Hydrometallurgy 59 2001.

217–231
www.elsevier.nlrlocaterhydromet

The design of bioreactors

G. Rossi)
Dipartimento Geoingegneria e Tecnologie ambientali, UniÍersita` degli Studi, Piazza d’Armi, 19, 09123 Cagliari, Italy
Received 19 October 1999; accepted 24 March 2000

Abstract

A survey of the literature on biohydrometallurgical topics revealed that the papers devoted to bioreactors amount to less
than 5% of the total and refer mainly to the analysis of the performance and to design guidelines of stirred tank and of bubble
column machines, the so-called Pachuca tanks. These machines can be defined conventional in the sense that they have been
borrowed from chemical engineering and hydrometallurgy and adapted to the requirements of biohydrometallurgi-cal
processes. However, past experience has shown that these types of reactors do not fully match the very particular conditions
that exist in biohydrometallurgical systems that, quite correctly, have been qualified as AhybridB owing to their specificity
since they are characterized by many of the features of hydrometallurgical operations and of biological conversions.
Literature data and the author’s personal experience demonstrate that one of the present major drawbacks of these reactors is
the power requirement that seriously affects the competitiveness of biohydrometallurgy with pyrometal-lurgy. The factors
affecting the performance of biohydrometallurgical reactors are discussed with special reference to the process parameters
and an analysis of the conditions to be satisfied by an ideal bioreactor is carried out. In the light of these considerations, the
reactors currently operating in commercial plants are examined. The new prospects opened up by recent developments are
finally discussed and, also on the grounds of experience recently gained on a laboratory scale, the potentials of machines
tailor-designed for the conditions reigning in biohydrometallurgical systems are outlined. q 2001 Elsevier Science B.V. All
rights reserved.

Keywords: Bioreactor; Biorotor; Stirred tank reactor; Pachuca; Solubilization rate; Design; Mass transfer coefficient

1. Introduction published results of investigations on reactors suit-

able to biohydrometallurgical processes only repre-
Compared to the great effort devoted worldwide sent a small minority. In fact, out of the total number
to the biology and physiology of microorganisms and of papers published in the volumes of Symposia
to biosolubilization kinetics in the light of mi- Proceedings and in the journals over the years, not
croberminerals interactions—where the influence of even 5% have been devoted to bioreactor technol-
reactors has unfortunately been overlooked—the ogy.
Most of these papers provide very good design
guidelines, and an indication of the excellent cultural
)
Tel.: q39-070-675-5528; fax: q39-070-675-5523.
and practical background of the authors in chemical
E-mail address: grossi@vaxcal.unica.it G. Rossi.. engineering. However, they are based on the implicit

0304-386Xr01r$ - see front matter q 2001 Elsevier Science B.V. All rights reserved.
PII: S 0 3 0 4 - 3 8 6 X 0 0 . 0 0 1 6 1 - 4
218 G. Rossi rHydrometallurgy 59 (2001) 217–231

assumption that the microflora is kind of a biological 3. The factors affecting biohydrometallurgical re-
catalyzer, whose overall performance is only moder- actor performance
ately dependent on the operating characteristics of
the machines where the process is carried out. In my Biohydromellurgical processes take place in three-
mind this is probably the Achille’s heel of this phase systems, consisting of i. an aqueous phase, that
approach. is a solution of salts providing the nutri-ents for a
Hence, the reason to review and develop the microflora that acts as a biological cat-alyzer of the
subject of biohydrometallurgical reactor design was metal sulphides oxidation processes, ii. a solid phase,
the desire to discuss the state of the art and to point to consisting of the finely ground ore that contains a
the need for further research aimed at providing our mixture of waste rock and metal values combined
technology with suitable machines where the with sulfur to form sulphides, and iii. a gaseous
potential of biohydrometallurgy can be fully ex- phase consisting of a mixture of atmospheric oxygen
ploited. and carbon dioxide. The aqueous phase is the
suspending medium where several elementary pro-
cesses occur: a. the growth of microorganisms, b. the
encounter of solid particles with microorganisms, c.
2. Expected performance of biohydrometallurgi-
the encounter of solid particles with chemically
cal reactors active molecules, d. the release of metal ions, e. the
uniform distribution and effective dissolution of
Biohydrometallurgical processes are very attrac- oxygen and carbon dioxide. The solid phase is the
tive insofar as they present few environmental haz- energy source for microbial biosynthesis, i.e. for
ards. However, they are still a long way from being microbial growth and continuously releases metal
able to compete with pyrometallurgical and pressure and sulphur ions in oxidized form. The gaseous
leaching processes mainly because of the unsatisfac- phase supplies the oxygen required for the oxidation
tory all-round performance of the reactors, the de- processes as well as the carbon dioxide that the
vices where the process is carried out. microflora uses for its biosynthesis w4x.
The performance of a reactor is considered eco-
nomically convenient when—for comparable quali-
tative and quantitative characteristics of the end 4. The general design procedure of a biohy-
product—the incidence of the investment and opera- drometallurgical reactor
tion costs on the unit product—in our case the tonne
of metal solubilized and recovered—is reasonably The design procedure shown in the block diagram
lower than the expected returns and, in any case, is of Fig. 1 is based on the well documented expected
lower than that of other processes. performance of a biohydrometallurgical reactor.
The parameters for assessing the biohydrometal- The inputs are represented by the AproductionB and
lurgical performance of a reactor for a given produc- Akinetic dataB. At this juncture it should be stressed
tion are i. tank size related to the dry mineral that in effect, the real independent outputs are the
throughput, ii. total power requirements, i.e. the production data, insofar as the kinetic data are strongly
power for mixing and aeration, referred to the unit dependent upon reactor type and envi-ronment. This has
mass of metal recovered in unit time in bioleaching been highlighted in recent papers w5,6 x and justifies
or to the unit mass of sulphur removed in unit time in the AfeedbackB line linking the AType of reactorB box
biodesulphurization, iii. the chemical compounds to the AKinetic dataB box. In fact, the microflora is
added to the aqueous phase as nutrients for the more than a simple catalyzer. In conventional chemical
microflora or as pH modifiers, iv. plant attendance engineering processes the amount of catalyzer does not
and supervision, v. effluents purification and vi. vary with time, whereas in biohydrometallurgical
maintenance. The first parameter affects investment processes the microbial population—whose growth has
costs, the other five determine operating costs. These been described as a Apseudo autocatalytic processB—
costs are of the same order of magnitude w1–3x. may vary with time
 G. Rossi rHydrometallurgy 59 (2001) 217–231
Fig. 1. Logical diagram for the design of a biohydrometallurgical reactor.

219
220 G. Rossi rHydrometallurgy 59 (2001) 217–231

depending on the conditions reigning in the bioreac- mission of European Communities, produced evi-
tor. The faster the microbial growth kinetics, the dence that, as far as the achievement of the desired
faster the oxidation and solubilization process. The levels of mass transfer, mixing and suspension is
most important factors influencing microbial growth concerned, the STR performs better than the Pachuca
are oxygen availability, determined by its mass trans- tank R. Meester, pers. written commun., 1992.. The
fer coefficient, and substrate availability, that de- experimental data reported by Acevedo and Aroca
pends upon the exposed mineral surface, both w12x also show the STR to be superior in this sense.
achieved by adequate mixing. The most significant The most important role is played by the impeller,
factors adversely affecting microbial growth are the which has to accomplish three major tasks, solids
shear stresses within the suspension, the accumula- suspension, mixing and dissolution of the required
tion of metabolites, incorrect temperature and hydro- atmospheric oxygen into the aqueous phase, maxi-
gen ion concentration of the aqueous phase and the mizing the interfacial area between the gaseous and
release of toxic substances by the surfaces minerals aqueous phases. Initially, and for many years, the
and equipment. with which the suspension comes Rushton-type turbine was the most widely used im-
into contact. peller for these reactors, but lately the curved blade,
The reactors most commonly employed in biohy- axial flow impeller has been shown to outperform the
drometallurgical processes are the Stirred Tank Re- Rushton turbine as it requires less power for
actor STR., and the Air Lift Reactor ALR. Pachuca achieving the same performance and induces smaller
tank. w7x. shear stresses in the suspensions w13,14 x. Chemical
Other reactors have been proposed recently but engineering has provided some correlations that help
are dealt within a separate section as they can be in establishing, as a first approximation, the ma-
regarded as typical examples of possible develop- chine’s characteristic parameters also with reference
ments of bioreactors tailor-made for biohydrometal- to the different types of agitators.
lurgy. As a specific design procedure has to be Hence, for the impeller speed, Njs , necessary to
followed for each type of reactor, in the following satisfy the just-suspended condition for solid parti-
sections the main types of reactors will be consid- cles in the vessel, AZwietering’s criterionB w15x is
ered separately. usually adopted, for which the author prefers the
Some time ago it was suggested that the plug-flow following expression:
reactor is the most suitable to the type of reactions
occurring in metal sulphide bioleaching w4,8–10x. 2 0 .1 2 0 .45 0 .2

/
_
However, on account of the almost unsormountable
practical problems that arise with this type of reactor,
ND

Reynolds number
N D
g _
l

Froude’s number
/
D
d /
it has to be simulated by a cascade consisting of a
suitable number of vessel reactors. It has been shown _
T
that at least six vessels are required w4,11 x as a
smaller number results in short-circuiting of the sus-
pension with loss of effectiveness. sk D /
B 0 .13 1.
where: N srotational frequency of the impeller, D
4.1. The stirred tank reactor (STR) s impeller diameter m., _sviscosity Pa s m., _l
sliquid density kg my3 ., g sacceleration of gravity m
This reactor, borrowed from chemical engineer-
ing, has been given priority since the early days of sy2 ., _ _sdensity difference, gas–liquid kg my3 ., d
biohydrometallurgy, in spite of the evident draw- sparticle diameter m., k sconstant see text.,
backs that have emerged in biohydrometallurgical _sconstant see text., T stank diameter, B sdistance
applications. A very interesting technico-economic from impeller midplane to tank bot-tom, m.
analysis, carried out by one of the European partners This equation highlights the relationship existing
in the coal biodepyritization pilot plant project at between Reynolds’ number, Froude number, the ra-
Porto Torres Sardinia, Italy., funded by the Com- tios of the impeller diameter to the particle diameter
 G. Rossi rHydrometallurgy 59 (2001) 217–231 221

and of the tank diameter to the impeller diamater and Table 1

the percentage ratio B, between liquid and solid Impeller speed calculated using Zwietering’s and Nienow’s corre-
lations
masses in the suspension w16x. Since all the
quantities in brackets only refer to the physical Rushton Zwietering’s correlation Nienow’s correlation
system, they can be symbolized by a single turbine rad sy1 m sy1 rpm rad sy1 m sy1 rpm
diameter m
dimensional settling parameter Y, while the impeller
is characterized by its dimensionless value S w16x, 1.00 2.59 2.59 25.00 4.45 4.45 42.50
0.67 3.65 2.44 35.00 10.00 6.68 95.60
where S sk TrD._ and k and _ depend on impeller
type and relative blade height.
Finally, the criterion can be written:

Njs sSYDy0 .85 2. _ svolume fraction of liquid in the suspension volume

concentration of solids., dimensionless, VG sair
When designing the Porto Torres reactors the superficial velocity based on tank cross-section gas
following correlation, proposed by Nienow w17x, flowrate divided by cross-sectional area of the tank
was used:
my1 ..
4 For a six flat-blade Rushton disc impeller, Neale
and Pinches w21x report the following application of
NCD s D 2 Q G0.5 T 0 .25 3. van’t Riet’s equation:
P 0 .52
where: NCD srotational frequency of the impeller at G
which the gas is just dispersed throughout the vessel.
QG svolumetric gas flowrate m 3 sy1 ., for cross- k L a s0.0069 V /
VG0 .24 6.
checking the results obtained with Zwietering’s crite- where: PG sagitation shaft power gassed., kW, and
rion. It was no surprise that the values obtained were for the BX04 Impeller:
quite different, as Table 1 shows. P 0 .79
The optimum speed, measured over one year’s G

pilot plant operation, was 5.65 rad sy1 54 rpm., 34%

higher than Zwietering’s correlation and 43% lower
k L a s0.0084 V /
VG0 .58 7.
It goes without saying that adequate experimentation
than w18x. is required.
For impeller power prediction either Mills et al’s.
Both correlations require the tank volume and the
correlation w19x: air superficial velocity to be known. The useful tank
1 .4925
volume is calculated by multiplying the suspension
PT kLa 1 volumetric flow rate, VP , by the residence time, __ p
; the latter, in turn, is derived by setting the desired
V s 0 .10 y0.0018 _ /VG0 .4627 4. percent sulphide removal, _ p, and introducing this
or van’t Riet’s correlation w20x: condition into the equation being K c skinetic coef-
ficient for pyrite solubilization, hy1 .:
P k L a.1 .4286 1 ln 1 y_ p .
Vs 0.0001 V 0 .2857 5. __ p s , 8.
G
yK c
can be used, where: PT stotal power input agitator which was derived for pyrite w22x, under the
power qbubble expansion power, kW., P spower to assump-tion that its oxidation reaction is first order,
produce suspension to height H kW.. V sliquid by integrating the following equation:
volume plus particle volume below air–liquid inter-
face m 3 ., k L soverall liquid phase mass transfer d w FeS2 x
syK c w FeS 2 x 9.
coefficient m sy1 ., a sinterfacial area m 2 my3 ., dt
222 G. Rossi rHydrometallurgy 59 (2001) 217–231

The kinetic coefficient, K c , is obtained empiri- 4.2. The Pachuca tank

cally, usually from bench scale tests carried out
either in agitated Erlenmeyer flasks or in STRs Several researchers w4,26–29x have developed
whose carefully optimized performance is taken as correlations for predicting power requirements and
the absolute optimum for pyrite oxidation kinetics. It oxygen mass transfer coefficients of this type of
is reasonable to assume that it is precisely this reactor. A great deal of work has been done by the
passage that is the pitfall behind the design proce- Delft University school. After Bos et al. w30x, for
dure; in effect, this assumption equates to stating that large Pachucas the following simple correlation
the pyrite oxidation kinetics in the STRs are the best holds:
achievable and are intrinsic of the biooxidation pro-
cess.
k L a s0.6VG 10.
Thus, the K c adopted for the Porto Torres plant whereas Boon et al. w29x proposed the following
bioreactors was 1.2 =10y2 hy1 w22x, but the value empirical correlation:
calculated from the results of one year’s plant opera- Q 0 .66

tion turned out to be considerably higher, 1.53 = 10 y2 G

hy1 w23x. k L a s 2 .29 =10y2 V / 11.

Tests carried out using bioreactors of new concept where: k L a svolumetric mass transfer in a slurry,
operated under the same conditions as the STRs sy1 , that, combined with Lamont’s power dissipation
w24x yielded much higher K c . law w26x yields:
In effect, the dependence of K c on several factors
is well documented though never explicitly stated. Of P 0 .66

these, solids concentration plays a major role in

STRs and in Pachuca tanks. The limit of about 20% k L a s2.29 =10y2 V /
solids concentration for metals sulphides bioleaching 1
in STRs was experimentally ascertained by several = 12.
researchers w25x as long as 30 years ago and today p 0 q_f gH
has become an accepted rule for commercial plants,
as shown by Table 2. This limitation is the major
p ln
0 p0 / 0 .66

drawback of STRs, since it affects both investment where: p0 satmospheric air pressure Pa., _f sfluid
costs size of the machinery. and operating costs phase pressure Pa., H ssuspension interface height
power and maintenance.. above vessel bottom m..

Table 2
Data of commercial and semicommercial biohydrometallurgical plants
Name of mine or plant Feed to Sulphur Reactor Reactor Daily Solids Residence Reference
the plant assay % type size m 3 . throughput concentration time h.
tonnes. %.
Fairview South Africa. GAP 22.6 STR 90 35 20 96 w 3,37x
Harbour Lights Australia. GAP 18 STR 160 40 w 3,37x
Sao Bento Brazil. GAPyP 24.9 STR 1 =580 150 20 24 w37,38x
Wiluna Western Australia. GAP 20–24 STR 9 =450 158 20 120 w39x
Youanmi Western Australia. GAP 20–30 STR 6 =480 120 18 91.2 w40,41x
Sansu Ashanti, Ghana. GAPyP 11.4 STR 3 modules 720 20 96 w 3,42,43x
6 =900
per module

GAP sgold-bearing arsenopyriterpyrite flotation concentrate; GAPyP sgold-bearing arsenopyriterpyrrhotite and pyrite flotation concen-trate.
 G. Rossi rHydrometallurgy 59 (2001) 217–231 223

According to Meester Meester, pers. written bioleaching—such as that produced by shear stresses
commun. 1992., an equation describing the correla- acting within the suspension—causes irreversible
tion among the variables influencing mixing can be damage to microbial cells that, despite still being
derived if it is assumed that mixing takes place in viable, lose their ability to adhere and oxidize the
two steps, and on this there is general consensus. On minerals.
a large scale, the flow of the reactor contents pro- Moreover, Hansford and Bailey w32,33,36 x pro-
duces mixing by convection, while further mixing duced evidence that the determining factor in bi-
takes place by diffusion on smaller scale. The state of oleaching is the proportion of oxidizable solids, i.e. the
mixing that thus is attained by turbulent flow is oxygen-consuming solid fraction in the suspen-sion, not
determined by the size of the smallest eddies in the simply the solids concentration. Those solids that are
flow pattern, for which the following definition is inert to oxidation only slightly affect the process. Under
proposed by Hinze w31x: the above assumptions, the presumed threshold value of
1
about 20% solids concentration would mean, quite
4 simply, that the oxygen mass transfer coefficient of the
_ f3
STRs and Pachucas cur-rently used in
le s P 13.
biohydrometallurgy does not provide, in the best
_m2 V / operating conditions, enough oxygen for oxidizing
where: le ssmallest turbulent eddy size m., _f s larger amounts of metal sulphides.
intrinsic fluid phase viscosity Pa s., _m sslurry Any attempt to enhance reactor performance by
increasing aeration inevitably results in increased
density below slurry-liquid interface kg my3 ..
agitation and greater shear stresses within the sus-
For suspensions, Zwietering’s correlation is valid
also for Pachucas, within the above mentioned lim- pension and possibly waste of injected air due to
its. partial flooding. Greater damage to microorganisms
The bioleaching performance of Pachuca tanks and insufficient oxygen mass transfer are probably
also rapidly decreases when the oxidizable solids the reasons behind what may be defined the A20%
concentration rises beyond 20% w25x with the same threshold solids concentrationB.
implications, as far as metal sulphide solubilization The experimental results of bench scale and pilot
rate constant is concerned, already discussed for the plant operation for coal biodepyritization support
STRs. these conclusions. At Deutsche Montan Technologie
DMT. in Essen, Germany, and at Delft University
4.3. The importance of solids concentration The Netherlands. bench scale tests were carried with
Pachuca tanks that were successfully operated up to
All endeavours to employ STRs and Pachuca 40% coal assaying 2% pyritic sulphur. w44,45 x; the
tanks for bioleaching metal sulphide suspensions 8-m3 STRs of the Porto Torres pilot plant per-formed
with oxidizable solids concentrations higher than very well with a 40% solids suspension of the same
about 20% have so far failed. coal w25x.
The reasons for this limitation have been the In this perspective, the difficulties in developing a
subject of much speculation, extensively discussed in satisfactory conventional bioreactor design are obvi-
earlier reviews w4,7,32,34 x. The problem has been ous and probably unsurmountable.
approached from the strictly abiotic viewpoint in
chemical engineering journals with several interest-
ing suggestions. However, only recently has some 5. The new developments in biohydrometallurgi-
light finally begun to be shed on these intriguing cal reactor design
issues thanks to the contributions of South African
and Australian researchers w35,36 x. The growing awareness of the inadequacy of STRs
Evidence was provided by Ragusa that mechani- and of ALRs to cope with the problems inherent in
cal and brutal detachment of microorganisms from biohydrometallurgical processing owing to the con-
the mineral surfaces to which they adhere during flicting requirements to enhance agitation, for Oxy-
224 G. Rossi rHydrometallurgy 59 (2001) 217–231

gen Transfer Rate OTR. increase, and the one hand, gular tank with V-shaped bottom, along which a
and of a quiescent environment with negligible shear perforated pipe runs that acts as an air sparger Fig. 2..
stresses on the other, has prompted the search for The reactor is structurally identical to the Callow-
new types of reactors where these conditions are type pneumatic flotation cell w49x. Its most
fulfilled to as great an extent as possible. significant feature resides in the fact that it operates
Following, these new developments will be re- in a manner very similar to the plug flow reactor,
viewed. hence complying with the kinetics of pyrite bioleach-
It is fair to mention first the contributions by ing w9x.
Andrews et al. w46,47 x who—with a view to develop- As far as aeration is concerned, it is doubtful that
ing a reactor suitable for coal biodesulphurization— there was any improvement in OTR, since the draw-
already in the late 1980s focussed their attention on the backs of pneumatic reactors, like the Pachucas, do
features that this new reactor should have had. Based on not seem to have been overcome. In effect, the k L a
the consideration that the relatively low value of coal in water was 0.070 sy1 , whereas it decreased to
called for low biodesulphurization plant investment and
0.012 sy1 in a 45% solids concentration pulp, i.e. by
operating costs, Andrews found, when investigating the
about one order of magnitude wAndrews, 1990x: this
kinetics of pyrite bioleaching, that small reactor suggests that aeration was unsatisfactory.
volumes, high solids concentra-tions of the suspensions
and the largest possible active microbial populations 5.2. The Low Energy Bioreactor [50]
were the objectives to be pursued. As for microbial
Almost contemporaneously to Andrews’ aerated
populations, Andrews committed himself to giving the
trough a research team of the CRA Company, active
size of the most suitable microbial population: 1018 in Australia, made a successful attempt to solve the
bacteriarcm3! A figure that then would have probably problem of maximizing OTR without inducing ex-
been consid-ered almost science fiction. cessively large shear stresses in the suspension.
The device they developed, called a A Low Energy
Bioreactor B, basically consists of three components
5.1. The aerated trough bioreactor [48] Fig. 3.: i. a tank, where bioleaching is carried out; the
Aoff-the-bottomB condition for the solids being
On these grounds, Andrews and his research team satisfied by an agitator, operating inside of a draft tube,
developed what they called A aerated trough bioreac- most likely for reducing the shear stresses; ii. an
tor B. The device basically consists of a long rectan- aeration device where the pulp is very energeti-

Fig. 2. Aerated trough bioreactor w 48x .

 G. Rossi rHydrometallurgy 59 (2001) 217–231 225

Fig. 3. Low energy bioreactor, after w50x.

cally aerated by means of a Venturi pipe; iii. a pump experimentally by Andrews and Quintana w48x. Nei-
for pulp recirculation. ther data concerning cell growth, cell numbers and
This device represents a major step forward in the residence times could be found. Furthermore, the
development of a tailor-made bioreactor for biohy- inventors claim that their machine was developed for
drometallurgical processes; certainly as far as power processing low-grade ores such as run-of-mine. ob-
requirements are concerned, as emphasized by the taining iron leaching rates from pyrite of about 6 g
authors. The resort to separate aeration is also a dmy3 dayy1 . However, this performance is not ex-
significant feature, insofar as it permits the pulp to be ceptional since in the STRs of the Porto Torres pilot
aerated as intensively as needed. plant the research team of the CEC project observed
However, the suspension is subjected to violent iron leaching rates from pyrite in coal, hence compa-
turbulence and shear during its residence inside the rable to a low-grade run-of-mine ore, as high as 9.2 g
aerator. dmy3 dayy1 .This result suggests that the objective of
Further doubts remain as to the shear stresses i. achieving an atmospheric oxygen mass transfer
generated within the pulp by the impeller, ii. pro- coefficient high enough to promote a drastic increase
duced inside of the annulus where, similarly to what in microbial population and finally in pyrite solubi-
happens in the Pachuca tank, the pulp flows upwards liztion kinetics has not been attained.
and iii. induced in the pulp by the pump, by means of
which the pulp is repeatedly circulated. 5.3. The falling laminar liquid film principle
It was not possible to find data concerning the k L
a of this device in the literature. It is a well known It has long been recognized w51x that mass
fact that air flowrate alone is not sufficient, insofar as transfer across gas–liquid free surfaces plays a very
it does not provide a measure of the air actually impor-tant role in several natural and technological
dissolved into the pulp, as demonstrated pro-cesses, ranging from oxygen supply and CO 2 re-
226 G. Rossi rHydrometallurgy 59 (2001) 217–231

moval from animal cells and cultures to aeration of the solids are kept in suspension by an inclined
the culture media of shaken flasks or STRs, wastew- agitator A.; a pumping device P., draws the sus-
ater treatment and industrial microbial processes, like pension from the main tank and conveys it, through a
wine–vinegar manufacture and respiration of aerobic distributing system, to the head tanks D., that feed
life occurring near the sea or lake surfaces w4,52 x. the inclined plates S p .; the suspension flows down
A free-falling laminar liquid film poured into a pool these slopes cataracting into the main tank. Accord-
of the same liquid can be designed in such a way as ing to the inventors, atmospheric oxygen transfer is
to cause very slight shear stresses in its bulk or in the produced by the sheets of suspension plunging into
liquid layers located near to the surface of the pool. the liquid surface of the main tank and entraining air
Therefore, it is only natural to consider investigating bubbles. The magnitude of oxygen transfer is ad-
the potentials of a three-phase biohydrometallurgical justed by setting the length and slope of the inclined
reactor designed in such a way as to utilize, for plates and of the free-fall height as well as of the
atmospheric oxygen transfer, the properties of falling suspension flow rate.
laminar liquid films. Based on this concept two The device has only been tested with plain water
independent research teams, operating at the Univer- and the authors claim lower power requirements and
sities of Cagliari Italy. and Delft The Netherlands. higher k L a’s than STRs and Pachucas. The highest k
endeavoured to develop this new type of bioreactor. y1
L a value in water is reported to be 0.0384 s , i.e.
138.24 hy1 , hence very close to the k L a of the STR
5.4. The Delft Inclined Plate (DIP) Bioreactor [53] Fig. 5.. The most attractive feature of the device
seems to be the low power demand and, possibly, the
This device is schematically shown in Fig. 4. It mild agitation required for mixing. However, the
consists of a main tank T., containing the slurry S.: presence of the agitator, A, and of the pumping

Fig. 4. Delft Inclined Plate Bioreactor. A sagitator; D shead tanks of inclined plates; S p sinclined plates for slurry cascading; S sslurry; T
smain tank; P spumping device.
 G. Rossi rHydrometallurgy 59 (2001) 217–231 227

Fig. 5. Mass transfer coefficients for atmospheric oxygen in water for the most common bioreactors. 1. plunging jet reactor w 56x ; 2. deep
jet fermenter w56x; 3. mechanically stirred tank w56x; 4. axial flow turbines w 56x; 5. multiple Rushton turbines w 56x ; 6. single Rushton
turbine w56x; 7. simple bubble reactor w56x; 8. deep shaft areator w 56x ; 9. air-lift Pachuca-type reactor w 28x ; 10. vortex agitation
reactor w56x; 11. Biorotor w 55x; 12. low energy bioreactor w 50x; 13. aerated trough bioreactor w 47x ; 14. Delft Inclined Plate Bioreactor
w 53x ; 15. Biorotor w57x; 16. aerated trough bioreactor measured in a 45% solids suspension w 47x..

device still appears to involve some shear stresses scale the Achille’s heel of this new, promising bio-
within the suspension, thus partially offsetting the technology was the inadequacy of the reactors used
advantage of less violent aeration. w7x. Previous experiences were very helpful: the
No testing on mineral suspensions has been car- sink-and-float drums were very effective for gravity
ried out so far with the DIP, hence it is not possible sepa-ration of ores their operation being very quiet
to properly assess its potential for biohydrometal- and smooth. The only drawback was that, in certain
lurgy. As far as the mode of oxygen transfer is conditions, the entrainment of air caused by part of
concerned, the contribution of the contact of the the suspension plunging into the pool that formed in
relatively large surface area of the cataracting water the lower half of the drum produced a certain insta-
sheet with the atmosphere should not be neglected, bility.
although very little is known on this topic w51,54 x, In the sink-and-float drum, this was an undesir-
and it warrants further investigation. able effect: the reverse would have been true had it
been intended to mix the air with the suspension.
The first prototype of the revolving drum bioreac-
5.5. The reÍolÍing drum bioreactor A Biorotor B [55]
tor was developed in 1991, but it was only in 1993
that the final version of what was baptized ABioro-
From the very first endeavours to transfer biohy- torB was thoroughly tested, first with plain water and
drometallurgy from bench to pilot to commercial finally with pure pyrite.
228 G. Rossi rHydrometallurgy 59 (2001) 217–231

An isometric view of the device is shown in Fig. out any Adead volumesB; ii. the only shear stresses
induced into the suspension are limited to the re-
6: it has been exhaustively described in earlier pa- lative motion of the plunging film with respect to the
pers, which the interested reader should consult for suspension in the pool: for a 2-m diameter bar-rel,
more details. In short, the reactor consists of a
'
cylindrical barrel whose inner wall is fitted with (
this relative speed is lower than 2 gHf s 2 P9.81
regularly spaced lifters L.. Each lifter forms, with the
P1.4 s5.24 m sy1 the free-fall height, Hf , being 0.7D.
inner wall of the barrel, a sort of tray where the and the suspension drops into the upper few
suspension collects. The suspension and the air plus
millimetres of the top layer, iii. a value of k L a of at
carbon dioxide mixture are conveyed through a pipe least one order of magnitude larger than that of an
T. fitted into one front head M. of the barrel the optimized STR Fig. 5..
Afeed headB.; a pipe fitted into the opposite head is The most interesting feature of this bioreactor is,
the exhaust outlet. however, that the OTR actually matches the k L a :
As the barrel revolves on its rollers R., the i.e. the oxygen is made available to the microflora
suspension is lifted upwards and when the tray that can attain much higher growth kinetics than
reaches the top position it is discharged as a thin those currently reported in the literature. Tests car-
cataracting film with length equal to height of the ried out with a 30% solids suspension, the solids
cylindrical barrel. being museum-grade pyrite, yielded solubilization
Oxygen mass transfer most occurs during the rates as high as about 600 g my3 hy1 . This means
cataracting and when the free falling film plunges that the rate constant is at least one order of magni-
into the pool of suspension in the lower part of the tude higher than the value considered until now as
barrel. the maximum obtainable.
This mode of operation ensures the satisfaction of Hence, the Biorotor exploits, at a very high level,
all the prerequisites listed above for the ideal biohy- the potentials of the microflora for enhancing pyrite
drometallurgical reactor: i. mixing is complete with- solubilization.

Fig. 6. Isometric view of Biorotor.

 G. Rossi rHydrometallurgy 59 (2001) 217–231 229

Fig. 7. Plot of k L a vs. rotation speed for Biorotor w55x.

Fig. 7 gives the values of k L a corresponding to points out the undeniable limitations of the reactors
the range of rotation speeds considered as most used so far for metal sulphide minerals bioleaching.
suitable for biohydrometallurgical purposes w55x. It These limitations are related to the intrinsic struc-
seems rather trivial, but some remarks that have ture and mode of operation of these machines insuf-
appeared in the literature seem to justify pointing out ficient mass transfer coefficient, high shear stresses
that at rotation speeds higher than _s 0.5 gD.0.5 induced into the mineral suspensions, to mention just
Acritical speedB, corresponding to centrifugation of the most significant. as well as to poor accuracy of
the suspension. the machine does not work. For the the design formulae.
Biorotor prototype, which has a diameter of 0.3 m, The new devices proposed over the last decade are
this critical speed is 7.62 rad sy1 corresponding to a clear demonstration that the designers have
72.8 rpm.. Investigations carried out on a refractory achieved a full understanding of the specific features
gold-bearing complex sulphides concentrate with a that should characterize the biohydrometallurgical
conceptually similar machine w57x confirmed the reactors: namely, effective Oxygen Transfer Rate
su-perior performance of Biorotor, although it was OTR., effective but mild mixing and, consequently,
only operated at 0.16 rad sy1 . In effect, extrapolating low retention times and low specific power require-
the lower branch of the diagram of Fig. 7 yields the ments.
same values for k L a. The experience gained with the latest develop-
ments, the revolving drum bioreactor, has clearly
demonstrated that with an efficient reactor microbial
6. Conclusions
growth can also be optimized.
The experience gained over more than 30 years of Most of these encouraging results have, however,
laboratory research and commercial operation clearly been obtained empirically: more theoretical research
230 G. Rossi rHydrometallurgy 59 (2001) 217–231

is needed so as to develop reliable design and scale- w15x T.N. Zwietering, Suspending of solid particles in liquid by
up procedures based on sound theoretical founda- agitators, Chemical Engineering Science 8 1958. 244–253. w16x
J.J. Frijlink, PhD Physical aspects of gassed suspension
tions. reactors, Thesis, University of Delft, 1989, 119 pp.
w17x C.M. Chapman, A.W. Nienow, M. Cooke, J.C. Middleton,
Particle-gas–liquid mixing in stirred vessels. Part IV: Mass
transfer and final conclusions, Chemical Engineering Re-
References search and Design 61 1. 1983. 182–185.
w18x G. Loi, A. Mura, P. Trois, G. Rossi, La biodepiritizzazione
del carbon fossile. L’impianto pilota di Porto Torres, Memo-
w1x A.J. Carter, Economic comparison of the alternative methods for rie dell’Associazione Mineraria Sarda-Iglesias, Italia, 1994,
the recovery of gold from refractory gold ores, Collo-quium Bacterial pp. 29–41.
Oxidation, S.A.I.M.M, Johannesburg, 1991. w2x J. Barrett, M.N. w19x D.B. Mills, R. Bar, D.J. Kirwan, Effects of solids on oxygen
Hughes, G.I. Karavaiko, P.A. Spencer, Metal Extraction by transfer in agitated three-phase systems, AIChe Journal 33 9.
Bacterial Oxidation of Minerals, Ellis 1987. 1542–1549.
Horwood, New York, 1993, 191 pp. w20x K. van’t Riet, Review of measuring methods and results in
w3x D.W. Dew, E.N. Lawson, J.L. Broadhurst, The Biox process gas–liquid nonviscous mass transfer in stirred vessels, Indus-
for gold-bearing ores or concentrates, in: D.E. Rawlings Ed. trial and Engineering Chemistry Process Design and Devel-
., Biomining, Springer Verlag, Berlin, 1997, pp. 45–80. w4x G. opment 18 1979. 357–364.
Rossi, Biohydrometallurgy, McGraw-Hill, Hamburg, w21x J.W. Neale, A. Pinches, Determination of gas–liquid mass
1990, 609 pp. transfer and solids suspension parameters in mechanically-
w5x G. Loi, P. Trois, G. Rossi, Biorotor: a new development for agitated three-phase slurry reactors, Minerals Engineering 7
biohydrometallurgical processing, in: T. Vargas, C.A. Jerez, 2r3. 1994. 389–403.
J.Y. Wiertz, H. Toledo Eds. ., Biohydrometallurgical Pro- w22x N. Orsi, G. Rossi, P. Trois, P.D. Valenti, A. Zecchin, Coal
cessing, vol. 1. The University of Chile, Santiago, 1995, pp. biodesulfurization: design criteria of a pilot plant, Resources
253–271. Conservation and Recycling 5 1991. 211–230.
w6x G. Loi, A. Mura, N. Passarini, P. Trois, G. Rossi, Modifica- w23x G. Loi, A. Mura, P. Trois, G. Rossi, The Porto Torres
tion of oxidizing activity of Thiobacillus ferrooxidans by biodepyritization pilot plant: light and shade of one year
some particulate solids, Fuel 72 1993. 1607–1611. operation, Fuel Processes and Technology 40 1994. 261–
w7x G. Rossi, ICP-La miscelazione nei reattori per bioidrometal- 268.
lurgia, Rivista dell’Industria Chimica 1993. 76–86. w24x G.S. Hansford, Recent developments in modeling the
w8x T.F. Huber, N.W.F. Kossen, P. Bos, J.G. Kuenen, Modelling kinetics of bioleaching, in: D.E. Rawlings Ed. ., Biomining,
design of a reactor for microbial desulphurization of coal, in: Springer, Berlin, 1997, pp. 153–175.
G. Rossi, A.E. Torma Eds. ., Recent Progress in Biohy- w25x G. Loi, A. Mura, P. Trois, G. Rossi, Bioreactor performance
drometallurgy, Associazione Mineraria Sarda, Iglesias, Italia, vs. solids concentration in coal biodepyritization, Fuel Pro-
1983, pp. 279–289. cesses Technology 40 1994. 251–260.
w9x P. Bos, C.H. Huber, C. Ras, J.G. Kuenen, A Dutch w26x A.G.W. Lamont, Air agitation and Pachuca tanks, Canadian
feasibility study on microbial coal desulphurization, in: R.W. Journal of Chemical Engineering 1958. 153–160.
Lawrence, R.M.R. Branion, H.G. Ebner Eds. ., Fundamental w27x N.N. Clark, Predicting the circulation rate in Pachuca tanks
and Ap-plied Biohydrometallurgy, Elsevier, Amsterdam, with full height draft tubes, Transactions of the American
1986, pp. 129–150. Institute of Mining Engineers 276 1984. 226–232.
w10x C.J. Stevens, K.S. Noah, G.F. Andrews, Large laboratory w28x G.G. Roy, R. Shekhar, Oxygen mass transfer in air-agitated
scale demonstration of combined bacterial and physical coal Pachuca tanks, Section C: Mineral Processing and Extraction
depyritization, Fuel 72 1993. 1601–1606. Metallurgy 105 1996. 9–21.
w11x C.G. Hill Jr., An Introduction to Chemical Engineering Ki- w29x M. Boon, T.A. Meeder, J.J. Heijnen, K.Ch.A.M. Luyben,
netics and Reactor Design, Wiley, New York, 1977, 594 pp. w12x Influence of oxygen adsorption on the dynamic k L a mea-
F. Acevedo, G. Aroca, Studies on the agitation and power surement in three-phase slurry reactors, Biotechnology and
characteristics of mineral slurries, in: R.W. Lawrence, R.M.R. Bioengineering 40 1992. 1097–1106.
Branion, H.G. Ebner Eds. ., Fundamental and Applied Bio- w30x P. Bos, T.F. Huber, K.Ch.A.M. Luyben, J.G. Kuenen, Feasi-
hydrometallurgy, Elsevier, Amsterdam, 1986, pp. 255–261. bility of a Dutch process for microbial desulphurization of
w13x P. Forschner, H. Jekat, R. Vetterman, Ekato-Interpump: a coal, Resources, Conservation and Recycling 1988. 279–
two-bladed propeller with improved hydrodynamic proper- 291.
ties, GVC Annual Meeting for Process Engineers, Freiburg, w31x J.O. Hinze, Turbulence: An Introduction to its Mechanism
West Germany, 1987, 8 pp. and Theory, McGraw-Hill, New York, 1959, 586 pp.
w14x P.A. Spencer, D.M. Satalic, Selection of a bacterial w32x G.S. Hansford, A.D. Bailey, Oxygen transfer limitation of
oxidation plant: is it the best alternative? International bio-oxidation at high solids concentration, in: A.E. Torma,
Biohydrometal-lurgy Symposium IBS’97 BIOMINE 97, J.E. Wey, V.I. Lakshmanan Eds. ., Biohydrometallurgical
1997, paper M3.1.1–M3.1.8. Technologies, vol. 1. 1993, pp. 469–478.
 G. Rossi rHydrometallurgy 59 (2001) 217–231 231

w33x A.D. Bailey, G.S. Hansford, Factors affecting bio-oxidation bioreactor design on microbial desulfurization of coals, Ap-
of sulfide minerals at high concentrations of solids: a review, plied Microbiology and Biotechnology 24 1986. 342–346.
Biotechnology and Bioengineering 42 10. 1993. 1164– 1174. w46x G.F. Andrews, Mixing effects in coal processing bioreactors,
in: M.E. McIlwain, P.R. Dugan Eds. ., Bioprocessing of Coal
w34x T. Oolman, Bioreactor design and scale up applications in Workshop—III Tysons Corner, Idaho National Engi-neering
mineral bioleaching, in: A.E. Torma, J.E. Wey, V.I. Laksh- Laboratory, Idaho Falls, ID, USA, 1988, pp. 234– 254.
manan Eds. ., Biohydrometallurgical Technologies, vol. 1,
1993, pp. 401–415. w47x C.J. Stevens, K.S. Noah, G.F. Andrews, Large laboratory
w35x S. Ragusa, PhD Thesis, University of N.S.W., Department scale demonstration of combined bacterial and physical coal
of Biotechnology, School of Biological Technologies, biodepyritization, Fuel 72 12. 1993. 1601–1606.
Sydney, Australia, 1990. w48x G.F. Andrews, J. Quintana, Mixing and mass transfer in an
w36x A.D. Bailey, PhD Thesis, University of Cape Town, South aerated trough bioreactor, First International Symposium on
Africa, 1993. the Biological Processing of Coal, E.P.R.I., Palo Alto, CA,
w37x P.G. van Answegen, Bio-oxidation of refractory gold ores— USA, 1990, pp. 5-69–5-85.
the Genmin experience, Biomine ’93, Australian Mineral w49x R.H. Richards, C.E. Locke, Textbook of Ore Dressing, Mc-
Foundation, Adelaide, SA, 1993, pp. 15.1–15.14. Graw-Hill Book, New York, 1940, 608 pp.
w38x D.W. Dew, Comparison of performance for continuous bio- w50x W. Hoffmann, R. Batterham, D.A. Conochie, A Novel, Low
oxidation of refractory gold ore flotation concentrates, in: Energy Bioreactor for Refractory Gold Ores, Randol, Beaver
J.V. Jerez, H. Toledo, T. Vargas, J.V. Wiertz Eds. ., Biohy- Creek, 1993.
drometallurgical Processing, vol. 1. University of Santiago, w51x G.V. Levich, Physicochemical Hydrodynamics, Prentice
Chile, 1993, pp. 239–251. Hall, Englewood Cliffs, NJ, USA, 1962, 700 pp.
w39x D. Stephenson, R. Kelson, Wiluna BIOX plant—expansion w52x A.D. Bailey, D.F. Ollis, Biochemical Engineering
and new developments, Biomine ’97, Australian Mineral Fundamen-tals, McGraw-Hill, New York, 1986, 984 pp.
Foundation, Glenside, SA, 1997, pp. M4.1.1–M4.1.8. w53x G. van Weert, J.A. Snoeck, Oxygen transfer from air in
w40x N. Bell, I. Quan, The application of BacTech Australia. Delft Inclined Plate D.I.P. . bioreactor, in: A.E. Torma, J.F.
Technology for processing refractory gold ores at Youanmi Wey, V.I. Lakshmanan Eds. ., Biohydrometallurgical
Gold Mines, Biomine ’97, Australian Mineral Foundation, Technolo-gies, vol. 1. The Metals, Minerals and Materials
Glenside, SA, 1997, pp. M2.3.1–M2.3.9. Society, Warrendale, Penna, USA, 1993, pp. 237–248.
w41x P.C. Miller, The design and operating practice of bacterial w54x R.B. Bird, W.E. Stewart, E.N. Lightfoot, Transport Phenom-
oxidation plant using moderate thermophiles the BacTech ena, Wiley, New York, 1960, 780 pp.
Process., in: D.E. Rawlings Ed. ., Biomining, Springer- w55x G. Loi, P. Trois, G. Rossi, The revolving drum bioreactor—
Verlag, Berlin, 1997, pp. 81–102. a new development for biohydrometallurgical processing,
w42x H. Nicholson, S. Oti-Atakorah, D.J. Lunt, I.C. Ritchie, Bioreactorrprocess Fluid Dynamics, BHR Group, 1997, pp.
Selec-tion of a refractory gold treatment process for the Sansu 183–195.
project, Biomine ’93, Australian Mineral Foundation, Ade- w56x D.E. Brown, A. Carilli, W. Crueger, P.J.B. Dusseljee, S.-D.
laide, SA, 1993, pp. 20.1–20.11. Enfors, R.I. Fox, K. Gerlach, P.A. van Hernert, M. Jarai, ` R.
w43x H.M. Nicholson, G.R. Smith, R.J. Stewart, F.W. Kock, Manfredini, H.-P. Meyer, H.J. Michalski, A. Moser, K. Polya,
Design and commissioning of Ashanti’s Sansu BIOX plant, N. Starjak, H. van Suijdam, A. Linden, Physical aspects of
Biomine ’94, Australian Mineral Foundation, Glenside, SA, bioreactor performance, DECHEMA, Frankfurt am Main,
1994, pp. 2.1–2.8. 1987, 159 pp.
w44x T.F. Huber, C. Ras, N.W.F. Kossen, Design and scale-up of w57x M.N. Herrera, B. Escobar, N. Parra, C. Gonzalez, ´ T.
a reactor for the microbial desulphurization of coal: a kinetic Vargas, Bioleaching of refractory gold concentrates at high
model for bacterial growth and pyrite oxidation, Proceedings pulp densities in a non-conventional rotating-drum reactor,
3rd European Congress on Biotechnology 3 1984. 151–159. Miner-als and Metallurgical Processing 15 2. 1998. 15–19.
w45x M. Beyer, H.G. Ebner, Klein, Influence of pulp density and