Hydrometallurgy 58 Ž2000.

269–275
www.elsevier.nlrlocaterhydromet

High efficiency reactor for the biooxidation of ferrous iron

A. Mazuelos ) , F. Carranza, I. Palencia, R. Romero
´ Quımica,
Departamento de Ingenierıa ´ ´
Facultad de Quımica, UniÕersidad de SeÕilla, 41012 SeÕille, Spain
Received 29 February 2000; received in revised form 27 July 2000; accepted 27 July 2000

Abstract

The biooxidation of ferrous iron is a potential industrial process in the regeneration of ferric iron, in hydrometallurgical
leaching operations and in the removal of H 2 S in combustible gases. Another field of application is the treatment of acid
mine drainage water. The aim of this work was the design of a continuous reactor for a high efficiency ferrous iron
biooxidation capable of attaining the oxidation rates demanded by industrial processes for a minimum-sized reactor. Some
criteria of design come from the data of former studies.
The bioreactor consists of a 1249 mm high and 84 mm diameter column randomly packed with siliceous stone particles
with inlets for fresh medium and air at the bottom. The present work studies the influence of air and liquid flow rates on the
ferrous iron biooxidation in a flooded packed bed bioreactor. Starting from the data of this study, kinetics, and oxygen
concentration in the outlet gas, it was deduced that the biooxidation process was limited by the availability of oxygen in the
liquid medium. The free-swimming cells concentration was measured in the outlet solution for different operating
conditions. It was observed that once the biofilm has been formed, a variation in the liquid flow rate does not alter the
biomass attached to the packing. The highest ferric productivity attained was 11.25 grL P h. q 2000 Elsevier Science B.V.
All rights reserved.

Keywords: Biooxidation; Thiobacillus ferrooxidans; Packed bed; Bioreactor

1. Introduction v
the removal of hydrogen–sulphide in combustible
gases w3x:

™
Biooxidation of ferrous iron Ž2Fe 2qq 1r2O 2 q
2Hq 2Fe 3qq H 2 O. by Thiobacillus ferrooxidans
is a potential industrial process in the regeneration of
2Fe 3qq SH 2 ™ 2Fe 2q
q S8 q 2Hq.

ferric iron used as an oxidizing agent in:

hydrometallurgical leaching operations w1,2x:

™M
v

In this way, the oxidant agent consumption is

MS q 2Fe 3q 2q
q S8 q 2Fe 2q ; theoretically zero, the two processes are closed treat-
ments, not polluting and without effluents.
Another field of application is the treatment of
)
Corresponding author. Tel.: q34-954-55-71-82; fax: q34-
acid mine drainage water w4x. These effluents contain
954-55-71-34. heavy metal ions and ferrous iron. Neutralization is
E-mail address: mazuelos@cica.es ŽA. Mazuelos.. needed in order to eliminate such metals by precipi-

0169-4332r00r$ - see front matter q 2000 Elsevier Science B.V. All rights reserved.
PII: S 0 3 0 4 - 3 8 6 X Ž 0 0 . 0 0 1 4 1 - 9
270 A. Mazuelos et al.r Hydrometallurgy 58 (2000) 269–275
tation. A previous oxidation of ferrous iron techni-
cally facilitates the subsequent solid–liquid separa-
tion due to the fact that ferric compound precipitates
are more voluminous and heavier than ferrous com-
pound precipitates Žcolloidal precipitates..
Taking into account the treatment volumes of
industrial processes, the bioreactor design has very
strong economic limitations that affect its size, com-
plexity, and type of bacterial support.
The natural tendency of T. ferrooxidans to grow
on surfaces makes it an ideal organism for cell
immobilization. Several reactor designs have been
studied where bacteria are attached to supports in
attempts to improve the rate of ferrous iron oxida-
tion, e.g., rotating biological contactor ŽRBC. w4x,
packed-bed reactor w5–13x, fluidized-bed reactor w14x.
RBC and fluidized-bed reactors become complex in
commercial operation. The packed-bed reactor is an
attractive choice because it is simple to install and
operate. In a packed-bed reactor, the ferrous iron
solution can be fed in either at the top of the bed
Žtrickle bed. or at the bottom of the bed Žflooded
packed bed.. If the ferrous iron solution is fed on to
the top of the packed-bed, there are some disadvan-
tages related to the presence of ‘dead zones’. In
flooded packed-bed reactors, the liquid and air are
fed in to the bottom. They go up through the bed,
occupying all the voids of the fixed bed. The liquid
flow distribution is more homogeneous than in a
trickle bed. In these reactors, dead zones are unlikely
at high liquid flow rates.
Packed beds with different packings, such as low
grade sulphide minerals w6,10x, calcium alginate w7x,
PVC w5,8x, polyurethane foam w9x, glass w5,11x, acti-
vated carbon w5,10,11x, exchange resin w11x, nickel
w12x, siliceous stone w5,13x, polyethylene, extruded
polystyrene, and expanded polystyrene w5x, have been
found in the literature. With the exception of low
grade sulphide minerals and siliceous stone particles,
all the named packings are manufactured materials
and so their use on a commercial scale would be
expensive.
The present work studies the influence of liquid
and air flow rates on the ferrous iron biooxidation in
a flooded packed-bed bioreactor. Other variables such
as pH, temperature, inert support material, support
particle size and type of air distributor have been
previously studied w5x.
2. Materials
2.1. Inoculum
The culture used as inoculum was originally iso-
lated from Riotinto mine drainage waters. It con-
sisted mainly of T. ferrooxidans, Leptospirillum fer-
rooxidans, and some heterotrophic bacteria
Žacidiphilium., adapted to pH 1.25. It was main-
tained at 318C and pH 1.25 in a modified Silverman
and Lundgren 9 K nutrient medium. In order to get
the inoculum volume required for biofilm formation,
successive transfers into a fresh medium were per-
formed by mixing 20% Žvolrvol. inoculum of a
spent, iron-grown culture with 80% Žvolrvol. of
nutrient medium at pH 1.25.
2.2. Inert support
Siliceous stone particles were used as bacterial
support. In order to prepare the particles, the follow-
ing operations were performed:
v
screening to provide particles between 6 and 7 mm
in size;
v
washing with water under pressure to remove
slimes; and
v
washing with 1 M sulphuric acid to remove the
soluble compounds.
2.3. Reactor
The bioreactor consisted of a 1249 mm high and
84 mm diameter polymethyl methacrylate column
ŽFig. 1.. The reactor was divided into four zones,
named zero, first, second and third floor. First, sec-
ond and third floors were packed-beds each 280 mm
high. A 45 mm high empty space separated each
floor from the next one. These empty spaces had
three holes for sampling and for possible additions of
acid and oxygen. The packing of each floor was
supported by plastic screens. The air diffuser was
placed inside the zero floor, a space without support
particles. The air diffuser was a woodprism Ž4.3 =
1.5 = 1.5 cm. from whose pores emerged the air
bubbles. Air was supplied under pressure Ž0.5 bar.
by a compressor. The ferrous sulphate solution was
 A. Mazuelos et al.r Hydrometallurgy 58 (2000) 269–275
Fig. 1. Schematic of flooded packed-bed bioreactor.
fed in at the bottom of the column by a peristaltic
pump. The fresh medium composition was 8 grL in
ferrous iron, and pH 1.25 adjusted with sulphuric
acid. The solution went out by a hole Žby overflow.
placed above the third floor, where the bacteria
concentration was measured. The reactor was placed
in a thermostatically controlled room, the tempera-
ture of which was maintained at 318C.
3. Methods
3.1. Biofilm formation procedure
The process of biofilm formation on the solid
support took place in the following stages.
v
Immobilization: the support particles were ran-
domly placed in the column, constituting the three
beds Žfirst, second, and third floors.. The void frac-
tion in each floor was 0.42. The inoculum Ž1050 mL.
and fresh medium Ž2000 mL. were added at the top
of the column. The air flow rate was 500 mLrmin.
The bioreactor operated in batch mode. After 50 h,
95% conversion of ferrous iron was achieved and
bacteria had colonized the surface of support parti-
cles.
271
v
Continuous recycling: once the last step was
finished, continuous recycling flow mode of opera-
tion was initiated. Ten liters of fresh ferrous iron
solution was fed in at the column. The liquid flow
rate was 450 mLrh, and the air flow rate was 500
mLrmin. After 54 h, 95% conversion of ferrous iron
was achieved. Then, another volume of fresh medium
Ž10 L. was fed into the column, and 100% conver-
sion of ferrous iron was achieved after 38 h.
The biofilm formation was then considered to be
complete, and the reactor could operate in continu-
ous flow mode.
3.2. Continuous flow mode
The column was continuously fed with fresh
medium, and the steady state was established when
the ferrous iron concentration varied less than 5% in
a period of time equal to 50 times the mean resi-
dence time for each flow rate. The liquid and air
flow rates were simultaneously increased stepwise. If
ferrous iron concentration in the outlet solution was
less than 0.5 grL, the liquid flow rate was increased,
and if ferrous iron concentration in the outlet solu-
tion was higher than 0.5 grL, the air flow rate was
increased.
3.3. Residence time distribution (RTD)
The RTD was experimentally studied using cop-
per ions as a tracer. At a specified time t s 0, 5 mL
of a 40 grL copper ions solution Žcopper sulphate.
was injected quickly in one shot into the feed stream.
The outlet copper concentration was then measured
as a function of time.
3.4. Bubble size
The mean air bubble diameter was determined at
zero and third floors. Zero and third floors were
filmed by a video camera. Two hundred air bubble
diameters were measured using a TV monitor. The
mean air bubble diameter was calculated as the
arithmetic average of these measurements.
3.5. Analytical methods
Ferrous iron concentration was determined by
end-point automatic titration with 0.0083 M
272 A. Mazuelos et al.r Hydrometallurgy 58 (2000) 269–275

K 2 Cr2 O 7 . Copper concentration was determined by
atomic absorption spectrophotometry. Oxygen con-
centration in the outlet gas stream was monitored
with a gas analyser. The number of bacteria per
milliliter was determined by visual count using a
Thoma camera with an optical microscope.
4. Results and discussion
4.1. The influence of liquid and air flow rates on the
ferrous iron biooxidation
Table 1 shows the ferric iron productivity in each
floor for different air and liquid flow rates. Fig. 2
shows the total system ferric productivity. Air flow
rate has a noticeable effect on ferric iron productiv-
ity, which increased as the air flow rate increased. In
all cases, oxygen concentration in the gas outlet
stream was higher than 18%. It can be postulated
that the biooxidation process was limited by the
availability of oxygen in the liquid medium. On the
other hand, the ferric iron productivity of the first
floor was always the highest. It was observed that
the air bubble diameters increased, by coalescence,
as they move up through the bed. The mean air
bubble diameter was 0.39 mm in zero floor and 2.8
mm in third floor. Therefore, the ferric productivity
depends strongly on the gas–liquid interfacial area,
and so, gas–liquid mass transfer dominates.
The RTD was experimentally studied for different
air and liquid flow rates. Fig. 3 shows the corre-
sponding RTD functions. Apparently, significant de-
viations from plug–flow behaviour are observed.
There are some regions of apparent back-mixing, as
evidenced by the exponential-decay shape of the
curves. The large spikes indicate channeling. The
bioreactor can be modelled as equal size continuous
stirred tank reactors in series. The number of tanks in
series Ž N . is determined by calculating the dimen-
sionless variance Ž su 2 . according to the following
equation:
1 t2
Ns s
su 2 s2
where t is the mean residence time and s 2 is the
variance. Table 2 shows the parameters used to

Table 1
The ferric productivity in each floor Žprod., free-swimming cells Ž c c ., the number of swimming cells per hour Ž Nc ., and ferrous iron
concentration Ž c Fe . in the outlet solution, for different air and liquid flow rates
Liquid flow Air flow prod 0 prod 1 prod 2 prod 3 cc Nc c Fe
rate ŽLrh. rate ŽLrmin. Žgrh. Žgrh. Žgrh. Žgrh. Žbact = 10 9 rL. Žbact = 10 9 rh. ŽgrL.
2.036 0.508 0.98 5.31 3.48 1.71 18.6 37.9 2.23
2.027 0.771 1.14 6.12 4.18 1.86 21.8 44.2 1.34
2.059 0.888 1.17 6.77 4.45 2.29 23.4 48.2 0.76
2.073 1.084 1.22 7.71 4.95 2.51 77.6 160.9 0.18
2.374 1.095 1.14 7.31 4.89 2.90 16.0 38.0 1.24
2.389 1.283 1.22 7.81 5.06 3.15 26.6 63.6 0.55
2.865 1.283 1.38 6.47 4.76 2.81 16.6 47.6 2.84
2.853 1.392 1.28 6.82 4.94 3.00 – 2.49
2.845 1.572 1.45 7.00 5.15 3.07 16.3 46.4 2.18
2.844 1.775 1.39 7.74 5.43 3.36 – 1.84
2.812 1.969 1.46 7.90 5.68 3.49 13.2 37.1 1.46
2.802 2.216 1.57 8.21 5.94 3.56 – 1.11
2.816 2.390 1.75 7.91 6.14 3.60 14.8 41.8 1.22
2.822 2.648 1.78 7.79 6.46 3.81 – 1.07
2.856 2.831 1.69 7.91 6.43 3.83 – 1.05
2.818 3.015 1.78 7.69 6.51 3.92 – 1.05
 A. Mazuelos et al.r Hydrometallurgy 58 (2000) 269–275 273

Fig. 2. Total system ferric productivity for different air and liquid flow rates.

characterize the RTD for each operating condition
studied. As the number of reactors increases, the
reactor hydrodynamic behaviour approaches plug–
flow. Therefore, as either the air or the liquid flow
rates increases, deviations from plug–flow model
behaviour decrease. It is observed that as the liquid
flow rate increases, the portion of the RTD curve at
long time Žthe tail of the curve. decreases, i.e.,
back-mixing effect decreases. Since the back-mixing
effect can be considered as an additional contribution
of oxygen to the biooxidation process, this could
explain the decrease of ferric productivity when the
liquid flow rate increases ŽTable 1.. If back-mixing
decreases, the probability that liquid globules from

Fig. 3. RTD curves for different air and liquid flow rates: A Žliquid: 2.8 Lrh, air: 3.015 Lrmin., B Žliquid: 2.8 Lrh, air: 1.987 Lrmin., C
Žliquid: 2.8 Lrh, air: 0.995 Lrmin., D Žliquid: 2.40 Lrh, air: 0.995 Lrmin., E Žliquid: 1.97 Lrh, air: 0.995 Lrmin..
274 A. Mazuelos et al.r Hydrometallurgy 58 (2000) 269–275

Table 2 cellrh. The presence of large quantities of bacteria

Parameters used to characterise RTD in the outlet liquid stream must be related to the
Liquid flow Air flow t s2 su2 N detachment of the attached cells. Literature reports
rate ŽLrh. rate ŽLrmin. Žmin. Žmin2 .
that the biomass changes depend on the turbulence
2.795 3.015 61.2 842.8 0.225 4.45 conditions due to streams of air bubbles and liquid in
2.795 1.987 65.9 1095.4 0.252 3.96
biooxidation reactors, and the decrease of the ferric
2.832 0.995 65.7 1154.0 0.268 3.73
2.406 0.995 79.1 1883.5 0.301 3.32 productivity is usually related to a wash-out effect
1.970 0.995 97.74 3136.8 0.328 3.05 occurring at a critical liquid flow rate w9,11x. How-
ever, Table 1 shows that high values of Nc are not
related to an increase of the liquid flow rate; more-
over, an increase in the liquid flow rate, that implies
first, second or third floor return to the zero floor an increase in c Fe , leads to a considerable decrease
decreases. The oxygen concentration in these liquid of Nc . On the other hand, an increase in the air flow
globules must be very low Žoxygen is the limiting rate did not always imply an increase in Nc . This
reagent in first, second, and third floors.. When fact only occurred significantly in two cases: when
those liquid globules moving down pass through the the air flow rate increased from 0.888 to 1.084
zero floor, their oxygen concentration increases since: Lrmin and when the air flow rate increased from
1.095 to 1.283 Lrmin. In both cases, c Fe decreases
v
in the zero floor, there are no inert support parti- to values less than 1 grL. These results suggest that
cles Žsmall biomass density. and the rate of oxy- once the biofilm has been formed, variation in the
gen consumption has to be small; liquid and the air flow rates does not alter the
v
in the zero floor, the oxygen transfer conditions biomass attached to the packing. When ferrous iron
are the best: concentration was lower than 1 grL in the outlet
the average size of the bubbles is the smallest in liquid stream, a fraction of attached cells was de-
the column; and tached. If ferrous iron concentration is low Žnear to 0
the level of turbulence is the highest in the grL. in the outlet liquid stream, cells in a part of the
column due to the proximity of the air diffuser. biofilm, that specially placed at the top of the biore-
actor, did not have a substrate. It can be postulated
The same argument may explain the fact that as that a part of these cells was detached from the
the air flow rate increases, the upward slope of the matrix of biofilm, by chemotaxis, and then they were
total system productivity curve decreases ŽFig. 2.. carried away by the liquid stream.
Two opposite effects can be considered: an increase
in the air flow rate leads to an enhancement of the 4.3. The highest ferric productiÕity
air volume fraction Žpositive effect., and it leads to a
decrease of the back-mixing effect Žnegative effect.. The highest ferric productivity was 19.9 grh
Ž2327 grh m2 of reactor cross-section. Žfor 2.818
4.2. Swimming cells concentration Lrh and 3.015 Lrmin of liquid flow rate and air

The free swimming cells concentration, c c , was

measured in the outlet solution for different operat-
Table 3
ing conditions. Table 1 shows c c , the number of The maximum productivities in zero, first, second, and third floors
swimming cells per hour in the outlet liquid stream Floor Maximum productivity
Ž Nc . and ferrous iron concentration Ž c Fe . in the
Žgrh. ŽgrL h.
outlet solution, for different air and liquid flow rates.
In all cases, c c is higher than 13 = 10 9 cellrL. Free First 7.81 11.25
Second 6.51 9.17
suspended microorganisms came from the division Third 3.92 5.59
of the cells in the biofilm w14x. There are two high Zero 1.78 1.80
values of Nc , 160.9 = 10 9 cellrh and 63.6 = 10 9
 A. Mazuelos et al.r Hydrometallurgy 58 (2000) 269–275
flow rate, respectively.. This productivity is equiva-
lent to 5.8 grL h of reactor void volume Žthe total
volume of reactor minus the volume of inert support
particles.. Table 3 shows the maximum ferric pro-
ductivities in zero, first, second and third floors, in
grh and grL h.
5. Conclusions
Air flow rate has a noticeable effect on ferric iron
productivity, which increased as air flow rate in-
creased. The biooxidation process was limited by the
availability of oxygen in the liquid medium. As
either the air or the liquid flow rates increases,
deviations from plug–flow behaviour decrease. Once
the biofilm has been formed, variation in the liquid
flow rate does not alter the biomass attached to the
packing. It can be postulated that if ferrous iron
concentration is low in a zone of reactor, a fraction
of attached cells is detached by chemotaxis. The
highest ferric productivity was 11.25 grL h.
Acknowledgements
The authors wish to acknowledge the financial
support of CICYT ŽComision ´ Interministerial de
´ Spain., Project BIO95 0482.
Ciencia y Tecnologıa,
References
w1x F. Carranza, N. Iglesias, R. Romero, I. Palencia, Kinetics
improvement of high grade sulfides bioleaching by effects
separation, FEMS Microbiol. Rev. 11 Ž1993. 129–138.
275
w2x N. Iglesias, I. Palencia, F. Carranza, IBES process: descrip-
tion and applications, S. Afr. Inst. Min. Metall. Ž1996.
181–185, Hidden Wealth.
w3x H. Magota, Y. Shiratori, Treatment of sour natural gas
containing hydrogen sulfide, Jpn. Kokai Tokkyo Koho,
Japanese patent no. 63 205 124, 1988.
w4x K. Nakamura, T. Noike, J. Matsumoto, Effect of operational
conditions on biological FeŽII. oxidation with rotating bio-
logical contactors, Water Res. 20 Ž1986. 73–77.
w5x A. Mazuelos, I. Palencia, R. Romero, G. Rodriguez, F.
Carranza, Design variables in high efficiency reactors for the
biooxidation of ferrous iron in solution, in: R. Amils, A.
Ballester ŽEds.., Biohydrometallurgy and the Environment
Toward the Mining of the 21st Century, Part A, 1999, pp.
501–510, Amsterdam ŽNetherland..
w6x M.J. Garcıa,
´ I. Palencia, F. Carranza, Biological ferrous iron
in packed-bed columns with low-grades sulphide mineral as
support, Process Biochem. 24 Ž1989. 84–87.
w7x E.D. Lancy, O.H. Tuovinen, Ferrous iron oxidation by
Thiobacillus ferrooxidans immobilized in calcium alginate,
Appl. Microbiol. Biotechnol. 20 Ž1984. 94–99.
w8x E. Livesey-Goldblatt, T.H. Tunley, I.F. Nagy, Pilot plant
film oxidation ŽBacfox process. of recycled acidified ura-
nium plant ferrous sulphate leach solution, in: Schwartz
ŽEd.., Conference Bacterial Leaching, Verlag Chemie, Wein-
heim, FRG, 1977, pp. 175–190.
w9x H. Armentia, C. Webb, Ferrous sulphate oxidation using
Thiobacillus ferrooxidans cells immobilized in polyurethane
foam support particles, Appl. Microbiol. Biotechnol. 36
Ž1992. 697–700.
w10x F. Carranza, M.J. Garcıa, ´ Kinetic comparison of support
materials in the bacterial ferrous iron oxidation in packed-bed
columns, Biorecovery 2 Ž1990. 15–27.
w11x S.I. Grishin, O.H. Tuovinen, Fast kinetics of Fe 2q oxidation
in packed-bed reactors, Appl. Environ. Microbiol. 54 Ž12.
Ž1988. 3092–3100.
w12x J.M. Gomez,
´ D. Cantero, Soporte para la inmovilizacion ´ de
´ 354 Ž1999. 171–175.
un biocatalizador, Ing. Quım.
w13x A. Mazuelos, R. Romero, I. Palencia, N. Iglesias, F. Car-
ranza, Continuous ferrous iron biooxidation in flooded packed
bed reactors, Miner. Eng. 12 Ž1999. 559–564.
w14x D.G. Karamanev, Model of the biofilm structure of
Thiobacillus ferrooxidans, J. Biotechnol. 20 Ž1991. 51–64.