Conceptual model for assessment of inhalation exposure to manufactured nanoparticles.

Conceptual model for assessment of inhalation exposure to manufactured
nanoparticles
THOMAS SCHNEIDERa,b, DERK HENRI BROUWERc, ISMO KALEVI KOPONENb, KELD ALSTRUP JENSENb,
WOUTER FRANSMANc, BIRGIT VAN DUUREN-STUURMANc, MARTIE VAN TONGERENd AND
ERIK TIELEMANSc
aK�gemestervej 13, Hiller�d DK 3400, Denmark
bNFA National Research Centre for the Working Environment, Lers� Parkalle´ 105, Copenhagen DK 2100, Denmark
cTNO Research group Quality & Safety, PO Box 360,Zeist 3700 AJ, The Netherlands
dIOM Institute of Occupational Medicine, Research Avenue North Riccarton, Edinburgh EH14 4AP, UK
As workplace air measurements of manufactured nanoparticles are relatively expensive to conduct, models can be helpful for a first tier assessment of
exposure. A conceptual model was developed to give a framework for such models. The basis for the model is an analysis of the fate and underlying
mechanisms of nanoparticles emitted by a source during transport to a receptor. Four source domains are distinguished; that is, production, handling of
bulk product, dispersion of ready-to-use nanoproducts, fracturing and abrasion of end products. These domains represent different generation
mechanisms that determine particle emission characteristics; for example, emission rate, particle size distribution, and source location. During transport,
homogeneous coagulation, scavenging, and surface deposition will determine the fate of the particles and cause changes in both particle size distributions
and number concentrations. The degree of impact of these processes will be determined by a variety of factors including the concentration and size mode
of the emitted nanoparticles and background aerosols, source to receptor distance, and ventilation characteristics. The second part of the paper focuses on
to what extent the conceptual model could be fit into an existing mechanistic predictive model for ‘‘conventional’’ exposures. The model should be seen as
a framework for characterization of exposure to (manufactured) nanoparticles and future exposure modeling.
Journal of Exposure Science and Environmental Epidemiology advance online publication, 2 March 2011; doi:10.1038/jes.2011.4
Keywords: nanoparticles, exposure modeling, coagulation, source–receptor, modifying factors.
Introduction
The number of published workplace measurements for
assessing occupational exposure to manufactured nanoparti-
cles, as part of the larger group of nano-objects as defined by
ISO (2008b), has increased substantially over the past 2 years
(Brouwer 2010). However, in view of the large variety of
possible exposure scenarios, the amount of available data is
still very scarce. The need for additional information on
occupational exposure will rapidly increase due to growing
production volumes and use of manufactured nanoparticles.
However, as workplace air measurements for manufactured
nanoparticles are relatively complex and expensive to
conduct, exposure models may be required to provide
estimates of exposure for a first tier assessment.
Mechanistic models for exposure to ‘‘conventional’’ air
contaminants have been developed for quantitative retro-
spective occupational exposure assessment to be used in
epidemiology, for designing and complementing workplace
exposure assessment and for regulatory risk assessments,
such as under the Registration, Evaluation, Authorisation
and Restriction of Chemical substances (REACH) regu-
lations in Europe (Tielemans et al., 2007). The Advanced
REACH Tool (ART) incorporates a mechanistic source–
receptor model that consists of three types of components:
(1) sources, (2) compartments through which the conta-
minants may pass during their transport from the source to
the receptor, and (3) the receptor (Tielemans et al., 2008;
Fransman et al., 2009). This paper describes a conceptual
model for occupational inhalation exposure to manufactured
nanoparticles, similar to and based on the mechanistic model
developed for ART (Tielemans et al., 2008).
An important question is which exposure metric should be
used for model predictions. Maynard and Aitken (2007)
argued for a greater emphasis on the physicochemical
properties of nanoparticles and proposed a framework for
exposure monitoring, based on classification of nano-objects
and identification of biological relevant attributes. Sampling
methods and models should ideally measure or estimate
aerosol number, surface, and mass concentration. In
principle, these metrics are interrelated, and given the number
concentration and size distribution, the surface area and massReceived 23 December 2009; accepted 20 July 2010
1. Address all correspondence to: Dr. D.H. Brouwer, TNO, PO BOX 360,
Zeist 3700 AJ, The Netherlands.
Tel.: þ 31888665126,
E-mail: dick.brouwer@tno.nl
Journal of Exposure Science and Environmental Epidemiology (2011), 1–14
r 2011 Nature America, Inc. All rights reserved 1559-0631/11
www.nature.com/jes

can be estimated. The accuracy of the estimate will depend on
the availability of additional information, such as particle-
specific density and shape.
In this paper we will discuss (i) the requirements for
proper source characterization; (ii) the role of coagulation,
scavenging, and surface deposition during the transport of
manufactured nanoparticles from the source to the receptor;
and (iii) the interrelations between the three exposure metrics.
Next, we will present the conceptual ‘‘nano’’-model and
discuss any modifications required to adapt the existing
mechanistic model from ART (www.advancedreachtool.
com) for conventional inhalation exposure into a mechanistic
model for nanomaterials.
Source Characterization
Exposure to manufactured nanoparticles can occur during
synthesis, downstream use, application or treatment of
products containing embedded manufactured nanoparticles,
and waste recycling/disposal. Different mechanisms will
determine emission rate and the transport of (nano) aerosols
during the various life cycle stages.
Manufacture of nanoparticles can be broadly categorized
into bottom-up and top-down processes. The most common
examples of bottom-up processes are wet chemistry synthesis,
gas phase synthesis based on homogeneous nucleation, and
chemical vapor deposition where gases react and the product
grows on a substrate. Size reduction such as ball milling or
planetary grinding is a more conventional, top-down,
method to produce (bulk) nanopowders. In the production
phase, the nano-objects will be formed as primary particles
(particles not formed from a collection of smaller particles
(ISO, 2007), but due to several different mechanisms the bulk
powder are usually harvested and packed in a largely
aggregated and/or agglomerated state (Schneider and Jensen,
2009). Fugitive diffuse and point source emissions may occur
during the various stages in the manufacturing process
(Demou et al., 2008, 2009; Park et al., 2009).
After harvesting of the nanoparticles further processing
may take place, such as surface modification, before the
product is transported to the downstream user. Generally,
downstream users will mix or disperse the manufactured
nanoparticles with other materials to form the end or
intermediate product. Bag emptying is one of the most
obvious activities during which aerosols can be generated
during the downstream use of manufactured nanoparticles.
Within the EU-sponsored project NANOSH B50% of 41
monitored activities with nanopowders were related to
transfer of the nanoparticles, such as bagging (12%), bag
emptying (20%), and pouring or scooping of small amounts
(20%; Van Duuren-Stuurman, 2009). At the time of writing,
few other studies of workplace exposure to manufactured
nanoparticles have been published (Brouwer et al., 2009;
Brouwer, 2010; Seaton et al., 2010). Ten papers were
identified that reported on exposure during the production
of manufactured nanoparticles, of which five were on a
commercial scale. Bagging or packing were the most
frequently studied activities, followed by harvesting of the
product from the reactor. Two studies focused on down-
stream use and included activities such as transfer and
pouring of small amounts of nanopowders and feeding of an
extruder (injection molding).
Exposure may also occur during the application, further
processing, or machining of end products containing
embedded manufactured nanoparticles. In an experimental
setting, N�rgaard et al. (2009) and Hagendorfer et al. (2010)
investigated the release of volatile organic solvents and
aerosols during application of commercially available nano-
film spray products. Koponen et al. (2009) and Bello et al.
(2009) investigated release of particles following sanding and
cutting of CNT composites, respectively. Finally, aerosols
may also be released during abrasion processes (e.g.,
wearing) as was demonstrated in an experimental study by
Vorbau et al. (2009).
In summary, we have identified the following source
domains that include the vast majority of current and near-
future exposure situations for manufactured nano-objects:
(1) Point source or fugitive emission during the production
phase (synthesis) before harvesting the bulk material; for
example, emissions from the reactor, leaks through seals
and connections,
(2) Handling and transfer of bulk manufactured nanomater-
ial powders, for example, bag emptying, dumping,
scooping, etc.,
(3) Dispersion of either intermediates containing highly
concentrated (425%) nanoparticles or application of
(relatively low concentrated o5%) ready-to-use pro-
ducts; for example, spraying of solutions that will form
nanosized aerosols after condensation,
(4) Activities resulting in fracturing and abrasion of
manufactured nanoparticle-enabled end products at
work sites, such as machining, for example, sanding,
milling, cutting, etc.
From source to receptor
Coagulation and Scavenging
Manufactured nanoparticles emitted before harvesting may
coagulate rapidly during the transport to the receptor and
manufactured nanopowders have a tendency to agglomerate
(Luther, 2004; Ma-Hock et al., 2007; Seipenbusch et al.,
2008; Brouwer et al., 2009; Schneider and Jensen, 2009). The
conceptual nano-model must thus take into account
� Homogeneous coagulation of manufactured nanoparti-
cles emitted from a production line or reactor before
Model for assessment of inhalation exposureSchneider et al.
2 Journal of Exposure Science and Environmental Epidemiology (2011), 1–14

harvesting, in which high concentrations of primary
manufactured nanoparticles or nanosized agglomerates
may initially be present,
� Scavenging of emitted nanoparticles by background or
associated larger particles (heterogeneous coagulation),
� The degree to which the agglomerates in bulk nanopowder
break during handling and the consequences for the size
distribution and structure (morphology) of the particles
released to the air.
The relative occurrence in the breathing zone of workers of
manufactured nanoparticles as primaries/agglomerates, or as
attached to larger background particles will depend on the
source characteristics and the coagulation and removal
processes during transport from the source to the receptor
(Seipenbusch et al., 2008; Schneider and Jensen, 2009).
For an instantly mixed room, the time-dependent change
of number concentration of particles with diameter Dp,
n(Dp,t), in the room can be described as
qnðDp;tÞ
qt ¼
SðDp;tÞ
V þ
qnðDp;tÞ
qt
� �
coag
þ qnðDp;tÞqt
� �
loss
; ð1Þ
where S is source rate (particles/s) of manufactured
nanoparticles and V is the room volume. The subscript
‘‘coag’’ refers to changes due to coagulation and ‘‘loss’’ to
changes due to surface deposition and ventilation. For
simplicity in Eq. (1), it has been assumed that the makeup air
is particle free.
The change in number concentration due to Brownian
coagulation can be described in continuous form as
qnðDp;tÞ
qt
� �
coag
¼ 1
2
ZDp
0
K
ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
D3p � q33
q
; q
� �
n
ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
D3p � q33
q
; t
� �
nðq; tÞdq
� nðDp; tÞ
Z1
0
Kðq; DpÞnðq; tÞdq
;
ð2Þ
where K(x, y) is the Brownian coagulation rate constant
between particles of diameter x and y. The first term on the right
accounts for the coagulation between two particles forming a
new particle with volume equivalent diameter Dp. The second
term accounts for particles with diameter Dp coagulating with all
other particles. The Fuchs interpolation formula is commonly
used as a starting point for calculating the coagulation rate
constant. It can be modified to take agglomerate structure and
interparticle and other forces into account:
� The agglomerate and aggregate particle structure can be
parameterized by the use of fractal dimensions. Increasing
fractal complexity increases the rate of coagulation
(Jacobson and Seinfeld, 2004).
� Interparticle forces
(a) The net effect of Van der Waals and viscosity force
increases the coagulation rate constant. The A-value
A/kBT¼ 200 is often used, where A is the Hamaker
constant, kB Boltzmann’s constant, and T is the absolute
temperature. The effect of fractal structure and the Van der
Waals/viscosity force can be calculated by equations given
in Jacobson and Seinfeld (2004). As an example, for
Df¼ 1.7 and A/kBT¼ 200 the coagulation rate constant
for two 10 nm particles is enhanced by a factor 3.75, and
by a factor of B16 for a 10 nm particle in a 1000 nm
background. For particles in air, A/kBTcan typically range
from 20 to over 200 (Tsai et al. 1991).
(b) There are effects of electric charges (Zebel, 1966)
(i) Unipolar charging of an aerosol decreases the coagu-
lation rate constant.
(ii) For symmetrically charged aerosol
(1) For weakly bipolar charged particles there is no
net effect
(2) For highly bipolar charged particles, the effect of
the coagulation rate constant increases. Increasing air
turbulence increases the rate of coagulation. For example,
stirring a chamber with a fan could increase the rate for
100 nm particles by a factor of two (Kim et al., 2006).
The rate constant for coagulation between particles A and
B is shown in Figure 1. The coagulation rate constant
between a smaller particle A and a larger particle B increases
strongly with increasing difference in particle size, and is also
proportional to the product of the concentration of particles
A and B. Number concentrations of nanosized particles
can be very large for even moderate mass concentrations; a
mass concentration of 1mg/m3 consisting of 10 nm particles
(specific density 1g/cm3) corresponds to 1.9� 109 particles/cm3.
Thus, homogeneous coagulation of nanosized particles or
1.E-10
1.E-09
1.E-08
1.E-07
1.E-06
1.E-05
1.E-04
1.E+00
Radius, nm
Co
ag
ul
at
io
n
ra
te
c
on
st
an
t,
cm
3 /s
Background
particle radius
1 nm
10 nm
100 nm
1000 nm
Homogeneous coagulation
1.E+01 1.E+02 1.E+03
Figure 1. Coagulation rate (cm3/s) at the left Y axis calculated using
the Fuchs interpolation between emitted manufactured nanoparticles
(x axis) and background particles of different sizes (right Y axis).
Model for assessment of inhalation exposure Schneider et al.
Journal of Exposure Science and Environmental Epidemiology (2011), 1–14 3

scavenging of nanosized particles with larger-size background
aerosol can be important and fast.
Koch (2002) calculated the coagulation of nanosized
particles in a background of 1 mm particles for mass
concentrations of nanosized particles from 0.01 to 1mg/m3,
background particle concentrations 0.5, 5, and 50mg/m3 and
for residence times 1–10,000 s, and assuming that there was
no loss of mass. For concentrations of nanosized particles
below 0.1mg/m3, homogeneous coagulation would be
negligible and for a background concentration of 5mg/m3
(B104 particles/cm3) heterogeneous coagulation (scaven-
ging) would be negligible for residence times less than about
100 s. For concentrations of nanosized particles above
0.1mg/m3 homogeneous coagulation would increase the size
of the original nanosized particles, thereby initially reducing
the rate of scavenging.
Seipenbusch et al. (2008) studied coagulation of nanosized
particles released from a source with and without the
presence of a coarser background aerosol. On the basis of
their experiments and on modeling, they concluded that
nanoparticles released from a fugitive source would not reach
the receptor in the form of the primary aerosol.
Assessing the governing state of agglomeration is an
important aspect of modeling. However, the underlying
theory is complex and the concentration dependence of
coagulation leads to a complex model structure. The
consequences of coagulation can be included in a simple
model if simplifying assumptions about coagulation are
made. One approach has been proposed by Seipenbusch
et al. (2008). They used dimensional analysis of the
parameters of a simple coagulation model for an instanta-
neously mixed single compartment to obtain a scaling
parameter, Z, being the ratio of the loss rate by coagulation
and the source rate
Z ¼ Kijnjni;SQSni;S
V
¼ VKijnj
QS
ð3Þ
It was found that the experimental data could be
approximated by the relation
ni
ni;S
¼ const
Z
ð4Þ
In the presently proposed nano-model, the determination of
the governing state of agglomeration will be based on direct
calculation of Eqs. (1) and (2). For this purpose, a
coagulation module has been developed. It is based on an
instantaneously mixed single compartment. Manufactured
nanoparticles and background aerosols of any size distribu-
tion can be introduced. At time t¼ 0, a given concentration
can be defined as the initial condition as the result of an
impulse injection of manufactured nanoparticles and with or
without a given constant concentration of a background
aerosol. Another option is to have a constant source active
for a total duration of 10min. Surface deposition loss rate
is neglected. The temporal evolution can be calculated for
duration of up to 1 h, based on an algorithm developed by
Miikka Dal Maso from the University of Helsinki, following
the UHMA atmospheric dynamic model (Korhonen et al.,
2004). The coagulation coefficient is calculated using Fuchs
interpolation. Corrections for fractal structure and Van der
Waals/friction forces can be included at a later stage of
development of the module.
Figures 2–5 have been obtained using this coagulation
module. The figures serve to demonstrate the influence on
coagulation of different initial particle concentrations and
sizes and of a continuous nanoparticle source. In real
occupational exposure scenarios, additional factors, for
example, surface and ventilation losses, should be taken into
account. For simulating such scenarios, additional informa-
tion is needed, such as room dimensions and ventilation rate
as will be discussed later. Figure 2 shows that in 100 s and a
particle mode with a 10 nm geometric mean diameter
(GMD), concentration must have a minimum of 106/cm3
to see an effect of coagulation. This agrees with the rule of
thumb of Hinds (1999) saying that coagulation can be
neglected if the concentrations are less than 106/cm3. By
increasing the concentration to 107/cm3, the size mode shifts
rapidly to larger diameters. The GMD of the 10 nm mode
grew to B15 nm in 100 s (Figure 2c).
In the simulations shown in Figure 3, the initial conditions
were mode 10 nm GMD and modal concentration 108/cm3.
Then a background mode with 300 and 1000 nm GMD and
Figure 2. Coagulation simulation with mode 10 nm (GMD) initial
particle size at different initial modal concentrations: (a) 105/cm3,
(b) 106/cm3, and (c) 107/cm3.
Model for assessment of inhalation exposureSchneider et al.
4 Journal of Exposure Science and Environmental Epidemiology (2011), 1–14

modal concentrations of 54,000/cm3 (¼ 0.77mg/cm3) and
2000/cm3(¼ 1.05mg/cm3) were introduced. In all three
cases, the original 10 nm mode manufactured nanoparticles
disappears and grows to approximately up to 70 nm size after
3600 s. In the case of the 300 nm background mode, the
10 nm mode grows to around 60–70 nm while the back-
ground mode is shifting very little towards bigger size. At the
end of the simulation, the number concentrations in the
10 nmmode is lower and the peak is narrower compared with
Figure 4. Simulation with different background conditions and
constant source mode 10 nm (GMD) initial particle size at 108/cm3
every second: over 600 s: (a) no background, (b) 300 nm (mode)
background particles 2000/cm3 (0.029mg/m3), and (c) 1000 nm
(mode) background particles 2000 cm�3 (¼ 1.05mg/cm3).Figure 3. Coagulations simulations with mode 10 nm (GMD) initialparticle size and modal concentration 108/cm3 with different back-
ground conditions: (a) no background, (b) mode 300 nm with modal
concentration 54,000/cm3(¼ 0.77mg/cm3), and (c) mode 1000 nm
with modal concentration 2000/cm3(¼ 1.05mg/cm3).
Model for assessment of inhalation exposure Schneider et al.
Journal of Exposure Science and Environmental Epidemiology (2011), 1–14 5