Aerosol Science 37 (2006) 1152 – 1163

www.elsevier.com/locate/jaerosci

A set-up for field studies of respiratory tract deposition of fine and

ultrafine particles in humans
Jakob Löndahla,∗ , Joakim Pagelsb , Erik Swietlickia , Jingchuan Zhoua , Matthias Ketzelc ,
Andreas Masslinga , Mats Bohgardb
a Div. of Nuclear Physics, Dept. of Physics, Lund University, P.O. Box 118, S-22100, Lund, Sweden
b Div. of Ergonomics and Aerosol Technology, Lund University, P.O. Box 118, S-22100, Lund, Sweden
c Dept. of Atmospheric Environment, National Environmental Research Institute, P.O. Box 358, DK-4000, Roskilde, Denmark

Received 1 July 2005; received in revised form 28 October 2005; accepted 7 November 2005

Abstract
Respiratory tract deposition data of ultrafine aerosol particles, hygroscopic particles and ambient particles in general are scarce.
Measurements are associated with several difficulties. The objective of this work was to design a method for fast determination
of highly size-resolved fine and ultrafine particle deposition, to be used on larger groups of human subjects in exposure studies
and in typical ambient and indoor environments. The particle size distributions in dried samples of the inhaled and exhaled air
are characterised with an electrical mobility spectrometer. A particle counter desmearing procedure reduces the spectrometer scan
time. The precision and sensitivity of the method was tested for hygroscopic sodium chloride (NaCl) and hydrophobic Di-Ethyl-
Hexyl-Sebacate (DEHS) aerosols in repeated identical experiments and experiments with different breathing frequencies on a single
subject. The accuracy of the method was estimated by comparing results from three subjects with previous data obtained with
monodisperse particles and with the well-established International Commission on Radiological Protection model (1994). Potential
errors due to size shifts between the inhaled and exhaled samples and coagulation were simulated. The system has low losses in
the studied particle size range (10–475 nm), typically 10% or less of the fraction deposited in the respiratory tract. Coagulation
is noticeable at 105 cm−3 but can be corrected for up to 5 × 105 cm−3 . The precision in the determined deposited fraction is
0.02–0.08. The method is sensitive enough to quantify differences between breathing patterns and differences between hygroscopic
and hydrophobic aerosols. Our results for NaCl and DEHS are in agreement with the ICRP 66 model [International Commission
on Radiological Protection. (1994). Human respiratory tract model for radiological protection (ICRP Publication 66). Oxford, UK:
Elsevier Science], and also suggest that the relative humidity in the respiratory tract is close to 99.5%. A respiratory tract deposition
measurement can be done in 15–30 min. Recommendations are given for field applications of the method.
䉷 2005 Elsevier Ltd. All rights reserved.

Keywords: Aerosols; Ultrafine particles; Fine particles; Respiratory deposition; Exposure; Lung; Number

1. Introduction

It has been estimated that urban particulate matter (PM10 ) leads to 800,000 deaths annually in the world (WHO, 2003).
The corresponding figure for Europe is 100,000 deaths (WHO & Ebrary, 2002; Ezzati, Lopez, Rodgers, Vander Hoorn,

∗ Corresponding author. Tel.: +46 46 2227682; fax: +46 46 2224709.

E-mail address: jakob.londahl@nuclear.lu.se (J. Löndahl).

0021-8502/$ - see front matter 䉷 2005 Elsevier Ltd. All rights reserved.
doi:10.1016/j.jaerosci.2005.11.004
 J. Löndahl et al. / Aerosol Science 37 (2006) 1152 – 1163 1153

& Murray, 2002). Suggestions have been made that fine and ultrafine particles, especially from combustion sources,
have a stronger association with adverse health effects than larger particles (Donaldson, Stone, Clouter, Renwick, &
MacNee, 2001; Ferin, Oberdorster, & Penney, 1992; Oberdorster, Gelein, Ferin, & Weiss, 1995). Susceptible sub-groups
such as people with pre-existing respiratory or cardiovascular diseases, elderly and children have been identified.
The term “respiratory tract deposited fraction” (DF) refers to the mean probability for an inhaled particle to be
deposited in the respiratory system. DF depends on its size, hygroscopicity, shape and density as well as on the
breathing pattern and airway geometry of the inhaling subject.
Most knowledge of respiratory tract deposition originates from measurements of particle concentration in inhaled
and exhaled air using monodisperse hydrophobic particles in healthy young subjects. There are few experimental data
available for respiratory tract deposition of ambient particles and ultrafine particles in susceptible sub-groups (e.g.
Chalupa, Morrow, Oberdorster, Utell, & Frampton, 2004; Kim & Jaques, 2005). For ultrafine hydrophobic particles
the deposition increases to around 0.8 at 20 nm. Alveolar deposition has a maximum at 20 nm.
An important parameter rarely taken into account is the hygroscopic growth into solution droplets that occurs in the
respiratory tract for soluble species. The equilibrium relative humidity (RH) in the alveolar region is ∼ 99.5% (Anselm,
Heibel, Gebhart, & Ferron, 1990). Experimental data suggest that the deposition minimum is moved, for sodium chloride
(NaCl) from 400–500 to 80–100 nm indicating a 5-fold particle growth (Tu & Knutson, 1984; Blanchard & Willeke,
1983). Typically the accumulation mode particles in ambient aerosols have a diametric growth factor corresponding to
3–4 at 99.5% RH. This is mainly caused by the high sulphate content. Thus, it is vital to study respiratory tract deposition
for large groups in various environments, e.g. at the low concentrations often encountered for ambient particles.
Measurements using monodisperse particles would be slow if a large size range should be covered. In addition,
if applied to ambient particles, long sessions would be needed to receive good statistics. To improve time resolution
and counting statistics, polydisperse methods incorporating a highly size-resolved size spectrometer can be used. The
deposited fraction is then determined for each size channel by comparing the concentration in inhaled and exhaled
samples. Particles with a well-defined electrical mobility can be selected with a differential mobility analyser (DMA)
and detected with a condensation particle counter (CPC). Combined in a scanning mobility particle sizer (SMPS) a
complete size distribution between for example 10 and 500 nm can be obtained on a time-scale of minutes.
Respiratory tract deposition experiments with polydisperse methods have been made in a few previous studies.
Anderson, Wilson, and Hiller (1990) measured 20–240 nm particles from bag samples of inhaled and exhaled air with an
electrical aerosol analyzer. Morawska, Barron, and Hitchins (1999) and Morawska, Hofmann, Hitchins-Loveday, Swan-
son, and Mengersen (2005) studied respiratory tract deposition of combustion aerosols using an SMPS (16–626 nm)
sampling from two large chambers for inhaled and exhaled air, respectively. They found that the average deposited
fraction was 1.4–2.1 times higher than predicted (Hofmann, Morawska, Hitchins, & Ahmed, 2004). Rosati, Brown,
Peters, Leith, and Kim (2002) presented a method where “inhaled” and exhaled aerosol samples were collected in
25 L latex bags and analysed with an SMPS system and an APS 3310 (0.015–15
m). However, no studies involving
human subjects have been reported with this set-up. Daigle et al. (2003) determined respiratory tract deposition of
artificially generated spark discharge soot particles (GMD 26 nm) at high particle concentrations (∼ 2 × 106 cm−3 )
using an SMPS (7.5–75 nm) in a flow-through set-up (Chalupa et al., 2002) with short residence time from exhalation to
sampling. Montoya et al. (2004) studied respiratory tract deposition of ambient particles in Boston, US, for six subjects.
Particle concentrations in dried samples of ambient and exhaled air were measured with an SMPS system and an API
Aerosizer (40–1800 nm) using a flow-through method, which involved dilution of the exhaled sample with filtered air.
They found that the deposition minimum was shifted towards smaller sizes and attributed this to hygroscopic growth
of the particles.
A number of difficulties have to be solved when using a polydisperse method instead of a monodisperse. The
polydisperse techniques are sensitive to even very small size shifts of the dried diameter between the inhaled and
exhaled sample (Pagels, Rissler, Swietlicki, & Bohgard, 2003). It is therefore crucial to prevent size shifts caused
by insufficient drying, agglomerate restructuring, coagulation or evaporation. Size-classification may be altered by
smearing of the output signal because of finite CPC response time (Collins, Flagan, & Seinfeld, 2002), particularly
if scan-time is short. Measured concentration could be influenced by pressure variations originating from breathing.
Particle losses in the measurement equipment may be interpreted as respiratory tract deposition. Electrostatic material
increasing these losses, should be avoided. If the concentration is too high, especially if the time between measurement
of inhaled and exhaled sample is too long, coagulation will lead to both unwanted particle size shift and particle loss.
These things are sometimes of high importance. However, previous studies only take them into account to a limited
1154 J. Löndahl et al. / Aerosol Science 37 (2006) 1152 – 1163

Inlet
Temp./RH-sensor

Inhale tank
10 L
DMA

CPC
2-way valve
Drier

1-way valves Heated area

Pneumotachograph
Mouthpiece Outlet
Exhale tank
2/5 L
Volume >40 cm3

Fig. 1. Schematic picture of the flow-through system.

degree. None presented lowest acceptable concentration limits due to counting statistics or upper concentration limits,
e.g. due to particle coagulation.
There is a need for methods to fill the gap in the database for respiratory tract deposition of real-world aerosols and
ultrafine particles in different groups of people. In this work a fast method for field measurements of particle deposition
in the respiratory tract is developed and characterised. Two polydisperse system designs have been considered. A
flow-through system and a bag sampling system similar to those described by Chalupa et al. (2002) and Rosati et al.
(2002), respectively. An algorithm is written to account for smearing and thereby reduce the scan time of the particle
size spectrometer (SMPS). Careful loss measurements are reported. Minimum and maximum concentration limits are
quantified. Uncertainties due to size shifts are modelled and discussed. Respiratory tract deposition experiments in
three subjects with hygroscopic and hydrophobic particles are used to estimate the accuracy, precision and sensitivity
of the method.
It turned out that the flow-through system was more suitable for the desired applications, and in most characterisation
experiments only the flow-through system is considered. With minor adjustments, described in Section 5, regarding
the aerosol and breathing pattern, this system can be used in typical ambient and indoor environments.

2. Respiratory tract deposition method

The system consists of three main parts: aerosol generation, breathing system and particle size spectrometer. The
two latter parts are shown in Fig. 1.

2.1. Aerosol generation

Three different sources of aerosols were used: ammonium sulphate ((NH4 )2 SO4 ), sodium chloride (NaCl) and di-
ethyl-hexyl-sebacate (DEHS). NaCl and DEHS were chosen to compare with previous deposition measurements of
monodisperse aerosols and to represent two extremes with respect to hygroscopic properties.
Polydisperse aerosol was generated with a nebuliser (Model 3076, TSI Inc., US). For (NH4 )2 SO4 and NaCl a distilled
water solution with 1% salt by mass was prepared and nebulised at a pressure between 1 and 3 bar, which resulted in
an aerosol with a number geometric mean mobility diameter (GMD) of 50–80 nm and geometric standard deviation
(
g ) 1.9–2.0. The DEHS was diluted in ethanol to a 0.5% solution and nebulised at a pressure of 1 bar. In experiments
involving human subjects the wet aerosol from the nebuliser passed a single nozzle impactor with a cut-off diameter
of 0.7
m (Fig. 2). This reduced the mass concentration by a factor of 5–10 and minimised errors in the inversion due
to doubly charged particles above the upper DMA size range. The DEHS aerosol, having a flow rate of 5.5 L/min,
was subsequently diluted with 12.5 L/min particle and oil free air. Of this 11 L/min was wasted and the ethanol in the
remaining flow of 7 L/min was adsorbed in two active charcoal denuders, replaced every hour. Finally the aerosol was
 J. Löndahl et al. / Aerosol Science 37 (2006) 1152 – 1163 1155

NaCl Box, 0.5 m3

Nebuliser Impactor
Clean air
Charcoal denuders
DEHS Box, 0.5m3
Nebuliser Impactor Waste Thermo-desorber

Fig. 2. Set-up for the aerosol generation system.

heated to 65 ◦ C in a thermo-desorber to evaporate remaining ethanol, while leaving most of the DEHS unaffected with
a GMD of 130–140 nm and
g 1.9–2.0. All aerosols were finally diluted with particle and oil free air in a 0.5 m3 box
in order to decrease the RH below 30% and achieve the desired number concentrations (5.0 × 104 –4.3 × 105 cm−3 ).
For most experiments a particle concentration of 9 × 104 cm−3 was chosen.

2.2. Flow-through system

The arrangement of the flow-through system is shown in Fig. 1. It basically consists of a mouthpiece, 2 stainless steel
tanks connected by a T-shaped brass piece with 2 1-way valves, an automatic 2-way valve, a drier and an SMPS system.
When a subject inhales, air is drawn actively from the larger tank for “inhaled” air, which is connected to the 0.5 m3 box.
A volume of 10 L is chosen for the inhalation tank in order to minimise fast variations in concentration during a mixing
time of about 1ṁin. Inhaled and exhaled air are separated by 2 1-way valves (Lip membrane, Laerdal medical, Norway),
of the “duck-bill” type, which opens with minimal flow obstruction. The instrumental dead-space that may be inhaled
multiple times is 18 cm3 . The valves are made of silicon rubber, but were covered with a 50 nm thick layer of gold
applied with vacuum deposition (Auto 306, Edwards, UK) to minimise losses by electrostatic deposition. The exhaled
aerosol enters a smaller tank that is heated to 34–40 ◦ C to prevent condensation. The “exhalation” tank has a volume of
2 or 5 L to minimise concentration gradients within and between single breaths, while keeping the residence time for
the aerosol as short as possible. The larger volume is for breathing during exercise. The tank for exhaled air is emptied
through a tube with volume large enough to ensure that no outside air is mixed with the exhaled air during the inhalation
phase. Both the “inhalation” tank and the exhalation tank are thus open to the surrounding atmosphere and operate
at pressures near ambient and thus reduce pressure variations induced by breathing. The breathing continues until a
sufficient amount of data are collected to calculate the deposition, normally between 10 and 30 min depending on the
aerosol concentration. A pressure transducer (PasCal 100, Hoffrichter, Germany) connected to a pneumotachograph
(Type 1 or 2, Dr. Fenyves und Gut, Germany) is used to register the inhaled and exhaled flow rate and breathing
frequency. Temperature and RH are measured in the aerosol sampling lines just after exit from the two tanks using
capacitive sensors (Hygroclip S, Rotronic, Switzerland).

2.3. Bag-system

In a few measurements a bag system was used. The main difference between the two systems is that the bag system
uses two flexible rubber bags (Anti-Static, 30 L, Rüsch, Germany), where inhaled and exhaled air are sampled, instead
of fixed tanks. The main purpose of using flexible bags is to avoid pressure gradients in the system during breathing.
Prior to an experiment the bag for inhaled air is filled with aerosol and the bag for exhaled air is emptied. When a
subject inhales, air is drawn from the filled bag and the exhaled air is directed into the emptied bag, which is heated to
34–40 ◦ C.

2.4. Measurement system

In both systems the aerosol concentration is measured in the reservoirs for inhaled and exhaled air. The sampling
location is chosen with a computer controlled 2-way valve (Solenoid Valve, type 330, Bürkert, Germany). The aerosol
is dried with a 48 Nafion single tube drier (MD-110-48S, Perma Pure, NJ, USA), which reduces the RH to below
20%. Measurement of the particle size distribution was made with a scanning mobility particle sizer (SMPS, Wang &
Flagan, 1989) consisting of a DMA (Long DMA, TSI Inc., US) and a CPC (CPC, Model 3010, TSI Inc., US). The
1156 J. Löndahl et al. / Aerosol Science 37 (2006) 1152 – 1163

DMA is operated in a closed loop set-up with a diffusion drier (Model 3062, TSI Inc., US) in the sheath flow to reduce
the RH in the DMA to below 5%. Critical orifices control the sheath and aerosol flow rates. Temperature and RH is
measured after the Nafion drier and in the DMA sheath flow loop.
A computer program (written in LabVIEW 7.1) was constructed in order to control the DMA and the CPC. The
program corrects for effects of the finite response time of the CPC using a desmearing procedure (Collins et al., 2002),
the CPC counting efficiency (Zhou, 2001) and the effects on the DMA transfer function caused by scanning the DMA
voltage (Collins, Cocker, Flagan, & Seinfeld, 2004). It furthermore enables particle size distribution measurements on
both up and down scan. The scan time could thereby be reduced to 45 s and the inhaled and exhaled size distribution
can be analysed in less than 2 min. The computer program also records the breathing pattern and suggests a pre-defined
pattern that the subject can follow during semi-controlled breathing. The program controls valves and sensors and
continuously calculates the respiratory tract deposition. The voltage ramps of the DMA are set according to desired
particle sizes and measured pressure and temperature.
The measurements are done in cycles consisting of four scans in the order:

1. down scan, inhaled tank

2. up scan, exhaled tank
3. down scan, exhaled tank
4. up scan, inhaled tank

To smooth the small differences originating from data inversion an equal number of up and down scans are made
in both fractions. To decrease systematic errors because of slow concentration gradients the inhaled air is measured at
the beginning and in the end of the cycles. One up scan in the inhaled tank is made before the first cycle but wasted in
the analysis since that time is needed for the inhaled air to mix in the airways and for the subject to achieve a steady
breathing.
When using an aerosol flow rate of 1.0 L/min and a sheath flow of 6.0 or 10 L/min, the SMPS measurable size range
is 10–475 and 10–320 nm, respectively. Comparisons of up and down scan were made and Polystyrene Latex (PSL)
spheres were used for the DMA size calibration.

3. Method characterisation

3.1. Inherent losses in the flow-through system

The flow-through method was characterised with reference to losses in the breathing system. When determining the
fractional respiratory tract deposition, DFhuman , corrections must be made for particle losses in the equipment, DFequip :
Cex (dp,i )
DFhuman (dp,i ) = 1 − . (1)
Cin (dp,i ) · (1 − DFequip (dp,i ))
Here dp,i is the particle diameter in size-channel i and Cin and Cex denote the particle concentrations in the inhaled
and exhaled samples, respectively. It can be shown that Eq. (1) is valid for deposition losses occurring in any part of
the system between the two sampling ports if the particles keep their original size.
To measure the particle losses in the system a sinus wave piston-type breath simulator (PARI GmbH, Germany) was
connected to the inlet of the breathing system together with 2 1-way valves as illustrated in Fig. 3, and the mouthpiece
was plugged. This was done to minimise the influence of losses within the breath simulator. Breathing flow rates of
6.8, 9 and 13.5 L/min were simulated using a “tidal volume” of 750 ml at 9, 12 and 18 cycles/min. (NH4 )2 SO4 particles
were used (number concentration 6 × 104 –9 × 104 cm−3 , GMD 50–60 nm and
g 2.0).

3.2. Coagulation in the flow-through system

Coagulation may be a reason both for particle losses and for changing the size of the particles. An upper limit was
determined below which these effects were insignificant. The set-up was similar to that used for loss measurements
with the flow pattern held constant at 12 cycles/min and a volume of 750 ml per cycle. The particle concentration was
 J. Löndahl et al. / Aerosol Science 37 (2006) 1152 – 1163 1157

Breath simulator

Inhale tank

Sample aerosol

Fig. 3. Set-up used for measurement of particle losses in the flow-through system and increased losses due to coagulation.

varied from 6.8 × 104 to 4.2 × 105 cm−3 . The particle losses in 20 s caused by coagulation was in addition simulated
with a sectional aerosol dynamics model (Ketzel & Berkowicz, 2004) for 2 aerosols having GMD 70 nm,
g 2.2 and
total number concentrations 1 × 105 and 5 × 105 cm−3 , respectively.

3.3. Modelling of respiratory tract deposition and the influence of size shifts

Respiratory tract deposition predicted with the empirical ICRP 66 model (ICRP, 1994), based on a large amount of
experimental data, was calculated for a healthy male, breathing through the mouth at 10 L/min with frequencies 6, 10
and 15 min−1 . Hygroscopic growth was incorporated in the model by assuming RH = 99.5% throughout the respiratory
tract and immediate growth to the equilibrium size. Growth factors for NaCl at RH = 99.5% were calculated using
Raoult’s law, taking the Kelvin effect into account.
Caluculations were also made to study the influence of a size shift between the dry inhaled and exhaled size
distributions on the determined respiratory tract deposition. In this work the effect is illustrated for a parameterisation
of the DEHS aerosol used in the experimental validation (GMD 130 nm,
g 1.9). The concentration of the exhaled
aerosol was estimated using the ICRP model (10 L/min, 10 min−1 ) and thereafter shifted towards 1% and 5% smaller
and larger particle diameters (Fig. 6a). The deposited fraction was calculated from the shifted exhaled distribution,
Cexhaled,shifted , and compared to the correct deposited fraction, i.e. the distribution that would be measured if no size-shift
would occur (neglecting inherent particle losses).

Cexhaled,shifted (dp,i )
DF(dp,i ) = 1 − . (2)
Cinhaled (dp,i )

3.4. Experiments with human subjects

Measurements in three subjects were made to study the precision, accuracy and sensitivity of the method. The main
part of the validation experiments was performed in one healthy male subject (26 years old) to minimise physiological
variations. Single measurements were also done on two other male subjects (both 31 years old) for comparison with
data reported in previous studies. All three subjects are never smokers without any known respiratory disease. The
experiments were approved by The Danish National Committee on Biomedical Research Ethics as a pilot study within
a larger project. The subjects followed suggested (square wave) breathing patterns on a computer screen.
To estimate the precision of the method, NaCl aerosol was inhaled in three “identical” experiments (on the same
day) with breathing frequency 10 ± 0.1 min−1 , minute volume 10 ± 0.1 L/ min, scan time 90 s and DMA sheath flow
10 L/min. The measurement time was 15 min for each of these sessions. This was also repeated with single experiments
using 90 s scan time with a sheath flow rate of 6 L/min and 45 s scan time with a sheath flow of 10 L/min.
To validate the sensitivity of the method, deposition of NaCl and DEHS was compared for three different breathing
patterns, frequency 6 ± 0.5, 10 ± 1 and 15 ± 1 min−1 and minute volume 10 L/min. Scan time was 90 s and sheath flow
10 L/min.
To get a rough estimate of the accuracy of the method, experiments were performed with three subjects breathing
NaCl particles for 15 min each at a frequency of 10 ± 0.5 min−1 and a minute volume of 10 ± 0.5 L/ min. The results
were compared with previous measurements and the ICRP model. The measurements were covering 4–12 inhalation
and 4–12 exhalation scans.
1158 J. Löndahl et al. / Aerosol Science 37 (2006) 1152 – 1163

0.07
9 cycles/min
0.06
12 cycles/min
18 cycles/min

Deposited Fraction
0.05

0.04

0.03

0.02

0.01

0
10 100 1000
Mobility diameter [nm]

Fig. 4. Experimentally determined losses in the flow-through system with the breath simulator, given for different breathing frequencies.

The experiments were performed at a concentration of 1 × 105 cm−3 . The inhaled mass concentration in experiments
with human subjects was below 500
g/m3 , as estimated with combined measurements with an APS 3321 and the
SMPS.
To confirm that no particles were produced by the subject or leaking at the mouthpiece, an experiment was made
with inhaled air filtered through a HEPA filter with 99.95% collection efficiency.

4. Results and discussion

4.1. Inherent particle losses

Results from measurements of losses in the flow-through system are shown in Fig. 4. After covering the one-way
valves with gold, the deposited number fraction decreased with a factor up to two and is below 0.06 for 20 nm particles
and approximately 0.01 for 100–300 nm. This is only 10% or less of the fraction deposited in the respiratory tract
(e.g. DF = 0.20 ± 0.02). The low inherent losses are attributed to the use of electrically conducting material and short
residence time between the inhaled and exhaled sample (∼ 10–20 s).
The bag system had losses of 0.30 for 20 nm particles and around 0.10 for 80–300 nm particles, i.e. more than five
times higher. A further difficulty with the bag system was that the deposition increased as the bags were emptied—i.e.
the losses as a function of fill volume had to be calculated.

4.2. Influence of coagulation

Enhanced losses due to coagulation are noticeable in the flow-through system at number concentrations above
1 × 105 cm−3 (GMD 50–80 nm and
g 2.1). As can be seen in Fig. 5a, losses of ultrafine particles increased from 0.06
to 0.14 for 20 nm particles when the concentration increased from 1.2 × 105 to 4.2 × 105 cm−3 . The residence time was
about 10–20 s between the two measurement points. The simulated particle losses with the aerosol dynamics model are
shown in Fig. 5b. The modelled increase in losses when the concentration was increased from 1 × 105 to 5 × 105 cm−3
was 0.15 for 10 nm particles and 0.05 for 20 nm particles. This is slightly lower but of the same magnitude as that found
in the experiments.
Because of the longer residence time of the aerosol in the bag system (120 s in the measurements), an upper limit
when coagulation starts to affect the results is 4 × 104 cm−3 .

4.3. Influence of particle size shifts

Fig. 6b shows the influence of 1% and 5% size shifts between the inhaled and the exhaled sample for a DEHS size
distribution with GMD 130 nm and
g 1.9. For 30 nm particles a 5% negative shift (e.g. by agglomerate restructuring
 J. Löndahl et al. / Aerosol Science 37 (2006) 1152 – 1163 1159

0.16 0.20
0.14 420,000 cm-3 0.18 500,000 cm-3, CMD 70 nm, GSD 2.2
280,000 cm-3 100,000 cm-3, CMD 70 nm, GSD 2.2
0.16

Deposited Fraction
Deposited Fraction

0.12 160,000 cm-3

120,000 cm-3 0.14
0.1 68,000 cm-3 0.12
0.08 0.10
0.06 0.08
0.06
0.04
0.04
0.02 0.02
0 0.00
10 100 1000 10 100 1000
(a) Mobility diameter [nm] (b) Diameter [nm]

Fig. 5. (a) Deposited fraction in the system as a function of concentration. Increased particle losses are assumed to be due to coagulation, and (b)
particle lost due to coagulation over a period of 20 s according to the aerosol dynamics model.

0.12 1
Inhaled distribution DF, 5 % positive shift
0.9
0.1 DF, 1 % positive shift
Number concentration

Exhaled distribution Deposited Fraction 0.8

DF, ICRP
0.08 0.7
[dN/dlogDp]

Size-shifted exhaled DF, 1% negative shift

distribution 0.6
DF, 5 % negative shift
0.06 0.5
0.4
0.04 0.3
0.2
0.02
0.1
0 0
10 100 1000 10 100 1000
(a) Diameter [nm] (b) Diameter [nm]

Fig. 6. (a) An inhaled size distribution similar to that used in the experiments. The exhaled distribution is calculated using the ICRP model assuming
hydrophobic particles. The influence of a 5% negative shift in the exhaled distribution is illustrated, and (b) effect of size shift between the inhaled
and the exhaled sample on respiratory deposition. DEHS aerosol with GMD 130 nm and
g 1.9.

or evaporation) lead to an underestimation of 0.11 in DF and a 5% positive shift (e.g. coagulation or insufficient
drying) to an overestimation of 0.06. The corresponding figures for 300 nm particles are 0.08 (i.e. 80%) higher and
0.09 (90%) lower DF. This illustrates that even small size shifts can give substantial errors in measured DF when using
a polydisperse method. With a decrease in
g the errors will increase, i.e. major deviation on DF may occur if the
size distribution is narrow. Effects of size shifts are large in regions with high gradients of the number distribution. A
polydisperse method cannot be used if a size shift larger than a few percent is suspected.

4.4. Experiments with human subjects

The deposition in the respiratory tract determined with the system was found highly repeatable in “identical” exper-
iments with a single subject (Fig. 7). Variations in DF were below 0.05 for the main part of the size interval when the
scan time and the sheath flow rate were held constant and only slightly larger when they were modified (Fig. 8). The
experiment with reduced scan time shows stronger scatter between points. This may be a result of larger corrections in
the desmearing procedure.
The method is clearly sensitive enough to quantify differences between the three different breathing patterns in single
experiment (Figs. 9 and 10). The variations are in relatively good agreement with the ICRP model. For the smallest
particles (< 25 nm) with the DEHS aerosol, the counting statistics is poor and the deposition is slightly lower than
expected. The lower deposition may be caused by uncertainties in the SMPS-inversion for sizes with low concentrations
very far from the geometric mean diameter (GMD 130 nm). Figs. 9 and 10 also illustrate the difference in deposition
between hygroscopic NaCl particles and hydrophobic DEHS particles. The agreement with the ICRP model is relatively
good, indicating that the RH is close to 99.5% in regions where particles were deposited.
1160 J. Löndahl et al. / Aerosol Science 37 (2006) 1152 – 1163

1
0.9 NaCl, 10 breaths/min, 10 L/min
0.8 NaCl, 10 breaths/min, 10 L/min

Deposited Fraction
0.7 NaCl, 10 breaths/min, 10 L/min
0.6
0.5
0.4
0.3
0.2
0.1
0
10 100 1000
Dry mobility diameter [nm]

Fig. 7. Respiratory deposition of NaCl particles in a single subject in three subsequent “identical” experiments. Breathing frequency, 10 min−1 ,
minute volume 10 L/min. Each experiment consists of 4 exhaled and 4 inhaled samples. Error bars show a 95% confidence interval due to variation
in DF between the pairs of inhaled and exhaled scans.

1
90 s scan time, 6 L/min sheath flow
0.9
45 s scan time, 10 L/min sheath flow
0.8 90 s scan time, 10 L/min sheath flow
Deposited Fraction

0.7
0.6
0.5
0.4
0.3
0.2
0.1
0
10 100 1000
Dry mobility diameter [nm]

Fig. 8. Respiratory deposition of NaCl in a single subject, showing the influence of varied scan time and DMA sheath flow rate. Breathing frequency
10 min−1 , minute volume 10 L/min. Error bars show a 95% confidence interval due to variation in DF for inhaled and exhaled scans.

The mean DF for the three subjects and previous data were in accordance with the ICRP model, except that the
DF for the subjects were 0.05 lower between 100 and 200 nm (Fig. 11). Note that the ICRP model does not explicitly
include hygroscopic aerosols and therefore has been shifted as if the NaCl particles immediately grow to equilibrium
size at 99.5% RH when inhaled. Our results are in good agreement with those of Tu and Knutson (1984) but lower than
those of Blanchard and Willeke (1983). The previous studies were performed with similar but not identical breathing
patterns using monodisperse particles. The measured DF for the three subjects showed a relatively low intersubject
variability, typically the difference in DF was 0.05–0.10.
In experiments inhaling filtered air the number concentration of exhaled particles was less than 0.2 cm−3 and no
systematic difference could be observed between inhaled and exhaled air.

5. Field applications

When using the described method in field measurements the experimental procedure must be modified depending
on the specific environment to avoid errors due to size shifts, particle concentration near the measurable limits and
concentration fluctuations.
As shown in Fig. 6b, the poly-disperse method is sensitive to small size shifts of the dried diameter, between the
inhaled and exhaled sample. This can be caused by insufficient sample drying, agglomerate restructuring at high RH,
 J. Löndahl et al. / Aerosol Science 37 (2006) 1152 – 1163 1161

1
6 breaths/min
0.9 10 breaths/min
15 breaths/min
0.8 ICRP, 6breaths/min
ICRP, 10 breaths/min

Deposited Fraction
0.7 ICRP, 15 breaths/min
ICRP, 10 breaths/min, hydrophobic
0.6
0.5
0.4
0.3
0.2
0.1
0
10 100 1000
Dry mobility diameter [nm]

Fig. 9. Respiratory deposition of NaCl particles in a single subject. Single measurements with breathing frequencies of 6, 10 and 15 min−1 and
an average flow rate of 10 L/min. Error bars show a 95% confidence interval due to variation in DF for inhaled and exhaled scans. The deposition
according to the ICRP model for hydrophobic particles is shown for comparison (flow 10 L/min, frequency 10 min−1 ).

1
6 breaths/min
0.9 10 breaths/min
15 breaths/min
0.8 ICRP, 6 breaths/min
Deposited Fraction

ICRP, 10 breaths/min
0.7 ICRP, 15 breaths/min
ICRP, 10 breaths/min, hygroscopic
0.6
0.5
0.4
0.3
0.2
0.1
0
10 100 1000
Dry mobility diameter [nm]

Fig. 10. Respiratory deposition of hydrophobic DEHS particles in a single subject. Breathing frequency 6, 10 and 15 min−1 , minute volume 10 L/min.
Error bars show a 95% confidence interval due to variation in DF for inhaled and exhaled scans. The deposition according to the ICRP model for
hydrophobic particles shifted with a 99.5% RH growth for NaCl is shown for comparison (flow 10 L/min, frequency 10 min−1 ).

1
0.9 Test subject 1, 26 years old ICRP, hygroscopic
Test subject 2, 31 years old Blanchard & Willeke, 1983
0.8
Deposited Fraction

Test subject 3, 31 years old Tu & Knutson, 1984

0.7
0.6
0.5
0.4
0.3
0.2
0.1
0
10 100 1000
Dry mobility diameter [nm]

Fig. 11. Deposition of NaCl particles in three healthy male subjects breathing at a frequency of 10 min−1 and an average flow rate of 10 L/min.
Comparison with the ICRP model and previous measurements using monodisperse NaCl particles and mouth breathing: Blanchard and Willeke
(15 L/min, 10 min−1 ) and Tu and Knutson (11 L/min, 15 min−1 ).
1162 J. Löndahl et al. / Aerosol Science 37 (2006) 1152 – 1163

coagulation or particle evaporation. Sulphuric acid and some water-soluble organic aerosols may retain water even at
very low RH. Weingartner, Burtscher, and Baltensperger (1997) and Rissler et al. (2005) showed that fractal-like spark
discharge soot particles and hygroscopic biomass combustion particles decreased their mobility diameter 8%–15%
when humidified to a high RH. Diesel soot particles showed only a weak restructuring potential. Errors because of
agglomerate restructuring may be avoided if the aerosol is subjected to a water condensation–evaporation cycle prior
to inhalation. Validation can be made by comparing the determined respiratory tract deposition with and without an
added aerosol manipulation cycle or with tandem DMA measurements, where monodisperse aerosol samples from the
first DMA are humidified or heated and then dried to a low RH before classification in the second DMA. A narrow size
distribution increases effects of size-shifts. Concentration and temperature have to be taken into account to prevent size
shifts caused by coagulation and evaporation. Coagulation depends strongly on the particle size distribution. It should
be noted that uncertainties due to size shifts do not affect the total deposition determined from integration of the inhaled
and exhaled size distributions as long as the distribution is essentially covered within the measurement range.
Particle concentration has to be within the limits of the system. Noticeable increase in particle losses due to coagulation
were found at concentrations above 1 × 105 cm−3 and the effects were moderate up to 5 × 105 cm−3 . When the number
concentration is low, the measurement time needs to be prolonged. If allowing a standard deviation of 0.05 in DF
the lower detection limit is 300–1000 counted particles in the CPC for each size bin, based on counting statistics and
analysis of variation between different scans. For an aerosol with GMD 50 nm,
g 2.0 and 30 size bins, as used in this
study, the lowest particle concentration acceptable becomes approximately 2.500 cm−3 for the size range 15–250 nm
and a breathing time of 30 min. By adding the size bins together and using slightly longer sessions, accurate data may
also be produced at lower concentrations. The number of size bins is proportional to the lower limit, i.e. with four times
fewer size bins; the minimum concentration would be 600 cm−3 . Depending on GMD and
g the measurable interval
and the lower concentration limit will be shifted.
Fluctuations in the measured concentration have to be dealt with in different ways with respect to its cause. In
environments where the aerosol concentration is varying, the scan time may have to be reduced. Large number of
scans reduces errors caused by variations in inhaled aerosol concentration or breathing pattern. Variations in pressure
influence the sample flow rate to the DMA and thereby the measured concentration. A large chamber for exhaled air
decreases the breathing resistance and the pressure variations and improves mixing, but the residence time increases
and thus coagulation.

6. Conclusion

When comparing the two tested systems, the flow-through method offers many advantages compared to the bag
sampling method. These include significantly shorter residence times, which lead to lower losses and a higher maximum
concentration (5 × 105 cm−3 ) before coagulation distorts the results. Further, the length of the breathing session can
be adjusted to achieve the requested statistical power. A lower concentration limit was estimated to 600 cm−3 .
The introduced desmearing procedure (Collins et al., 2002), decreases the minimum scan time to approximately 45 s,
which makes it possible to use the set-up in environments with modest variations in particle concentration.
In conclusion, the described flow-through method incorporating an electrical mobility spectrometer offers a fast
and relatively simple way to determine respiratory tract deposition in human subjects. It can give quantitative results
at concentrations in most environments. However, several mechanisms, such as particle size shift between inhaled
and exhaled sample, particle losses in the instrument and pressure variations caused by the breathing, may produce
potentially large errors if not considered properly. We plan to use the described method, or variations thereof, together
with hygroscopic tandem DMA (H-TDMA) measurements on larger groups of subjects with varying respiratory status
in different environments, including street canyon, fresh wood smoke and indoor aerosols.

Acknowledgements

This study was supported by The Building and its Indoor Environment, a research school at Lund University, The
Research Council for Environment, Agricultural Sciences and Spatial Planning (FORMAS) and the Swedish National
Air-pollution Programme (SNAP, Swedish EPA). We greatly acknowledge AstraZeneca RnD, Lund, for letting us use
 J. Löndahl et al. / Aerosol Science 37 (2006) 1152 – 1163 1163

their breath simulator, Bo Olsson for his program for calculation of respiratory tract deposition and Elvira Vaclavik
and Peter Vinzents for the work with ethical application.

References

Anderson, P. J., Wilson, J. D., & Hiller, F. C. (1990). Respiratory-tract deposition of ultrafine particles in subjects with obstructive or restrictive
lung-disease. Chest, 97, 1115–1120.
Anselm, A., Heibel, T., Gebhart, J., & Ferron, G. (1990). “In vivo”—studies of growth factors of sodium chloride particles in the human respiratory
tract. Journal of Aerosol Science, 21, S427–S430.
Blanchard, J. D., & Willeke, K. (1983). Determination of the total lung deposition of ultrafine aerosols. Aerosol Science and Technology, 2, 292.
Chalupa, D. C., Gibb, F. R., Morrow, P. E., Oberdorster, G., Riesenfeld, E., Gelain, R. et al. (2002). A facility for controlled human exposures
to ultrafine particles. In: U. Heinrich, & U. Mohr (Eds.), Crucial issues in inhalation research—mechanistic, clinical and epidemiologic
(pp. 241–253). Washington, DC: ILSI Press.
Chalupa, D. C., Morrow, P. E., Oberdorster, G., Utell, M. J., & Frampton, M. W. (2004). Ultrafine particle deposition in subjects with asthma.
Environmental Health Perspectives, 112, 879–882.
Collins, D. R., Cocker, D. R., Flagan, R. C., & Seinfeld, J. H. (2004). The scanning DMA transfer function. Aerosol Science and Technology, 38,
833–850.
Collins, D. R., Flagan, R. C., & Seinfeld, J. H. (2002). Improved inversion of scanning DMA data. Aerosol Science and Technology, 36, 1–9.
Daigle, C. C., Chalupa, D. C., Gibb, F. R., Morrow, P. E., Oberdorster, G., Utell, M. J. et al. (2003). Ultrafine particle deposition in humans during
rest and exercise. Inhalation Toxicology, 15, 539–552.
Donaldson, K., Stone, V., Clouter, A., Renwick, L., & MacNee, W. (2001). Ultrafine particles. Occupational and Environmental Medicine, 58,
211–216.
Ezzati, M., Lopez, A. D., Rodgers, A., Vander Hoorn, S., & Murray, C. J. L. (2002). Selected major risk factors and global and regional burden of
disease. Lancet, 360, 1347–1360.
Ferin, J., Oberdorster, G., & Penney, D. P. (1992). Pulmonary retention of ultrafine and fine particles in rats. American Journal of Respiratory Cell
and Molecular Biology, 6, 535–542.
Hofmann, W., Morawska, L., Hitchins, J., & Ahmed, J. (2004). Total deposition of combustion aerosols in human respiratory tract: comparison
between experimental results and model predictions. Journal of Aerosol Science, 34, S1421–S1422.
International Commission on Radiological Protection. (1994). Human respiratory tract model for radiological protection (ICRP Publication 66).
Oxford, UK: Elsevier Science.
Ketzel, M., & Berkowicz, R. (2004). Modelling the fate of ultrafine particles from exhaust pipe to rural background: An analysis of time scales for
dilution, coagulation and deposition. Atmospheric Environment, 38, 2639–2652.
Kim, C. S., & Jaques, P. A. (2005). Total lung deposition of ultrafine particles in elderly subjects during controlled breathing. Inhalation Toxicology,
17, 387–399.
Montoya, L. D., Lawrence, J., Murthy, G. G. K., Sarnat, J. A., Godleski, J. J., & Koutrakis, P. (2004). Continuous measurements of ambient particle
deposition in human subjects. Aerosol Science and Technology, 38, 980–990.
Morawska, L., Barron, W., & Hitchins, J. (1999). Experimental deposition of environmental tobacco smoke submicrometer particulate matter in the
human respiratory tract. American Industrial Hygiene Association Journal, 60, 334–339.
Morawska, L., Hofmann, W., Hitchins-Loveday, J., Swanson, C., & Mengersen, K. (2005). Experimental study of the deposition of combustion
aerosols in the human respiratory tract. Journal of Aerosol Science, 36, 939–957.
Oberdorster, G., Gelein, R. M., Ferin, J., & Weiss, B. (1995). Association of particulate air-pollution and acute mortality—involvement of ultrafine
particles. Inhalation Toxicology, 7, 111–124.
Pagels, J., Rissler, J., Swietlicki, E., & Bohgard, M. (2003). Using polydisperse SMPS samples for fast determination of respiratory
deposition—Influence of small size-shifts between the inhaled and the exhaled sample. Journal of Aerosol Science, 34, S591–S592.
Rissler, J., Pagels, J., Swietlicki, E., Wierzbicka, A., Strand, M., Lillieblad, L. et al. (2005). Hygroscopic behaviour of aerosol particles emitted from
biomass fired grate boilers. Aerosol Science and Technology, 39, 919–930.
Rosati, J. A., Brown, J. S., Peters, T. M., Leith, D., & Kim, C. S. (2002). A polydisperse aerosol inhalation system designed for human studies.
Journal of Aerosol Science, 33, 1433–1446.
Tu, K. W., & Knutson, E. O. (1984). Total deposition of ultrafine hydrophobic and hygroscopic aerosols in the human respiratory system. Aerosol
Science and Technology, 3, 453–465.
Wang, S. C., & Flagan, R. C. (1989). Scanning electrical mobility spectrometer. Journal of Aerosol Science, 20, 1485–1488.
Weingartner, E., Burtscher, H., & Baltensperger, U. (1997). Hygroscopic properties of carbon and diesel soot particles. Atmospheric Environment,
31, 2311–2327.
WHO Working Group Meeting. (2003). Health aspects of air pollution with particulate matter, ozone and nitrogen dioxide. In WHO Working Group
Meeting (Ed.), World Health Organization, (pp. 10–34).
World Health Organization, & Ebrary I. (2002). The world health report 2002 Reducing risks, promoting healthy life. World Health Organization,
Geneva.
Zhou, J., (2001). Hygroscopic properties of atmospheric aerosol particles in various environments. Ph.D. thesis, Lund University, Dep. of Nuclear
Physics, Lund, Sweden, ISBN 91-78 4-120-3.