Gas/Particle Partitioning

Why is gas/particle partitioning important?

Dispersion of
Pollutants Introduced
into the Atmosphere
as Determined by
Residence Times
(Husar and Patterson, 1980)

Washout by rain:
Crain
WT
Wg (1 ) W p
Cair
Washout consists of capture of gas and particle phases
Gases equilibrate with rain via Henrys Law
(reversible air-water exchange)
Particles are irreversibly captured
For PCBs: Wg ~ 200-400, Wp ~ 104 105
For PAHs: Wg ~ 10-500, Wp ~ 103 106

Atmospheric reactions
Many (most?) organic species are reactive in the
atmosphere via direct or indirect photolysis
Half-lives of PAHs:
Gas phase = minutes-hours
Particle phase = hours-days
How does this contradict figure 1?

Health effects:
Particles can deposit in lungs.
What role does chemical makeup of these particles play?

Background of gas/particle distribution

Parameters affecting G/P
distribution:
1.

Chemicals properties
(liquid VP, Koa)

2.

particle characteristics:
size distribution,
carbon content,
composition

3.

Atmospheric
conditions: T, RH

How do we measure G/P partitioning?

Traditional hi-vol air
sampler uses a filter
to capture particle
phase and then an
adsorbent to capture
gas phase
Filters:
GFF, QFF, PTFE
Adsorbents:
PUF, polymer resin,
tenax

Denuder
under laminar
flow, gases
diffuse faster, so
adsorb to
surface
best for hi conc
and lo MW
gases (NOx, etc)

Impactor
MOUDI
micro-orifice
uniform deposit
impactor
nozzle size and thus
air velocity changes
in each chamber
8-10 stages
0.2 0.3 um cutoff
filter gets the rest

Cyclone

sampling artifacts?
Sorption of gases to filters or particles during
sampling
Volatile losses from filter, particles, impactor stages
especially at high T, high flow (I.e. large pressure
drop
Effect on partition coefficient?

Theory: adsorption vs. absorption

Particle phase concentration dynamics?

Adsorption
(Pankow, 1987)
Absorption
(Pankow, 1994)

N S aTSP T e ( Ql Qv ) / RT
CP
KP

CGTSP
1600 p Lo
f om 760 RT
CP
KP

o
6
CGTSP MWom om p L 10

Kp = mass normalized partition coefficient

Cg, Cp = gas and particle phase concentrations
NS = # of adsorption sites
Ql, Qv = enthalpy of desorption and vaporization
aTSP = aerosol surface area

TSP = total suspended particle

mass/volume
fom = fraction of organic matter
MWom = molecular wt of NOM
om = activity coeff. of SOC in OM

If absorption is important, Koa may be a better

predictor of Kp
those previous equations used p as the descriptor variable.
we could also use Koa :

K P f om

MWoct oct
K oa
12
MWom om oct 10

Koa = octanol-air partition coefficient

fom = fraction of organic matter
MWom, MWoct = molecular wt of NOM or octanol
om , oct = activity coeff. of SOC in NOM or octanol
oct = density of octanol

Temperature dependence
Yamasaki (1982)

Cg

A
log
B
C p / TSP
T

semi-volatile
compounds have
significant conc in
both the gas and
particle phases

What is observed in the field?

Mader and Pankow, 2002

Kp alone often
gives slopes of -1

normalizing Kp to
om, OC, or EC
improves R2, but
slopes still -1

accounting for OC
and EC usually
gives best R2,
slopes still -1

difference between PAHs and PCBs in same air mass sampled twice
July 22-23, 1994
1

Chicago PAH
LM PAH
Chicago PCB
LM PCB

log KP (m /g)

0
-1

Slopes:
Chicago
PAH = -0.734

-2

PCB = -0.517
-3

LM

-4

PAH = -0.811

-5

PCB = -0.634

Simcik et al, 1998

-6
-6

-5

-4

-3

log p0L (Pa)

-2

-1

Despite a transport time of ~3.5 hrs and reduction in

concentration to 1/3, Kp and slopes have not changed.
Slopes not necessarily 1 at eqbm (?)