is characterized by its amount Nx and its growth rate µ;
simple model equations are used for acetic acid production rate and
substrate uptake rate. Cell loss depends on the rate of cell lysis as well
4
In reality, at least a source of nitrogen, sulfur, and phosphorus must be provided in
addition to the substrate for growth.
 93

as on process characteristics like dilution rate D = F/V , cell immobi-

lization, or cell recycling. In a continuous fermentation, it is possible
to establish a steady state with constant Nx (growth equals cell loss).

4. The solids phase consists of the undissolved lime or limestone particles.

Three different cases will be considered,

• the use of dolomite CaMg(CO3 )2 ,

• the use of selectively calcined dolomite CaCO3 · MgO, and

• the use of type S dolime Ca(OH)2 · Mg(OH)2 .

Dissolution kinetics are described by the available empirical rate ex-

pressions, and in case of the carbonates also precipitation may occur
(“negative” dissolution rate, modeled by the same equations). How-
ever, the direct transformation of solid hydroxides into carbonates by
CO2 could not be described dynamically and is neglected. The solids
phase is characterized by the total amount Ni and the effective surface
area Āi for each of the solid species present. Fresh particles are fed to
the system to make up for the amount dissolved, and there may be also
a discharge of unreacted solids. The overall neutralization rate may be
controlled by changing the available effective surface area.

Although the pH of the liquid phase is an important variable which greatly

influences the dissolution rates and also affects the organism, it has not been
listed above as a relevant dynamic variable, because it will be shown that
T T T
the pH at each instance of time is determined by the variables CCa , CMg , CAc ,
and PCO2 . The same is true for all other individual species concentrations in
the liquid phase. These dependencies shall be discussed next.
 94

Liquid Phase Equilibrium Speciation

Homogenous ionic reactions in solution are very fast compared to dissolution

reactions and microbial product formation. Hence it can be assumed that
at each point of time all homogenous reactions in the liquid phase are at
equilibrium. In addition, dissolved CO2 shall be in equilibrium with the gas
phase as already stated above. No reaction shall occur between the dissolved
substrate (glucose) and any of the other species in solution.5 The presence of
glucose therefore does not have any influence on the ionic equilibrium and can
simply be neglected at this point. In order to find the liquid phase equilibrium
speciation, only a few modifications have to be made to the approach already
used in Section 3.3 for the carbonate system (see Table 3.3):

• The reactions involving solid species are omitted because solids are not
considered to be part of the equilibrium model any more (there is no
equilibrium between liquid and solid phase). This reduces the number
of species as well as the number of independent reactions by two.

• Although the number of principal components remains the same, now

other species must be chosen for CaCO3 (s) and MgCO3 (s), for instance
Ca2+ and Mg2+ . Also Ac− may be replaced by HAc, resulting in the
list of components H2 O, H+ , HAc, Ca2+ , Mg2+ , and CO2 (g).

• There is no longer a fixed “recipe”—instead, the total concentrations

of dissolved calcium, magnesium, and acetate are given as functions of
T T T
time (CCa , CMg , and CAc , respectively). Nevertheless, the equilibrium
5
This is a potentially “dangerous” simplification which has not been checked thoroughly
because of time constraints. In fact, it is not unlikely that certain reactions may occur.
 95

composition can be calculated at each point of time, in a similar manner

as it was done for the titration discussed in Chapter 3.

The species and their stoichiometric formulae in terms of the components

can be found in Table 4.4 along with the corresponding log K values for the
mass laws; the tableau was calculated based on the following independent
reactions6
H2 O = H+ + OH− , log K = −14.0
HAc = H+ + Ac− , log K = −4.76
CO2 (g) + H2 O = H2 CO∗3 , log K = −1.5
H2 CO∗3 = H+ + HCO−
3, log K = −6.3
HCO−
3 = H+ + CO2−
3 , log K = −10.3
CaOH+ = Ca2+ + OH− , log K = −1.15
CaAc+ = Ca2+ + Ac− , log K = −1.2 (4.33)
CaHCO+
3 = Ca2+ + HCO−
3, log K = −1.26
CaCO3 = Ca2+ + CO2−
3 , log K = −3.2
MgOH+ = Mg2+ + OH− , log K = −2.6
MgAc+ = Mg2+ + Ac− , log K = −1.3
MgHCO+
3 = Mg2+ + HCO−
3, log K = −1.01
MgCO3 = Mg2+ + CO2−
3 , log K = −3.4.
Only log K at zero ionic strength was included in the tableau, since activity
coefficients will be used in the mass laws instead of corrected equilibrium
constants (see equation (3.11)). Each non-component row in the upper part
of the table corresponds to a mass law, hence thirteen mass law equations
can be written

γOH− [OH− ] = 10−14.0 γH−1 + −1

+ [H ] (4.34)
6
CaCO3 and MgCO3 stand for the ion pairs in solution, not the solids.
 96

Table 4.4: Tableau-representation of the equilibrium model used for the liquid
phase. Equilibrium constants at 25◦ C and zero ionic strength.

Species H+ HAc Ca2+ Mg2+ CO2 (g) log K

H+ 1
OH− −1 −14.0
HAc 1
Ac− −1 1 −4.76
CO2 (g) 1
H2 CO∗3 1 −1.5
HCO− 3 −1 1 −7.8
CO2−
3 −2 1 −18.1
Ca2+ 1
CaOH+ −1 1 −12.85
CaAc+ −1 1 1 −3.56
CaHCO+ 3 −1 1 1 −6.54
CaCO3 −2 1 1 −14.9
Mg2+ 1
MgOH+ −1 1 −11.4
MgAc+ −1 1 1 −3.46
MgHCO+ 3 −1 1 1 −6.79
MgCO3 −2 1 1 −14.7
T
HAc 1 CAc
T
CaCO3 (s)/Ca(OH)2 (s) −2 1 1/0 CCa
T
MgCO3 (s)/Mg(OH)2 (s) −2 1 1/0 CMg
CO2 (g) 1 ?
 97

γAc− [Ac− ] = 10−4.76 γH−1 + −1

+ [H ] γHAc [HAc] (4.35)
γH2 CO∗3 [H2 CO∗3 ] = 10−1.5 PCO2 (4.36)
γHCO−3 [HCO−
3 ] = 10
−7.8 −1
γH+ [H+ ]−1 PCO2 (4.37)
γCO2−
3
[CO2−
3 ] = 10
−18.1 −2
γH+ [H+ ]−2 PCO2 (4.38)
γCaOH+ [CaOH+ ] = 10−12.85 γH−1 + −1
+ [H ] γCa2+ [Ca2+ ] (4.39)
γCaAc+ [CaAc+ ] = 10−3.56 γH−1 + −1
+ [H ] γHAc [HAc]γCa2+ [Ca2+ ] (4.40)
γCaHCO+3 [CaHCO+
3 ] = 10
−6.54 −1
γH+ [H+ ]−1 γCa2+ [Ca2+ ]PCO2 (4.41)
γCaCO3 [CaCO3 ] = 10−14.9 γH−2 + −2
+ [H ] γCa2+ [Ca2+ ]PCO2 (4.42)
γMgOH+ [MgOH+ ] = 10−11.4 γH−1 + −1
+ [H ] γMg2+ [Mg2+ ] (4.43)
γMgAc+ [MgAc+ ] = 10−3.46 γH−1 + −1
+ [H ] γHAc [HAc]γMg2+ [Mg2+ ] (4.44)
γMgHCO+3 [MgHCO+
3 ] = 10
−6.79 −1
γH+ [H+ ]−1 γMg2+ [Mg2+ ]PCO2 (4.45)
γMgCO3 [MgCO3 ] = 10−14.7 γH−2 + −2
+ [H ] γMg2+ [Mg2+ ]PCO2 . (4.46)

The activity coefficients γi can be estimated from a semiempirical expression

like Davies’ formula

I 1/2
à !
ln γi = −Azi2 − bI , (4.47)
1 + I 1/2

where the ionic strength I is defined by

1X 2
I= z Ci , (4.48)
2 i i

in the way it was discussed in Section 4.1 (see equations (3.4) and (3.6) for
further explanation). The pH is defined as the negative logarithm of H+ -
activity, i.e., it can be calculated using

pH = − log γH+ [H+ ] (4.49)

 98

from concentration and activity coefficient of H+ .

The information given in the lower part of the tableau shows how the equi-
librium model is linked to the total concentrations of acetic acid (or acetate),
dissolved calcium (sum of all calcium species), and dissolved magnesium (sum
of all magnesium species) which are time-dependent functions denoted by
T T T
CAc , CCa , and CMg , respectively. In the case of acetic acid this is straightfor-
ward, for calcium and magnesium it is a little bit more complicated—here
one has to consider that all the calcium and magnesium present in the liquid
phase stems from the dissolution of either carbonates or hydroxides or both.
Therefore, solid species had to be included at this point, although they are
not considered part of the equilibrium system and consequently were omitted
from the top of the tableau. This is just a matter of accounting: one may
think of the system as being prepared by completely dissolving the solids
with an excess of acetic acid. Fortunately, carbonates as well as hydrox-
ides have the stoichiometric coefficient −2 in the H+ -column of the tableau,
hence the TOT H balance can be written without knowledge about the ori-
gin of the dissolved material (carbonate, hydroxide, or both). Only in the
CO2 (g)-column different coefficients occur, but since the TOT CO2 balance
is of no interest (PCO2 in the gas phase is assumed to be fixed anyway), this
does not cause any difficulties. The same reasoning can be applied to the
double carbonate dolomite CaMg(CO3 )2 , because it may be simply consid-
ered as CaCO3 · MgCO3 for this purpose. Hence from the first four columns
of the tableau, the mole balances for the components H+ , HAc, Ca2+ , and
Mg2+ are

2−
TOT H = [H+ ] − [OH− ] − [Ac− ] − [HCO−
3 ] − 2[CO3 ]

− [CaOH+ ] − [CaAc+ ] − [CaHCO+

3 ] − 2[CaCO3 ]
 99

− [MgOH+ ] − [MgAc+ ] − [MgHCO+

3 ] − 2[MgCO3 ]
T T
= −2(CCa + CMg ) (4.50)
TOT HAc = [HAc] + [Ac− ] + [CaAc+ ] + [MgAc+ ]
T
= CAc (4.51)
TOT Ca = [Ca2+ ] + [CaOH+ ] + [CaAc+ ] + [CaHCO+
3 ] + [CaCO3 ]
T
= CCa (4.52)
TOT Mg = [Mg2+ ] + [MgOH+ ] + [MgAc+ ] + [MgHCO+
3 ] + [MgCO3 ]
T
= CMg . (4.53)
T T T
Once CAc , CCa , and CMg are known, the concentrations of all seventeen
individual species—excluding the “special” species H2 O and CO2 (g)—can
be calculated by solving the system of seventeen algebraic equations given
by (4.34)–(4.46) and (4.50)–(4.53) in conjunction with the expressions for
activity coefficients (4.47) and ionic strength (4.48).

Acetic Acid Production and Substrate Consumption

T
In order to derive equations for CAc and CST the empirical expressions for
the rates of acetic acid production (2.10) and substrate uptake (2.13) by C.
thermoaceticum from Section 2.3.1 are used. Simple mole balances for acetic
acid and glucose in the liquid phase lead to the ordinary differential equations
T
dCAc 1 ³ T
´
= (αP µ + βP )Nx − F CAc (4.54)
dt V
and
dCST 1 ³ ´
= F CSf − (αS µ + βS )Nx − F CST , (4.55)
dt V
where the total amount of biomass Nx in the cell phase is described by
dNx
= µNx − (loss)x , (4.56)
dt
 100

and µ, αP , βP , αS , and βS are assumed to be known parameters. This as-

sumption might be justified under steady state conditions, but certainly not
during start up, since all these parameters are expected to be functions of
pH, CST , CAc
T
, and perhaps some other variables. A quantitative description
of these dependencies seems remote, it would require a lot more of experi-
mental and theoretical work than has been done so far. Similarly, the loss
of cells is difficult to quantify; in a more comprehensive modeling approach,
biological factors (e.g., cell lysis) as well as certain process characteristics
(e.g., effects of dilution rate, cell recycling, cell immobilization) would have
to be considered. Therefore, the use of the above equations is strictly limited
to cases where only minor changes occur in the dynamic variables, i.e., essen-
tially to steady state operation. Appropriate values for the parameters would
have to be determined experimentally under steady state process conditions.
T
The initial values of CST , CAc , and Nx have no influence on the steady state
and would only have to be considered, if equations (4.54)–(4.56) could also
describe the dynamics of start up.

Dissolution of Solids
T T
Differential equations for the remaining dynamic variables CCa and CMg can
be found using the results on dissolution kinetics from the beginning of this
chapter. First, dolomite shall be considered as neutralizing agent—the cor-
responding liquid phase mole balances for calcium and magnesium yield
T
dCCa 1 ³ T
´
= Ādol Rdol − F CCa (4.57)
dt V
T
dCMg 1 ³ T
´
= Ādol Rdol − F CMg , (4.58)
dt V
 101

where Ādol denotes the effective surface area of dolomite and Rdol is obtained
by combining equations (4.30) and (4.31) to

Rdol = k1 γHn+ [H+ ]n + k2 γHn2 CO∗3 [H2 CO∗3 ]n + k3 − k4 γHCO−3 [HCO−

3 ]. (4.59)

Since calcium and magnesium are released at the same rate Rdol after a very
short initial stage of dissolution, one should expect that

T T
CCa = CMg (4.60)

is automatically satisfied in this case, i.e., there should be no difficulty in

maintaining the desired 1:1 molar ratio of Ca:Mg. The amount of dolomite
present in the solids phase can be described by

dNdol
= (feed)dol − Ādol Rdol − (discharge)dol . (4.61)
dt

Much more difficult to quantify is the effective surface area Ādol since it
depends on the particle size distribution which in the most general case
would have to be modeled using a population balance approach. The size of
each individual particle depends on its initial size, on the time it entered the
system, and on the conditions it met during its lifetime, i.e., on its entire
“history”. A simple shrinking sphere model like the one used by Barton
and Vatanatham [Bar76] is certainly not adequate because it is based on
the assumption that all particles start dissolving at the same time, and there
is no continuous feed of “fresh” particles. Population balance models have
been applied to processes like crystallization, but they involve rather complex
mathematics, hence using such a sophisticated approach is not justified in our
case considering the other serious limitations of our model. However, under
steady state conditions, the effective surface area as well as the amount of
 102

dolomite are constant, and they are related by

2/3
Ādol = cNdol , (4.62)

where c is a constant depending on the steady state particle size distribution.

If selectively calcined dolomite or type S dolime are used for neutraliza-
tion, the governing equations are
T
dCCa 1 ³ T
´
= ĀCaCO3 RCaCO3 + ĀCa(OH)2 RCa(OH)2 − F CCa (4.63)
dt V
T
dCMg 1 ³ T
´
= ĀMgCO3 RMgCO3 + ĀMg(OH)2 RMg(OH)2 − F CMg , (4.64)
dt V

where again Āi denotes the effective surface area of solid species i. The
carbonate rates may be derived from equation (4.29)

RCaCO3 = k1 γH+ [H+ ] + k2 γH2 CO∗3 [H2 CO∗3 ] + k3

− k−3 γ Ca2+ [Ca2+ ]γCO2−
3
[CO2−
3 ] (4.65)
RMgCO3 = k1 γH+ [H+ ] + k2 γH2 CO∗3 [H2 CO∗3 ] + k3
− k−3 γ Mg2+ [Mg2+ ]γCO2−
3
[CO2−
3 ] (4.66)

using the proper sets of rate constants ki for CaCO3 and MgCO3 , respec-
tively. The hydroxide rates can be estimated from the linear log R vs. pH
relationship discussed in Section 4.2.2 which leads to expressions of the form

log R = c1 − c2 pH, (4.67)

where the constants c1 and c2 can be determined for Ca(OH)2 and Mg(OH)2
from the data presented in Table 4.3. Of course, the simple rate expres-
sions (4.67) are only very crude approximations, and more accurate relations
would definitely be desirable. The effective surface area Āi and the total
 103

amount Ni of each solid species i behave in the same way as it was explained
above for dolomite, so no further discussion is necessary here. Ideally, ĀCaCO3
and ĀMgCO3 are zero for type S dolime, and ĀCa(OH)2 and ĀMgCO3 are zero
for selectively calcined dolomite, but this is only true as long as there is
no precipitation of carbonates or direct transformation of solid hydroxides
into carbonates. Especially in case of selectively calcined dolomite, the CO2
generated by dissolution of CaCO3 might react with Mg(OH)2 to form some
MgCO3 . Furthermore, it is unlikely that the solid rawmaterials are “pure”,
i.e., selectively calcined dolomite will almost certainly contain Ca(OH) 2 and
MgCO3 , while type S dolime might contain some CaCO3 and MgCO3 , al-
though in this case the residual carbonates can be effectively eliminated by
grit removal after slaking. As can be seen from the rate expressions, now
generally
T T
CCa 6= CMg , (4.68)

therefore, it might be more difficult to achieve a 1:1 molar ratio of Ca:Mg

with type S dolime or selectively calcined dolomite than it was in case of
dolomite.
Finally it should be pointed out again that also this part of the dynamic
model suffers from serious limitations:

1. The underlying empirical expressions for the dissolution rates are not
necessarily adequate for the chemical and hydromechanical conditions
encountered in the continuous bioreactor.

2. Possible effects of inhibitors or accelerators on dissolution rates have

been neglected since no appropriate models were available.

3. No attempt has been made to develop a comprehensive model for the

 104

effective surface area Āi of the various solids, because this would have
been too much effort considering the limited reliability of other model
parts.

Out of these reasons, the results obtained from the model should not be
expected to carry too much information about the real process, even if con-
ditions of steady state are assumed.

Mathematical Model Structure

At this point a brief discussion of the mathematical model structure and

the methods available for numerical solution is in place, although no actual
simulation was made. All the equations derived in this section together form
a system of semi-explicit nonlinear differential-algebraic equations (DAEs)
which can be written as
dx
= f (x, y) (4.69)
dt
0 = g(x, y), (4.70)

where the vector x contains all the dynamic variables of the model, while
y consists of the remaining algebraic variables. DAEs arise naturally in
many applications, and their study is an active field in applied mathematics.
Numerical methods to solve DAEs directly—i.e., without transformation into
a system of ordinary differential eqations (ODEs)—have been developed since
the early 1970s, and reliable algorithms exist. DAEs present numerical and
analytical difficulties which do not occur with ODEs, for instance the initial
values must be consistent with the algebraic constraints (4.70). An excellent
review of recent developments in theory and numerical methods for DAEs is
provided by Brenan et al. [Bre89].
 105

4.3.2 Model Uncertainties and Conclusions

Originally, it was planned to use the dynamic model for process simulation
and as a basis to develop a control strategy. The idea was to maintain a
constant pH by controlling the total amount of lime solids present in the
system—so that the overall dissolution rate in the reactor just balances the
acetic acid production rate—and at the same time to ensure a 1:1 molar
ratio of Ca:Mg by adjusting the partial pressure of CO2 so that calcium and
magnesium compounds essentially dissolve at the same rate. Although some-
thing like this could be possibly done, the process model discussed above does
not provide the necessary basis for such a “sophisticated” approach, because
too many and too large uncertainties are involved which would almost cer-
tainly make the results useless for practical application. To summarize, the
apparent weak spots of the model are

• the limited reliability of the underlying expressions for dissolution rates,

• the difficulties to calculate the effective surface areas,

• the assumption that no reactions between substrate and other species

occur,

• the limited understanding of the factors influencing the rate of substrate

uptake and acetic acid production by the organism,

• the assumption that hydroxides are not directly transformed into car-
bonates by CO2 .

Nevertheless, the model was helpful to gain some insight into the dynamic
structure of the process and to determine important variables and their in-
terdependencies. The two main implications for practical applications are:
 106

1. The control of pH in the range between pH 6 and 6.8 is not expected to

cause major difficulties. For the fast reacting type S dolime, the amount
of undissolved solids in the system would be relatively small, and the pH
might overshoot if too much lime is added at a time, but this problem
could be easily eliminated by using a slow continuous feed of lime slurry
instead of separate additions. The very slow reacting dolomite could
pose the problem of excessively high solid concentrations that would be
necessary to maintain the required overall neutralization rate. In case
of selectively calcined dolomite none of these problems is expected.

2. It will be much more difficult to maintain the proposed 1:1 mole ra-
tio of Ca:Mg (or any other fixed mole ratio, if desired). A solution to
this problem will have to be found especially for type S dolime, since
Ca(OH)2 dissolves about ten times as fast as Mg(OH)2 . For selectively
calcined dolomite, the dissolution rates derived in Section 4.2.2 seem
to indicate that CaCO3 and Mg(OH)2 dissolve at rates of at least the
same order of magnitude under process conditions (see Table 4.3). In
case of dolomite only the initial dissolution stage could cause difficul-
ties, because ultimately calcium and magnesium should be released
stoichiometrically. If very fine particles are used in order to achieve a
large surface area, this initial non-stoichiometric release could actually
have a significant effect on product composition.

In the next chapter, a more practical approach will be made to specifically

adress the mole ratio problem, and it will be shown how a 1:1 CMA can
be produced using a continuous process under steady state conditions. The
preliminary design of a small CMA plant will be presented, including a brief
economic analysis.
Chapter 5

Process Design Considerations

Various fermentation processes for CMA production have been proposed, but
none of them has been commercialized or even tested at a pilot plant level
yet. This is kind of surprising since the authors have all claimed signifi-
cant cost advantages for their processes compared to the conventional route
which is currently used for small scale production. A real breakthrough in
the commercialization of CMA just did not happen, probably for following
reasons:

• CMA is supposed to be a “cheap” bulk chemical rather than a valuable

speciality product; therefore, large plants are required to make its pro-
duction economically feasible, and high profits are not to be expected.
The plant capacities considered by the FHWA range from 100 tons per
day minimum to preferably 1,000 tons per day.

• Better organisms might be isolated or existing ones might be geneti-

cally improved in the near future—this would significantly reduce the
cost of fermentation since it would allow for instance higher produc-

107
 108

tivity and higher final product concentrations. People tend to wait for
these possible improvements rather than to take the risk of an early
investment.

• There is still not much experimental data available in the literature;

the findings are often inconclusive, and some of them are not very
encouraging. Many design proposals are based on wishful thinking
rather than on experimental facts, thus making it sometimes difficult
to assess their real potential.

On the other hand, a case can be made also for rather small alternative CMA
plants, at least with respect to the next five to ten years, since it is not very
likely that large plants will be built during this time. There definitely is a
small market for CMA at prices of $ 500 per ton or even higher—for instance,
it can be sold to the State Highway Administrations for use on bridges, and
to environmentally concerned private consumers who would probably like to
use it on their sidewalks and driveways—and if a small scale fermentation
process can actually do better than the conventional route, it should really
be worth trying. One possibility to initiate such a small scale production of
CMA would be to establish CMA as a new byproduct of the pulp and paper
industry, made from unpurified waste sugars which otherwise would just be
burned. Quite much research has been done on new organosolv wood pulping
methods recently, and one of the most promising processes of this kind uses
concentrated acetic acid as cooking liquor. From the spent liquor of this
acetic acid pulping process an aqueous solution containing xylose, glucose,
and some acetic acid could be obtained after recovering other byproducts
and most of the acetic acid—a “waste” stream that would be ideally suited
for fermentation to CMA. The preliminary design for a CMA plant based on
 109

this feedstock will be presented at the end of the chapter together with an
economic analysis.
None of the previously published CMA fermentation processes offered a
solution to the mole ratio problem which provided the starting point for this
study. In the next section, it will be shown that an amazingly simple solution
exists. First, the key features of a continuous process capable of producing
the desired mole ratio of Ca:Mg will be discussed, then the experimental
realization of such a process will be described, and results will be discussed
for three different types of raw materials (type S dolime, selectively calcined
dolomite, and dolomite). Finally, the potential of the new process as a whole
and of each of the raw materials will be assessed.

5.1 How to Solve the Mole Ratio Problem

5.1.1 Stationary Solids Phase

When it became obvious that the “sophisticated” approach to solve the mole
ratio problem would fail because of the large uncertainties involved in the
model, an alternative solution was sought which does not require exact knowl-
edge about the real system. Just by considering the mass balances for calcium
and magnesium, a very promising new approach was actually found: if solids
are fed to a dissolver with a 1:1 molar ratio of Ca:Mg and a solids-free prod-
uct solution is withdrawn, i.e., essentially all solids are kept in the system,
then the system must eventually reach a steady state which is characterized
by the same 1:1 molar ratio of Ca:Mg for the solutes in the product stream—
what is put into the system in form of solids is what comes out as dissolved
material. In case of type S dolime for instance, assmuming that Ca(OH)2 (s)
 110

in fact dissolves ten times as fast as Mg(OH)2 (s)at pH 6, the magnesium

mineral would build up in the solids phase until it has ten times the effective
surface area of the calcium mineral, and then the overall release of calcium
and magnesium from the solids phase would be the same. Theoretically this
should work regardless how different the individual dissolution rates might
be. In practice, however, there are certain limitations when at least one of the
components reacts very slowly, because this would lead to excessive amounts
of solids present in the system. As the dissolution of the double carbonate
dolomite shows, the buildup of the slower reacting component may occur
only on the particle surface and hence the stationary solids phase may have
essentially the same bulk composition as the feed solid. This could be true at
least to some extent for the selectively calcined material; it is rather unlikely
for the type S dolime because of its very small particle size and the porous,
microcrystalline structure of the particles.
Apart from possible limitations due to low reactivity, no exceptional prob-
lems are expected for the practical application of continuous steady-state
dissolution in the context of CMA production:

• The neutralization can take place right in the fermentor if a continuous

flow, stirred tank reactor (CSTR) is used; this fermentor type was
already assumed for the dynamic model in Section 4.3 and will also
be part of the proposed plant design. Cells and undissolved solids are
recycled together so that the stationary solids phase and a high cell
density can be maintained at the same time.

• If batch fermentation or a continuous packed bed reactor were used,

the in-fermentor neutralization approach would probably not be ap-
propriate, but it would still be possible to use a separate steady-state
 111

dissolver in a recirculation loop and thus allow pH control while main-

taining the desired product composition. This design modification shall
not be discussed any further here.

• Either a centrifuge or a membrane filter can be used to separate the

cells and undissolved solids from the product solution; the separator
should be designed in a way that allows continuous recirculation of the
retentate. For instance, a disc-nozzle centrifuge or a cross flow filter
would meet this requirement. The decision between these two separator
types is quite difficult, especially since membrane filters have recently
found interesting new applications for product recovery in the context
of commercial fermentations (not only on a lab scale). In our case, the
decision was finally made in favour of the centrifuge because

– a centrifuge can be operated in a more flexible way, e.g., it is

possible to retain essentially all solids as well as to discharge a
certain fraction of them with the product stream, thus providing
an additional control variable,

– a membrane filter might get clogged during longer periods of op-

eration and then require cleaning or replacement, and

– there is still not very much experience with large scale applications
of membrane filters.

However, for the experiments that will be described in the next section,
a cross flow filter was used since it could be obtained more easily and
also provided simpler operation than a continuous centrifuge.

There might be two reasons for not recycling all solids—to prevent the
 112

buildup of insolubles like silica which are always present in lime and lime-
stone, and to influence product composition, if the desired mole ratio does
not match the composition of the feed solids. This can be done by discharg-
ing solids directly with the product stream (a certain amount of solids in
the product could actually improve the deicing properties) or by splitting
the retentate stream and thus sending only part of it back to the fermentor
(again the solids could be added to the deicer product, and it would actually
be preferable to do that after the evaporation step in order to avoid scaling
problems in the evaporator system). In practice, it would be most desirable
to specifically separate out unwanted components from the recycle stream
rather than simply split the whole stream into two substreams of the same
composition. It is not unlikely that such a specific separation can be achieved
since there will probably be sufficient differences in size and density among
the solid components.
How will the three different kinds of solid base perform in the proposed
new process? Ultimately, this question can be answered only based on exper-
imental results, but a preliminary assessment of these raw materials can be
made by considering their general properties and the rate data from Chap-
ter 4:

1. In type S dolime, high dissolution rates for both minerals Ca(OH)2 and
Mg(OH)2 are combined with a very large effective surface area due to
small particles (most of them in the range 2–5 µm) of porous structure.
Therefore, a high overall reactivity and only a relatively small accumu-
lation of solids is expected. Although the dissolution rate of Ca(OH) 2 is
about one order of magnitude higher than that of Mg(OH)2 , a station-
ary state should be established quickly, because Mg(OH)2 still dissolves
 113

at a fairly high rate. The stationary solids phase is expected to con-

tain mainly Mg(OH)2 , hence the discharge of undissolved solids would
produce a high calcium CMA.

2. Selectively calcined dolomite will react more slowly, but the dissolution
rates for both CaCO3 and MgO should still be relatively high and of
the same order of magnitude. The material has to be ground to obtain
a powder because it does not decompose automatically when mixed
with water. Even in finely pulverized form it will probably have a
smaller effective surface area than type S dolime due to the less porous
structure. Although the MgO should be only light-burned and reactive,
it is difficult to predict whether the step

MgO + H2 O −→ Mg(OH)2

is rate determining or not, i.e., whether the MgO will dissolve at es-
sentially the same rate as the Mg(OH)2 from type S dolime or rather
more slowly. From the rate data in Table 4.3 it would be expected that
the CaCO3 dissolves slightly faster than the MgO, but the difference
(factor two) is small compared to possible errors in the data, and the
opposite might be true. As a matter of fact, if MgO would turn out
to be the faster dissolving component, this would be a very interest-
ing property of the material, since it would allow the production of a
high magnesium CMA by discharging part of the undissolved solids.1
Compared to type S dolime, the necessary accumulation of solids in
the system will be higher, but it is not expected to be excessive. Since
1
As discussed in Section 1.2, there exists a CMA double salt with a Ca:Mg mole ratio
between 2:3 and 3:7 which may have superior deicing properties compared to the 1:1 CMA.
 114

the dissolution rates of CaCO3 and MgO should not differ too much,
the steady state is supposed to be established after a relatively short
period of time.

3. Dolomite dissolves at a very low rate compared to the other minerals

(at least ten times slower than CaCO3 and MgO), and like the selec-
tively calcined material it has to be pulverized mechanically. Calcium
and magnesium should be released stoichiometrically after the initial
dissolution stage; therefore, the discharge of solids is not expected to
significantly change the composition of the product (this might or might
not be an advantage). Although the high process temperature of 60◦ C
might accelerate dissolution considerably, the accumulation of solids
will be very high, and problems are likely to occur. A steady state will
be established soon after the stationary solids concentration is reached,
because the initial non-stoichiometric dissolution stage of dolomite does
not last very long.

A comparative experimental study of these materials in the new process is

clearly necessary in order to decide which one would be most appropriate for
the production of CMA. From an economic point of view, all three raw mate-
rials should be available at a comparable cost, since the cheaper materials—
selectively calcined dolomite and dolomite—would have to be mechanically
crushed and ground, which is not required for the more expensive type S
dolime.
 115

5.1.2 Experimental Realization

The Model System and Its Performance

Using a simplified model system, an experimental study of the new process

outlined in the last section was done; all three raw materials were tested. The
experiments were performed without the organism—instead, acetic acid was
fed into the system. An attempt was made to simulate the typical process
conditions that were chosen in Section 2.3.1 for a commercial CMA fermen-
tation, i.e.,

• acetic acid productivity 4 g l−1 h−1 , corresponding to a CMA produc-

tion rate of 5 g l−1 h−1 ,

• CMA concentration of 50 g l−1 in the product solution,

• pH controlled at slightly above pH 6, and

• temperature of about 60◦ C.

Unfortunately, the temperature controller failed during the first hour of op-
eration, so that all experiments had to be done at 26–31◦ C (without tem-
perature control), since time constraints did not allow waiting for repair or
replacement of the broken unit. A schematic diagram of the experimental
system is given in Figure 5.1; some less important parts (temperature control,
volume control, stirrer, and fraction collector) are not shown.2 At the begin-
ning of each experiment, the dissolver (“fermentor”) was filled with V = 1.5 l
deionized water. The recirculation loop (pump P2 ) was operated at relatively
high flow rates of 0.27–0.43 l min−1 in an attempt to prevent the buildup
2
For a complete list of equipment, see Appendix C.
 116

CO2
6

P1 - ¾
P4 neutralizing
HAc feed agent
(40 g l−1 )
6
(slurry)
dissolver
(“fermentor”)
- pH control
unit
¾ ¾
6

P2 - cross flow -
? filter

P3 - CMA product
? (≈ 50 g l−1 )

Figure 5.1: Schematic diagram of the experimental system used for continu-
ous production of CMA under steady state conditions.
 117

of a filter cake in the cross flow filter, and the contents of the dissolver was
well mixed by stirring at 800–1000 rpm. The feed and the product stream
had exactly the same flow rate F = 0.15 l h−1 ; this was accomplished by
using one pump drive with two identical peristaltic pump heads P1 and P3 .
Nevertheless, a volume control was required, because some additional water
was introduced with the neutralizer slurry. Whenever the liquid level in the
dissolver became too high, excess fluid was removed automatically by a pump
operating parallel to pump P3 (not shown in the diagram). Without these
additional streams the dilution rate would have been D = F/V = 0.1 h−1 ;
the actual value was slightly higher. The pH was controlled at a setpoint
of about pH 6 using a rather unsophisticated on/off-controller which actu-
ated the neutralizer feed pump P4 . A high speed stirrer was used for the
slurry storage vessel to prevent settling of the particles. The system was
operated at atmospheric pressure; several openings in the headplate of the
dissolver provided a means for the discharge of CO2 generated by carbonate
dissolution.
The feed stream contained 40 g l−1 acetic acid, thus simulating an acetic
acid productivity of 4 g l−1 h−1 in the “fermentor”. Under conditions of
steady state, this translates into approximately 50 g l−1 CMA in the prod-
uct stream, if the dilution by the slurry feed can be neglected. The total
concentration of acetate ion in the dissolver CAc can be described by
f
dCAc /dt = D(CAc − CAc )
(5.1)
CAc (0) = 0,
f
where CAc is the feed concentration of acetate ion, and D = F/V the dilution
rate. Equation (5.1) has the solution
f
CAc (t) = CAc (1 − e−Dt ), (5.2)
 118

f
i.e., the steady state acetate concentration CAc will be approached very closely
after 40–50 h of operation. Therefore, a duration of 50 h was chosen for the
experiments, and it was hoped that this amount of time would also allow the
solids phase to get close to steady state.
Hourly samples of the solids-free product solution were taken with an
automatic fraction collector; these samples were later analyzed for calcium,
magnesium, and several other dissolved ions by inductively coupled plasma
emission spectroscopy (ICP). From the analyses, the total concentration of
CMA in the product solution and the Ca:Mg mole ratio could be determined
as functions of time.
The cross flow filter with pore size 1.2 µm performed well in retaining even
the finest lime particles—a perfectly clear product solution was obtained.
However, there were problems with the accumulation of a filter cake in all
experiments, and it was not possible to prevent this by increasing the flow
rate in the recirculation loop (perhaps it just could not be increased enough).
Shaking of the filter capsule from time to time allowed recirculation of part
of the accumulated solids, but new cake formed quickly. The amount of
solids “trapped” in the cake (and hence essentially removed from the active,
dissolving part of the solids phase) was significant compared to the total
amount of solids present in the system. As will be seen from the discussion
of results, it was mainly this effect of “quasi solids removal” which made it
difficult to approach the steady state. Another problem that occurred only
in the experiment with selectively calcined dolomite was the clogging of the
neutralizer feed mechanism due to fast settling of the slurry. One time the
neutralizer feed was cut off for several hours, which produced quite interesting
effects, but of course seriously disturbed the course of the experiment from
 119

the way it was planned. Time constraints made it impossible to repeat the
experiment with an improved system; therefore, the presumed undisturbed
behaviour had to be extrapolated from the available data.

Discussion of Results

The first experiment was done with type S dolime supplied by The Western
Lime & Cement Co. in West Bend, Wisconsin; it had a Ca:Mg mole ratio
of 1.04:1.3 A slurry was prepared from one part dry lime and two parts deion-
ized water. The total amount of dolime consumed during 50 h of operation
was about 170 g; therefore, approximately 340 g of water were introduced
with the slurry—this amounts to less than 5% of the total feed of water,
thus increasing the dilution rate only slightly. As already mentioned, the
temperature controller failed during the first hour after start up, and it was
decided to continue the experiment without temperature control. The tem-
perature fell to the range of 26–31◦ C, about 5 K above ambient temperature
due to the heat generated by stirrer and pumps. During the first 10 min
of the experiment, the pH varied between pH 4.87 and 9.81, i.e., there was
some overshooting each time when dolime was added, but after 25 min it
had already stabilized in the range of pH 6.00–6.11. As more solids built
up in the system and the total acetate concentration increased, the range
narrowed further so that an essentially constant value of pH 6.02 could be
maintained. Starting after a few hours of operation, a brownish, gel-like filter
cake accumulated in the filter capsule which could not be removed during
the experiment. Backflushing at high flowrates had to be applied afterwards
in order to clean the filter for the next experiment.
3
Analyses of this material and the mole ratio calculation can be found in Appendix B.1.
 120

2.5
CMA from type S dolime 3
2 3

Ca:Mg 1.5
3
mole 333 33 3 3 3 3
3 3 3 3
ratio 1 33

0.5

0
0 5 10 15 20 25 30 35 40 45 50
time (h)

Figure 5.2: Mole ratio vs. time for CMA produced from type S dolime;
determined from ICP analyses of the product samples.

The Ca:Mg mole ratio in the product was calculated for some of the
hourly samples; the results are shown in Figure 5.2.4 As expected, the
calcium hydroxide turned out to dissolve much faster than the magnesium
hydroxide—at the beginning the Ca:Mg mole ratio in the product was as high
as 2:1. Thereafter, magnesium hydroxide started to build up in the dissolving
solids phase, and the mole ratio decreased rapidly. After only seven hours
of operation, the theoretical value 1.04:1 was reached, although there had
already been some minor accumulation of filter cake which caused the curve
to turn up slightly between hour 3 and 5. Obviously, serious cake accumula-
tion started just before hour 8, because there suddenly was a jump towards
4
Original results of the sample analyses are given in Appendix A.2.
 121

higher calcium content, and for the rest of the experiment the mole ratio
remained in the range between 1.20:1 and 1.35:1. The filter cake contained
high amounts of magnesium, alumina, and iron (which was responsible for
the brown color). The amount of solids that made up the solids phase in the
dissolver was very small (probably less than 5 g l−1 ) due to the fast reaction
of both minerals. At the end of the experiment, the concentration of CMA
in the product stream was about 45 g l−1 . Dry CMA was recovered from 2 l
product solution collected during the last 13 h of operation.
In the second experiment, selectively calcined dolomite was used as neu-
tralizer. The uncalcined stone was again supplied by The Western Lime
& Cement Co.; it had a Ca:Mg mole ratio of 1.03:1, i.e., it was essentially
pure dolomite.5 Samples of this stone were calcined for half an hour at
about 700◦ C in an electrical furnace. Ideally, only the magnesium part of
the dolomite should be decomposed under these conditions according to the
reaction
CaMg(CO3 )2 −→ CaCO3 + MgO + CO2 ↑, (5.3)

and the theoretical weight loss can be calculated from the amount of CO2 re-
leased. For the selective calcination of pure dolomite, the theoretical weight
loss is 23.9%, and if all carbonate is decomposed (full calcination), the weight
loss would be 47.7% instead. The actual weight losses of the calcined sam-
ples used in this experiment ranged from 17.7% to 23.6%, indicating that
there remained a small portion of unreacted dolomite. Before the slurry was
prepared, all samples were mixed and pulverized; the powder was screened
using a 35 mesh sieve. A total amount of 377.4 g dry powder was obtained.
The first batch of slurry was prepared from one part selectively calcined ma-
5
A chemical analysis can be found in Appendix B.2.
 122

terial and two parts deionized water. No observable reaction occurred when
the water was added—this confirmed that the material was practically free
of CaO which would have been hydrated in a fast and highly exothermal
reaction. Unfortunately, in this experiment the slurry feed mechanism did
not work as well as it had done with the type S dolime because the solids
tended to settle in the tubing. In an attempt to prevent clogging, an ele-
vated neutralizer storage vessel was used, and the slurry was diluted with
more water. Thus a relatively high amount of about 1.18 l water was added
to the system with the slurry, corresponding to almost 14% of the total feed
of water (this caused a significant increase of the overall dilution rate, and a
more dilute product solution was obtained). But still the clogging problem
was not solved completely, and one time the neutralizer feed was blocked for
several hours before the failure was discovered. Definitely, it would be a bet-
ter idea to feed the dry powder or a putty instead of the slurry using a screw
conveyor or similar piece of equipment when repeating the experiment. The
theoretical amount of selectively calcined dolomite which had to be dissolved
during 50 h of operation was 175.5 g, but the actual amount fed into the
system was higher, since much more undissolved solids accumulated than in
the first experiment due to the lower reactivity of the material. Right after
the experiment had been started, the pH fell from about pH 7 to 5.07, and
then started rising slowly, as slurry was added continuously. About 10 min
later the setpoint of pH 6.00 was reached. Enough solids had built up in
the dissolver to maintain the required overall neutralization rate at around
pH 6, and the neutralizer feed became discontinuous. There was no signif-
icant overshooting like with type S dolime, the pH just varied in the range
of pH 5.96–6.03. Within a few hours of operation this range narrowed fur-
 123

2.5
CMA from selectively calcined dolomite 2
2

Ca:Mg 1.5 2
mole
ratio 1 2
2 2
2 2 2
2 2
0.5 2222 2
222222 2 2 2 2
0
0 5 10 15 20 25 30 35 40 45 50
time (h)

Figure 5.3: Mole ratio vs. time for CMA produced from selectively calcined
dolomite; determined from ICP analyses of the product samples.

ther due to the buffering effect of the increasing acetate concentration. The
problem with the accumulation of filter cake started even earlier than in the
first experiment, although the cake was less “sticky” and could be removed
more easily during operation. Again, some nonideality in the behaviour of
the system had to be expected.
Figure 5.3 shows the Ca:Mg mole ratio for the product during the course
of the experiment.6 The two most important observations are:

1. The magnesium oxide clearly dissolved faster than the calcium carbon-
ate at pH 6, in contradiction to the rate estimates from Section 4.2.2
(Table 4.3). After eleven hours of operation the Ca:Mg mole ratio was
6
See Appendix A.2 for the original results of the analyses.
 124

as low as 1:3.5, and then started moving up slowly towards a higher

calcium content. The two sudden peaks at hour 24 and hour 50 were
both caused by failures of the slurry feed—in the first case the feed
pump got clogged sometimes after hour 17 (when it was repaired at
hour 26, the pH had already dropped to pH 5.07), in the second case
the system simply ran out of neutralizer slurry (the pH was down at
pH 5.66 when the experiment was finished). If the first failure had not
occurred, the system would probably have come much closer to a true
steady state. Now the planned course of the experiment was seriously
disturbed, but the system behaved exactly as one would have expected:
once the feed of fresh neutralizer was cut off, the mole ratio markedly
increased as the faster reacting magnesium oxide got more and more
depleted in the solids phase. Another effect which may have acceler-
ated the calcium surge is the decrease of pH. If the rate data presented
in Table 4.3 are at least qualitatively correct, then the rate of calcite
dissolution should increase significantly relative to the magnesium ox-
ide rate with decreasing pH. This could mean that already at pH 5 the
calcium carbonate actually dissolves faster than the magnesium oxide
which would reverse the outcome of the dissolution experiment if done
at lower pH.

2. It is not possible to explain the significant decrease in the mole ratio

between start up and hour 11, if the difference between the dissolu-
tion rates of the two minerals is assumed to remain constant through-
out the experiment—although a decrease in the mole ratio could be
caused by filter cake accumulation (removal of calcium carbonate from
the actively dissolving solids phase), this alone would never lead to a
 125

mole ratio below the start up level. Therefore, “quasi solids removal”
might be responsible for delaying the upturn towards steady state after
hour 11, but definitely not for the observed minimum itself. A likely
explanation for this strange behaviour is that the dissolution rate of
the magnesium oxide significantly increased during the first hours of
operation. As briefly mentioned in Section 4.2.2, acetate acts as an
accelerator in the dissolution of magnesium hydroxide, and the same
might be true for bicarbonate. The concentrations of both acetate and
and bicarbonate increased after start up,7 hence an accelerating effect
might very well occur. Since for selectively calcined dolomite the disso-
lution rates are of the same order of magnitude, even a relatively small
acceleration of magnesium oxide dissolution is observable which is not
necessarily true for type S dolime due to the much higher reactivity of
calcium hydroxide.

The chemical analysis of the solids phase showed that it contained calcium
and magnesium in a molar ratio of 3.07:1, and also significant amounts of
sulfur, alumina, and iron. The solids concentration in the dissolver was
approximately 45 g l−1 —higher than in the first experiment, but not unrea-
sonably high. A final CMA concentration of about 38 g l−1 was reached, and
again dry CMA was recovered from 2 l product solution.
In the third experiment, the same dolomitic limestone was used as in the
second experiment, but without the calcining. The stone was just pulverized
(less than 35 mesh), and a slurry was prepared from one part dolomite and
about three parts water. Some difficulties with a high accumulation of solids
7
The bicarbonate formed by dissolution of calcium carbonate is likely to accumulate
to a certain level due to slow carbon dioxide exsolution.
 126

had been expected, but it was still surprising how soon the system had to
be shut down because of complete failure. Immediately after the experiment
had been started, the pH dropped to pH 4.70, then stabilized and started
rising slowly as limestone was continuously fed into the system. However,
about 20 min later there was already an excessive buildup of dolomite in
the filter capsule, and the pH had just reached pH 5.40—still far below the
setpoint. The solids concentration was so high that even shaking of the filter
capsule could not prevent further accumulation of cake. Only 35 min after
start up the cross flow filter got clogged, and the recirculation pump had to
be turned off. The acetic acid feed was stopped as well, only the pH con-
trol was left operating. It took another 10 min of continuous limestone feed,
before the setpoint pH was reached. Clearly, the uncalcined material was
too slow to maintain pH 6 at a reasonable concentration of solids under the
chosen process conditions. Perhaps the performance would have been better
at higher temperature (60◦ C) and for a smaller particle size. High solids con-
centrations could probably be handled using a more sophisticated equipment
(e.g., a centrifuge instead of the filter), but the other two materials which
can do the job at much lower accumulations would definitely be preferable
in the context of a fermentation process. An analysis of the product solution
collected during the first 35 min of operation showed that it had a Ca:Mg
mole ratio of 1.12:1; another sample taken at pH 8.14 after solids and liquid
in the dissolver had equilibrated for several hours had an even higher ratio of
1.50:1. These results confirm again the significance of the initial dissolution
stage which already had been found in the equilibrium experiment described
in Section 3.3.
Based on the results of these experiments, a better evaluation of the pro-
 127

2.5
CMA from type S dolime 3
CMA from selectively calcined dolomite 2
2 3

Ca:Mg 1.5
mole 3
ratio
33
1 333333333333333333333
22222
2 2 2 2222
0.5 2 2 22
2 22
222222
0
0 5 10 15 20 25 30 35 40 45 50
time (h)

Figure 5.4: Ideal mole ratio curves.

posed process and the tested raw materials could be made. It should be
possible to avoid the adverse effects of filter cake accumulation by using a
continuous centrifuge instead of the filter—such a system will be able to op-
erate under conditions of steady state for long periods of time. Although the
experimental system described above was far from ideal, it performed well
enough to confirm the feasibility of the new process. A more ideal system
would probably have shown a behaviour similar to that given in Figure 5.4;
the curves for both type S dolime and selectively calcined dolomite were
obtained from the available data by correcting for the effects of cake accu-
mulation and control system failure. The type S curve approaches the 1:1
steady state mole ratio from above, and the curve for the selectively calcined
material comes from below (it initially turns down due to acceleration of
 128

magnesium oxide dissolution, and then turns up again because of the accu-
mulation of calcium carbonate in the solids phase). A curve for uncalcined
dolomite would probably look quite similar to the type S curve, if this ma-
terial could be successfully applied in spite of its low reactivity.
The potential of the three raw materials tested can now be assessed in
the following way:

1. Type S dolime performes well at very low solids accumulation. A steady

state is established within a few hours of operation, and it should not
be too difficult to produce a 1:1 CMA by completely recycling the
undissolved solids. High calcium CMA could be made by discharging
some of the solids, but if such a product is desired (e.g. for flue gas
desulfurization), it would be better to use a high calcium lime instead
of the dolime, unless the magnesium-rich solids discharge can be sold
as a byproduct.

2. Selectively calcined dolomite requires a higher accumulation of solids

(about ten times that of type S dolime), but this is not expected to
cause any difficulties. It may take also about ten times longer to reach
the steady state (perhaps 50–70 h), but again this is not judged to
be a problem, and it should be possible to produce a 1:1 CMA using
the proper equipment. Even more interesting is the option to make a
high magnesium CMA by discharging part of the solids. Any Ca:Mg
mole ratio between 1:3.5 and 1:1 can be achieved in the product through
controlling the solids discharge, thus it would be no problem to produce
the mole ratio of 4:6 to 3:7 which is required for the calcium magnesium
double salt described by [Tod90].
 129

3. Uncalcined dolomite did not perform well enough to be considered as

a real alternative to the other two materials. The main problem is
its very low reactivity which would possibly require a solids concen-
tration of 400–500 g l−1 for 35 mesh particles. Of course, a smaller
particle size could be used, and perhaps the material would dissolve
significantly better at higher temperature, but it is not expected that
a performance comparable with the other materials can be achieved.
Therefore, chances are slim that dolomite will ever be used as a neutral-
izer in CMA fermentations—in fact the same conclusion was already
made by Marynowski et al. [Mar85].

In the next section, the proposed process will be incorporated in the prelim-
inary design of a fermentation plant for the production of CMA from wood
hydrolyzate, and the economics of such a plant will be briefly discussed.
Then a summary of the advantages of this new process compared to other
proposed processes will be given.

5.2 Preliminary Plant Design

As already mentioned in the introduction to this chapter, CMA shall be
examined as a valuable new byproduct of the pulp and paper industry, hence
not a large stand-alone CMA plant will be discussed here, but rather a small
scale operation that could easily be added to an acetic acid wood pulping
plant. It was decided to use selectively calcined dolomite as neutralizer,
because this material offers interesting design options, and it has never been
used in a commercial process before. The design for a process using type S
dolime would be quite similar.
 130

5.2.1 Process Description and Flowsheet

Conventional wood pulping processes like the kraft process utilize only the
cellulosic fraction of the wood (about 50 wt%) which is transformed into
pulp—the organics dissolved during the pulping operation, especially sugars
from hemicelluloses and lignin, are usually burned, providing just enough
energy for the evaporators that are needed to concentrate them before they
can be sent to the furnace. Furthermore, the kraft process has serious envi-
ronmental drawbacks, e.g. the emission of malodorous sulfur compounds like
hydrogen sulfide and methyl mercaptan. Hence there is a growing interest in
alternative pulping methods that allow the recovery of valuable byproducts
and are environmentally safe. Various organosolv processes which use organic
solvents instead of aqueous solutions of sodium sulfide and sodium hydroxide
have been proposed since the 1930s. One of these processes uses concentrated
acetic acid as solvent, and this makes it particularly “compatible” with the
production of CMA. A simple block flow diagram of such an operation is
given in Figure 5.5; more details can be found in [DeH71, Nim86]. Dry
hardwood consists of 40–45 wt% cellulose, 20–30 wt% hemicelluloses, and
20–30 wt% lignin, so that for each ton of pulp produced, roughly half a ton
of sugars (mainly xylose) and half a ton of lignin can be recovered as byprod-
ucts. The sugars are obtained in form of an aqueous solution which may also
contain some traces of acetic acid, furfural, levulinic acid, and minerals re-
leased from the wood. It is not expected that these impurities would cause
any difficulties in the context of CMA fermentation or have an adverse effect
on product properties (in fact, the acetic acid will react to CMA); there-
fore, it should be possible to feed this aqueous sugar solution directly to the
fermentor.
 131

wood chips

?
acetic acid - - pulp
(from recovery) digestion

?
acetic acid,
evaporation - byproducts
and drying
(to recovery)

?
- lignin - sugar solution
water precipitation (to fermentation)

?
lignin

Figure 5.5: Simple block flow diagram of the acetic acid pulping process; the
solvent recovery system which is essential to make this process economically
viable is not shown in the diagram.
 132

Another advantage of oganosolv pulping is that the plants can be smaller

than conventional kraft mills which usually have a capacity of more than
1000 tons per day pulp. Hence a rather small pulping plant was chosen as
basis for the design. The following major assumptions and design specifica-
tions were made:8

• The pulping process has a capacity of 100 t per day pulp. About 330 t
per day hardwood chips (40 wt% moisture) are required as feed, and
it is assumed that among other byproducts 50 t per day sugars can be
recovered for fermentation to CMA.

• The fermentor is operated at 60◦ C and at slightly below atmospheric

pressure (in order to facilitate the exsolution of carbon dioxide). It
is assumed that no sterilization of fermentor and feeds is necessary to
prevent contamination.

• Of the sugars fed to the fermentor, 95% are utilized by the organism,
and only 5% are lost with the product stream (and form part of the
deicer product). For 1 g of sugar consumed by the organism, 0.85 g
of acetic acid are produced, i.e., the yield on a weight basis is 85%.
Therefore, 0.95 × 0.85 × 50 t = 40 t per day acetic acid are produced;
that corresponds to 50 t per day pure CMA product.

• As before, a concentration of 50 g l−1 CMA in the product stream

is assumed. This requires the same concentration of 50 g l−1 for the
sugars in the feed stream. Feed and product stream have a volumetric
flowrate of 1000 m3 per day (or 41.7 m3 h−1 or 11.6 l s−1 ), hence about
8
Metric units were used consistently in the design calculations, e.g. metric tons (t)
instead of U.S. tons; the main results will be given in both unit systems.
 133

1000 t water have to be evaporated per day in order to recover the

product.

• Again, a value of 5 g l−1 h−1 is used for the CMA production rate per
unit volume of the fermentor, and hence a fermentor volume of 416 m3
is required.

• Although the sugar feed stream might already contain some nutrients in
addition to the substrate itself (e.g. minerals from the wood), and others
might be supplied by the lime, it is most likely that at least additional
sources of nitrogen, sulfur, and phosphorus have to be provided in the
medium for growth of the organism. These nutrients are assumed to
be already added to the sugar feed stream.

• The fermentor effluent at pH 6 still contains 2.18 g l−1 undissoci-

ated acetic acid as calculated by the Henderson-Hasselbalch equa-
tion (2.15). This remaining acid is neutralized by adjusting the pH of
the product stream to pH 8–9 before evaporation.

• Selectively calcined dolomite is used as neutralizer in the fermentor,

and the pH of the product solution is adjusted using a small amount
of calcium hydroxide (high-calcium hydrated lime). Two cases are con-
sidered for the product composition:

1. Production of a 1:1 CMA which would most likely consist of a

physical mixture of the two individual acetate salts. Figure 5.6
shows a mass balance for this case on a basis of metric tons per day.
The unfinished product stream leaving the fermentor at pH 6 has
to contain slightly more magnesium than calcium, since only cal-
 134

H2 O 1, 000.00 CaCO3 17.12

HAc 40.00 MgO 6.90
Ca:Mg
1:1
? ?
fermentor (pH 6)

Ca:Mg Ca:Mg
1:1.12 5:1 CO2 6.51
? - CaCO3 2.30
H2 O 1, 005.67 MgO 0.18
HAc 2.18
CaAc2 23.46
MgAc2 23.71 Ca(OH)2 1.35

? ?
neutralizer tank (pH 8–9)

Ca:Mg
1:1
?
H2 O 1, 006.32
CaAc2 26.34
MgAc2 23.71

Figure 5.6: Mass balance for the production of a 1:1 CMA from selectively
calcined dolomite; all figures are given in units of metric tons per day.
 135

cium hydroxide is used in the neutralization tank for final pH ad-

justment. Therefore, a small amount of solids must be discharged;
that may also help to get rid of inerts contained in the lime. In
order to determine the necessary amounts of lime feed and solids
discharge, the composition of the stationary solids phase has to
be known—from the experimental results discussed above, it was
estimated that for a 1:1.12 product, the mole ratio in the solids
phase has to be at least 5:1, perhaps even higher. Based on this
assumption, the feed of selectively calcined dolomite is 24.02 t per
day, and 2.48 t per day solids have to be discharged. For the
pH adjustment, 1.35 t per day calcium hydroxide are required. Of
course, the solids can be added to the deicer product as a traction
aid and would probably enhance its properties.

2. Production of a 2:3 CMA suitable for crystallization as a double

salt. Now a much higher fraction of the solids has to be discharged
in order to maintain the high magnesium concentration in the
product. The corresponding mass balance is given in Figure 5.7;
a 4:1 mole ratio was assumed for the stationary solids phase. The
feed of selectively calcined dolomite is 32.07 t per day, and 12.50 t
per day solids have to be discharged that can be added to the
product. Again, 1.35 t per day calcium hydroxide are required for
the pH adjustment.

In both cases, 2.50 t per day sugars would also be contained in the
deicer. Therefore, the total amount of deicer material produced would
be 55.03 t per day in case 1 and 64.51 t per day in case 2 if all discharged
solids were added to the product.
 136

H2 O 1, 000.00 CaCO3 22.87

HAc 40.00 MgO 9.20
Ca:Mg
1:1
? ?
fermentor (pH 6)

Ca:Mg Ca:Mg
1:1.74 4:1 CO2 5.08
? - CaCO3 11.36
H2 O 1, 005.67 MgO 1.14
HAc 2.18
CaAc2 18.18
MgAc2 28.45 Ca(OH)2 1.35

? ?
neutralizer tank (pH 8–9)

Ca:Mg
2:3
?
H2 O 1, 006.32
CaAc2 21.06
MgAc2 28.45

Figure 5.7: Mass balance for the production of a 2:3 CMA from selectively
calcined dolomite; all figures are given in units of metric tons per day.
 selectively
sugar carbon calcined
feed dioxide water dolomite

calcium
hydroxide

5 LP
4 1
steam

cond.

LP
steam
cond. dry CMA
product

Figure 5.8: Process flow diagram for small-scale CMA production from waste sugars and selectively
calcined dolomite.
137
 138

A process flow diagram is given in Figure 5.8; the process flow through all
of the important pieces of equipment shall be briefly described at this point.
The fresh medium containing the sugars and additional nutrients is first
heated to 60◦ C and then enters the 416 m3 fermentor which is a continuous-
flow stirred-tank reactor (CSTR). A heating coil—or, alternatively, a steam
jacket—provide a means to maintain the process temperature inside the fer-
mentor. Selectively calcined dolomite is purchased in crushed form; it is first
pulverized using a grinder and then mixed with a very small amount of wa-
ter. The resulting putty is directly fed to the fermentor by a screw conveyor
which is actuated by the pH controller. Fermentation broth containing prod-
uct, undissolved lime solids, and cells is sent to a disc nozzle type centrifuge
capable of producing an overflow essentially free of solids and cells while dis-
charging these continuously along with a variable portion of the liquid phase
in the underflow. The solids concentration in the underflow does not have to
be exceptionally high, as a matter of fact, it would probably be preferable
to maintain a high flowrate at relatively low solids concentration to avoid
settling problems. A second, smaller centrifuge is used to separate out the
fraction of lime solids which has to be discharged; essentially all cells and
the remaining lime solids are recycled back to the fermentor with the over-
flow. The overflow of the first centrifuge is sent to a 40 m3 neutralizer tank,
where a small amount of calcium hydroxide is added in order to raise the
product pH to about pH 8–9. Complete reaction of the calcium base in the
neutralizer can be assumed, and the essentially solids-free product stream is
then sent to a multi-effect evaporator for concentration of the CMA to about
40 wt%. The available total temperature difference is constrained by the de-
composition temperature of CMA which is about 160◦ C. For instance, if the
 139

evaporator was operated in the range between 60◦ C and 140◦ C, the number
of effects would be limited to five assuming a minimum approach temperature
of 16 K. The design could be further improved by using an adiabatic flash in
order to bring the temperature of the evaporator feed down to about 45◦ C
(this would widen the available temperature difference for the evaporator
and at the same time already remove some of the water), and the maxi-
mum number of effects could be increased by thermal or mechanical vapor
recompression. These possibilities shall not be considered in this preliminary
design, although they could significantly reduce the cost of evaporation. A
five-effect evaporator with a steam efficiency of 4.4 is assumed, i.e., each ton
of low pressure steam at 3.614 bar, 140◦ C evaporates 4.4 tons of water. The
concentrated CMA solution is then mixed with the solids discharged from
the second centrifuge and sent to a steam heated drum dryer/flaker, where
the dry deicer product is recovered in the form of coarse flakes.
At this point, the major design specifications shall be summarized, and
the results shall be given in both metric and U.S. units:

• The required amounts of raw materials are

– 50.0 t (55.1 ton) per day waste sugars as feedstock (additional

nutrients are not exactly known),

– 24.0 t (26.5 ton) per day and 32.1 t (35.4 ton) per day selectively
calcined dolomite as neutralizer for the production of 1:1 CMA
and 2:3 CMA, respectively,

– 1.35 t (1.49 ton) per day calcium hydroxide for the pH adjustment.

• The plant output of deicer product (including discharged solids and

sugars) is
 140

– 55.0 t (60.6 ton) per day for 1:1 CMA,

– 64.5 t (71.1 ton) per day for 2:3 CMA.

These specifications will be used in the next section as a basis for the eco-
nomic evaluation of the process.

5.2.2 Economic Analysis and Discussion

An economic analysis shall be presented only for the CMA process, not for
the entire acetic acid pulping plant. Furthermore, only the case of producing
a 2:3 CMA shall be considered here, the other case could be examined in a
similar manner. In addition to the design specifications already discussed,
the following assumptions were made:

• The equipment is installed in an existing plant, where auxiliary facilities

are existing and adequate.

• Carbon steel is used for the evaporator and drum dryer/flaker, but all
other major equipment items have to be made from stainless steel.

• Neither the waste acetic acid contained in the sugar feed stream nor
the additional amount of CMA product formed from this acid are con-
sidered in the calculations.

• The sugars are available at their fuel value, i.e. for about $ 25.00/t
($ 22.70/ton).

• Other nutrients required by the organism are assumed to cost $ 15.00

per metric ton of deicer produced ($ 13.60 per U.S. ton of product).
 141

Table 5.1: Capital requirement for a small CMA plant with 64.5 t (71.1 ton)
per day output of 2:3 CMA deicer (including lime solids and sugars).

Item Cost (U.S. $)

lime storage 40,000

grinder 60,000
screw conveyor 20,000
fermentor (416 m3 ) 500,000
agitator (750 kW) 860,000
disc-nozzle centrifuge (760 mm Ø) 150,000
disc-nozzle centrifuge (400 mm Ø) 80,000
neutralizer tank (40 m3 ) 50,000
evaporator (5 effects) 2,940,000
drum dryer/flaker 1,790,000
product storage 80,000

total bare module capital 6,570,000

total fixed capital 7,750,000

• Low pressure steam at 3.614 bar, 140◦ C is available at a cost of $ 5.00/t

($ 4.54/ton).

• Electric energy costs $ 0.06 per kWh; total power requirements for
stirrer, centrifuges, and pumps are estimated at 1,250 kW.

An estimate of the capital requirement for such a plant addition was made
based on the cost data provided by [Ulr84]; all prices were escalated to
1991 using the CE Plant Cost Index. Table 5.1 lists the prices found for the
major individual equipment items. These prices are bare module prices, i.e.,
installation is already included. Their sum has just to be multiplied by the
 142

Table 5.2: Annual operating cost for the same CMA plant, assuming a
90% plant service factor, i.e., an output of 21,190 t (23,360 ton) per year
of 2:3 CMA deicer.

Item Cost per Year

(U.S. $)

raw materials
sugar feedstock ($ 25/t) 410,600
other nutrients 317,800
selectively calcined dolomite, crushed ($ 50/t) 527,200
high-calcium hydrated lime ($ 60/t) 26,600
utilities
electricity ($ 0.06/kWh) 591,300
low pressure steam at 3.614 bar, 140◦ C ($ 5/t) 373,300
labor 500,000
fixed charges on capital 542,500

total annual operating cost 3,289,300

factor 1.18 to account for contingency and fee, yielding a total fixed capital
of $ 7,750,000.
Table 5.2 shows the annual operating costs for the production of 2:3 CMA
deicer, determined on the basis of a 90% plant service factor (21,290 t per
year output). If a 30% simple rate of return (ROR) before taxes is required
on the total fixed capital, the deicer product can be sold for $ 264/t or
$ 239/ton (at plant gate). This compares well to the cost estimates for other
fermentation processes, which range from about $ 260 to $500/ton, depending
on the feedstock and process scale. It also should be pointed out that this
low production cost could be achieved on a very small plant scale—all other
 143

proposed CMA plants were much bigger; most of them were designed for a
capacity of 500–1,000 tons per day.
The new process for the production of a 2:3 CMA deicer from selectively
calcined dolomite has several advantages compared to previously proposed
processes (and even to the production of 1:1 CMA using the “same” process):

• Only inexpensive raw materials are used to make the 2:3 CMA which
is supposed to have superior deicing properties. In the conventional
Chevron process, expensive magnesia ($ 265–366/ton) has to be added
in order to achieve the 2:3 mole ratio, and together with the acetic acid
this contributes significantly to the high cost of the product (selling
price $ 657/ton).

• No waste streams are generated throughout the whole process, since

all solids that have to be discharged are included in the final product.
The resulting deicer might perform even better than “pure” CMA, and
the amount of product that can be sold is considerably higher.

• Inhibition of the organism is expected to be less severe than it would

be for the production of 1:1 CMA, because calcium is much more toxic
than magnesium (see Section 2.3.1).

• The solids discharge is roughly one third of the solids feed—this effec-
tively limits the undesired buildup of inerts in the solids phase.

Therefore, it may be concluded that the production of 2:3 CMA from se-
lectively calcined dolomite has the greatest potential and would definitely
justify further investigation.
Chapter 6

Analysis of Results

At this point, a brief summary of the most important results of this study
shall be given, and the major conclusions shall be discussed from the stand-
point of the original objectives of the research. This discussion will quite
naturally lead to the formulation of concrete recommendations for the prac-
tical application of some of the reported findings, and it will also lead to
specific suggestions for future work.

6.1 Results and Conclusions

In Chapter 1, CMA was introduced as a promising alternative deicer chemi-
cal: it is very effective as an deicer, non-corrosive, less damaging to concrete,
and environmentally safe. Furthermore, this chemical can be used as an ad-
ditive to coal combustion, acting as a catalyst and “sulfur grabber” at the
same time. However, it has not yet become a bulk chemical due to its pro-
hibitively high cost when produced from synthetic glacial acetic acid. The
deicing properties of the material depend on its chemical composition (Ca:Mg

144
 145

mole ratio), on whether or not it has been crystallized as a double salt, and
also on its physical form (coarseness and form of the flakes). From the avail-
able literature it was concluded that suitable compositions for a CMA deicer
would be either a mole ratio of 1:1 which most likely results in a physical
mixture of the two acetate salts, or some ratio between 3:7 and 4:6 which
may lead to crystallization as a double salt. For the other possible applica-
tion as combustion aid, a calcium acetate (CA) might actually be preferred
to CMA, since the magnesium does not react.
The raw materials for the production of CMA—acetic acid, and various
kinds of dolomitic lime or limestone—were discussed in Chapter 2, and it
was found that acetic acid cost is the key factor determining the cost of the
final product, since CMA contains almost 80 wt% acetate. Synthetic glacial
acetic derived from petroleum or natural gas costs more than $ 500 per ton,
hence CMA has to be expensive unless a cheaper source of acetate can be
found. Compared to the acid, the cost of the possible neutralizer materials is
almost insignificant, and a choice should be made based on their properties
and suitability for the process rather than on their cost.
Three basically different processes for the manufacture of CMA were re-
viewed,

• conventional processes starting from purchased glacial or concentrated

acetic acid,

• fermentation processes employing bioconversion of glucose, corn, or

even organic waste materials into acetic acid, and

• alkaline fusion processes converting cellulose to acetate in an excess of

alkali under conditions of high temperature and pressure.
 146

Fermentation was found to be an advantageous route compared to conven-

tional processes, especially if organic wastes can be used as a feedstock. As
one of several possible modifications of this route, in-fermentor production
of CMA was picked for further investigation.
Some important features of the anaerobic homoacetate fermentation with
Clostridium thermoaceticum were summarized, and optimal fermentation
conditions as well as performance limits of the organism were discussed.
The fermentation should be carried out at 60◦ C and slightly above pH 6;
it is expected that—using a continuous process with cell recycling—a CMA
productivity of 5 g l−1 h−1 and a final CMA concentration of 50 g l−1 can be
achieved with currently available strains of C. thermoaceticum.
The starting point for the main part of this study was the problem of con-
trolling the composition of the CMA product which arises in this context. As
pointed out above, the chemical composition of the CMA strongly influences
its properties, and, of course, a product of uniform quality is desired. There-
fore, the specified mole ratio should be closely maintained, and this turned
out to be quite difficult due to the different reactivities and solubilities of
the various corresponding calcium and magnesium compounds, particularly
in dilute acetic acid solutions such as those present in the context of fer-
mentation. The available literature was very inconclusive regarding to this
problem; obviously, no serious effort had been made before to investigate it
further and to find a solution.
As a first step towards this goal, equilibrium models were examined both
theoretically and experimentally in order to obtain some information about
thermodynamic limitations which have to be expected for the relevant chem-
ical systems (Chapter 3). For the hydroxide system composed of HAc,
 147

Ca(OH)2 , and Mg(OH)2 no such limitations were found at pH < 9; only

at higher pH was Mg(OH)2 practically inert, while Ca(OH)2 could dissolve
readily up to pH > 12. Therefore, no difficulties should arise for the dissolu-
tion reaction at pH 6 in the fermentor. The observation reported by [Mar85]
that “MgO becomes essentially inert at pH > 6” was probably due to hard-
burned (sintered) MgO with low reactivity which did not hydrate readily at
pH > 6 to form Mg(OH)2 , thus greatly inhibiting further dissolution. Three
possible solutions to this problem were found, namely

• dissolving MgO at pH < 6,

• using fully hydrated type S dolime Ca(OH)2 · Mg(OH)2 , or

• using selectively calcined dolomite CaCO3 · MgO which contains light-

burned MgO with high reactivity.

In the carbonate system made from HAc, CaCO3 , and MgCO3 there was no
profound difference between the behaviour of the calcium and the magnesium
compounds; both carbonates dissolved up to about pH 8. Experiments were
also made with dolomite (which showed a significantly different behaviour,
e.g. much lower reactivity, so that no true equilibrium was established within
24 h) and with a CO2 -bubbled hydroxide system (which was essentially trans-
formed into a carbonate system within a few hours). From the latter result
it was concluded that CO2 bubbling is likely to have a strong influence on
product composition in a CMA fermentation.
The kinetics of the various dissolution reactions relevant for CMA pro-
duction were reviewed in Chapter 4. Rate data and models were collected
from the available literature for carbonates (calcite, magnesite, dolomite)
and oxide minerals (CaO, MgO). Included was also the discussion of several
 148

engineering publications about the application of limestone for acid neutral-

ization (acidic waste water treatment). Estimates for the dissolution rates of
all minerals in the interesting pH range (pH 5–7) could be made as a result
of this literature search; the rate estimates were summarized in Table 4.3.
However, large uncertainties were expected in these results, since all of the
data and model equations had been derived for other acids than acetic acid
(mostly for strong mineral acids), and some of the sources were judged to be
not very reliable. Nevertheless, a dynamic model for a hypothetic continuous
CMA fermentation was developed, but the original plan to use this model
for simulations in order to examine process dynamics and control was aban-
doned in favour of a more practical approach. The continuous reactor was
modeled as an open system with lumped parameters; four phases were dis-
tinguished (liquid phase, solids phase, cells, and gas phase). All homogenous
chemical reactions in the liquid phase were assumed to be at equilibrium and
hence could be described by a system of algebraic equations. Acetic acid
production, substrate consumption, and solids dissolution were described by
ordinary differential equations. Together these equations formed a system of
semi-explicit nonlinear differential-algebraic equations (DAEs).
In Chapter 5, a simple solution to the mole ratio problem was presented—
it was shown that a stationary solids phase allows producing a stoichiometric
solution of CMA, and, more generally, that it is possible to achieve a wide
range of other product compositions if desired. Feasibility was confirmed by
an experimental continuous dissolution process, and the performance of three
different neutralizer materials was compared: type S dolime and selectively
calcined dolomite were both found to be suitable choices for future applica-
tions; however, the use of uncalcined dolomite is not recommended due to
 149

its very low reactivity. Selectively calcined dolomite provides the option of
producing high-magnesium CMA without the addition of expensive magne-
sia; most interestingly, a 2:3 CMA can be made which might be suitable for
crystallization as a double salt. It has been claimed that this double salt has
superior deicing properties [Tod90]. The preliminary design of a small-scale
process for the production of CMA from wood hydrolyzate (waste sugars)
and selectively calcined dolomite was presented, and an economic analysis
showed that this new process compares well to previously proposed ones. It
was concluded that particularly the production of 2:3 CMA from selectively
calcined dolomite has much potential, and that further investigation of this
specific route would definitely be justified.

6.2 Future Work and Recommendations

Although the concept of continuous steady-state dissolution developed in
Chapter 5 seems to be promising and the experimental results were quite
encouraging, much more work has to be done before a pilot plant can be built.
All experiments were done without the organism so far, and not even the
physical and chemical process conditions expected in a “real” fermentation
were maintained. Further research is required in the following areas:

1. The experimental results without organism should be confirmed using

a more sophisticated model system (e.g. the cross flow filter should
be replaced by a continuous centrifuge). Definitely, data should be
obtained at process conditions, i.e., at 60◦ C, pH 6, and in the presence
of substrate. It would also be interesting to study the effects of different
pH values and temperatures on the process behaviour. Other problems
 150

which could be addressed include the exsolution of carbon dioxide, the

accumulation of inerts in the solids phase, possible interactions between
ions and substrate, and the causes for the observed acceleration of
magnesium oxide dissolution.

2. The organism should be included in the experiments, and the validity of

the “performance assumptions” (attainable productivity and product
concentration) should be tested. Medium composition, cell density,
and several other fermentation conditions have to be optimized—for
instance, it has to be ensured that not too much oxygen is present in
the fermentor, since this would have a toxic effect on the organism.

3. CMA produced under realistic conditions should be recovered and its

deicing properties should be tested—since the material contains dis-
charged solids as well as substrate, it is not likely to have the same
properties as pure CMA.

If the lab scale results with the organism are still promising, the process
could be scaled up to pilot plant level. Chances are slim that an acetic acid
pulping plant will be built at this time because of high overcapacities in the
pulp and paper industry. However, should the need for a new plant arise,
this would be a very interesting alternative to a conventional kraft plant, and
CMA production could be tested at the pilot scale. Here in Wisconsin, an
almost ideal location for such a plant would be the Green Bay area, since
it has a long tradition in the manufacture of pulp and paper, large natural
supplies of dolomite, and lots of ice and snow to melt during the winter.
Appendix A

Original ICP Data

A.1 Equilibrium Experiments (Chapter 3)

The solution samples obtained from the equilibrium experiments described in
sections 3.2 and 3.3 were submitted to the University of Wisconsin-Madison
Soil & Plant Analysis Laboratory for the analysis of total soluble calcium
and magnesium concentrations by inductively coupled plasma emission spec-
troscopy (ICP). The original results as given in Table A.1 are on the ppm
scale (mass fractions for Ca and Mg). The mass fractions of H2 O and Ac−
could be determined from the known initial composition of the system, as-
suming that the filtered solution consists only of the four components H2 O,
Ac− , Ca2+ , and Mg2+ . From these mass fractions ξi the corresponding mole
fractions xi had to be calculated in order to determine the molar concentra-
tions Ci . This was done using the formula

M
xi = ξi (A.1)
Mi

151
 152

Table A.1: Original results of the ICP analyses on six filtered solution samples
from equilibrium experiments; source: Soil & Plant Analysis Laboratory,
University of Wisconsin-Madison.

Sample Mass Fraction (ppm)

Ca Mg

blank deionized water < 0.043 < 0.104

Ca(OH)2 – Mg(OH)2 – HAc (Recipe 1) 11075 1.790
type S dolime – HAc 9928 0.940
CaCO3 – MgCO3 – HAc (Recipe 3) 139.5 5144
Ca(OH)2 – Mg(OH)2 – CO2 – HAc 179.7 6504
dolomitic limestone – HAc 4683 2000

where Mi is the molecular weight of component i and

1
M=P −1 (A.2)
j Mj ξ j

is the molecular weight of the mixture. With the assumption that 1 l solution
contains 55.6 mol H2 O, the molar concentrations in Tables 3.2 and 3.4 were
obtained.
 153

A.2 CMA Production (Section 5.1.2)

The CMA solution samples obtained from the continuous dissolution exper-
iments described in section 5.1.2 were again submitted to the University of
Wisconsin-Madison Soil & Plant Analysis Laboratory for the analysis of to-
tal soluble calcium and magnesium concentrations by inductively coupled
plasma emission spectroscopy (ICP). The original results (mass fractions of
calcium and magnesium on the ppm scale) are given in Tables A.2 and A.3
for the CMA made from type S dolime and selectively calcined dolomite, re-
spectively. From these mass fractions, the Ca:Mg mole ratio could be easily
determined using the formula

xCa MMg ξCa

= , (A.3)
xMg MCa ξMg

where xi are the mole fractions, ξi the mass fractions, and Mi the molecular
weights. Table A.4 shows the detected ranges of concentration (minimum
and maximum) for several other elements that were present as impurities in
the product solution.
 154

Table A.2: Original results of the ICP analyses on CMA solution samples
made from type S dolime; source: Soil & Plant Analysis Laboratory, Univer-
sity of Wisconsin-Madison.

Time Mass Fractions Ca:Mg Mole Ratio

(h) (ppm)

1 758.1 225.6 2.04

2 1278 505.9 1.53
3 1600 754.0 1.29
4 2280 1033 1.34
5 2526 1246 1.23
6 2668 1502 1.08
7 2796 1628 1.04
8 3929 1841 1.29
9 4154 1900 1.32
11 4566 2160 1.28
14 4843 2313 1.27
19 5707 2753 1.26
24 6377 2891 1.34
29 6057 3034 1.21
34 6336 3102 1.24
39 6153 3091 1.21
44 6524 3129 1.26
52 6247 3127 1.21
 155

Table A.3: Original results of the ICP analyses on CMA solution samples
made from selectively calcined dolomite; source: Soil & Plant Analysis Lab-
oratory, University of Wisconsin-Madison.

Time Mass Fractions Ca:Mg Mole Ratio

(h) (ppm)

1 474.9 388.7 0.74

2 785.2 748.2 0.64
3 1004 1164 0.52
4 1177 1512 0.47
5 1426 1748 0.49
6 1443 2035 0.43
7 1505 2461 0.37
8 1484 2686 0.33
9 1535 3066 0.30
10 1525 3189 0.29
11 1623 3473 0.28
12 1577 3328 0.29
14 2141 4207 0.31
17 3136 3848 0.49
18 3525 3641 0.59
19 4083 3632 0.68
20 4575 3229 0.86
22 5083 2909 1.05
24 6112 2746 1.35
30 5091 3579 0.86
36 2946 4727 0.38
42 2553 5116 0.30
46 2326 4950 0.28
50 4231 3542 0.72
 156

Table A.4: Ranges of concentration detected for other elements that were
present as impurities in the CMA solution; source: Soil & Plant Analysis
Laboratory, University of Wisconsin-Madison.

Element Concentration Range (ppm)

type S dolime selectively
calc. dolomite

P < 0.217 < 0.217–0.316

K < 0.621 < 0.621–3.852
S 7.591–48.80 4.665–31.42
Zn < 0.010–0.089 < 0.010–0.086
B 0.073–0.269 0.098–0.373
Mn 0.076–0.265 0.051–0.895
Fe < 0.011–0.543 < 0.011–0.920
Cu < 0.025–0.120 < 0.025–0.327
Al < 0.352–4.937 1.096–8.680
Na < 0.611–1.601 1.977–5.832
Appendix B

Lime and Limestone Analyses

B.1 Type S Dolomitic Hydrated Lime

The type S dolime used in some of the experiments in sections 3.2 and 5.1.2
was supplied by The Western Lime & Cement Co., West Bend, Wisconsin.
The company also provided a data sheet with the results of some analyses.
In addition to particle size distribution and chemical composition (as given
in Tables B.1 and B.2), the neutralizing value of the lime was reported to be
165% of the calcium carbonate equivalent. Using the formula

xCa(OH)2 MMg(OH)2 ξCa(OH)2

= , (B.1)
xMg(OH)2 MCa(OH)2 ξMg(OH)2

where xi are the mole fractions, ξi the mass fractions, and Mi the molec-
ular weights, the mole ratio of Ca:Mg can easily be calculated—the value
found from the given data is 1.04:1, i.e., the material has almost the desired
composition.

157
 158

Table B.1: Sieve analysis of type S dolime; source: The Western Lime &
Cement Co.

Sieve Size Passing Specification

30 mesh 100.0% 100.0%

100 mesh 99.0% 95.0% min
200 mesh 96.0% 90.0% min

Table B.2: Chemical analysis of type S dolime; source: The Western Lime &
Cement Co.

Compound Mass Fraction (wt%)

calcium as CaO 42.40

magnesium as MgO 28.68
combined water 25.18
silica and insolubles 1.01
loss on ignition (CO2 , SO3 ) 0.77
free water 0.50
alumina as Al2 O3 0.42
iron as Fe2 O3 0.25

calcium hydroxide Ca(OH)2 54.7

magnesium hydroxide Mg(OH)2 41.6
 159

B.2 Dolomitic Limestone

The dolomitic limestone used in experiments in sections 3.3 and 5.1.2 was
also supplied by The Western Lime & Cement Co., West Bend, Wisconsin.
The stone had a slight yellow tint and seemed to be of sedimentary origin; it
was already finely crushed. A limestone analysis performed by the University
of Wisconsin-Madison Soil & Plant Analysis Laboratory confirmed that this
material was essentially pure dolomite. The mass fractions were found to be
22.3% for calcium and 13.1% for magnesium; the theoretical values for pure
dolomite are 21.7% and 13.2%, respectively. The neutralizing value of the
stone was reported to be 103.5% of the calcium carbonate equivalent. Again,
the mole ratio of Ca:Mg can be determined from the given mass fractions
using the formula
xCa MMg ξCa
= , (B.2)
xMg MCa ξMg
and the result obtained was 1.03:1 (essentially the same as for the type S
dolime).
Appendix C

Equipment List (Section 5.1.2)

The following equipment was used for the continuous dissolution experiment
described in section 5.1.2; brand names are given for identification purposes
only:

• glass/steel fermentor 3.7 dm3 (Bioengineering KLF 2000)

• fermentation control system (Valley Instrument Co.), consisting of

– support unit (Mod. 520MC)

– pH controller (Mod. 506M)

– temperature controller (Mod. 503M)

– speed controller (Mod. 508M)

– foam/pump controller (Mod. 517M)

• pH glass electrode (Ingold)

• heating element 800 W (Bioengineering 30355)

160
 161

• temperature sensor (Bioengineering 30266)

• cross flow filter capsule with 1.2 µm pores, 900 cm2 area (Gelman
Acroflux)

• variable speed peristaltic pumps (Cole-Parmer)

– slow pump (Mod. 7520-35) with two identical heads (Mod. 7013-
21)

– fast pump (Mod. 7520-25) with head (Mod. 7016-21)

• flexible tubing (Masterflex 6411-13 and 6402-16)

• automatic fraction collector (Gilson FC 203)

• slurry stirrer (Glas-Col S25)

Bibliography

[Bar76] Barton, P., and T. Vatanatham: Kinetics of limestone neu-

tralization of acid waters. Environmental Science and Technology
10 (3), pp. 262–266, March 1976.

[Ber74] Berner, R. A., and J. W. Morse: Dissolution kinetics of

calcium carbonate in sea water IV. Theory of calcite dissolution.
American Journal of Science 274, pp. 108–134, 1974.

[Bir52] Bircumshaw, L. L., and A. C. Riddiford: Transport con-

trol in heterogenous reactions. Quarterly Review of the Chemical
Society London 6, pp. 157–185, 1952.

[Bog85] Boguslavskii, A. N., V. L. Zelenko, E. D. Vertuzaev,

and G. E. Petrova: Kinetics of neutralization of technical alkyl-
benzenesulfonic acid with milk of lime. Journal of Applied Chem-
istry of the U.S.S.R. 58 (6), pp. 1224–1228, June 1985.

[Bol22] Bole, G. A., and J. B. Shaw: The caustic calcination of

dolomite and its use in sorrel cements. Journal of the American
Ceramics Society 5, pp. 817–822, 1922.

162
 163

[Boy80] Boynton, R. S.: Chemistry and Technology of Lime and Lime-

stone (2nd ed.). Wiley, New York, 1980.

[Bus82] Busenberg, E., and L. N. Plummer: The kinetics of dissolu-

tion of dolomite in CO2 – H2 O systems at 1.5 to 65◦ C and 0 to
1 atm PCO2 . American Journal of Science 282, pp. 45-78, January
1982.

[Bus83] Busche, R. M.: Recovering chemical products from dilute fer-

mentation broths. Biotechnology and Bioengineering Symposium
No. 13, pp. 597–615, 1985.

[Bus90] Busche, R. M.: Acetic acid from corn—economic trade-off be-

tween yield of Clostridium and concentration of Acetobacter. Corn
Utilization Conference III, Proceedings. National Corn Growers
Association, June 1990.

[Bre89] Brenan, K. E., S. L. Campbell, and L. R. Petzold:

Numerical Solution of Initial-Value Problems in Differential-
Algebraic Equations. North-Holland, New York, 1989.

[Cab87] Cabaniss, S. E.: TITRATOR—Chemical Equilibrium Specia-

tion (Version 2.2). User’s Manual. May 1987.

[Car32] Carlsmith, L. E., B. J. Boeckeler, and I. J. Krchma: Ap-

paratus for the Recovery of Ammonia from Fermentation Mashes.
United States Patent No. 1 885 003, October 1932.
 164

[Cas91] Casey, W. H.: On the relative dissolution rates of some ox-

ide and orthosilicate minerals. Journal of Colloid and Interface
Science 146 (2), pp. 586–589, October 1991.

[Cho84] Chou, L., and R. Wollast: Study of the weathering of albite

at room temperature and pressure with a fluidized bed reactor.
Geochimica et Cosmochimica Acta 48, pp. 2205–2217, 1984.

[Cho89] Chou, L., R. M. Garrels, and R. Wollast: Comparative

study of the kinetics and mechanisms of dissolution of carbonate
minerals. Chemical Geology 78, pp. 269–282, 1989.

[Cho91] Chollar, B. H., D. L. Smith, and J. A. Zenewitz: The

involvement of the Federal Highway Administration with cal-
cium magnesium acetate. In Wise, D. L., Y. A. Levendis,
and M. Metghalchi (ed.): Calcium Magnesium Acetate, An
Emerging Bulk Chemical for Environmental Applications. Indus-
trial Chemistry Library, Vol. 2. Elsevier, Amsterdam, 1991.

[Com90a] Compton, R. G., and P. R. Unwin: The dissolution of cal-

cite in aqueous solution at pH < 4: kinetics and mechanism.
Philosophical Transactions of the Royal Society London A 330,
pp. 1–45, February 1990.

[Com90b] Compton, R. G., and K. L. Pritchard: The dissolution of

calcite at pH > 7: kinetics and mechanism. Philosophical Trans-
actions of the Royal Society London A 330, pp. 47–70, February
1990.
 165

[Cro92] Cross, C. F., E. J. Beven, and J. F. V. Isaac: On the pro-

duction of acetic acid from carbohydrates. Journal of the Society
of Chemical Industry 2, pp. 966–969, December 1892.

[Dav62] Davies, C. W.: Ion Association. Butterworths, Washington,

D.C., 1962.

[DeH71] De Haas, G. G., and C. J. Lang: Non-Catalytic Process for

the Production of Cellulose from Lignocellulosic Materials Using
Acetic Acid. United States Patent No. 3 553 076, January 1971.

[Dun80] Dunn, S. A., and R. U. Schenk: Alternate Highway Deicing

Chemicals. Federal Highway Administration, Report No. FHWA-
RD-79-108, March 1980.

[Fan89] Fan, L. S.: Gas-Liquid-Solid Fluidization Engineering. Butter-

worths Series in Chemical Engineering, Boston, 1989.

[Gan83] Gancy, A. B.: Process of Making Two Uniform Grades of Cal-

cium Magnesium Acetate. United States Patent No. 4 389 323,
June 1983.

[Gan84a] Gancy, A. B.: Process of Making Calcium Magnesium Acetate

Deicing Agents and Product. United States Patent No. 4 444 672,
April 1984.

[Gan84b] Gancy, A. B.: Course-Particle Calcium Magnesium Acetate

Suitable for Roadway and Highway Deicing, and Process for Its
Manufacture. United States Patent No. 4 488 978, December 1984.
 166

[Gan86] Gancy, A. B.: Continuous Process for the Manufacture of

Calcium Magnesium Acetate Deicer. United States Patent
No. 4 606 836, August 1986.

[Gan87a] Gancy, A. B.: Low-Energy Process for the Manufacture of Cal-

cium Magnesium Acetate-Chloride Deicers and Freezing Point
Depressants. United States Patent No. 4 673 519, June 1987.

[Gan87b] Gancy, A. B.: Magnesium Calcium Acetate Products, and Pro-

cess for Their Manufacture. United States Patent No. 4 699 725,
October 1987.

[Gho85] Ghose, T. K., and A. Bhadra: Acetic Acid. In: Moo-Young,

M. (ed.): Comprehensive Biotechnology, Vol. 3, Pergamon Press,
New York, 1985.

[Gor84] Gorichev, I. G., and N. A. Kipriyanov: Regular kinetic

features of the dissolution of metal oxides in acidic media. Russian
Chemical Reviews 53 (11), pp. 1039–1061, 1984.

[Hoa45] Hoak, R. D., C. J. Lewis, and W. W. Hodge: Treatment

of spent pickling liquors with limestone and lime. Industrial and
Engineering Chemistry 37 (6), pp. 553–559, June 1945.

[Hud88] Hudson, L. R.: Calcium magnesium acetate (CMA) from low

grade biomass. In Klass, D. L. (ed.): Energy from Biomass and
Wastes XI, pp. 893–901. Institute of Gas Technology, Chicago,
1988.
 167

[Lin65] Linke, W. F.: Solubilities of Inorganic and Metal Organic Com-

pounds, Vol. 1–2. Van Nostrand, Princeton, N.J., 1958–1965.

[Lju85] Ljungdahl, L. G., L. H. Carreira, R. J. Garrison,

N. E. Rabek, and J. Wiegel: Comparison of three ther-
mophilic acetogenic bacteria for production of calcium magnesium
acetate. Biotechnology and Bioengineering Symposium No. 15,
pp. 207–223, 1985.

[Mac71] Macdonald, D. D., and D. Owen: The dissolution of magne-

sium oxide in dilute sulfuric acid. Canadian Journal of Chemistry
49, pp. 3375–3380, 1971.

[Mar89] Martell, A. E., and R. M. Smith: Critical Stability Con-

stants, Vol. 1–6. Plenum Press, New York, 1974–1989.

[Mar85] Marynowski, C. W., R. L. Boughton, D. Tusé, and

J. L. Jones: Fermentation as an advantageous route for the
production of an acetate salt for roadway deicing. Industrial En-
gineering Chemistry—Product Research and Development 24 (3),
pp. 457–465, 1985.

[Mer83] The Merck Index, An Encyclopedia of Chemicals, Drugs, and

Biologicals (10th ed.). Merck & Co., Inc., Rahway, N.J., 1983.

[Mor72] Morel, F., and J. Morgan: A numerical method for comput-

ing equilibria in aqueous chemical systems. Environmental Sci-
ence and Technology 6 (1), pp. 58–67, January 1972.
 168

[Mor83] Morel, F.: Principles of Aquatic Chemistry. Wiley, New York,

1983.

[Nim86] Nimz, H. H., and R. Casten: Chemical processing of lignocel-

lulosics. Holz als Roh- und Werkstoff 44, pp. 207–212, 1986.

[Par65] Parsons, W. A.: Chemical Treatment of Sewage and Industrial

Wastes. National Lime Association, Washington, D.C., 1965.

[Par90] Parekh, S. A., and M. Cheryan: Acetate Production from

glucose by Clostridium thermoaceticum. Process Biochemistry In-
ternational 25 (8), pp. 117–121, August 1990.

[Pea75] Pearson, F. H. and A. J. McDonnell: Use of crushed

limestone to neutralize acid wastes. Journal of the Environmental
Engineering Division, American Society of Civil Engineers 101
(EE1), pp. 139–158, February 1975.

[Plu76] Plummer, L. N., and T. M. L. Wigley,: The dissolution

of calcite in CO2 -saturated solutions at 25◦ C and 1 atmosphere
total pressure. Geochimica et Cosmochimica Acta 40, pp. 191–
202, 1978.

[Plu78] Plummer, L. N., T. M. L. Wigley, and D. L. Parkhurst:

The kinetics of calcite dissolution in CO2 -water systems at 5 to
60◦ C and 0.0 to 1.0 atm CO2 . American Journal of Science 278,
pp. 179–216, February 1978.

[Plu79] Plummer, L. N., D. L. Parkhurst, and T. M. L. Wigley:

Critical review of the kinetics of calcite dissolution and precip-
 169

itation. In Jenne, E. A. (ed.): Chemical Modeling in Aqueous

Systems. ACS Symposium Series, Vol. 93. American Chemical
Society, Washington, D.C., 1979.

[Ric83] Rickard, D., and E. L. Sjöberg: Mixed kinetic control of cal-

cite dissolution rates. American Journal of Science 283, pp. 815–
830, October 1983.

[Rie91] Riet, K., and J. Tramper: Basic Bioreactor Design. Marcel

Dekker, New York, 1991.

[Rip86] Rippie, W.: Noncorrosive De-Icing Composition and Method

Comprising Calcium and Magnesium Carboxylate. United States
Patent No. 4 588 512, May 1986.

[San84] Sandvig, R. L., W. A. Klemm, J. R. Gaines, and

R. W. Looyenga: De-Icing Chemicals and Their Preparation
from Polysaccharide Sources. United States Patent No. 4 430 240,
February 1984.

[Sch82a] Schwartz, R. D., and F. A. Keller, Jr.: Isolation of a strain

of Clostridium thermoaceticum capable of growth and acetic acid
production at pH 4.5. Applied and Environmental Microbiology
43 (1), pp. 117–123, January 1982.

[Sch82b] Schwartz, R. D., and F. A. Keller, Jr.: Acetic acid pro-

duction by Clostridium thermoaceticum in pH-controlled batch
fermentations at acidic pH. Applied and Environmental Microbi-
ology 43 (6), pp. 1385–1392, June 1982.
 170

[Sch91] Schenk, R. U.: Physical and chemical properties of Calcium

Magnesium Acetate. In Wise, D. L., Y. A. Levendis, and
M. Metghalchi (ed.): Calcium Magnesium Acetate, An Emerg-
ing Bulk Chemical for Environmental Applications. Industrial
Chemistry Library, Vol. 2. Elsevier, Amsterdam, 1991.

[Seg78] Segall, R. L., R. S. C. Smart, and P. S. Turner: Ionic ox-

ides: distinction between mechanisms and surface roughening ef-
fects in the dissolution of magnesium oxide. Journal of the Chem-
ical Society, Faraday Transactions I 74, pp. 2907–2912, 1978.

[Seg88] Segall, R. L., R. S. C. Smart, and P. S. Turner: Ox-

ide Surfaces in Solution. In Nowotny, J., and L. C. Dufour:
Surface and Near-Surface Chemistry of Oxide Materials. Materi-
als Science Monographs, Vol. 47, Elsevier, Amsterdam, 1988.

[Sha22] Shaw, J. B., and G. A. Bole: New developments in oxychloride

stucco and flooring. Jornal of the American Ceramics Society 5,
pp. 311–321, 1922.

[Sjö76] Sjöberg, E. L.: A fundamental equation for calcite dissolu-

tion kinetics. Geochimica et Cosmochimica Acta 40, pp. 441–447,
1976.

[Sjö84] Sjöberg, E. L., and D. Rickard: Calcite dissolution kinetics:

surface speciation and the origin of the variable pH dependence.
Chemical Geology 42, pp. 119–136, 1984.
 171

[Sta53] Stansel, T. B., and W. H. MacIntire: Steam catalysis in

calcinations of dolomite and limestone fines. Industrial and Engi-
neering Chemistry 45 (7), pp. 1548–1555, 1953.

[Stu81] Stumm, W., and J. J. Morgan: Aquatic Chemistry (2nd ed.).

Wiley, New York, 1981.

[Sua84] Suarez, D. L., and J. D. Wood: Simultaneous determination

of calcite surface area and content in soils. Soil Science Society of
America, Journal 48, pp. 1232–1235, 1984.

[Tod90] Todd, H. E., and D. L. Walters: Deicing Compositions Com-

prising Calcium Magnesium Acetate Double Salt, And Processes
for Their Production. United States Patent No. 4 913 831, 1990.

[Tra90] Trantolo, D. J., J. D. Gresser, D. C. Augenstein, and

D. L. Wise: The feasibility of a residue biomass bioconversion
process to prepare calcium magnesium acetate deicing salt. Re-
sources, Conservation and Recycling 4, pp. 215–232, 1990.

[Tra91] Trantolo, D. J., J. D. Gresser, D. C. Augenstein, and

D. L. Wise: Calcium magnesium acetate from the bioconver-
sion of residue biomass. In Wise, D. L., Y. A. Levendis,
and M. Metghalchi (ed.): Calcium Magnesium Acetate, An
Emerging Bulk Chemical for Environmental Applications. Indus-
trial Chemistry Library, Vol. 2. Elsevier, Amsterdam, 1991.

[Ulr84] Ulrich, G. D.: A Guide to Chemical Engineering Process De-

sign and Economics. Wiley, New York, 1984.
 172

[Ver69] Vermilyea, D. A.: The dissolution of MgO and Mg(OH)2 in

aqueous solutions. Journal of the Electrochemical Society: Elec-
trochemical Science 116 (9), September 1969.

[Wag78] Wagner, F. S.: Acetic Acid. In Kirk-Othmer: Encyclopedia

of Chemical Technology Vol. 1, Wiley, New York, 1978.

[Wan78] Wang, D. I. C., R. J. Fleishchaker, and G. Y. Wang:

A novel route to the production of acetic acid by fermentation.
AIChE Symposium Series 74 no. 181, pp. 105–110, 1978.

[Wan84] Wang, G. Y., and D. I. C. Wang: Elucidation of growth inhi-

bition and acetic acid production by Clostridium thermoaceticum.
Applied and Environmental Microbiology 47 (2), pp. 294–298,
February 1984.

[Wis88] Wise, D. L., and D. Augenstein: An evaluation of the biocon-

version of woody biomass to calcium magnesium acetate deicing
salt. Solar Energy 41 (5), pp. 453–463, 1988.

[Wie91] Wiegel, J., L. H. Carreira, R. J. Garrison,

N. E. Rabek, and L. G. Ljungdahl: Calcium magne-
sium acetate (CMA) manufacture from glucose by fermentation
with thermophilic homoacetogenic bacteria. In Wise, D. L.,
Y. A. Levendis, and M. Metghalchi (ed.): Calcium Mag-
nesium Acetate, An Emerging Bulk Chemical for Environmental
Applications. Industrial Chemistry Library, Vol. 2. Elsevier, Am-
sterdam, 1991.
 173

[Wis91] Wise, D. L., Y. A. Levendis, and M. Metghalchi (ed.):

Calcium Magnesium Acetate, An Emerging Bulk Chemical for
Environmental Applications. Industrial Chemistry Library, Vol. 2.
Elsevier, Amsterdam, 1991.

[Yan92] Yang, S. T., I. C. Tang, and H. Zhu: A novel fermenta-

tion process for Calcium Magnesium Acetate (CMA) produc-
tion from cheese whey. Applied Biochemistry and Biotechnology
34/35, pp. 569–583, 1992.