Modeling SOA Formation: Modeling SOA Formation: New Insights and More New Insights and More

QuestionsQuestions??

Department of Environmental Science and Engineering

UNC, Chapel Hill

MasteryMastery of Fire of Fire

400,000 years ago in Europe 400,000 years ago in Europe

100,000 years ago in Africa100,000 years ago in Africa

M. N. Cohne, 1977M. N. Cohne, 1977

From a global perspective, fire From a global perspective, fire results in huge emissions of black results in huge emissions of black carbon into the atmospherecarbon into the atmosphere

Biomass burningBiomass burning 6x106x1012 12 gg Fossil fuel burningFossil fuel burning 7x107x1012 12 gg

Biogenic aerosolsBiogenic aerosols 13-60x1013-60x101212gg((presentations by: presentations by: Schnaiter and Jackobson)Schnaiter and Jackobson)

What are Organic AerosolsWhat are Organic Aerosols??

organic liquid layer

inner solid core inorganic/carbon

H2O

H2SO4

Semi-volatile organics

Fresh wood soot (0.5 m scale)

Composition of LA Particulate Matter (adjusted for smoggy days)((Rogge &Cass et al, 1993, Turpin et al, 1991)

NH4 10nitrate 20sulfate 11EC 6other 23OC 30

Percent mass

PM10 Chemical Characterization in PM10 Chemical Characterization in

BeijingBeijing Xiao-Feng, Min Hua, Ling-Yan Hea, Xiao-Yan Xiao-Feng, Min Hua, Ling-Yan Hea, Xiao-Yan Tang, Tang, Atmos. Environ. 39 (2005) 2819–2827Atmos. Environ. 39 (2005) 2819–2827

Characteristics of carbonaceous aerosols in Beijing, ChinaYele Suna, Guoshun Zhuang, Ying Wang, Lihui Han, Jinghua Guo, Mo Dan, Wenjie Zhang, Zifa Wang, Zhengping Hao, Atmos, Environ. 38 (2004) 5991–6004

coal burning, traffic exhaust, and dustcoal burning, traffic exhaust, and dust from the long-range transportfrom the long-range transport

Mineral aerosol from outsideMineral aerosol from outside Beijing Beijing accounted for 79% of the total PM10 accounted for 79% of the total PM10 minerals and 37% of the PM2.5 in minerals and 37% of the PM2.5 in winter. It was 19% and 20% in summerwinter. It was 19% and 20% in summer

Characteristics of carbonaceous aerosols in Beijing, ChinaFengkui Duan, Kebin He, Yongliang Ma, Yingtao Jia,Fengkui Duan, Kebin He, Yongliang Ma, Yingtao Jia,Fumo Yang, Yu Lei, S. Tanaka, T. Okuta,Fumo Yang, Yu Lei, S. Tanaka, T. Okuta, Chemosphere 60 (2005) 355–364Chemosphere 60 (2005) 355–364

OC/EC ratio (on a 1.5 basis showed that OC/EC ratio (on a 1.5 basis showed that SOC accounted more than SOC accounted more than 50%50% for the total for the total organic carbon. In winter, the SOC organic carbon. In winter, the SOC contribution to OC was also significant, and contribution to OC was also significant, and as high as as high as 40%.40%.

Secondary organic Secondary organic aerosolaerosol (SOA)(SOA) Material as Material as organic compoundsorganic compounds that resides in that resides in the the aerosol phase as a result of aerosol phase as a result of atmospheric reactionsatmospheric reactions that occur in that occur in either the either the gasgas or or particle particle phasesphases..

Do we see any Do we see any chemical chemical evidence for SOA formation?evidence for SOA formation?

Leonardo Da Vinci describes blue haze and thinks that plant emissions are its source. (F. W. Went, 1959)

Da Vinci believes that it was due to water moisture emitted from the plants

F.W.Went published papers on biogenic emissions from vegetation over 40 years ago.

He posed the question, “what happens to the 17.5x107 tons of terpene-like hydrocarbons or slightly oxygenated hydrocarbons once they are in the atmosphere each year?”

Went suggests that terpenes are removed from the atmosphere by reaction with ozone

attempts to demonstrate “blue haze” formation

Went suggests that terpenes are removed from the atmosphere by reaction with ozone

attempts to demonstrate “blue haze” formation by adding crushed pine or fir needles to a jar with dilute ozone.

Over a eucalyptus forest in Over a eucalyptus forest in Portugal Portugal Kavouras et al.Kavouras et al. (1998,1999)(1998,1999) show evidence for show evidence for terpene reaction products in terpene reaction products in aerosolsaerosols

Terpenes products

Kavouras et al, 1998 ng m-3

pinic acid 0.4 - 85pinonic acid 9 - 141norpinonic acid 0.1 - 38Pinonaldehyde 0.2 - 32

Nopinone 0.0 - 13

-pinene -pinene

Turpin and co-workersTurpin and co-workers

In the LA area (estimated on smoggy In the LA area (estimated on smoggy

days from days from OC OC //ECEC ratios ratios), as much as ), as much as 50 - 50 - 80%80% of the of the aerosolaerosol organic carbonorganic carbon comes from comes from secondary aerosol secondary aerosol formationformation (1984 and 1987 samples) (1984 and 1987 samples)

In Atlanta in 1999, SOA averaged 46% of the In Atlanta in 1999, SOA averaged 46% of the total OC but with highs of 88% total OC but with highs of 88%

Turpin Approach for SOA formationTurpin Approach for SOA formation The primary aerosol elemental carbon The primary aerosol elemental carbon (EC)(EC)pripri and and

particle organic content particle organic content (OC)(OC)pripri in an un-reacted in an un-reacted

airshed are measured and a primary ratio of airshed are measured and a primary ratio of {{OC OC //ECEC}}pripri is determined is determined (Turpin et al for 1984 and 1987 aerosol (Turpin et al for 1984 and 1987 aerosol samples)samples)

Under SOA formation OCUnder SOA formation OCtottot and EC and ECtottot are measured are measured

OCOCsecsec= = OCOCtottot- - OCOCpri pri

OCOCpripri = EC = EC {{OCOC /EC} /EC} pripri

On smoggy days in California ~50 - 80% of the organic On smoggy days in California ~50 - 80% of the organic carbon comes from secondary aerosol formationcarbon comes from secondary aerosol formation

Spyros Pandis also recently looked Spyros Pandis also recently looked at OC/EC ratios (Pittsburgh area)at OC/EC ratios (Pittsburgh area)

He estimates that SOA formation can He estimates that SOA formation can account for 35-50% of the organic account for 35-50% of the organic carboncarbon

OC/EC Ratio and Photochemical Activity

0

2

4

6

8

10

12

14

15-Jul 16-Jul 17-Jul 18-Jul 19-Jul

OC

/EC

Ra

tio

0

10

20

30

40

50

60

70

80

90

100

O3

(p

pb

)

OC/ECO3

Pittsburgh, 2001

If we look at the IR spectra of aerosols collected from the smoky mountains, they look like lab aerosols from acid catalyzed

particle phase reactions of carbonyls…

0

0.001

0.002

0.003

0.004

0.005

5001000150020002500300035004000

wavelength (cm-1)

ab

so

rba

nc

e (

gly

ox

al)

glyoxal/acid-catalyst

Heterogeneous reactions as seen in Heterogeneous reactions as seen in the IR regionthe IR region

0

0.001

0.002

0.003

0.004

0.005

5001000150020002500300035004000

wavelength (cm-1)

ab

so

rba

nc

e (

gly

ox

al)

-0.1

-0.05

0

0.05

0.1

0.15

ab

so

rba

nc

e (

Sm

ok

y M

ou

nta

ins

)

glyoxal/acid-catalyst

Smoky Mountains SOA

C-O-C bonds

In the 1980s In the 1980s Yamasaki, Bidelman, Yamasaki, Bidelman, PankowPankow began to investigate the began to investigate the equilibrium distribution ofequilibrium distribution of PAHs, PAHs, alkanes, and chlorinated organicsalkanes, and chlorinated organics between the gas and the particle between the gas and the particle phases.phases.

K PAH

PAH TSPp

part

gas

PAHPAHgas gas + surface + surface PAH PAHpartpart

log Klog Kp p = -log P= -log Pssoo + const. + const.

Relate solid saturated vapor pressures with Kp

log Pso

log Kp

naphthalenenaphthalene

BaPBaP

PyrenePyrene

log Klog Kp p = -log P= -log PooLL + const. + const.

PAHs,PAHs, alkanesalkaneschlorinatedchlorinated organics organics

slope = -1

log Po(L)

log Kp

Problems with the theoryProblems with the theory

many aerosols are composed of 40-100% many aerosols are composed of 40-100% organicsorganics

This gives much more than a mono-layer This gives much more than a mono-layer of coverageof coverage

log Klog Kpp= m log P= m log Poo(L)(L)+ c+ c

KR T

p M wpLo

7 5 0 1

1 0 9

. fom

In 1994 James Pankow fixes the theory for liquid particles

Can we chemically / kinetically Can we chemically / kinetically model SOA Formation???model SOA Formation???

Numerical fittingNumerical fitting Semi-explicitSemi-explicit

From a modelingFrom a modeling perspective perspective Equilibrium Organic Gas-particle Equilibrium Organic Gas-particle partitioningpartitioning provides a context for provides a context for addressing SOA formationaddressing SOA formation

Gas/Particle PartitioningGas/Particle Partitioning

particleParticle typeCompound Temperature

Humidity

gas

Thermodynamic Equilibrium?

TSPC

CK

gas

partp

Cgas +surf Cpart

Kp will vary with 1/Po

Odum-Seinfeld Model SOA modelOdum-Seinfeld Model SOA model

Y= MY= Moo / / HC HC

Y Y MK

K Mii

o

i om i

om i oi

,

,( )1

Odum theory

- pinene- NOx experiments by Odum

Y Mo(g/m3) 1 0.012 1

2 0.028 7

3 0.059 22

4 0.067 34

5 0.078 38

6 0.122 83

7 0.125 94

Y MK

K MM

K

K Mo

om

om oo

om

om o

1 1

1

2 2

21 1,

,

,

,( ) ( )

Y = M= Moo / / HC HC

-pinene

Y MK

K MM

K

K Mo

om

om oo

om

om o

1 1

1

2 2

21 1,

,

,

,( ) ( )

Y MK

K MM

K

K Mo

om

om oo

om

om o

1 1

1

2 2

21 1,

,

,

,( ) ( )

Numerical fitting values for Kom and for OH, O3, and NO3 reactions with terpenes and sesquiterpenes were developed by Griffin and Sienfeld et al.

From the averages for OH, O3, and NO3 , the amounts of atmospherically reacted terpenes and sesquiterpenes were estimated ( HC HC ) ) by Griffin and Sienfeld et al.

Y= MY= Moo / / HC HC

Globally, biogenic emissions

13-24x1012g y-1 of aerosol mass

Gives little insight into the chemical nature of products involve in SOA formation

From a global perspective, fire From a global perspective, fire results in huge emissions of black results in huge emissions of black carbon into the atmospherecarbon into the atmosphere

Biomass burningBiomass burning 6x106x1012 12 gg Fossil fuel burningFossil fuel burning 7x107x1012 12 gg

Biogenic aerosolsBiogenic aerosols 13-60x1013-60x101212gg((presentations by: presentations by: Schnaiter and Jackobson)Schnaiter and Jackobson)

Semi explicit models link gas and particle phases

C=OO

cis-pinonaldhyde

particleC=OO

Gas phase reactions

K

R T

p M wpLo

7 5 0 1

1 0 9

. fom

Kp = kon/koff

[ [ iigasgas] + [part] ] + [part] [ [ iipartpart]] kon

koff

particle

kon

koff

C=OO

Kp = kon/koff

koff = kbT/h e -Ea/RT

Particle formation-self nucleationParticle formation-self nucleation

Criegee’s can react with aldehydes and Criegee’s can react with aldehydes and carboxylic groups to form secondary carboxylic groups to form secondary ozonides and ozonides and anhydridesanhydrides..

O=C

C=OCH3

+C

C=O.

CH3

oo.

C

C=OCH3

C

C=OCH3

O

oo

Secondary Organic Aerosol Secondary Organic Aerosol (SOA) Formation of Toluene(SOA) Formation of Toluene

CH3

+ OH Highly oxygenated gas phase products

Sunlight

NONOxx

NucleationNucleation

Klotz et al. observed a rapid particle Klotz et al. observed a rapid particle formation from the photolysis of formation from the photolysis of hexendiendial.hexendiendial. CH3

O

O

C7KETENEC7KETENEO

C O

+

C14KETENE

OC

O

OO

O C

O

C14KETNE + C14KETENEC14KETNE + C14KETENE SEED1 SEED1

2+2

Cycloadditon

PoL ~ 10-21 torr

Particle Growth from Toluene Particle Growth from Toluene Reaction with Background OHReaction with Background OH

bkg

6 min

10 min

3 min

CHOOO

CH3

OO

O

Criegee2

Criegee1OO O

-pinene

O3

COOHCOOH

pinic acid

+ otherproducts

O

pinonic acid

CHOO

COOH

+ CO, HO2, OH

COOHO

norpinonaldehyde

norpinonic acid

Mechanism

pinonaldehyde

OH

OO

O2

+

(a)(b)

(c)

(d)

(e)

pinonaldehyde

acetone

O

OO.

NO2NO

O

O.

pinald-oo

OH

pinonic acid

O

pinO2

OO.

NO2

NO

organic nitrate

+HO2

+NO2

pinald-PAN

=o

=o

=o

=o

=o

=o

=o

OO.

O2

=o

OO=C8=O

C8-oo.

O2

NO2NO

O

+ h

+

+CO+HO2=o

OO.

NO2NO

=o

=o+HO2

+ h

NO2NO

=o

OO.

C8-oo. (C8O2)

+CO+HO2

NO2

NO

(f)

(g)

CO2+

pinO2H2O+

+HO2

O2

OO

H3C-OO.

+oxygenated products

+NO2

+H3C-OONO2PAN

(stab-oxy)

+HO2

norpinonaldehyde

OOH

O=o+

pin-ooH

+OH

O

OO.

=o

NO2

NO

+CO2

norpinaldPAN

+NO2

+HO2

norpinonic acid+norpin-ooH

O

OONO2

=o

+O2

ONO2

=o

+

=o

ONO2

+

organic nitrate

Overall kinetic MechanismOverall kinetic Mechanism

linked gas and particle phase rate linked gas and particle phase rate expressionsexpressions

Particle Phase reactions

particle

C=OO

cis-pinonaldhyde

C=OO

polymers

Gas phase reactions

Particle Phase reactions

particle

C=OO

cis-pinonaldhyde

C=OO

polymers

Gas phase reactions

Particle Phase reactions

C=OO

cis-pinonaldhyde

C=OO

polymers

Gas phase reactions

A

B

C

D

[H3O+]

O

OOH

O

O

OH

O

OH

O

[H3O+]

O

OH

OH

OO

O

O

O O

HO

O

O

1

[H3O+]

4

2

OH

H2C

O

O O

[H3O+]

O

O

O

HO

O O

HO

3

O

O

OH

O

O CH2

O

O

OH

2

2

9

O O

HO

4

O

6

O

HOb

a

a

5

b

7 8

O

O

2

2

10 11

A

B

C

D

[H3O+]

O

OOH

O

O

OH

O

OH

O

[H3O+]

O

OH

OH

OO

O

O

O O

HO

O

O

1

[H3O+]

4

2

OH

H2C

O

O O

[H3O+]

O

O

O

HO

O O

HO

3

O

O

OH

O

O CH2

O

O

OH

2

2

9

O O

HO

4

O

6

O

HOb

a

a

5

b

7 8

O

O

2

2

10 11

pinonaldehyde

2 x

Pinonaldehyde dimerization

ESI-QTOF mass spectrum of SOA from ESI-QTOF mass spectrum of SOA from reaction of reaction of -pinene + O-pinene + O33 + acid seed + acid seed

aerosolaerosol (Tolocka et. al (Tolocka et. al., ., 20042004))

200 300 400 500 600 700 800 900 1000

m/z

337.

18 351.

18

361.

21

377.

2

393.

2

407.

2

423.

2

439.

2

453.

21 489.

32

300 320 340 360 380 400 420 440 460 480 500

321.

21

m/z

17

7.0

7

19

1.1

2 20

7.1

1

22

5.1

12

33

.14

24

5.1

2

25

5.1

82

61

.11

28

9.1

8

30

1.1

8

31

3.2

3

32

7.1

6

34

1.2

35

9.2

36

0.2

150 200 250 300 350

Inte

nsity,

A.U

.

m/z

O

OH

O

H2C

O

177

207

341

261289

91

77

.07

19

1.1

2 20

7.1

1

22

5.1

12

33

.14

24

5.1

2

25

5.1

82

61

.11

28

9.1

8

30

1.1

8

31

3.2

3

32

7.1

6

34

1.2

35

9.2

36

0.2

150 200 250 300 350

Inte

nsity,

A.U

.

m/z

O

OH

O

H2C

O

177

207

341

261289

9

M Na+ (ESI-QTOF Tolocka et al, 2003)

Particle phase pinonaldehyde dimers Particle phase pinonaldehyde dimers from from -pinene +O-pinene +O3 3 on on acid particlesacid particles

Similar results were obtained by Hartmut Herrmann’s Similar results were obtained by Hartmut Herrmann’s group group (Atmos Envir, 2004)(Atmos Envir, 2004)

Chemical SystemChemical System

-pinene

+ NOx+ sunlight + ozone----> aerosols

0.95 ppm -pinene + 0. 44ppm NOx

O3NO

NO2

NO2

model

data

Time in hours EST

pp

mV

Gas phase pinonaldehdye

OO

mg

/m3

Time in hours EST

Particle phase

model TSP

mg

/m3

Particle phase

model TSP

mg

/m3

Measured particle mass vs. model

data

Time in hours EST

Much lower terpene concentrations Much lower terpene concentrations

Different background aerosols which Different background aerosols which have different chemical and physical have different chemical and physical propertiesproperties

Low volatility gas phase products will Low volatility gas phase products will have different interactions with have different interactions with different pre-existing particlesdifferent pre-existing particles

The Real AtmosphereThe Real Atmosphere

New UNC aerosol smog chamberNew UNC aerosol smog chamber

Dual 270mDual 270m33 chamber chamber fine particle t fine particle t 1/21/2 >17 h >17 h

0.1 ppmV Toluene 0.1 ppmV Toluene + 0.1 ppm NOx+ 0.1 ppm NOx

072705S

0.0

0.1

0.2

0.3

0.4

0.5

8:00 10:00 12:00 14:00 16:00

LDT (hours)

To

luen

e, N

Ox-

O3

con

c (p

pm

)

-20

-10

0

10

20

Par

ticl

e m

assc

on

c (

g/m

3)

TSP

O3

Toluene

TotNO3

NO

NO

2

-pinene SOA formation in -pinene SOA formation in the presence of dilute diesel the presence of dilute diesel and woods soot particlesand woods soot particles

different solubilities of gas different solubilities of gas phase products in the phase products in the different soot matrices different soot matrices

ss of of -pinene products in diesel -pinene products in diesel

and woods soot and woods soot particles(UNIFAC)particles(UNIFAC)

Compounds Diesel Wood

cis-pinonaldehyde ~5 ~1

pinalic-4-acid ~8 ~1

cis-pinonic acid ~8 ~1

10-hydroxypinonaldehyde ~25 ~1

cis-pinic acid ~11 ~2

* Jang et al. 1997. Envr.Sci.Tech. 31, 2805-2811

Activity Coefficients were estimated at 298 K and 50 RH% by UNIFAC.

[ [ iigasgas] + [part] ] + [part] [ [ iipartpart]]

KR T

p M wpLo

7 5 0 1

1 0 9

. fom

Kp = kon/koff

kon

koff

kon = koff 7.5 RTfom / { poL Mw109}

The bottom line???The bottom line???

We could not even come close to We could not even come close to predicting predicting -pinene SOA in the -pinene SOA in the presence of background presence of background diesel seeddiesel seed aerosolaerosol

Our model was consistently Our model was consistently under under predictingpredicting observed SOA formation observed SOA formation by a factor of 5 to 10 by a factor of 5 to 10

50 50 g/mg/m33 of Diesel Soot Particles of Diesel Soot Particles

0

50

100

150

200

250

8 10 12 14 16 18Time (EDT)

Par

ticl

e m

ass

(ug/

m3 )

North SMPS

North filter

Inject diesel

250 250 g/mg/m33 of Diesel Soot Particles of Diesel Soot Particles

0

50

100

150

200

250

300

350

400

450

500

8 10 12 14 16 18Time (EDT)

Par

ticl

e m

ass

(ug/

m3 )

South SMPS

South filter

Inject diesel

Is the polarity of diesel exhaust particles Is the polarity of diesel exhaust particles changing as it ages changing as it ages SOA?? SOA?? Sangdon LeeSangdon Lee (Atmos. Envirn 2004) a(Atmos. Envirn 2004) added dded

deuterated alkanes to the chamber deuterated alkanes to the chamber atmosphere followed by the addition of atmosphere followed by the addition of diesel exhaustdiesel exhaust

Measured gas and particle phase Measured gas and particle phase concentrations and calculated a measured concentrations and calculated a measured

KKpp: : KKpp = d 42 = d 42partpart// {d 42 {d 42gasgasxTSP}xTSP}

Compared to theoryCompared to theory KR T

p M wpLo

7 5 0 1

1 0 9

. fom

-4.0

-3.5

-3.0

-2.5

-2.0

11 12 13 14 15 16Time in hours (EDT)

Log

Kp

(m3 /u

g)

d42-eicosane_estimated

d42-eicosane_observed

Predicted Kp

Observed Kp

KR T

p M wpLo

7 5 0 1

1 0 9

. fom

Kd42

d42 xTSPp

part

gas

When When --pinene is present the pinene is present the effect is even greatereffect is even greater

Diesel particle polarity increases as it Diesel particle polarity increases as it ages and reacts in the presence of ages and reacts in the presence of --

pinenepinene

-3.8

-3.6

-3.4

-3.2

-3.0

-2.8

-2.6

-2.4

-2.2

-2.0

8 9 10 11 12 13 14Time in hours (EDT)

Log

Kp (

m3 /u

g)

d42-eicosane_estimated

d42-eicosane_observed

predicted

observed

50 50 g/mg/m33 diesel exhaust diesel exhaust + 0.13 + 0.13 ppmppm -pinene-pinene in sunlight in sunlight

Add 0.13 ppmV -pinene

model

data

Where do we go from here?Where do we go from here?

Begin integrating single compound mechanisms

Expand mechanisms to take into account longer atmospheric aging times

Build better nucleation representations

Build a particle size model which shows the distribution of products with size

Where do we go from here?Where do we go from here?

Investigate the compounds that are resulting in SOA formation from diesel exhaust

AcknowledgementsAcknowledgements

Grants from Grants from National Science FoundationNational Science Foundation

USEPA STAR RESEARCH GRANT programUSEPA STAR RESEARCH GRANT program

Gifts of a GC-FTIR-MS system (HP 5890 GC & Gifts of a GC-FTIR-MS system (HP 5890 GC & HP 5965B FT-Infrared Detector) from the HP 5965B FT-Infrared Detector) from the Hewlett Packard Corporation Hewlett Packard Corporation and the Saturn and the Saturn GC-ITMS fromGC-ITMS from the Varian Corpthe Varian Corp..

Dept. Eviron Sci and Eng,NC Chapel HillNORTH CAROLINA

kamens@unc.eduhttp://airsite.sph.unc.edu/~kamens