117

forum
described the evolution of the N I H
'Points to Consider' document on
the 'Design and Submission of
Protocols for the Transfer of
Recombinant D N A into the
Genome of Human Subjects'. This
document has formed the basis of the
N I H guidelines for gene-therapy
protocols. It is important to realize
that these are guidelines and not
regulations and, as such, the points
raised are flexible and easily amended
as the pace of scientific advance
quickens. (Indeed, since their incep-
tion, they have already been
amended more than 100 times.)
Although these guidelines were
drafted to cover protocols carried out
under the auspices of the NIH, many
other individuals and institutions
have voluntarily submitted their pro-
tocols to the N I H for approval. The
committee considering new gene-
therapy protocols includes scientists,
lawyers and lay people and has done
much to dispel the myths surround-
ing gene therapy. The committee
considers all aspects of the protocols
including scientific validity, efficacy,
safety considerations and the risks
versus the possible benefit to the
patient. Wivel emphasized the need
to be flexible in the approach to
regulatory issues and to allow them
to evolve with the pace of the
research.
This conference brought together
the science, ethics and regulatory
issues of human somatic cell gene
therapy. This type of therapy is now
a reality and will shortly become
more widespread.
Fiona Watson
Trends in Biotechnology

Biofilm bioreactors for waste-

water treatment
Mark C. M. Van Loosdrecht and Sef J. Heijnen

Immobilization of microorganisms in biofilms is particularly appropriate for use in

environmental biotechnology processes. In biofilm reactors, high biomass
concentrations (up to 50 g 1-1)can easily be maintained, although in many cases
medium-to-biofilm mass transfer of substrate is the rate-limiting process.
Processes using thin biofilms (< 0.5 ram) with large specific surface areas
(> 1000 m2 m-3 bioreactor volume) are being developed for this purpose.

A major advantage o f biofilm bioprocesses is that,
unlike suspension cultures, there is no need to incor-
porate special measures (for example, the use o f m e m -
brane filters or centrifugation) to retain the biomass in
culture. This feature makes the use ofbiofilms particu-
larly appropriate in bioreactor systems where large
substrate flow-through is required - in the processing
of waste streams, for example.
A key difference between suspension culture and
biofilm reactors is the presence o f distinct phases -
bulk-culture medium and immobilized microorgan-
isms - in the latter system. Substrates (oxygen, carbon
and nitrogen sources) have to cross the biofilm-liquid
interface by diffusion and, as a result, conditions in a
bioreactor are not homogeneous. Organisms in the
biofilm will experience conditions different from those
in the bulk liqnid, and within the biofilm there
will be considerable differences in the organisms'
micro-environment, depending on the distance from
the surface o f the biofilm, as a result o f the diffusion
gradients which occur. In the design ofbiofilm reac-
tors the following key factors need to be considered:
(1) a large biofilm-liquid interface; (2) the diffusion
and metabolic characteristics of immobilized cell cul-
tures; and (3) biofilm stability - the exchange o f cells
between the medium and immobilized biofilm phases.
Natural occurrence ofbiofilms
Biofilms are a natural phenomenon: most micro-
organisms in nature are associated with solid surfaces I .
In nutrient-poor environments, biofilms tend to exist
as a monolayer of attached cells, whereas in nutrient-
rich environments, more extensive biofilms build up.
One reason for this p h e n o m e n o n is probably that in
most ecosystems the streaming conditions have plug-
flow characteristics, or have quite a high dilution rate
(for example, a river, soil, or the mouth). T o survive
in such environments, microorganisms have to attach
to solid surfaces, or immobilize themselves in the form
ofbiofilms.

M. C. M. Van Loosdrecht and].]. Heijnen are at the Department of Biotechnological application ofbiofilms
Biochemical Engineering, Delft University of Technology,.]ulianalaan For the majority o f industrial fermentation
67, 2628 BC Delft, The Netherlands. processes, high substrate concentrations are used and

© 1993, ElsevierScience Publishers Ltd (UK) TIBTECHAPRIL1993 (VOL 11)

 118
fOCUS
Box 1. Comparison of artificially immobilized cells and
biofUms
Naturally (biofilm), or artificially (e.g. alginate beads)immobilized cells
experience the same restrictions on substrate transport over the
solid-liquid interface, and inside the immobilization matrix. Growth,
however, will be different.
In alginate beads, the cells are initially homogeneously dispersed
throughout the gel matrix. However, as a result of preferential growth
of cells near the surface of the biocatalyst, the cell density near the
surface will increase 6. At the same time, cells inside the beads will die
from a lack of substrate. When the carrier is stable, loss of cells by
detachment does not occur.
Biofilm formation on small suspended carriers can be initiated on
bare carriers. After the cells have attached, they grow outwards,
increasing the thickness of the biofilm. The build up of the biofilm is
counteracted by cell detachment, which eventually leads to a steady
state, in which growth of the biofilm is equal to cell detachment. The
steady-state thickness of the biofilm will depend primarily on the shear
forces acting on the biofilm 7.
Whereas growth on an artificial matrix leads to increased cell den-
sity, growth in natural biofilms leads to increased volume.
Immobilized biocatalyst Biofilm on carrier
3mm diameter O.5mm diameter
Initial stage
Detachment of cells
due to shear forces
. Jlly-developeC
stage
biofilm formation is either unnecessary, or even dis-
advantageous. Most commonly-used microorganisms
have been selected for their tendency to grow in sus-
pension with minimal aggregation, or wall growth.
Pure cultures o f such organisms do not easily form
biofilms, and this has led to the development o f arti-
ficial immobilization techniques, although the range
of applications of(artificially) immobilized-cell systems
in industry is limited. Only within the field o f en-
vironmental technology is extensive use made of
biofilms. This is because: (1) compared with most
other industrial bioprocesses, large volumes of dilute,
aqueous solutions (or gases) have to be treated; (2) natu-
ral, mixed populations o f microorganisms, which
readily form biofilms are used; and (3) the process can
be operated at high biomass concentration in the
bioreactor, without the need to separate the biomass
and treated effluent.
General reviews 2,3 of biofilm formation and biology,
and overviews o f biofilm processes4,s are available.
TIBTECH APRIL 1993 (VOL 11)
This article focuses on the application ofbiofilm bio-
reactors in environmental biotechnology, emphasiz-
ing factors that need to be considered in the design of
high-conversion-capacity bioreactors, and high-
activity, high-biomass biofilm processes. Although
many aspects of the use of artifcially, or naturally
immobilized-cell bioprocesses are similar, there are
differences between biofflms and immobilized cells
(Box 1).
In general, the physiology of immobilized, or
attached cells is not greatly different from that o f free
cells 8. However, immobilized cells are subject to dif-
fusion gradients (substrates, dissolved gases, metabolites
and even temperature) and therefore experience con-
ditions different from those in the bulk medium. T o
calculate the kinetics of conversion in biofilm systems,
a combination of two important processes that
occur in the system has to be taken into account:
(1) transport of solutes over the biofilm surface; and
(2) combined reaction and diffusion inside the biofilm
(Box 2).
With oxygen as the limiting substrate, and assuming
typical values for the different constants (qs =
1 kgO 2 kg biomass -1 day-1, Cxr = 100 kg m 3, D =
2 x 10 -9 m 2 S-1, Csf = 3 x 10-3 kg m -3) (Box 2), the
active-layer thickness ofa biofilm surface is calculated
as 100 txm, with a specific surface-consumption rate
of 10 g O 2 m -2 film day-1. These values are in close
agreement with typical observed experimental values
for the thickness of the aerobic-active layer
(75-200 Ixm), and oxygen-uptake rate (7-15 gO 2
m -2 day-l). For other (methanogenic, or denitrifying)
biofflm systems, comparable calculations show that the
active-biofilm thickness is always in the range o f
100-400 Ixm, provided that the limiting substrate
concentration is in the milligram range. Such calcu-
lations emphasize that, for biofilm processes, one
should focus on maximizing the surface area (with a
relatively thin biofilm) to maximize reactor capacity.
Biofilm growth
Biofilms will develop wherever microbial growth is
possible, provided suspended-cell growth is mini-
mized by a high dilution rate (J.J. Heijnen, PhD the-
sis, Delft University of Technology, The Netherlands,
1984). The main factor governing biofilm formation,
therefore, is the dilution rate o f the process7, i.e. the
dilution rate should be greater than the maximal
growth rate of the population.
It is impossible to consider the growth rate of a
biofflm perse. As a consequence of the concentration
gradients inside the biofilm, a growth-rate gradient will
also exist inside the biofilm. In multi-species biofilms,
where different microorganisms have to compete, not
only for substrates, but also for space, this will lead to
a biofilm with a layered structure. The organisms with
the highest growth rate will be found at the outside of
the biofilm, whereas slower-growing organisms will
be found inside 9,1°. As a result of this organization,
slower-growing organisms will be protected, and are
less likely to be lost due to detachment and wash-out.
 119
focHs
Biofilm formation is also influenced by the surface
characteristics o f the carrier: surface roughness
enhances biofilm formation 7,11,12, whereas surface
chemistry seems to be less important 13,14.
The overall accumulation ofbiofilms in a reactor is
determined by the average growth rate o f the biofilm
cells, minus the rate of detachment of cells from the
biofilm. In fact, the detachment process is equivalent
to the dilution rate in a chemostat; i.e. during steady
state, the specific detachment rate is equal to the aver-
age growth rate of the biofilm. It is important to note
that as a result o f detachment, the average growth rate
of the biofilm cells is higher than might be inferred
from a direct observation ofbiofilm accumulation. At
high shear rates, detachment of cells from the biofilm
surface can be almost equivalent to the growth rate,
resulting in a slow biofilm-accumulation rate 7.
Biofilrn reactors
A variety of reactor types have been developed to
exploit biofilm processes. O u r discussion focuses on
applications for the aerobic treatment of waste water.
Since biomass is retained in biofilms, their use is par-
ticularly advantageous when dilute waste waters with
high flow-rates are to be treated, or when the mi-
crobial population to be used has a relatively low
growth rate. With waste waters where the pollutant
(substrate) concentration is high (> 10 g chemical-
oxygen-demand 1-1), and rapidly growing organisms
(growth rate > 0.1 h -1) are to be used, there is no
advantage in biomass retention, as sufficient biomass
will be formed to metabolize the substrate within
relatively short residence times. However, such con-
centrated (industrial) waste-water streams will usually
first require treatment in a methanogenic process - one
that uses relatively slow-growing organisms, and
where biofilms are beneficial.
The oldest forms ofbiofllm reactors are trickling fil-
ters and biorotors. In both these types o f reactor, the
biomass is attached to a carrier that is retained in the
reactor. The biofilm surface area of these reactors does
not exceed 200 m 2 m-3; and thus the overall aerobic
capacity will n o t exceed 2 kgO 2 m -3 day q. This is low
when compared with industrial fermentation pro-
cesses, where conversion capacities of 80 kgO 2 m -3
day q may be reached. T h e benefit of biorotors, or
trickling filters, is their simplicity and low maintenance
requirements. Fixed-carrier reactors are therefore use-
ful if it is biomass retention, and not the mass (or oxy-
gen) transfer that is the main requirement, such as in
the removal ofxenobiotics (e.g. pesticides, herbicides,
or organic solvents) from ground water. In this case,
large volumes o f water with low substrate concen-
trations (< 10-50 m g 1-1) have to be treated. The fact
that the biomass is attached to a fixed carrier eliminates
a biomass-separation step in the process. For more
concentrated waste waters [chemical-oxygen-demand
(COD) > 100 mg 1-1], the enlargement o f the specific
surface area o f the biofilm can lead to a substantial
reduction in the reactor volume and the area require-
ments of the treatment process 15. The most obvious
Box 2. Mass transfer in a biofilm system
Substrates are transported into the biofilm by diffusion. Outside the
water-biofilm interface there will be a 'diffusion layer' through which
substrates have to diffuse to the biofilm surface. The thickness of this
layer depends on the local mixing conditions. Inside a biofilm, solutes
will diffuse down the concentration gradient inside the biofilm. These
substrate gradients are due to the process of metabolic conversion of
the substrate. At a certain distance (8) from the surface, the concen-
tration of the limiting substrate will become zero. In aerobic processes,
the limiting substrate will generally be oxygen, due to its poor solubility
in water (several miligrams per litre). The other substrates are
usually present at higher concentrations. The depth of the active layer
of the biofilm can be estimated from a macroscopic balance of the
substrate over the biofilm. The substrate flux follows from the cell den-
sity and activity in the biofilm and the thickness of the active layer.
Diffusion
Carrier Biofilm layer
~ 1~
Surface substrate flux:
U s = (2DqsCsfgxf)°.5
Cs
Active-layer ~'
thickness: 4:--~- Diffusion
= (2DCJqsCxf) °.5 ,"
-;-,, Cs~
.-;---- Diffusion and reaction
Distance
Abbreviations: D, diffusion coefficient (m2 s-]); Cs, substrate concen-
tration in the bulk liquid; Csf, substrate concentration at biofilm surface
(g m-3); qs, specific substrate conversion rate (g h-l); Cxf, biomass den-
sity inside biofilm (g m-3).
method is to grow the biomass in the form o f flocs,
and retain the flocs in the reactor by introducing a sep-
aration step in the process. At present, the only prac-
tical separation method in this context is gravity
settling. This is, however, also the main limitation of
the process. As a result o f the low sedimentation rate
o f floes (< 1 m hq), large settlers are required. The
spacious structure of the floes leads to a low concen-
tration o f suspended solids in the settled sludge, which
in turn limits the biomass concentration in the reac-
tor to ~ 3 kg m -3. The conversion capacity of the sys-
tem is, therefore, not significantly different from that
o f biorotors, or trickling filters. T o enhance the separ-
ation of sludge it has to be grown in the form o f dense
spherical aggregates. These aggregates can form spon-
taneously [as in upflow-anaerobic-sludge beds
(UASB)], but there are also examples o f granules
formed by nitrifying or denitrifying microorganisms t6
(D. De Beer, PhD thesis, University of Amsterdam,
T h e Netherlands, 1990), or they can be formed as
biofilms on small suspended particles (1KeE 11 ; J. J.
Heijnen, PhD thesis, Delft University o f Technology,
The Netherlands, 1984). A dramatic increase in biofilm
surface area can be obtained by growing biofilms as
discrete, small particles. The choice is a compromise
TIBTECHAPRIL1993(VOL11)
 120

fOCUS
a
5 ///1 Air
Degassing space ~' Foam destruction
Q. i
iI i
4
~E ' R" // >
0
,,'F," I I T Waste water

o 3
Pump s p a c e -
g~
Settler - -
oa. 2 Annulus - -
GILLS

Conversion limited Secondary air,

= 1
\
J
,v by biornass
1(
recycle carrier I
Conversion limited lr
. .,/ ~ r a n s t e r
E 0 ~" - - 0 A
v
O.0 0.3 0.6 0.9 1.2 1.5
Particle radius (mm)

Figure 1 reactor
Relationship of particle radius and mass transfer to the particle (solid lines) and ter-
minal sedimentation velocity (dotted lines). Mass transfer of oxygen in aerobic pro-
cesses (with carrier A, without carrier o) and acetate in methanogenic processes
(without carrier •). Terminal sedimentation velocity for biofilm particles with (ram),or
without (T) a carrier. Carrier: radius, 0.1 mm; density, 2.9 kg I-t. Specific conver-
sion rates: 1 gO 2 g-t day-t, or 1 g HAc g-~ day4. Biomass concentration inside
biofilm: 100 g biomass H. Mass transfer: 10 gO2 m-2 day-1, 35 g HAc m-2 day4.
Specific density biomass; 1020 kg m-3.

between conversion and the sedimentation rate of the

granules (Fig. 1). The latter depends on the particle
radius and density. In the case of anaerobic-conver-
sion processes, the active-layer thickness is larger than
that for aerobic processes, therefore mass-transfer
limitation will only occur for relatively large particles
(radius > 1-2 mm). These particles have sufficiently
high settling rates, and there is no direct need to
increase the particle density. For aerobic processes, the
radius at which conversion becomes rate limiting is
much lower, due to the low concentration of oxygen.
For optimal process conditions, the settling rate o f
the particles has to be improved by incorporating a
dense carrier inside the granule (J. J. Heijnen, PhD
thesis, Delft University of Technology, The Nether-
lands, 1984). The particles have to be formed in a re-
actor in which they are kept in suspension. Suspen-
sion cannot be achieved by mechanical stirring
because of the detrimental effects of the biofilm
particles caused by high shear forces around the stir-
rer blades. The reactor types applicable for these
processes are fluidized-bed reactors, or air-lift-suspen-
sion reactors. For anaerobic reactors, fluidized-bed
reactors, or upflow-sludge-bed reactors are most feas-
ible. For aerobic processes, (aerated) fluidized-bed
reactors, as well as air-lift reactors are used. The aer-
ation o f the fluidized bed is, however, problematic. Figure 2
Large amounts of water have to be recirculated over (a) Air-lift reactor with three-phase separator on top of the reactor.
the bed and oxygenated in the recycle stream. W h e n (b) Full-scale biofilm air-lift suspension reactor for the treatment of
there is an oxygen demand o f 500 mg 1-1, the recycle anaerobically pre-treated waste water, at Gist Brocades, Delft, The
flow rate for the use of air or pure oxygen is, respect- Netherlands. (c) Biofilm on small carriers as formed in a biofitm air-
ively, 62 and 12.4 times the influent flow rate. This lift-suspension reactor. Radius of one particle is _+0.3 ram.

TIBTECHAPRIL 1993 (VOL 11)

 121

foclAS
Table 1. Comparison of the flocculated-sludge process and a
recirculation causes (besides extra pumping costs) a process based on biofilms on small suspended carriersa,b
hydraulic problem for the fluidized bed. Therefore,
airlift reactors are more applicable for the suspension Biofilms on
Characteristics Flocculatedsludge carriers
of biofilm particles. These particles can then be
retained in the reactor by a simple three-phase sep- Settling velocity < 1 m h-1 < 50 m h-1
arator on top of the reactor (Fig. 2) 1537.
Biofilms on suspended carriers are a very good Inert sediment Accumulated in sludge Not accumulated in
alternative to the flocculated-sludge process. A com- sludge
parison o f both processes is shown in Table 1. It is
Biomass content < 3 kg m-3 < 30 kg m-3
evident from the data shown in Table 1 that the use
ofbiofilms in waste-water treatment can lead to a great Residence time 8-16 h 1-2 h
reduction in required reactor volume and, thus, the
industrial plant required. As high biomass concen- Sludge age Several days Several weeks
trations are present, it is possible to maintain a high
sludge age (low growth rate) in the reactor, which in Reactor geometry Flat Tall and narrow
turn leads to a minimization of the sludge production
Area requirement Large Small
and a stable nitrification, even at lower temperatures
(in the range of 5 - 1 0 ° C ) 17. Nitrification capacity 0.5 kg m-3 day-I 5 kg m-3 day<

Conclusions Oxidation capacity 3 kg m-3 dayq 20 kg m-3 dayq

Biofilm processes are especially advantageous when
large, diluted streams have to be treated. As this is gen-
erally the case for environmental processes, biofilms
are widely used in this field, whereas there are very
few industrial applications ofbiofilms. The maximiz-
ation o f biofilm-reactor productivity has to focus on
increasing biofilm surface-area, rather than biomass
concentration. In this context, fluidized-bed or air-lift
reactors, wbere biofilms grow on small, suspended
particles are the most suitable.
As the benefits o f biofilm processes (such as the
possibility of working at high biomass concentrations
and/or low growth rates, and the ease of separation of
biomass and product) may also apply to some indus-
trial processes, it will be advantageous to study the for-
mation of biofilms by pure cultures, and to develop
techniques for the isolation o f biofilm-forming
microorganisms.
References
1 Marshall, K. C. (1976) Interfaces in Microbial Ecology, Harvard Univer-
sity Press
2 Blenkinsopp, S. A. and Costerton, J. W. (199I) Tren& Bioteehnol. 9,
138-143
3 Breyers, J. D. (1991) Trends Biotechnol. 9, 422M26
4 Characklis, W. G. and Marshall, K. C. (1990) Biofilms, Wiley
5 Characklis, W. G. and Wilderer, P. A. (1989) Structure and Function of
Biofilms, Wiley
"From Ref. 17.
bj. j. Heijnen, PhD thesis, Delft University of Technology, The Netherlands,
1984.
6 Wijffels, R. H., de Gooijer, C. D., Kortekaas, S. and Tramper, J.
(1991) Biotechnol. Bioeng. 38, 232-240
7 Heijnen, J. J., Van Loosdrecht, M. C. M., Mulder, A. and Tijhuis, L.
(1992) Water Sci. TechnoI. 26, 647-654
8 Van Loosdrecht, M. C. M., Lyklema, J., Norde, W. and Zehnder,
A.J.B. (1990) Microbiol. Rev. 54, 75-87
9 Wanner, O. and Gujer, W. (1986) Biotech. Bioe~tj~.28, 314-328
10 Watanabe, Y., Lee, C., Koike, M. and Ishiguro, M. (I 990) Water &i.
Technol. 22, t69-178
11 Heijnen, J. J., Mulder, A., Enger, W. and Hoeks, F. (1989) Chem.
Eng..J. 41, B37-BS0
12 Mulder, A. and Heijnen, J. j. (1988) Proc. 2nd Netherlands Biotechnol.
C0nf (Breteler, H., ed.), pp. 102-110, Netherlands Biotechnology
Society
13 Van Loosdrecht, M. C. M., Lyklema, J., Norde, W. and Zehnder,
A.J.B. (1988) Mierob. Ecol. 17, 1-15
14 Baker, J. H. (1984) Can.]. Micwb. 30, 511-515
15 Heijnen,J.J., Mulder, A., Weltevrede, P,.., Hols,J. and Van Leeuwen,
H. L.J.M. (1990) Chem. Eng. Technol. 13, 145-160
16 Lettinga, G., Van Velsen, A. F. M., Hobma, S. W., De Zeeuw, W.
and Klapwijk, A. (1980) Biotech. Bioeng. 22, 699-734
17 Heijnen, J. J., Van Loosdrecht, M. C. M., Mulder, R., Mulder, A.
and Weltevrede, R. (1991) in Proc. Int. Syrup. Env. Biotechnot.
(Verachtert, H. and Verstraete, W., eds), pp. 211 217, KVI

• Bioremediation - a Special Issue from TIBTECH

ill soon be publishing a Special Issue looking at the techno-
logical, regulatory and commercial aspects of biological degradation and detoxification
of h a z a r d o u s w a s t e . A r t i c l e s cover:
• process strategies • monitoring bioremediation efficacy • ecology and evolution of
microbial populations • impact of regulatory policies on commercial development
° prospects for profit in the global bioremediation industry • exploiting microbial
metabolism for bioremediation of organic contaminants, heavy metals and nitrogenous
wastes.

TIBTECHAPRIL.1993 (VOL 11)