Gas-Particle Partitioning of Carbonyls inSimulation Chamber Studies of
Secondary Organic Aerosol Formation
John Wenger, Robert Healy, Kristina Kuprovskyte, Shouming Zhou
Department of Chemistry and Environmental Research InstituteUniversity College Cork
Ireland
Gas/Particle Partitioning
• Many organic compounds partition between gas and particle phase
• Kp=[particle phase]/[gas phase][aerosol]
VOC
Gas phase products
oxidized
KpOrganic aerosol
Denuder-Filter Sampling
Particle phase
Gas phase
Air Flow
Denuder tube coated with XAD-4 resin
Filter
Sorbent
Typically used for non-polar organic compounds
• Aim: to apply denuder-filter sampling to studies of SOA formation
• Secondary organic fraction represents up to 70% of the organic fraction of fine aerosols
• Composition of SOA ?
• Formation mechanisms ?
• Main species contributing to SOA ?
Secondary Organic Aerosol (SOA)
Biogenic Precursors of SOA
Seinfeld & Pankow, Annu. Rev. Phys. Chem. 2003. 54:121–40
Anthropogenic Precursors of SOA
Seinfeld & Pankow, Annu. Rev. Phys. Chem. 2003. 54:121–40
SOA Formation
Emission ofVolatile Organic
Compounds
SOA PrecursorsAlkanes (>C7)
AromaticsAlkenes (>C6)
~100 compounds
High volatility productsCarbonyls
Low volatility productsMultifunctional oxygenates
Reaction withOH, O3, NO3 Gas-particle
Partitioning
Derivatization of oxygenated organics
Yu et al. ES&T 1998, 32, 2357-2370
FF
F
F F
O N
N O
F
F
F
F
F
MW = 448
Fragment mass = 181
Glyoxal derivatizedfragments here
More than one isomer possible- (multiple peaks)
O
Oglyoxal derivatizes
Example: OH + Limonene SOA
Jaoui et al., ES&T, 2006, 40, 3819-3828
Denuder-Filter Sampling at UCC
• 5-channel glass denuder
• Coated with XAD-4 resin
• Denuder and filter doped with pentafluorobenzyl hydroxyl amine (PFBHA) to convert polar carbonyls to oximes
Denuder Tube(gas collection)
Filter pack (particle collection)
Air flow
Trapping Efficiency Tests• Range of carbonyls introduced to chamber (10 - 200 ppbv)
To pump To pump
Denuder tube coated withXAD-4 and PFBHA
Sampling ports
Impinger system
PFBHA in H2O
6000 L FEP Teflon chamber
Trapping efficiency =100*(1-Cout/Cin)
XAD-4 vs XAD-4/PFBHABenzaldehyde
0
20
40
60
80
100
120
10 20 30 40 50
Time (min)
Trap
ping
Effi
cien
cy (%
) XAD-4
XAD-4 +PFBHA
O
XAD-4 vs XAD-4/PFBHAMethylglyoxal
0
20
40
60
80
100
120
10 20 30 40 50
Time (min)
Trap
ping
Effi
cien
cy (%
) XAD-4
XAD-4 +PFBHA
O
O
Temime et al., ES&T, 2007, 41, 6514-6520
Atmospheric Simulation Chamber at UCC
• FEP foil (6000 litres)• Dry purified air• Atmospheric P and T• Humidity control
• on-line GC• NOx and O3 analysers• denuder – filter, GC-MS• Particle Sizer and counter
p-xylene photo-oxidation experiment
• Aerosol mass yield=3.84%
0
100
200
300
400
500
600
700
800
-30 70 170 270 370 470
Time (min)
ppbv
0
50
100
150
200
250
300
350
400
450
ugm
-3
Lights OffLights On
NO NO2
Aerosol mass
O3
p-xylene/5
p-xylene photo-oxidation
OH
O
O
O
O
O
O
O
O
O
OH additionH-abstraction
ringcleavage
OH addition
O
O
GC-MS Analysis of p-xylene extracts (filter)
1
2
3
4
5
6
7
8
Reconstructed ion chromatogram (m/z 181) of a filter extract from XYL_NOx_11: glyoxal 2: methylglyoxal 3: oxopropanedial 4: 2,3-dioxobutanal 5: 3-hexene-2,5-dione6: 2-hydroxy-3-oxobutanal 7: 2,3-dioxobutanal 8: oxopropanedial.
Denuder-Filter vs Filter Alone
0
10000
20000
30000
40000
50000
60000
70000
80000
glyoxal methylglyoxal hexenedione p-tolualdehyde
GC
-MS
resp
onse
Denuder FilterFilter alone
GC-MS Analysis of p-xylene extracts (filter)
XYL_NOx_1 XYL_NOx_2
Relative humidity (%) <5 24
glyoxal 3.88 ± 0.26 4.37 ± 0.70
methylglyoxal 3.72 ± 0.12 3.47 ± 0.74
oxopropanedial 0.66 ± 0.06 0.79 ± 0.14
2,3-dioxobutanal 5.22 ± 0.11 6.19 ± 0.57
2-hydroxy-3-oxobutanal 0.38 ± 0.04 0.31 ± 0.03
p-tolualdehyde not observed 0.84
3-hexene-2,5-dione 2.06 ± 0.28 2.44 ± 0.34
Total identified 15.92 ± 0.87 18.41 ± 2.52
% Contribution to SOA mass
Gas/Particle Partitioning Values
• Kp calculated both theoretically:
610760
×°×××××
=Lomom
omltheoretica PMW
TRfKpγ
• MWom= Average molecular weight of organic species in particles (=120)
• fom= fraction of particle that is organic (= 1)• γom = Activity coefficient (assumed =1)• Po
L = sub-cooled vapour pressure
Pankow,Atmos Environ,1994
Gas/Particle Partitioning Values
• Kp calculated both theoretically:
• and experimentally:
][exp aerosolCC
Kpgas
particle
×=
610760
×°×××××
=Lomom
omltheoretica PMW
TRfKpγPankow,
Atmos Environ,1994
Gas/Particle Partitioning Values
• Glyoxal and methylglyoxal Kp several orders of magnitude higher than expected
p-tolualdehyde Hexenedione Glyoxal Methylglyoxal
Kptheoretical 3.2x10-07 1.3x10-06 9.8x10-10 2.0x10-09
Kpexperimental 4.3x10-06 3.8x10-05 4.2x10-05 3.3x10-05
Kpexp/theory 13 29 43238 16963
(vapour pressures from SPARC on line calculator- University of Georgia)
Experiments at PSI Chamber
Healy et al., ACP, 2008, 8, 3215-3230
Denuder-Filter Configurations
• Setup 2 allows for the trapping efficiency of the tube to be tested for each experiment
Tube 1
Tube 2
FilterTube 1
Filter
Tube 2
AIR
FLOW
Setup 1(Typical)
Setup 2(Gas-phase breakthrough)
AIR
FLOW
Gas phase
Particle phase
Gas phase breakthrough
glyoxal Methylglyoxal
Denuder vs Filter Extracts
Photooxidation of Isoprene
O
methacrolein
O
methyl vinyl ketoneO
Oglyoxal
O
O
methylglyoxal
OHO
glycolaldehyde
O
HO
hydroxyacetone
OHO
C4 hydroxycarbonyl
O
C5 carbonyl
Photooxidation products
Gas/Particle Partitioning Values
• methacrolein, methylvinylketone not detected in particle phase
• Glyoxal and methylglyoxal Kp several orders of magnitude higher than expected
glycolaldehyde Hydroxyacetone Glyoxal Methylglyoxal
Kptheoretical 3.6x10-07 7.2x10-07 9.8x10-10 2.0x10-09
Kpexperimental 2.2x10-05 1.5x10-05 4.4x10-05 6.7x10-06
Kpexp/theory 59 20 45538 3476
Photooxidation of 1,3,5-TMB
Gas/Particle Partitioning Values
2-methyl-4-oxo-2-pentenal Methylglyoxal
Kptheoretical 9.3x10-07 2.0x10-09
Kpexperimental 1.8x10-04
*1.3x10-04
1.2x10-05
*2.0x10-05
Kpexp/theory 190 6256
*Obtained using PTR-MS with denuder and heated inlet to vapourize SOA;Hellen et al, ES&T, (2008), 42, 7347-7353.
SOA Formation mechanisms
Emission ofVolatile Organic
Compounds
SOA PrecursorsAlkanes (>C7)
AromaticsAlkenes (>C6)
~100 compounds
High volatility productsCarbonyls
Low volatility productsMultifunctional oxygenates
Reaction withOH, O3, NO3 Gas-particle
Partitioning
Kp exp ≈ Kp calc
Kp exp >> Kp calc
HeterogeneousReactions*
Acid-catalyzed Oligomerization-uptake of glyoxal to particles
Liggio et al.,ES&T, 2005, 39, 1532-1541
O
H
OH
OH
H
OH
OH
H
OH
OH
OH
HOOH
OH
H
HOOH
OH
H
OOH
H2O
H+
H2O
- H+
H
HOOH
+
HO
OH
OH
O
H
OH
OH
OH
- H+
HO
OH
OH
O
OH
OH
OH
H+
Conclusions• Small dicarbonyls (glyoxal and methyl glyoxal)
partition to the particle phase much more than expected from vapour pressure calculations. Consistent with oligomerization hypothesis (in chambers at least!).
• Monofunctional carbonyl compounds much less likely to undergo reactive uptake
• Models that incorporate Kp for oxidation products should use compound-specific values (Johnson et al., 2005, 2006; Jenkin et al., 2004).
Future Directions
• On-tube derivatization of acids/phenols
• Further chamber experiments on oxygenated aromatics, PAHs and BVOCs
• What happens in the real atmosphere?
36
Sampling site
500 m
N
City centre
Tivoli Docks
50 m
Sampling point
Denuder-filter sampling
To pump
Quartz fibre filters (2)(PFBHA-treated)
Annular denuder(XAD-4 coated & PFBHA-treated)
(KI-coated denuder)
Cyclone(PM2.5 fraction)
Inlet
GC-MS data (denuder & filter extracts) after sampling for 24 hr 23-24th September 2008
* Impurities or column peaks
Carbonyls at Tivoli Docks
Sampling for 24 hours at 16.7 L/min. Weather conditions dry, mainly cloudy, lightwind from SW
Gas phase conc. Particle phase conc.
Detection limit for standard
ng/m3 (ppbv) ng/m3 ngMVK 1087 (0.38) * 3.5Methacrolein 380 (0.13) * 1.0Glycolaldehyde 1006 (0.48) * 4.0Hexanal 409 (0.10) 2.0Heptanal 144 (0.03) 2.0Benzaldehyde 263 (0.06) 2.0p-Tolualdehyde 55 (0.01) 0.6Nonanal 404 (0.07) 5.0Decanal 343 (0.05) 5.0Glyoxal 77 (0.03) ? 0.6Methylglyoxal 69 (0.02) ? 0.3Dimethylglyoxal 102 (0.03) ? 0.6
3,5-dimethylbenzaldehyde 73 (0.01) 0.6
First studies on SOA formation from naphthalene
Odum yield curves
SOA mass concentration (M0)
0 100 200 300 400 500
SOA
yie
ld (Y
)
0.00
0.05
0.10
0.15
0.20
0.25
RH=0%, HC/NOx=1.0-2.2
RH=0%, HC/NOx=3.0-4.4
RH=25%, HC/NOx=1.0-1.8
RH=50%, HC/NOx=1.0-1.8
HONO as OH source
∑∑==
⎟⎠⎞
⎜⎝⎛
×+×
==n
1i i0
ii0
n
1ii
KM1KαMYY
Aerosol yield parameters
One compound model
Two compounds model
a K a1 K1 a2 K2
RH=0, HC/NOx=1.0-1.8
0.1636 0.0113 0.1477 0.013 1 3.19E-5
RH=0, HC/NOx=3.0-4.4
0.2198 0.0125 0.2198 0.0125 3.52E-7 1.32E-8
RH=25, HC/NOx=1.0-1.8
0.2324 0.0081 0.2324 0.0081 3.65E-5 1.22E-6
RH=50, HC/NOx=1.0-1.8
0.2548 0.0095 0.2548 0.0095 1.96E-7 6.44E-9
Acknowledgements
• Robert Healy, Shouming Zhou
• Kristina Kuprovskyte, Ashley Allshire
• Brice Temime (now in Marseille)
• Josef Dommen, Axel Metzger et al. (PSI)
XAD-4 vs XAD-4/PFBHA2,6-dimethylbenzoquinone
0
20
40
60
80
100
120
10 20 30 40 50
Time (min)
Trap
ping
Effi
cien
cy (%
) XAD-4
XAD-4 +PFBHA
O
O
Glyoxal trimer dissolved in solvent mix vs dissolved in methanol
Kp calculations
• MWom= Average molecular weight of organic species in particles (=120)
• fom= fraction of particle that is organic (= 1)• γom = Activity coefficient (assumed =1)• Po
L = sub-cooled vapour pressure • Adsorption to particle surface not considered
for Kp values
610760
×°×××××
=Lomom
omltheoretica PMW
TRfKpγ
Polymerization reactions in aromatics-SOApolymer-”backbone”
Gas-phase Carbonyl yields for p-xylene
methylglyoxal 35%methylbutenedial 3%
Glyoxal 30%Hexenedione 5%
P-tolualdehyde 10%
Minimize Artefacts
• Direct filter sampling is prone to artefacts• Adsorption of gases to filter - Positive artefact• Desorption of semi-volatiles – Negative artefact• Denuder-filter sampling minimizes artefacts• Mainly used for non-polar organic compounds
that partition between both gas and particle phases, e.g. PAHs