APPLIED AND ENVIRONMENTAL MICROBIOLOGY, Mar. 2007, p. 1742–1752 Vol. 73, No.

6
0099-2240/07/$08.00⫹0 doi:10.1128/AEM.01521-06
Copyright © 2007, American Society for Microbiology. All Rights Reserved.

Improved Experimental and Computational Methodology for Determining

the Kinetic Equation and the Extant Kinetic Constants of Fe(II)
Oxidation by Acidithiobacillus ferrooxidans䌤
Sharon Molchanov, Yuri Gendel, Ilya Ioslvich, and Ori Lahav*
Faculty of Civil and Environmental Engineering, Technion, Haifa 32000, Israel
Received 1 July 2006/Accepted 6 January 2007

The variety of kinetics expressions encountered in the literature and the unreasonably broad range of values
reported for the kinetics constants of Acidithiobacillus ferrooxidans underscore the need for a unifying exper-
imental procedure and for the development of a reliable kinetics equation. Following an extensive and critical
review of reported experimental techniques, a method based on batch pH-controlled kinetics experiments
lasting less than one doubling time was developed for the determination of extant kinetics constants. The Fe(II)
concentration in the experiments was measured by a method insensitive to Fe(III) interference. Kinetics
parameters were determined by nonlinear fitting of the integrated form of the Monod equation to yield a KS of 31 ⴞ
4 mg Fe2ⴙ literⴚ1 (mean ⴞ standard deviation), a KP of 139 ⴞ 20 mg Fe3ⴙ literⴚ1, and a ␮max of 0.082 ⴞ 0.002
hⴚ1. The corresponding kinetics equation was as follows:
dS
冉 0.082
冊 S䡠X

冉 冊
⫽ ⫺
dt 2.3 䡠 107 P0 ⫹ S0 ⫺ S
31 1 ⫹ ⫹S
139
where S represents the Fe(II) concentration in mg literⴚ1, P0 represents the initial Fe(III) concentration in mg
literⴚ1, X represents the suspended bacterial cell concentration in cells mlⴚ1, and t represents time in hours.
The measured data fit this equation exceptionally well, with an R2 of >0.99. Fe(III) inhibition was found to be
of a competitive nature. Contrary to previous reports, the results show that the concentration of Acidithioba-
cillus ferrooxidans cells has no affect on the kinetics constants. The kinetics equation can be considered
applicable only to A. ferrooxidans cells grown under environmental conditions similar to those of the inoculum
tested in the study. In contrast, the experimental and computational procedure is completely general and can
be applied to A. ferrooxidans irrespective of the culture history.

Acidithiobacillus ferrooxidans is a chemolithotrophic aerobic
bacterium capable of oxidizing both ferrous iron to ferric iron
and sulfide to sulfate, utilizing the energy derived from the oxi-
dation to support carbon dioxide fixation and cell growth (17).
These characteristics are particularly fitting for the processes
of mineral bioleaching, desulfurization of sour and flue gases,
and the treatment of Fe(II)-containing acidic solutions,
namely, acid mine drainage streams (28, 33).
The importance of these applications is the main incentive
behind the intensive studies performed on the kinetics related
to biological Fe(II) oxidation. However, despite a considerable
effort in the last decades, examination of published kinetics
constants reveals large discrepancies, of up to 3 orders of
magnitude, between data sets derived from different sources.
The large variations in published results may perhaps be ex-
plained by the variety of experimental conditions employed
(temperature, pH, substrate concentration, etc.), differences in
the experimental techniques used for the determination of the
kinetics parameters (batch, continuous or “initial rate” exper-
iments), the use of different mathematical models for kinetics
data interpretation, the analytical techniques employed, the
* Corresponding author. Mailing address: Faculty of Civil and En-
vironmental Engineering, Technion, Haifa 32000, Israel. Phone: 972 4
9292191. Fax: 972 4 9228898. E-mail: agori@tx.technion.ac.il.
䌤
Published ahead of print on 19 January 2007.
parameter that was monitored [Fe(II) concentration, Fe(III)
concentration, dissolved oxygen concentration, or A. ferrooxi-
dans cell concentration], and also perhaps the fact that differ-
ent A. ferrooxidans strains were examined. Nevertheless, as-
suming that A. ferrooxidans strains do not differ significantly
from each other and that, therefore, the range of kinetics
constants for Fe(II) oxidation obtained in different studies
should at least be of the same order of magnitude, it is not
unlikely that at least a few of the techniques used for deter-
mining these kinetics constants in some of the previously pub-
lished studies were inadequate. Moreover, it is clear from the
diversity of experimental procedures proposed in the literature
that a unified methodology for the proper determination of A.
ferrooxidans kinetics constants is needed.
As a result of the large differences between kinetics data
reported in the various sources, considerable difficulty exists in
predicting a priori the Fe(II) oxidation rate under operational
conditions in which both the substrate (Fe2⫹) and the product
(Fe3⫹) are present at various concentrations. Such a situation
is not uncommon in processes that utilize A. ferrooxidans for
Fe(II) oxidation. For example, processes aimed at H2S re-
moval from biogas in which A. ferrooxidans is used for Fe3⫹
bioregeneration are often operated at a relatively high Fe(III)
concentration (several g Fe3⫹ per liter) along with a varying
Fe(II) concentration, which is a function of both the H2S(g)

1742
VOL. 73, 2007
load into the system and the rate of the concurrent biological
Fe(II) oxidation.
Intrinsic versus extant kinetics data. In a widely cited pub-
lication, Grady et al. (14) coined a commonly used nomencla-
ture for the interpretation of kinetics data results. Under this
terminology, kinetics constants are divided into “intrinsic” and
“extant.” Intrinsic kinetics data are defined as “kinetic con-
stants that represent the maximum capability of the members
of the microbial community with the fastest growth kinetics.”
In a long kinetics experiment (multiple duplication times), the
composition of the microbial community changes due to an
enrichment of the fastest-growing species at the expense of
slower-growing species, resulting in “intrinsic” kinetics data
(14). In contrast, when a short batch experiment is performed,
this leads to the determination of “extant” kinetics constants,
i.e., “only limited changes in the protein synthesizing system
and synthesis of new enzymes can take place before the sub-
strate is depleted, and changes in the physiological state will be
minimal.” For practical and design purposes both Grady et al.
(14) and Chudova et al. (7) recommended using short batch
kinetics tests for the determination of the kinetics data set,
because they provide an “insight into the immediate response
of the continuous culture from which the cells are obtained”
(7).
Accordingly, the goals of the present work were twofold: to
determine the appropriate kinetics equation and a credible
range for A. ferrooxidans extant kinetics constants and to es-
tablish a rigorous empirical/computational procedure to deter-
mine the extant kinetics constants for A. ferrooxidans while
minimizing the common errors encountered in this respect in
the literature. In order to establish the most suitable experi-
mental technique, a comprehensive literature survey was per-
formed.
Literature survey of kinetics data relating to A. ferrooxidans.
Table 1 summarizes the Monod-based kinetics terms found in
the literature to describe the growth rate and Fe(II) oxidation
rate of suspended cultures of A. ferrooxidans. The models differ
from each other by the types of coefficients used, the experi-
mental system and environmental conditions under which they
were determined, the computational methods used to interpret
the data, the raw parameters monitored, and the analytical
methods used to determine them. In the following paragraphs
each of these parameters is critically reviewed, and associated
advantages/disadvantages are discussed.
Monod terms suggested for describing the kinetics of Fe(II)
oxidation by A. ferrooxidans. Monod’s kinetics equation is the
most widely used expression to represent both the growth rate
of A. ferrooxidans and the rate at which it oxidizes the substrate
[Fe(II)]. The simplest equation, which describes the growth
rate of A. ferrooxidans solely as a function of the ferrous iron
concentration (41, 44), is shown in equation 1.
␮max[Fe2⫹]
␮⫽ (1)
KS ⫹ 关Fe2⫹兴
where ␮ is the specific growth rate [i.e., (dX/dt)/X, where X ⫽
the cell concentration], ␮max is the maximum specific growth
rate, [Fe2⫹] is the ferrous iron concentration, and KS⫺1 is the
substrate affinity constant.
The reliability of equation 1 for predicting the Fe(II) oxida-
KINETICS OF Fe(II) OXIDATION BY A. FERROOXIDANS 1743
tion rate is doubtful, since it has been extensively and repeat-
edly reported that both the Fe(II) oxidation rate and A. fer-
rooxidans cell growth are sensitive to inhibition caused by both
the reaction product [Fe(III)] and also, above a certain Fe(II)
concentration, by the substrate itself. To overcome this prob-
lem, more complex Monod terms (Table 1) have been intro-
duced to describe the inhibition effect of the substrate (Fe2⫹),
product (Fe3⫹), bacterial cells, and heavy metal concentrations
on the rate of Fe(II) oxidation (or the corresponding dissolved
oxygen concentration reduction) and bacterial growth.
With regard to the inhibition caused by the Fe(III) concen-
tration on the rate of Fe(II) oxidation, both “competitive” and
“noncompetitive” models have been suggested. Noncompeti-
tive inhibition to Fe(II) oxidation by Fe(III) has been reported
by Jones and Kelly (19) and by Nemati and Webb (31); how-
ever, most of the works published to date on A. ferrooxidans
have reported competitive product inhibition rather than non-
competitive (5, 12, 13, 15, 20, 23, 27, 34).
Substrate inhibition at ferrous iron concentrations higher
than 2 to 3 g Fe2⫹ liter⫺1 was reported previously (1). Con-
versely, Jones and Kelly (19) reported an inhibition effect only
at Fe(II) concentrations above 5 g Fe2⫹ liter⫺1. Another type
of competitive inhibition to Fe(II) oxidation, reported to be
caused by a high concentration of mine-isolated bacterial cells
(but not by laboratory-cultivated cells), was observed by Suzuki
et al. (47). Those authors hypothesized that A. ferrooxidans
cells have the ability to compete with Fe2⫹- for Fe2⫹-binding
sites of neighboring A. ferrooxidans cells. They explained this
observation by the unusual surface properties of A. ferrooxi-
dans cells, which are responsible for the well-known ability of
the bacterium to adsorb to solid surfaces. Subsequent to this
observation, Nemati and Webb (29) reported that a linear
dependency exists between the value of KS and the biomass
concentration, corroborating thus the observation made by
Suzuki et al. (47). These authors also reported that the inhi-
bition effect of A. ferrooxidans concentration on the Fe(II)
oxidation rate was not restricted to mine-isolated strains.
Other kinetics models that are not based on the Monod
model usually describe only particular cases, such as zero-
order kinetics at high Fe(II) concentrations (35) or first-order
kinetics at low Fe(II) concentrations (38), and will not be
discussed further in this paper.
Assessment of the experimental techniques used for A.
ferrooxidans kinetics parameters determination. Three basic
techniques have been used in the studies described in Table 1
to determine the kinetics parameters of A. ferrooxidans: initial
rate measurements, continuous culture, and batch culture.
In the technique termed “initial rate measurements” (20, 29,
30, 34), the initial biomass concentration and the duration of
the batch experiments are chosen in such a way that the oxi-
dation rate of the substrate is negligible relative to the sub-
strate concentration and, thus, the substrate concentration can
be considered constant. Cell growth rate and/or Fe(II) oxida-
tion rate are subsequently measured at various initial substrate
concentrations, and each experiment constitutes a point on the
kinetics graph. Kinetics parameters are either derived from the
linear form of Monod’s equation (20, 34) or by using nonlinear
regression techniques (29, 30). The main drawback of the ini-
tial rate measurement technique is that the assumption of
constant substrate concentration is bound to be incorrect at
 TABLE 1. Monod expressions and kinetics constants reported for suspended A. ferrooxidans cultures 1744
Expt. Comput. Expt. Monitored
Reference Kinetics model Kinetics constantsa Comments
conditions method system parameter

19 ␮max关Fe2⫹兴 ␮max, 1.25 h⫺1; KS, 0.048 pH 1.6, 30°C Linweaver-Burk Chemostat 关Fe(II)兴 Competitive inhibition by
␮ ⫽ g liter⫺1; KP, 0.06–0.11 Fe3⫹
关Fe3⫹兴
KS 1 ⫹ ⫹ 关Fe2⫹兴 g liter⫺1
冉 KP 冊
19 ␮max关Fe2⫹兴 ␮max, 1.78, 1.33 h⫺1; KS, pH 1.6, 30°C Linweaver-Burk Chemostat 关Fe(II)兴 Noncompetitive
␮ ⫽ 0.04, 0.13 g liter⫺1; KP, inhibition by Fe3⫹
关Fe3⫹兴
1⫹ 共KS ⫹ 关Fe2⫹兴兲 0.06, 0.15 g liter⫺1
冉 KP 冊
23 ␮max关Fe2⫹兴 ␮max, 0.11 h⫺1; KS, 0.05 g pH0 1.8, 35°C Linweaver-Burk Batch 关Fe(II)兴 Competitive inhibition by
␮ ⫽ liter⫺1; KP, 0.44 g Fe3⫹, no pH control
关Fe3⫹兴
KS 1 ⫹ ⫹ 关Fe2⫹兴 liter⫺1
MOLCHANOV ET AL.

冉 KP 冊
2⫹
23 ␮max关Fe 兴 ␮max, 0.12 h⫺1; KS, 0.09 g pH 1.8, 35°C Linweaver-Burk Chemostat 关Fe(II)兴 Competitive inhibition by
␮ ⫽ liter⫺1; KP, 2.32 g Fe3⫹
关Fe3⫹兴
KS 1 ⫹ ⫹ 关Fe2⫹兴 liter⫺1
冉 KP 冊
47 d关O2兴 qmax关Fe2⫹兴X qmax (nmol O2 min⫺1 mg pH 2.3, 25°C Linweaver-Burk Initial rates Oxygen Competitive inhibition by
⫽ cell⫺1), (a) 88 and 143, consumption A. ferrooxidans cells
dt X
KS 1 ⫹ ⫹ 关Fe2⫹兴 (b) 125 and 100; KS (g rate
冉 冊 Ki
liter⫺1) (a) 0.02 and
0.04, (b) 0.01 and 0.02;
Ki (mg cell 䡠 ml⫺1), (a)
not determined, (b)
0.33 and 0.11
24 d关O2兴 qmax关Fe2⫹兴X qmax, 200 nmol O2 min⫺1 pH 2.3, 25°C Linweaver-Burk Initial rates Oxygen Synergistic competitive
⫽ mg cell⫺1; KS, 0.004 g consumption inhibition by A.
dt X 关Fe3⫹兴 X关Fe3⫹兴
KS 1⫹ ⫹ ⫹ ⫹ 关Fe2⫹兴 liter⫺1; Ki, 0.135 (mg rate ferrooxidans cells and
冉 Ki KP ␣KiKP 冊 cell) ml⫺1; KP, 0.036 g Fe3⫹
liter⫺1; ␣, 5.0
32 ␮max关Fe2⫹兴 ␮max, 0.23 h⫺1; KS(KP, pH 1.8, 18–37°C Nonlinear Chemostat 关Fe(II)兴 Competitive inhibition by
␮ ⫽ 0.94; Ks, 12.0 g liter⫺1 regression Fe3⫹; Fe2⫹ inhibition
关Fe3⫹兴 关Fe2⫹兴2
⬘KS 1⫹ ⫹ 关Fe2⫹兴 ⫹
冉 KP 冊 Ksi
2⫹ 2⫹
34 d关Fe 兴 Vmax关Fe 兴X KS (g liter⫺1), (a) 0.037, pH 2.0, 27°C Linear Initial rates 关Fe(II)兴 Competitive inhibition by
⫽ (b) 0.028; KP (g liter⫺1) regression Fe3⫹; strains of A.
dt 关Fe3⫹兴
KS 1⫹ ⫹ 关Fe2⫹兴 (a) 0.35, (b) 0.4 ferrooxidans used were
冉 KP 冊 ATCC 13598 (a) and
ATCC 13661 (b)
12 ␮max关Fe2⫹兴 ␮max, 0.14 h⫺1; KS, 0.94 g pH0 2.0, 30°C Nonlinear Batch Biomass concn Competitive inhibition by
␮ ⫽ liter⫺1; KP, 0.315 g regression Fe3⫹; no pH control;
关Fe3⫹兴
KS 1⫹ ⫹ 关Fe2⫹兴 liter⫺1 关Fe2⫹兴 at end around
冉 KP 冊 0.5 g liter⫺1
2⫹
29 d关Fe 兴 K0e⫺E/RTX关Fe2⫹兴 K0, 6,438 g liter⫺1 h⫺1 pH 2.0, 20–35°C Nonlinear Initial rates Redox Competitive inhibition by
⫽ (cells ml)⫺1; KS, 0.0672 regression A. ferrooxidans cells;
dt X X 关Fe2⫹兴2
KS 1⫹ ⫹ 关Fe2⫹兴 ⫹ 1 ⫺ g liter⫺1; Ki, 2.68 ⫻ 107 Fe2⫹ inhibition
冉 冊 Ki 冉 冊␤ ␣
cells ml⫺1; E, 68.4 kJ
mol⫺1; ␣, 26.1 g liter⫺1;
␤, 7.8 ⫻ 108 cells ml⫺1
15 ␮max关Fe2⫹兴 ␮max, 0.16 h⫺1; KS, 0.073 pH 1.8, 35°C Nonlinear Redox Redox Competitive inhibition by
␮ ⫽ g liter⫺1; KP, 0.78 g regression controlled Fe3⫹, As3⫹ inhibition
关Fe3⫹兴 关As3⫹兴
KS 1⫹ ⫹ 关Fe2⫹兴 ⫹ liter⫺1; Ka, 21.75 g reactor
冉 KP 冊 Ka
liter⫺1
31 d关Fe2⫹兴 K0⬘e⫺E/RTX关Fe2⫹兴 K0⬘, 3,077 g liter⫺1 h⫺1 pH 2.0, 35°C linear Initial rates Redox Noncompetitive
⫽ (cells ml)⫺1; KS, 0.258 g regression inhibition by Fe3⫹
dt 关Fe3⫹兴
1⫹ 共KS ⫹ 关Fe2⫹兴兲 liter⫺1; KP, 2.06 g
冉 KP 冊 liter⫺1; E, 68.4 kJ
mol⫺1
APPL. ENVIRON. MICROBIOL.
VOL. 73, 2007 KINETICS OF Fe(II) OXIDATION BY A. FERROOXIDANS 1745

low initial substrate concentrations, especially when a relatively

␮max, maximum specific growth rate; Vmax; maximum specific ferrous iron oxidation rate; qmax, maximum specific oxygen consumption rate. The substrate affinity constant, KS, and product inhibition constant, KP,
cannot be evaluated on a common basis but must be considered in relation to the specific Monod terms. Models for which two sets (a and b) of constants are reported included a laboratory strain (a) and a mine-isolated
Competitive inhibition by

Competitive inhibition by

Competitive inhibition by

Competitive inhibition by

ferrooxidans used were

ATCC 23270 (a) and
Fe3⫹, including cells’

Fe3⫹, including cells’

high biomass concentration is applied, resulting in high Fe(II)

Fe3⫹; strains of A.
requirements; pH

requirements; pH
oxidation rates. This drawback is particularly problematic
when the value of the substrate affinity constant (KS) is low, as
maintenance

maintenance
controlled

controlled

T23-3 (b)
is apparently the case with A. ferrooxidans. Apart from this
difficulty, this technique often neglects the possible lag phase in

Fe3⫹
the growth of A. ferrooxidans, which frequently occurs follow-
ing the introduction of a cell suspension into an oxidation cell.
Such neglect may lead to a further error.
A technique based on steady-state chemostat culture was
analysis of

analysis of

used by Jones and Kelly (19), Nikolov and Karamanev (32),

关Fe(II)兴

关Fe(II)兴
Off-gas

Off-gas

and Gomez and Cantero (13). The biomass concentration in a

O2

O2

continuous flow completely mixed bioreactor can be described

by the following equations (11):
Initial rates
Chemostat

Chemostat

dX
⫽ 共␮ ⫺ D兲X (2)
Batch

dt

dS
⫽ D共S0 ⫺ S兲 ⫺ ␮X/Y (3)
dt
regression

regression

regression

regression
Nonlinear

Nonlinear

Nonlinear

where D (h⫺1) is the dilution rate, defined as D ⫽ F/V, which

Linear

is the ratio between the influent flow rate F (in liters h⫺1) and
the volume of the reactor V (in liters), and S0 is the influent
substrate concentration (in mg liter⫺1).
From equations 2 and 3, it follows that at steady state (i.e.,
pH 1.8, 30°C

pH 1.8, 30°C

pH 1.8, 30°C

pH 1.5, 30°C

dX/dt ⫽ 0 and dS/dt ⫽ 0), ␮ ⫽ D and X* ⫽ Y(S0 ⫺ S*), where

X* and S* are the biomass and substrate concentrations in the
reactor at steady state. By running the reactor at different
dilution rates, the corresponding S* values may be measured,
the dependency of ␮ on S* can be obtained, and the kinetics
(a) 0.279, (b) 0.268; KP⬘

(b) 0.352; n, (a) 2.4, (b)

␮max, 0.22 h⫺1; KS, 0.92 g

3.8 ⫻ 10⫺15, (b) 2.5 ⫻

, 0.052 C-mol

, 0.051 C-mol

constants may be derived from the Monod equation. Such an

Vmax 关kg (cell h)⫺1兴, (a)
KS/KP, 0.08; ␮max, 0.096

KS/KP, 0.04; ␮max, 0.096

(mol O2)⫺1; mo, 0.24

10⫺15; KS (g liter⫺1),

(g liter⫺1), (a) 0.061,

mol O2 (C-mol h)⫺1

mol O2 (C-mol h)⫺1

(mol O2)⫺1; mo, 0.1

approach, however, may and frequently does encounter serious

liter⫺1; KP, 4.38 g

technical problems associated with maintaining a biological

reactor at a constant low substrate concentration. This is par-
h⫺1; Ymax

h⫺1; Ymax
ox

ox

ticularly problematic when the bacteria in question have a high

liter⫺1

affinity towards a limiting substrate, or in other words, a low KS

1.9

value (37). When applying the continuous culture technique,

there is also a need to eliminate apparatus-related artifacts,
strain (b); see the comments for this model for additional strain information.

such as nonperfect mixing and fluctuations in nutrient medium

supply. However, the greatest objection to this technique is the
关Fe2⫹兴 ⫺ 关Fe2⫹兴t KP 关Fe2⫹兴 ⫺ 关Fe2⫹兴 t
关Fe2⫹兴 ⫺ 关Fe2⫹兴t KP 关Fe2⫹兴 ⫺ 关Fe2⫹兴 t

fact that bacteria tend to alter their kinetics properties under

关Fe3⫹兴
关Fe3⫹兴

the different steady-state conditions applied in a chemostat-

type kinetics experiment (i.e., each point in the kinetics graph
␮max ⫹ 共m0 䡠 Ymax 兲

is obtained under a different set of operational conditions)

ox
max
␮max ⫹ m0 䡠 Yox

(37).
⫹ 䡠

⫹ 关Fe2⫹兴
KS

KS

⫹ 关Fe2⫹兴

A technique for estimating kinetics constants from the

⫹

␮max关Fe2⫹兴关Fe3⫹兴

growth kinetics of batch cultures was employed by Liu et al.

Vmax关Fe2⫹兴X

冊
冊

(23), Gomez et al. (12), and Boon et al. (4). In this technique,
关Fe3⫹兴n
关Fe3⫹兴

KP⬘
KS

KS

KP

cells harvested during the logarithmic growth phase are intro-

max

max
⫺ m0 䡠 Yox

⫺ m0 䡠 Yox

duced into a fresh medium with a known initial Fe(II) concen-

1⫹

1⫹

tration. The substrate (Fe2⫹) or the biomass concentration are

冉
冉
1⫹

1⫹

KS

KS

measured as a function of time until Fe(II) is completely oxi-

⫽

⫽
␮ ⫽

␮ ⫽

dized. A chosen kinetics model is subsequently fit (by either a

d关Fe 兴

d关Fe 兴
2⫹

2⫹

linear or a nonlinear curve fitting technique) to the measured

dt

dt

Fe(II) concentrations to determine the kinetics constants. The

main disadvantage that has been associated with this technique
is that experimental conditions change continuously during a
13

20

relatively long batch experiment. As mentioned above, changes

4

in community structure and physiological adaptation of the

1746 MOLCHANOV ET AL.
bacteria to the changing environmental conditions during the
experiment have been reported to significantly affect the values
of the kinetics constants (14). If the duration of the batch
experiment is sufficiently long to allow for several cell divisions,
a change in physiological state is expected to occur. Moreover,
the composition of the microbial community would change due
to an enrichment of the fastest-growing species at the expense
of slower-growing species. The kinetics measured under such
conditions will represent the maximum capability of the mem-
bers of the microbial community with the fastest growth kinet-
ics (intrinsic kinetics) rather than that of the initial community.
However, this problem mainly affects mixed cultures and cul-
tures that are characterized by a high cell yield (such as het-
erotrophic, aerobic populations). Although the problem tends
to be less troublesome in homogenous autotrophic communi-
ties, there is still a need to manage the duration of the batch
experiment so that population multiplication does not occur
and an extant, rather that an intrinsic, kinetics data set is
determined.
Batch tests: linear versus nonlinear data interpretation and
differential versus integral curve-fitting techniques. In some of
the batch culture and initial rate experiments reported in the
literature and shown in Table 1, the interpretation of observed
data was based on the linear form of Monod’s equation (Line-
weaver-Burk’s technique) (23, 34, 41). However, the reliability
of the linearization method has been seriously and repeatedly
criticized (3, 39). It has been shown that this technique suffers
from inherent systematic errors, which can be explained as
follows. Equation 4 shows Monod’s equation to which an ex-
plicit error term has been added.
␮max[Fe2⫹]
␮⫽ ⫹e (4)
KS ⫹ 关Fe2⫹兴
where e is a value that represents a deviation from the theo-
retical model rather than a true error.
Equation 4 is clearly not linear with respect to [Fe2⫹]. Al-
though the original Monod equation can be easily transformed
into a straight line equation when it is written without the error
term (e), this cannot be done when e is included. If an error
term is included after linearization, as shown in equation 5,
then the result is not a true transformation of equation 4, elin
is not the same as e, and minimizing the linear sum of squares
(SSlin ⫽ ⌺elin
2
does not provide the same parameter values as
minimization of the true sum of squares, SS (8).
1
␮
⫽ 冉
KS
⫻
1
␮max 关Fe2⫹兴
⫹
1
冊
␮max
⫹ elin (5)
Because of the inherent errors associated with the linear re-
gression technique, the more recent works that have addressed
bacterial kinetics determinations have made use of nonlinear
regression analysis to interpret kinetics data. This technique is
particularly convenient when the monitored parameter reflects
the oxidation rate itself, for example, when the oxygen con-
sumption rate is measured. However, usually the rate is not
directly measured but rather calculated by differentiating the
measured parameter. As a result of numerical differentiation,
significant errors may be introduced due to even small fluctu-
ations in the measured parameter. To overcome this problem,
an integrated form of Monod’s equation has been introduced
APPL. ENVIRON. MICROBIOL.
for estimating the kinetics parameters from a single-substrate
depletion curve (40, 42, 43). This technique, which appears to
be the most appropriate from the mathematical standpoint,
has been applied to determine the kinetics constants of various
bacterial populations, such as a methanogenic mixed culture
utilizing acetate (44) and Sphingomonas chlorophenolica grown
on pentachlorophenol (9). However, to the best of the writers’
knowledge, no attempt has been made to obtain the values of
the kinetics constants of A. ferrooxidans by applying a nonlinear
fitting technique to the integrated form of Monod’s equation.
Choosing the measured experimental parameter. Since a
direct measurement of the biomass concentration is cumber-
some and has a low accuracy, most of the published kinetics
studies rely on the measurement of indirect parameters to
represent A. ferrooxidans oxidation activity and growth. Such
an indirect approach may include measurement of oxygen up-
take rate, measurement of the change in Fe(II) and Fe(III)
concentrations with time, or the change with time of the redox
potential of the solution. Another option may be to monitor
the change in solution turbidity, which has been shown to be
correlated with A. ferrooxidans cell concentration (35). The
choice between the different monitoring parameter(s), along
with the reported use of diverse analytical techniques, is prob-
ably one of the main reasons for the large variation between
the reported kinetics constants. For example, the extremely
high KS value (0.94 g liter⫺1) reported by Gomez et al. (12)
appears to have resulted from the use of the standard o-phen-
anthroline method for the determination of Fe(II) concentra-
tion in the presence of a high Fe(III) concentration. In this
regard, Herrera et al. (16) showed that the error in measured
Fe(II) concentration is 14% when Fe(II) constitutes 10% of
the total iron concentration and may reach 150% when the
Fe(II) is 0.5% of the total iron concentration. Moreover, those
authors found that the interference of the Fe(III) concentra-
tion in the o-phenanthroline analysis is due to direct color
formation caused by the reaction of Fe(III) with o-phenanthro-
line and that the nature of the response of o-phenanthroline to
Fe(III) is nonlinear. Therefore, correcting the measurements
by comparing the absorbance to a blank to which no o-phen-
anthroline has been added, as was done, for example, by Boon
et al. (4), is clearly incorrect. In another study, Nyavor et al.
(34) did not report at all on the Fe(II) analysis method that was
used, a fact which tends to detract from the reliability of the
results. In order to use the o-phenanthroline method for fer-
rous iron analysis, one needs to overcome the interference
caused by the presence of Fe(III). One way to do this is by
adding a Fe(III) complexing agent. One of the well-known
chelating agents is nitrilotriacetic acid (10); however, its appli-
cation is cumbersome. Herrera et al. (16) proposed the use of
sodium fluoride as a Fe(III) complexing agent. The procedure
is simple and allows an accurate Fe2⫹ determination (relative
errors ranging from 0.0% to 2.4%) at Fe(II) concentrations of
ⱖ5% of the total iron concentration. When Fe(II) amounts to
0.5% of the total iron concentration, the relative error is
around 12%.
Another problematic measurement technique that has been
proposed for A. ferrooxidans kinetics constant determinations
is based on redox readings. Pesic et al. (38) suggested using the
redox potential of the solution as a measure of the change in
the ratio between Fe(II) and Fe(III) concentrations with time
VOL. 73, 2007
and thus an indirect measure of the Fe(II) oxidation rate. The
same method was later used by Nemati and Webb (29, 31) and
Harvey and Crundwell (15). However, it appears that this
technique may also lead to erroneous results: Nernst’s equa-
tion is used in this method for calculating the Fe(II) concen-
tration from measured redox potentials and a known total iron
concentration:
冉 冊
2.3RT ␣Fe3⫹
E ⫽ E0 ⫹ log (6)
nF ␣Fe2⫹
Where, ␣Fe3⫹ and ␣Fe2⫹ are Fe(III) and Fe(II) activities, F is
Faraday’s constant (in kJ V⫺1 equiv⫺1), n is the number of
electrons transferred per molecule, E0 is the standard potential
of the Fe3⫹/Fe2⫹ couple (in V), and E is the potential of the
solution (in V).
However, the redox potential reading depends not only on
the ratio between the dissolved Fe(II) and Fe(III) in solution
but also on the total iron concentration and on the ionic
strength of the solution [which is itself dependent on the ratio
between the Fe(III) and Fe(II) concentrations]. Because of the
logarithmic scale of the abscissa used in forming the linear
calibration curve in this approach, even an apparent small
difference between calibration curves may cause a large error
in the calculation of the Fe(II) concentration. To overcome
this problem, redox calibration curves must be made for each
particular total iron concentration used in the kinetics exper-
iments. This was not reported to be carried out in the studies
quoted above, a fact which tends to detract from the reliability
of their results.
Experimental conditions chosen for determining the extant
kinetics constants of A. ferrooxidans. Based on the discussion
thus far and in order to overcome the variety of technical,
analytical, and mathematical problems associated with the
studies reported in the literature, we opted in the present work
to conduct the A. ferrooxidans kinetics experiments under the
following experimental conditions: (i) batch pH-controlled ex-
periments lasting less than one doubling time; (ii) measure-
ments of the change in Fe(II) concentration with time by
applying the o-phenanthroline method with sodium fluoride as
a Fe(III) complexing agent, to avoid Fe(III) interference, as
proposed by Herrera et al. (16). Interpretation of the results
was performed by applying nonlinear fitting of the integrated
form of the Monod equation.
The fundamental Monod equation considered in this study
consists only of a substrate affinity term and a product inhibi-
tion term. Neither a bacterial cells inhibition term nor a sub-
strate inhibition term was included, as these were found not to
affect the Fe(II) oxidation rate under the experimental condi-
tions tested (see Results and Discussion, below). The experi-
ments were conducted at various cell concentrations, and the
value of the yield coefficient was incorporated into the kinetics
terms. With regard to product inhibition, it was impossible to
determine the type of inhibition a priori, and therefore both
competitive and noncompetitive models were examined.
It should be emphasized that, by definition, any extant ki-
netics data set depends on the physiological state of the bac-
terial population tested. Therefore, the results presented in the
paper must be considered particular to the environmental con-
ditions to which the bacteria were subjected prior to the ex-
KINETICS OF Fe(II) OXIDATION BY A. FERROOXIDANS 1747
periments. However, the methodology presented is general
and can be applied regardless of the “culture history”.
MATERIALS AND METHODS
Preparation of A. ferrooxidans cell suspension. A pure culture of Acidithioba-
cillus ferrooxidans was obtained from the German Collection of Microorganisms
and Cell Cultures (DSM 14882). The bacteria were grown in a 1-liter Erlenmeyer
flask in a medium comprised of 0.4 g liter⫺1 (NH4)2, 0.4 g liter⫺1
MgSO4 䡠 7H2O, 0.1 g liter⫺1 K2HPO4, and 12 g liter⫺1 FeSO4 䡠 7H2O (48).
Initial pH of the growth medium was adjusted to 1.7 by the addition of 2 N
H2SO4. Cell suspensions used in the experiments were taken from the flask and
filtered during the logarithmic phase of growth by using a Whatman no. 1 filter
paper in order to remove suspended iron-containing solids. Subsequently, 100 ml
of the filtrate was refiltered through a 0.45-␮m filter. The filter was then washed
with 50 ml basal salts solution (pH 1.5) and washed again with 50 ml basal salts
solution at pH 2. The cells were then resuspended by backwashing the filter with
10 ml basal salts at pH 2. At the end of the experiments, samples from the growth
flask were sent to Hy-labs (Rehovot, Israel) for rRNA gene cloning, sequencing,
and identification. All of the clones that were returned exhibited greater than
97% similarity to the original culture (expect value, 0).
Procedure of batch experiments. For batch experiments, cells (in suspension)
were added to a basal salt solution comprising FeSO4 and Fe2(SO4)3 salts to
yield a 310-ml solution with known Fe(II), Fe(III), and cell concentrations.
Oxygen was supplied in excess by bubbling air through the medium. Dissolved
oxygen was maintained at ⬎6 mg/liter throughout the experiments. To improve
mixing, a linear shaker at 125 excursions per minute was used. Temperature was
maintained at 25 ⫾ 0.1°C by a water bath. Samples of 1 to 2 ml were taken from
each batch every few minutes using a syringe, and Fe(II) concentration was
determined following 0.22-␮m filtration. Since Fe(II) oxidation consumes pro-
tons, the pH of the medium was periodically adjusted back to pH 2.00 by the
controlled manual addition of 2 N H2SO4.
Analyses. Fe(II) concentration was determined by the modified o-phenanthro-
line method proposed by Herrera et al. (16). Absorbance was measured with a
Genesys 10 spectrophotometer (Spectronics). Fe(III) concentration was deter-
mined by the sulfosalicylic acid technique (48). A. ferrooxidans cells were stained
by using a standard Lugol solution and counted in an improved Neubauer
counting chamber (0.1-mm depth, 0.0025-mm2 area) with an optical microscope
at ⫻400 magnification. Each count was performed at least three times.
RESULTS
Preliminary tests were carried out in order to determine the
most appropriate experimental technique for the determina-
tion of A. ferrooxidans extant kinetics constants. The goal of
this step was to estimate the approximate range of the kinetics
parameters by applying the initial rate measurement tech-
nique. The results (not shown) showed that the initial Fe(II)
oxidation rate is roughly constant at an initial Fe(II) concen-
tration range between 0.2 g Fe2⫹ liter⫺1 and 20 g Fe2⫹ liter⫺1.
Notwithstanding the fact that these results were not useful for
accurate determination of the kinetics parameters, they clearly
indicate that KS is, in all likelihood, lower than 200 mg Fe2⫹
liter⫺1 and also that inhibition caused by Fe(II) is not signifi-
cant at Fe(II) concentrations below 20 g Fe2⫹ liter⫺1.
Based on the knowledge from the preliminary experiments
that KS is ⬍200 mg Fe2⫹ liter⫺1, batch experiments appeared
to be the most appropriate for further investigation. Four sets
of batch experiments were performed, each comprising six
310-ml batch containers with an identical initial biomass con-
centration (X0) and varied initial Fe(II) concentrations (S0).
The initial conditions applied in all the experiments are listed
in Table 2. In all experiments, air was added in excess and
Fe(II) oxidation was continued until at least 97% of the initial
Fe2⫹ was oxidized.
Assuming that cell lysis was negligible during the relatively
short duration of the experiments, the equation for the change
1748 MOLCHANOV ET AL. APPL. ENVIRON. MICROBIOL.

TABLE 2. Batch experiments: initial conditions Marquardt modifications for global convergence. Calculations
Exp nos. Avg X0 (SD) (cells ml ⫺1
) S0 (mg Fe2⫹
liter ⫺1
) were made using the function “nlinfit” within the Statistics
Toolbox of the Matlab software (26). The 95% confidence
1–6 30.9 ⫻ 10 (1.5 ⫻ 10 )
7 7
1,000–2,000
intervals for parameter estimates were calculated using the
7–12 19.2 ⫻ 107 (9.1 ⫻ 106) 1,000–2,000
13–18 7.5 ⫻ 107 (3.0 ⫻ 106) 600–1,800 statistical Matlab function “nlparci” (26). The values (means ⫾
19–24 2.8 ⫻ 107 (3.0 ⫻ 106) 200–700 standard deviation) of the calculated extant kinetics constants
and associated errors estimates were as follows: KS, 31 ⫾ 4 mg
Fe2⫹ liter⫺1; KP, 139 ⫾ 20 mg Fe3⫹ liter⫺1; ␮max, 0.082 ⫾
in biomass concentration in a batch culture can be expressed as 0.002 h⫺1.
follows: The results from 16 out of 24 experiments were used for
calibration, while the remaining 8 were used for validation. R2
␮X ⫽
dX
dt
⫽ ⫺Y
dS
dt 冉 冊 (7)
values were 0.9913 for the calibration procedure and 0.9917 for
the validation. The parameters KS and KP were found to be
highly correlated with each other, with a correlation coefficient
Separating the variables in equation 7 and integrating gives the
of ⫺0.99. The linear combination of these parameters was
following:
obtained from a Fisher matrix: ⌬KP ⫽ 5.38⌬KS, where ⌬ is the
X ⫺ X0 difference between the value of a parameter and its optimal
Y⫽ (8)
S0 ⫺ S value. This information means that for each value of KS inside
the confidence interval, there is a value of KP that produces
The Monod kinetics expression that includes a term for cell with equation 11 and the measured data approximately the
growth and assuming competitive product inhibition is as fol- same sum square errors (18).
lows: To assess the accuracy of the kinetics constants obtained in
dS
冉 ␮max
冊 S䡠X the study, as well as to extend the upper Fe3⫹ concentration

冉 冊
⫽ ⫺ (9) limit within which the calibrated equation 10 can be used, a
dt Y P0 ⫹ S0 ⫺ S
KS 1 ⫹ ⫹S series of batch experiments were performed with an identical
KP
initial Fe(II) concentration (1.2 g liter⫺1 of Fe2⫹), an identical
where P0 is the initial Fe(III) concentration. biomass concentration (17.9 ⫻ 107 cells ml⫺1), and an initial
Combining equations 8 and 9 gives: Fe(III) concentration varying from 0 to 1.2 g liter⫺1 of Fe3⫹.

冉 冊
Because of space limitations, only the results of two of these
dS ␮max S 䡠 关Y共S0 ⫺ S兲 ⫹ X0兴

冉 冊
⫽ ⫺ (10) experiments are presented in Fig. 2. However, for all six runs
dt Y P0 ⫹ S0 ⫺ S the model predicted the measured data extremely well (R2 of
KS 1 ⫹ ⫹S
KP all six runs was between 0.98 and 0.99).
According to reference 46, equation 10 can be integrated. This To determine whether the bacterial culture used in the
integration yields the following: present study was sensitive to the experimental procedure ap-
plied, two tests were performed. First, two identical batch
t ⫽ t0 ⫹ 再 KS Y 关1 ⫹ 共P0 ⫹ S0兲/KP兴
␮max YS0 ⫹ X0
ln 冎再
关S0共YS0 ⫹ X0 ⫺ YS兲兴
X 0S 冎 culture experiments were carried out on different days to test
the reproducibility of the results. The results are shown in Fig.

再 冋 册冎
3. Second, six identical runs were performed with starvation
1 ⫺ 共KS/KP兲 共YS0 ⫹ X0 ⫺ YS兲
⫹ ln (11) times (defined as the time interval between cell inoculation
␮max X0 and ferrous iron addition to the culture) varying from 0 to 6 h.
According to equation 8, a plot of the values of (X ⫺ X0) versus The results of these experiments are shown in Fig. 4.
(S0 ⫺ S) obtained during the batch experiments should yield a
straight line with a slope that equals the yield coefficient of A.
ferrooxidans (i.e., Y) and an intercept that equals zero. As
shown in Fig. 1, a linear curve was indeed obtained (R2 ⫽
0.975), which validates equation 8. The value of the yield co-
efficient, which was obtained by least-square linear regression
analysis, was 2.3 ⫻ 1010 cells (g Fe2⫹)⫺1. This value is similar
to A. ferrooxidans yield values reported in previous works: 2.5 ⫻
1010 cells (g Fe2⫹)⫺1 (2), 2.4 ⫻ 1010 cells (g Fe2⫹)⫺1 (25),
2.23 ⫻ 1010 cells (g Fe2⫹)⫺1 (6), and (1.7 ⫾ 0.4) ⫻ 1010 cells (g
Fe2⫹)⫺1 (27).
Based on the raw results [i.e., Fe(II) concentrations versus
time at 25°C and pH 2.0] obtained in the 24 batch experiments
and the knowledge of the cell concentration calculated based
on Y and the initial cell concentration, the parameters KS, KP,
and ␮max were obtained by applying nonlinear least-squares FIG. 1. Data used for determining the value of the yield coefficient
data fit using the Gauss-Newton algorithm with Levenberg- of A. ferrooxidans.
VOL. 73, 2007 KINETICS OF Fe(II) OXIDATION BY A. FERROOXIDANS 1749

FIG. 4. Effect of starvation period (defined as the time interval

between cell inoculation and ferrous iron addition to the culture) on
the rate of Fe(II) oxidation of A. ferrooxidans.

FIG. 2. Change in Fe(II) concentration with time in batch culture

experiments with different initial ferric iron concentrations: F, 0 mg
liter⫺1 of Fe3⫹; ■, 1,200 mg liter⫺1 of Fe3⫹. Solid lines represent
model predictions.
DISCUSSION
Using the values of the extant kinetics constants derived
from the kinetics data by nonlinear fitting to an integrated
form of the Monod equation, the following equation was de-
veloped to predict the rate of ferrous iron oxidation by a cell
suspension of A. ferrooxidans:
completely general and can be applied irrespective of the cul-
ture history.
In all 30 runs carried out in this study, equation 12 fit the
empirical results exceptionally well. Equation 12 does not in-
clude a Fe(II) concentration inhibition term because it was
found in “initial rate” experiments that the Fe(II) concentra-
tion does not inhibit the oxidation rate up to a concentration of
(at least) 20 g Fe2⫹ liter⫺1.
In order to assess the effect on the oxidation of Fe(II) when
the initial Fe(III) concentration was other than zero, ferric iron

冉 冊
was added to the reaction mixture in six batch experiments at
dS 0.082 S䡠X an initial concentration ranging from 0 to 1,200 mg liter⫺1. The

冉 冊
⫽ ⫺ (12)
dt 2.3 ⫻ 107 P0 ⫹ S0 ⫺ S very high prediction accuracy obtained in these experiments
31 1 ⫹ ⫹S
139 corroborated both the correctness of equation 12 and the ac-
curacy of each of the three kinetics constants obtained in the
where S is the Fe(II) concentration in mg liter⫺1, X is the study and also showed that equation 12 can be considered
suspended bacterial cell concentration (in cells ml⫺1), P0 is the applicable up to (at least) Fe(III) concentrations of around
initial Fe(III) concentration in mg liter⫺1, and t is time in 2.4 g Fe3⫹ liter⫺1.
hours. Since one set of parameters was found to fit exceptionally
Note that strictly speaking equation 12 can be considered well the data from all the experiments (which were run with
applicable only to A. ferrooxidans cells grown under conditions different bacterial cells concentrations), it was concluded that
similar to those of the inoculum tested in this study. In con- the value of KS does not depend on the A. ferrooxidans cell
trast, the experimental and computational procedures are concentration. This conclusion can be considered valid for the
cell concentration range used in the experiments, i.e., from
2.77 ⫻ 107 cells ml⫺1 to 3.1 ⫻ 108 cells ml⫺1. This finding
totally contradicts the conclusion made in reference 29 that a
linear dependency exists between the values of KS and the
biomass concentration (those authors applied a cell concentra-
tion range between 3.25 ⫻ 107 cells ml⫺1 and 4.47 ⫻ 108 cells
ml⫺1, i.e., practically the same cell concentration range as in
the current study).
A comparison between the kinetics constants obtained in
our study with previously published constants, some of which
were obtained under entirely different experimental conditions
or interpreted using different models, is clearly problematical.
The maximum specific growth rate (␮max) values found in the
literature vary between 0.047 h⫺1 (36) and 1.78 h⫺1 (19), i.e.,
a difference of 2 orders of magnitude. Excluding values derived
FIG. 3. Results of two pairs of identical batch culture tests held on
from the application of linear regression techniques (which can
different days to check the effect of “culture history” on the reproduc- be assumed to be inherently inaccurate), the range narrows to
ibility of the kinetics results. ⫻, run I; ■, run II. between 0.047 h⫺1 and 0.23 h⫺1 (32). The ␮max value found in
1750 MOLCHANOV ET AL. APPL. ENVIRON. MICROBIOL.

the study (0.082 h⫺1) lies inside this range. Note that the ␮max
value found in the study corresponds to a doubling time of
around 10 h.
The values of KS obtained in the literature by applying the
competitive Fe(III) inhibition models vary from 0.028 g liter⫺1
(34) to 0.09 g liter⫺1 (23), if one excludes the exceptionally
high value of 0.94 g liter⫺1 reported by Gomez et al. (12). This
value is probably incorrect, because Gomez et al. (12) per-
formed batch experiments without pH control, the Fe(II) ox-
idation was measured by the standard phenanthroline method
without compensation for Fe(III) interference, and the exper-
iments were completed at a high Fe(II) concentration of 0.5 g
liter⫺1, i.e., well above the expected KS value. As was the case
with the maximum specific growth rate, the KS value found in
the present study (0.031 g liter⫺1) conforms well to the range
of previously reported values.
With respect to A. ferrooxidans kinetics constants reported in
the literature, the largest discrepancy appears between the FIG. 5. Comparison between competitive (solid lines) and non-
values of KP (product inhibition coefficient). In the models that competitive (dashed lines) inhibition model predictions. (a) Run 3; (b)
run 11; (c) run 18; (d) run 20 (see Table 2).
assume competitive inhibition, KP was reported to be as low as
0.06 g liter⫺1 (19) and as high as 4.38 g liter⫺1 (13), i.e., a

冋 册
difference of 2 orders of magnitude. The value of KP (0.37 g
Y KS关1 ⫹ 共P0 ⫹ S0兲/KP兴 S0共YS0 ⫹ X0 ⫺ YS兲
liter⫺1) found in the present study lies within this range. How- t ⫽ t0 ⫹ ln
ever, considering the differences in the methods and the prob- ␮max YS0 ⫹ X0 X 0S

lematic nature of many of the techniques used (as explained
before), a considerable number of the reported KP values can-
not be regarded as reliable.
Competitive or noncompetitive inhibition? A disagreement
exists in the literature regarding the use of competitive versus
noncompetitive inhibition models to describe the inhibiting
effect of the Fe(III) concentration on A. ferrooxidans kinetics.
To determine the type of inhibition, the common technique
is to compare the values of KS and ␮max that are determined in
the absence and presence (at a constant total iron concentra-
tion) of the inhibiting substance. According to this technique,
if competitive inhibition dominates, KS, but not ␮max, should
change. In contrast, in the case of noncompetitive inhibition,
␮max should decrease and KS should remain unchanged. How-
ever, such a procedure cannot be carried out in the course of
batch experiments, since the concentration of the inhibiting
substance [Fe(III) in this case] changes constantly during the
experiment. Therefore, to establish the type of Fe(III) inhibi-
tion in the current investigation, the integrated form of the
noncompetitive inhibition model shown in equation 13 was
also applied to the raw results, and the most appropriate inhi-
bition model was chosen by comparing (i) the R2 values, which
quantify the goodness of fit of the prediction equations to the
measured data, and (ii) the standard deviations of the esti-
mated constants.
冉 冊
Y共P0 ⫹ KP ⫺ KS兲 ⫺ X0 YS0 ⫹ X0 ⫺ YS
⫹ ln
␮maxKPY X0
S0 ⫺ S
⫹ (15)
␮maxKP
When applying the raw data from the kinetics experiments to
equation 15, the following results were obtained: KS, 151.4 ⫾
26.7 mg Fe2⫹ liter⫺1; KP, 3,839 ⫾ 898 mg Fe3⫹ liter⫺1; ␮max,
0.11 ⫾ 0.01241 h⫺1. The R2 values derived from applying
equation 15 to the raw data were 0.9549 and 0.9565 for the
calibration and validation procedures, respectively, i.e., the
“noncompetitive” model fit the experimental data much less
well than the “competitive” model (R2 of 0.9913 and 0.9917,
respectively). In addition, the standard deviations of the pa-
rameters fit by the noncompetitive model (8.8%, 11.7%, and
5.6% for KS, KP, and ␮max, respectively) were considerably
larger than those obtained by fitting the competitive model
(6.4%, 7.2%, and 1.2% for KS, KP, and ␮max, respectively).
Figure 5 shows measured versus predicted results of four
individual runs (run numbers 3, 11, 18, and 20 in Table 2). It
can be seen that the prediction of the noncompetitive inhibi-
tion model did not fit well the experimental results (the R2
values for these runs were 0.8873, 0.8908, 0.9320, and 0.8629,
respectively), whereas the R2 values for the same runs when

冉 冊
applying the competitive model were much higher: 0.9862,
dS ␮max S䡠X 0.9686, 0.9805, and 0.9944, respectively. Further examples for
冉 冊
⫽ ⫺ (13)
dt Y P0 ⫹ S0 ⫺ S the inadequacy of the noncompetitive model to the empirical
1⫹ 共KS ⫹ S兲
KP results are not shown in order to conserve space, but the
inevitable conclusion is that for A. ferrooxidans kinetics the
Combining equations 8 and 13 gives the following:
competitive inhibition model results in a much better predic-
dS
冉␮max
冊 S 䡠 关Y共S0 ⫺ S兲 ⫹ X0兴 tion of measured data than the noncompetitive model. More-

冉 冊
⫽ ⫺ (14) over, extrapolation of the noncompetitive model by applying
dt Y P0 ⫹ S0 ⫺ S
1⫹ 共KS ⫹ S兲 equation 15 to the results from the experiments in which Fe3⫹
KP
was added initially to the reaction mixtures resulted in very
Integration of equation 14 yields the following: inaccurate prediction with the respect to the measured data.
VOL. 73, 2007
For example, the prediction of the experiment with an initial
Fe(III) concentration of 1,200 mg liter⫺1 gave an R2 of 0.795,
whereas the same data, when applied to the competitive inhi-
bition model, yielded an R2 of 0.9887.
Repeatability of the kinetics results. Several studies con-
cerned with determining bacterial kinetics constants have re-
ported that a change in “culture history” was one of the major
reasons for the large variations observed in the values of re-
ported Monod kinetics parameters (14, 21, 22). The accepted
hypothesis is that the way in which a culture is grown often
determines the nature of the enzymatic systems that are ex-
pressed and also the physiological state, which is the sum total
of a cell’s macromolecular composition. The physiological
state may further determine how rapidly the bacteria can syn-
thesize enzymes, as well as how rapidly these enzymes would
react. In this respect, Sommer et al. (45) showed poor repro-
ducibility of biodegradation experiments (Pseudomonas cepa-
cia with toluene as the only carbon and energy source) and
concluded that the reason was that they were performed on
different days.
Figure 3 shows that the duplicates of two similar experi-
ments (each experiment was initiated with a different initial
Fe2⫹ concentration, and both experiments were repeated on
two different days) were practically identical, the slight varia-
tion being attributable to normal analytical fluctuations. Tak-
ing into account that it is almost impossible to prepare two
solutions with an identical biomass concentration (X0 differed
slightly between the runs, i.e., 4.41 ⫻107 cells ml⫺1 in run I
versus 4.39 ⫻ 107 cells ml⫺1 in run II), it can be safely con-
cluded that at least in the case of A. ferrooxidans, kinetics
experiments carried out on different days can be very well
reproduced, provided that prior to the kinetics experiment the
inocula are prepared in an identical fashion (see Materials and
Methods), resulting in populations with closely similar physi-
ological conditions.
Since the time between the preparation of cell suspensions
(by filtration) and the beginning of the experiments varied in
the different experiments and sometimes even between the
various runs within the same experiment, a second experiment
was conducted to validate that the time lag within the experi-
mental procedure itself [defined here as a “starvation period,”
because during this time no Fe(II) was supplied to the bacte-
ria] did not have an effect on the kinetics constants deter-
mined. Figure 4 clearly shows that a starvation period, within
the 6 hours tested, had no effect on the value of the kinetics
constants. This finding has obvious practical implications on A.
ferrooxidans kinetics-related laboratory procedures.
ACKNOWLEDGMENT
This research was supported by research grant IS-3522-04 from
BARD, the United States-Israel Binational Agricultural Research and
Development Fund.
REFERENCES
1. Barron, J. L., and D. R. Luecking. 1990. Growth and maintenance of Thio-
bacillus ferrooxidans cells. Appl. Environ. Microbiol. 56:2801–2806.
2. Beck, J. V., and F. M. Shafia. 1964. Effect of phosphate ion and 2,4-dinitro-
phenol on the activity of intact cells of Thiobacillus ferrooxidans. J. Bacteriol.
88:850–857.
3. Blunk, M., and T. P. Mommsen. 1978. Systematic errors in fitting linear
transformations of the Michaelis-Menten equation. Biometrika Vol. 65:363–
368.
KINETICS OF Fe(II) OXIDATION BY A. FERROOXIDANS 1751
4. Boon, M., C. Ras, and J. J. Heijnen. 1999. The ferrous iron oxidation kinetics
of Thiobacillus ferrooxidans in batch cultures. Appl. Microbiol. Biotechnol.
51:813–819.
5. Boon, M., T. A. Meeder, C. Thöne, C. Ras, and J. J. Heijnen. 1999. The
ferrous iron oxidation kinetics of Thiobacillus ferrooxidans in continuous
cultures. Appl. Microbiol. Biotechnol. 51:820–826.
6. Braddock, J. F., H. V. Luong, and E. J. Brown. 1984. Growth kinetics of
Thiobacillus ferrooxidans isolated from arsenic mine drainage. Appl. Environ.
Microbiol. 48:48–55.
7. Chudova, P., B. Capdeville, and J. Chudova. 1992. Explanation of biological
meaning of the S0/X0 ratio in batch cultivation. Water Sci. Technol. 26:743–
751.
8. Cornish-Bowden, A. 1995. Analysis of enzyme kinetic data. Oxford Univer-
sity Press, New York, N.Y.
9. Cort, T. L., and A. R. Bielefeldt. 2002. A kinetic model for surfactant
inhibition of pentachlorophenol biodegradation. Biotechnol. Bioeng. 78:606–
616.
10. Fadrus, H., and J. Maly. 1975. Suppression of iron(III) interference in the
determination of iron(II) in water by the 1,10-phenanthroline method. Analyst
100:549–554.
11. Flickinger, M. C., and S. W. Drew. 1999. Encyclopedia of bioprocess tech-
nology: fermentation, biocatalysis, and bioseparation. John Wiley & Sons,
Inc., New York, N.Y.
12. Gomez, J. M., I. Caro, and D. Cantero. 1996. Kinetic equation for growth of
Thiobacillus ferrooxidans in submerged culture over aqueous ferrous sulphate
solutions. J. Biotechnol. 48:147–152.
13. Gomez, J. M., and D. Cantero. 2003. Kinetic study of biological ferrous
sulphate oxidation by iron-oxidising bacteria in continuous stirred tank and
packed bed bioreactors. Process Biochem. 38:867–875.
14. Grady, C. P. L., Jr., B. F. Smets, and D. S. Barbeau. 1996. Variability in
kinetic parameter estimates: a review of possible causes and a proposed
terminology. Water Res. 30:742–748.
15. Harvey, P. I., and F. K. Crundwell. 1997. Growth of Thiobacillus ferrooxi-
dans: a novel experimental design for batch growth and bacterial leaching
studies. Appl. Environ. Microbiol. 63:2586–2592.
16. Herrera, L., P. Ruiz, J. C. Aguillon, and A. Fehrmann. 1989. A new spec-
trophotometric method for the determination of ferrous iron in the presence
of ferric iron. J. Chem. Technol. Biotechnol. 44:171–181.
17. Ingledew, W. J. 1982. Thiobacillus ferrooxidans: the bioenergetics of an acido-
philic chemolitotroph. Biochim. Biophys. Acta 683:89–117.
18. Ioslovich, I., I. Seginer, and P.-O. Gutman. 2004. Dominant parameter
selection in the marginally identifiable case, mathematics and computers in
simulation (MATCOM). 65:127–136.
19. Jones, C. A., and D. P. Kelly. 1983. Growth of Thiobacillus ferrooxidans on
ferrous iron in chemostat culture: influence of product and substrate inhi-
bition. J. Chem. Tech. Biotechnol. 33B:241–261.
20. Kawabe, Y., C. Inoue, K. Suto, and T. Chida. 2003. Inhibitory effect of high
concentrations of ferric ions on the activity of Thiobacillus ferrooxidans.
J. Biosci. Bioeng. 96:375–379.
21. Kovarova-Kovar, K., and T. Egli. 1998. Growth kinetics of suspended mi-
crobial cells: from single-substrate-controlled growth to mixed-substrate ki-
netics. Microbiol. Mol. Biol. Rev. 62:646–666.
22. Liu, C., and J. M. Zachara. 2001. Uncertainties of Monod kinetics param-
eters nonlinearly estimated from batch experiments. Environ. Sci. Technol.
35:133–141.
23. Liu, M. S., R. M. R. Branion, and D. W. Duncan. 1988. The effects of ferrous
iron, dissolved oxygen, and inert solids concentrations on the growth of
Thiobacillus ferrooxidans. Can. J. Chem. Eng. 66:445–451.
24. Lizama, H. M., and I. Suzuki. 1989. Synergistic competitive inhibition of
ferrous iron oxidation by Thiobacillus ferrooxidans by increasing concentra-
tion of ferric iron and cells. Appl. Environ. Microbiol. 55:2588–2591.
25. MacDonald, D. G., and R. H. Clark. 1970. The oxidation of aqueous ferrous
sulphate by Thiobacillus ferrooxidans. Can. J. Chem. Eng. 48:669–676.
26. MathWorks Inc. 2004. Statistical toolbox for use with Matlab: user’s guide.
MathWorks, Inc., Natick, Mass.
27. Mesa, M. M., M. Macı́as, and D. Cantero. 2002. Mathematical model of the
oxidation of ferrous iron by a biofilm of Thiobacillus ferrooxidans. Biotech-
nol. Prog. 18:679–685.
28. Nakamura, K., T. Noike, and J. Matsumoto. 1986. Effect of operation con-
ditions on biological Fe2⫹ oxidation with rotating biological contactors.
Water Res. 20:73–77.
29. Nemati, M., and C. Webb. 1997. A kinetic model for biological oxidation of
ferrous iron by Thiobacillus ferrooxidans. Biotechnol. Bioeng. 53:478–486.
30. Nemati, M., S. T. L. Harrison, G. S. Hansford, and C. Webb. 1998. Biological
oxidation of ferrous sulphate by Thiobacillus ferrooxidans: a review on the
kinetics aspects. Biochem. Eng. J. 1:171–190.
31. Nemati, M., and C. Webb. 1998. Inhibition effect of ferric iron on the kinetics
of ferrous iron. Biotechnol. Lett. 20:873–877.
32. Nikolov, L. N., and D. G. Karamanev. 1992. Kinetics of the ferrous iron
oxidation by resuspended cells of Thiobacillus ferrooxidans. Biotechnol. Prog.
8:252–255.
33. Nikolov, L. N., D. G. Karamanev, L. Dakov, and V. Popova. 2001. Oxidation
1752 MOLCHANOV ET AL. APPL. ENVIRON. MICROBIOL.

of ferrous iron by Thiobacillus ferrooxidans in a full-scale rotating biological
contactor. Environ. Progr. 20:247–250.
34. Nyavor, K., N. O. Egiebor, and P. M. Fedorak. 1996. The effect of ferric ion
on the rate of ferrous oxidation by Thiobacillus ferrooxidans. Appl. Microbiol.
Biotechnol. 45:688–691.
35. Okereke, A., and S. E. Stevens, Jr. 1991. Kinetics of iron oxidation by
Thiobacillus ferrooxidans. Appl. Environ. Microbiol. 57:1052–1056.
36. Pagella, C., P. Silvestri, and D. M. De Faveri. 1996. Hydrogen sulfide re-
moval with a biochemical process: the biological step. Chem. Biochem. Eng.
Q. 10:165–174.
37. Panikov, N. S. 1995. Microbial growth kinetics. Chapman & Hall, New York,
N.Y.
38. Pesic, B., D. J. Oliver, and P. Wichlasz. 1988. An electrochemical method of
measuring the oxidation rate of ferrous to ferric iron with oxygen in the
presence of Thiobacillus ferrooxidans. Biotechnol. Bioeng. 33:428–439.
39. Raaijmakers, J. G. W. 1987. Statistical analysis of the Michaelis-Menten
equation. Biometrics 43:793–803.
40. Robinson, J. A., and J. M. Tiedje. 1983. Nonlinear estimation of Monod
growth kinetic parameters from a single substrate depletion curve. Appl.
Environ. Microbiol. 45:1453–1458.
41. Savı́c, D. S., M. L. Lazı́c, V. V. Veljkovı́c, and M. M. Vrvı́c. 2005. A kinetic
model of ferrous iron oxidation by Acidithiobacillus ferrooxidans in a batch
culture. Chem. Ind. Chem. Eng. Q. 11:59–62.
42. Schmidt, S. K., S. Simkins, and M. Alexander. 1985. Models for the kinetics
of biodegradation of organic compounds not supporting growth. Appl. En-
viron. Microbiol. 50:323–331.
43. Simkins, S., and M. Alexander. 1984. Models for mineralization kinetics with
the variables of substrate concentration and population density. Appl. En-
viron. Microbiol. 47:1299–1306.
44. Smith, L. H., P. L. McCarty, and P. K. Kitanidis. 1998. Spreadsheet method
for evaluation of biochemical reaction rate coefficients and their uncertain-
ties by weighted nonlinear least squares analysis of the integrated Monod
equation. Appl. Environ. Microbiol. 64:2044–2050.
45. Sommer, H. M., H. Spliid, H. Holst, and E. Arvin. 1998. Examination of
reproducibility in microbiological degradation experiments. Biodegradation
9:65–82.
46. Spiegel, M. R., and J. Liu. 1999. Mathematical handbook of formulas and
tables. McGraw-Hill, Columbus, Ohio.
47. Suzuki, I., H. M. Lizama, and P. D. Tackaberry. 1989. Competitive inhibition
of ferrous iron oxidation by Thiobacillus ferrooxidans by increasing concen-
tration of cells. Appl. Environ. Microbiol. 55:1117–1121.
48. Zoltov, Y. A. 2001. Fundamentals of analytical chemistry. Practical guide.
Visshaya Shkola, Moscow, Russia.