Biotechnology Advances 36 (2018) 1079–1093

Contents lists available at ScienceDirect

Biotechnology Advances
journal homepage: www.elsevier.com/locate/biotechadv

Research review paper

Biofiltration of volatile organic compounds using fungi and its conceptual T

and mathematical modeling
⁎
Alberto Vergara-Fernándeza, , Sergio Revahb, Patricio Moreno-Casasa, Felipe Scotta
a
Green Technology Research Group, Facultad de Ingeniería y Ciencias Aplicadas, Universidad de los Andes, Chile
b
Departamento de Procesos y Tecnología, Universidad Autónoma Metropolitana-Cuajimalpa, Mexico

A R T I C LE I N FO A B S T R A C T

Keywords: Volatile organic compounds (VOCs) are ubiquitous contaminants that can be found both in outdoor and indoor
Biofiltration air, posing risks to human health and the ecosystems. The treatment of air contaminated with VOCs in low
VOC concentrations can be effectively performed using biofiltration, especially when VOCs are hydrophilic. However,
Fungi the performance of biofilters inoculated with bacteria has been found to be low with sparsely water soluble
Bacteria
molecules when compared to biofilters where fungi develop. Using conceptual and mathematical models, this
Modeling
review presents an overview of the physical, chemical and biological mechanisms that explain the differences in
Computational fluid dynamics
the performance of fungal and bacterial biofilters. Moreover, future research needs are proposed, with an em-
phasis on integrated models describing the biological and chemical reactions with the mass transfer using high-
resolution descriptions of the packing material.

1. Introduction adults (Elliott et al., 2006), induce inflammatory reaction of airways,

Atmospheric pollution has become one of the major causes of pre-
mature death in developed and developing countries. Using data from
the Global Burden of Diseases Study 2015, Cohen et al. (2017) reported
that fine particulate material (PM2.5) alone was responsible for 4.2
million deaths in 2015, compared to 3.5 million in 1990. Nearly 60% of
these deaths took place in east and south Asia. While the role of other
contaminants is well understood, such as ozone causing an additional
0.25 million deaths, the impact of volatile organic compounds (VOCs)
has not been fully quantified, despite having a role in ozone formation
and that often they are associated to PM and polycyclic aromatic hy-
drocarbons (PAHs) (Kundu and Stone, 2014). VOCs comprise a group of
organic chemicals with a high vapor pressure at room temperature and
are commonly present in indoor and outdoor air (Khan and Ghoshal,
2000). VOCs are emitted from industrial sources, such as petroleum
refineries and chemical plants (Hoyt and Raun, 2015), commercial
activities, such as gasoline refilling (Harley et al., 2006); and at the
residential level, wood and fossil fuels burning for heating and cooking
(Evtyugina et al., 2014) and products used indoor like aerosols, paints
and cleaners (Bernstein et al., 2008). VOCs such as the low-boiling
hydrocarbons (pentane, hexane, heptane, etc.), alcohols, aldehydes,
halogenated hydrocarbons, ethers and PAHs represents 7% to 10% of
all atmospheric pollutants emitted (Delhoménie and Heitz, 2005).
These compounds have been shown to reduce pulmonary function in
immunological abnormalities and neurological symptoms (Win-Shwe
et al., 2013) and an increased carcinogenic risk for formaldehyde,
benzene, acetaldehyde and naphthalene (Sarigiannis et al., 2011). The
changes and mechanisms behind some of these responses to VOCs ex-
posure have been recently studied in mice (Wang et al., 2012) and lung
cells (Gostner et al., 2016). However, the lack of understanding of the
mechanisms involved, did not preclude establishing links between
VOCs exposure and health issues. In this regard, a study lasting 12 years
and including nearly 60,000 Toronto residents, found that the exposure
to ambient levels of benzene, n-hexane and total hydrocarbons were
associated with an increased risk of cancer mortality (Villeneuve et al.,
2013).
Depending on the airflow and their concentration, VOCs can be
abated using purely physico- chemical technologies such as gas mem-
brane separation, condensation and adsorption (typically used for high
VOCs concentrations and low to medium air flows), incineration and
catalytic oxidation (for high concentration of VOCs and large air flows)
and biotechnological processes for dilute VOCs streams. The reviews by
Revah and Morgan-Sagastume (2005) and Delhoménie and Heitz
(2005) represent comprehensive assessments of the applicability of the
different technologies which can be complemented with the work of
Estrada et al. (2012) to gain a broad perspective of the applicability and
costs of the most commonly applied VOCs abatement technologies.
Biofiltration has become a viable economic and technical

⁎
Corresponding author at: Mons. Álvaro del Portillo 12455, Las Condes, Santiago 7620001, Chile.
E-mail address: aovergara@miuandes.cl (A. Vergara-Fernández).

https://doi.org/10.1016/j.biotechadv.2018.03.008
Received 9 January 2018; Received in revised form 9 March 2018; Accepted 14 March 2018
Available online 17 March 2018
0734-9750/ © 2018 Elsevier Inc. All rights reserved.
A. Vergara-Fernández et al.
alternative for the treatment of airstreams with low concentrations of
volatile organic and inorganic compounds (such as H2S and NH3) in
waste gas streams. It is based on the capacity of an active microbial
population, usually growing on a solid support in a biofilm, to trans-
form the pollutants into less harming molecules, namely carbon di-
oxide, sulfates, nitrates, etc. (Revah and Morgan-Sagastume, 2005) in a
process that shares similarities with the microbial degradation of pol-
lutants in soils (Ren et al., 2018).
The inherent heterogeneity and complexity of biofilters make the
analysis and mathematical modeling of these systems a challenging
task. In such models, not only mass transport, fluid flow through porous
beds and biochemical reactions are entangled; but also, the properties
of the biofilm, such as the thickness of the active biomass layer, and of
the solid inert support play a role, for example in determining the area
of the biofilm (Spigno and Tronci, 2015; Vergara-Fernández et al.,
2008).
As the deployment of industrial biofilters has consistently grown
since the late 80’s, so has been the increasing complexity of the con-
ceptual and mathematical models developed for VOCs abatement in
fixed bed biofilters. In this regard, reviews in the area are devoted to the
conceptual analysis of biofiltration systems (Devinny and Ramesh,
2005), to the use of fungi in biofilters (Kennes and Veiga, 2004), the
mass transfer aspects in biofilters (Kraakman et al., 2011) and on how
to reduce the mass transfer limitations (Cheng et al., 2016; Ferdowsi
et al., 2017). However, a revision of the action of fungi in the treatment
of volatile organic compounds and its modeling, including detailed
descriptions of the phenomena occurring in the reactor (mass, heat and
momentum transport) and of the microbial growth and biodegradation
kinetics, is missing.
In this review, results from biofiltration experiments are expressed
in terms of the empty bed residence time (EBRT, s), VOCs inlet load (L,
g m−3
reactor h
−1
), biofilters elimination capacity (EC, g m−3
reactor h
−1
) and
removal efficiency (RE, %) according to:
Vr
EBRT =
Q (1)
Q contaminants are transferred from the gas phase to the liquid phase
EC = (CG0 − CG (out ) )
Vr (2)
Q fungal biofilters. The contaminants diffuse or are transported to the
L= (CG0)
Vr (3)
(CG0 − CG (out ) )
RE = 100
CG0 (4)
where CG0 and CG(out) are inlet and outlet VOCs concentration, re-
spectively, Q is airflow and Vr is the reactor volume.
This review is organized as follows: the individual biological and
physicochemical phenomena involved in fungal biofiltration are re-
viewed and compared with the bacterial case; (Section 2) next, the
modeling efforts to integrate these aspects, understanding of the ex-
perimental information and optimization of biofilters design are re-
viewed in Section 3. Finally, the characteristics and performance, in
terms of elimination capacity, are compared and analyzed for fungal
and bacterial biofilters treating several VOCs.
The remaining of this paper is devoted to reviewing the current
understanding of the physical, chemical and biological phenomena in-
volved in fungal biofiltration and its modeling compared to bacterial
biofiltration.
2. Physicochemical and biological phenomena in a fungal biofilter
and its mathematical modeling
Biofilters are multiphasic reactors that respond to diverse spatial
and temporal scales. From molecular diffusion and reaction to micro-
bial growth and gas or liquid channeling, modeling of phenomena in-
volved in the operation of biofilters presents challenges to the modeler
1080
Biotechnology Advances 36 (2018) 1079–1093
on whether using an approach tailored to each spatial and temporal
scale or using an aggregated (or lumped) model. Despite the approach
used, models must produce macroscopic outputs, often the only mea-
sured quantities in an experimental system, to compare model predic-
tions, such as elimination capacity for contaminants and pressure drops,
with the operation of the system being modeled. This section aims at
introducing the general principles involved in the phenomena of mass
and energy transport, as well as the biomass growth and biodegradation
kinetics. Each aspect will be the subject of the forthcoming sections,
dedicated to reviewing the application of these principles to the mod-
eling of fungal biofilters. Before introducing the physical, chemical and
biological mechanisms involved in biofiltration of air, a precision re-
garding biofilters and biotrickling filters needs to be made. While bio-
filters have a static water phase associated to the biofilm and the sup-
port; in biotrickling filters water is constantly supplied producing a thin
water film flowing over the packing and the biofilm. However, in the
laboratory and industrial practice, biofilters are intermittently supplied
with water, or a nutrient solution, to ensure appropriate moisture
contents at regular time intervals, some are washed for as much as
20 min in each hour (Devinny and Ramesh, 2005). On the other hand,
because unsaturated gravitational flow in porous media is difficult to
control, biotrickling filters are likely to include patches that are not
properly covered with water.
For highly soluble organic compounds, such as alcohols, the parti-
tion coefficient favors a higher concentration in the water phase, which
promotes the biodegradation of the contaminant by increasing its
concentration in the environment surrounding the biofilm (Cohen,
2001). On the other hand, in fungal biofilters, if a layer of flowing water
would be created over the solid support, then it would decrease the
surface area created by the aerial hypha, thus impairing the transfer of
VOCs, especially hydrophobic ones, to the biomass phase (Sugai-
Guérios et al., 2015).
Fig. 1 shows a conceptual model of a biofilter. Contaminants are
transported into the biofilter by the air at rates that justify assuming
that the flow is laminar. However, dispersion occurs due to the tortu-
osity of the porous packing. As the air flows through the packing,
embedded in the biofilm (for bacterial biofilms), or to the fungal hypha
and to the water embedded in the non-aerial portion of the biomass in
biofilm and then degraded by the microbial catabolic activity. As most
biofilters are used to treat polluted air and thus are based on the activity
of aerobic microorganisms, the availability of oxygen is also an im-
portant issue as well as the retro-diffusion of metabolic products such as
CO2. Relevant variables affecting the mass transfer of contaminants and
oxygen in the reactors include the flow regimes, pressure drop, void
fraction of the packed-bed, biofilm wetting; and for trickle-bed bio-
filters, the flow of liquid and liquid hold-up (Lobo et al., 1999) and
therefore are often considered in models of biofilter operation.
In biofilters, the mass transfer parameters associated with the fluid
mechanics of the system, e.g. the diffusion coefficient of the con-
taminant in air, have been traditionally estimated using the correlation
proposed by Fuller et al. (1966), based on the molecular diffusivity of
the species. Similarly, the diffusion coefficients of the nutrients from the
liquid media in the biofilm have been estimated using de Nernst-Haskell
equation (Vergara-Fernández et al., 2008) and the axial dispersion
coefficient, a measure of the deviation from plug-flow regime in a re-
actor, has been approximated using the correlation proposed by
Edwards and Richardson (1968). However, the aerial mycelia and the
colonization of the void space in the biofilter, when filamentous fungi
are used, might lead to discrepancies between the predicted and ob-
served values of these coefficients.
2.1. Mass transfer of species in the gas phase and reaction in the biofilm
The temporal variation of the concentration of a compound; either a
A. Vergara-Fernández et al.
Fig. 1. Scheme of a conceptual model of a fungal biofilter showing the different scales involved.
pollutant, oxygen or any other chemical species, in a differential vo-
lume of gas can be obtained by assuming that there is no radial dis-
persion, that a species enters the control volume by advection (bulk
transport in the air) but its transport is delayed by axial dispersion
(accounted for using an effective dispersion coefficient, DDz) leaving the
control volume at a rate governed by the specific mass flux of compo-
nent i from the gas phase to the interface (NG,i), the specific area of the
gas-liquid or gas-biofilm interface (a, interface surface per unit volume
of biofilter bed), and the void fraction occupied by the gas (ε). These
purely physical transport mechanisms are regarded as the mass balance
for component i in the gas phase (Eq. 5).
∂CG, i ∂CG, i ∂2CG, i a
= vg − DDz − NG, i
∂t ∂z ∂z 2 ϵ (5)
where, CG,i = CG,i(t,z) represents the concentration of contaminant i in
the gas phase at a given position in the axial direction (z) of the reactor
at a certain time (t) and υg is the gas velocity. Considering that com-
ponent i is being transferred from the bulk of the gas phase to a stagnant
layer of gas, the specific mass flux (NG,i) can be calculated using the
Fick's law:
Cb, i
NG, i = Db, i ∂
∂x x=0
Where, Db,i is the diffusion coefficient in the biofilm, Cb,i is the
concentration of contaminant i in the biofilm, and coordinate x de-
scribes the thickness of the stagnant layer of gas and x = 0 denotes the
interface between the stagnant layer of gas and its bulk (Fig. 1).
The next transfer step delivers the contaminant from the stagnant
layer of gas to a layer of water in biotrickling filters, or to the biofilm in
biofilters. In the latter, and when bacteria are the oxidizing micro-
organisms, the transport across a layer of water can be omitted if the
biofilm is assumed to be a pseudo-homogeneous phase composed of
water and microorganisms (Fig. 2.b). The mass balance of any com-
ponent, considering reaction and diffusion within the biofilm according
to Fick's Law, can be written as:
Biotechnology Advances 36 (2018) 1079–1093
∂CG, i ∂2CG, i
= Db, i +r
∂t ∂x 2 (7)
where, Cb,i = Cb,i(t, xb) is the concentration of contaminant i at a given
time and thickness in the biofilm (x). The reaction rate (r) accounts for
the catabolic reaction that allows the microorganisms embedded in the
biofilm to oxidize the contaminant.
In fungal biofilters (Fig. 2.a), a contaminant is transported directly
from the gas phase to the aerial hyphae (Vergara-Fernández et al.,
2016). This situation can be represented by a simplified form assuming
a negligible concentration gradient within the hyphae and an overall
mass transfer coefficient (KG):
∂CG, i CG, i
= K G a ⎛⎜ − Cb, i⎞⎟ + r
∂t m
⎝ i, G , b ⎠ (8)
where, mi,G,b represents the partition coefficient of contaminant i be-
tween the gas phase and the biomass.
As it can be seen, the driving force for the transfer of the con-
taminant to the biofilm is determined by the difference between the
equilibrium concentration at the gas-biofilm interface, where the par-
tition coefficient plays a key role, and the concentration of the pollutant
in the biofilm, controlled by the catabolic activity of the microbial
population.
(6)
The diffusivities of low molecular weight contaminants in the gas
phase are in the range of 1⋅10−5 m2 s−1 (Tang et al., 2015) while in
water is close to 1⋅10−9 m2 s−1 (Harms and Bosma, 1997). The diffu-
sivity within biofilms can be expected to be lower than in pure water. If
the biofilm is conceptualized as a cluster of cells surrounded by water
and exopolymers, then, transfer of substances within the biofilm can be
regarded as a combination of advection within voids and water chan-
nels permeating the biofilm and diffusion within cell clusters and the
exopolymeric matrix (Guimerà et al., 2016). Thereby, since a biofilm
cannot be regarded as a homogeneous system, the measured diffusion
coefficient is an effective diffusion coefficient (De).
In fungal biofilters the aerial mycelium is in direct contact with the
gas phase while the non-aerial mycelium lies within the watery biofilm.
The formation of the aerial hyphae requires decreasing water surface
1081