Improved Prediction of In-Cloud Biogenic SOA: Experiments and CMAQ Model Refinements

Barbara Turpin, PISybil Seitzinger, Co-I

Rutgers UniversityNew Brunswick, NJ

SOA –

a major fine particle constituent

Zhang GRL 2007

organicsulfateammoniumnitrate

OOAHOA

Aerosol Mass Spectrometer Data

CondensationEvaporation

VolatileOrganicCompound>C7

Oxidation (+OH, NO3 , O3 )

Semi-volatile Products

pre-existing particlePartitioning into OM

SOAVolatileproducts

Gas-phaseReactions

Pankow 1994, AE

Odum 1996, ES&T

Traditional “smog chamber”

SOA

Precursors must be large (>C7) to produce high yields because partitioning depends on product vapor pressure

Dept of Environmental SciencesWhile important, fails to explain:

•Predictive models capture organic aerosol poorly (Hallquist ACP 2009)

•Atmospheric O/C > smog chamber O/C (Aiken EST 2008)

atmos. HOA = 0.06 – 0.1atmos. OOA-SV = 0.5 – 0.6atmos. OOA-LV = 0.8 – 1.0

sm. chamber SOA = 0.3 – 0.5

•Liquid water may be more accessible than OMe.g., high humidities of eastern US

Cloud evaporation

SOA

·OH

·OH

·OH

NOx

·OH

·OH

·OH

NOx

NOx, alkenes, aromatics

O

OO

O

·OH O3

·OH

Isoprene

In Clouds and Fogs:Hypothesis: Blando and Turpin, AtmosEnv, 2000

Gelencser and Varga, ACP, 2005

Organic gases are oxidized forming water-soluble compounds.

These partition into cloud droplets and react further (e.g., by ·OH).

Cloud droplets evaporate, lower volatility products remain, forming SOA.

and in Aerosol Water:Hypothesis: Carlton AtmosEnv, 2007

Volkamer GRL, 2007

Water-soluble gases partition into aerosol water and react.

Lower volatility products remain, forming SOA.

SOA through Aqueous Chemistry: Partitioning driven by water solubility

isoprene

Wet aerosol

SOAglyoxal OH

PartitioningOH

oligomerization

Aerosol water(LWC)

Organic/inorganic

constituents

Partitioning

low volatilityproducts

OH

oligomerization

Clouddroplet

SOACloud Droplet Evaporation

Volatile butHighly water soluble

~M

~0.1- 10M

glyoxal uptake

glyoxal evaporation

Implications•organic PM loading •vertical distribution•different precursors

(Volkamer ACP, 2009)

•product O/C

1

Carlton GRL 2006, AE 2007; Altieri EST 2006, AE 2008; Perri AE 2009

Previous research:Major products of OH radical oxidation verified (≥1mM)

Phase transfersRxs with •OH

Aqueous phase

ISOPRENE + oxidants (·OH, O3, NO3)

HOCH2CHO CHOCHO CH3COCHO(glycolaldehyde) (glyoxal) (methylglyoxal)

HOCH2CH(OH)2 (OH)2CHCH(OH)2 CH3COCH(OH)2

HOCH2COOH CH3COCOOH(glycolic acid) (pyruvic acid)

(OH)2CHCOOH CH3COOH(glyoxylic acid) (acetic acid)

HOOCCOOH CH2(OH)2(oxalic acid) (formaldehyde - hydrated)

CO2 HCOOH(formic acid)

Gas phase

(hydrated glyoxal)

Demonstrate kinetic feasibility of SOA formation through cloud processing using (mostly) measured rate const.

Rate constants available from cloud and wastewater literature Herrmann, Monod, Stefan and Bolton

Questions:How does precursor concentration matter?How do high molecular weight products form?Can the chemical model reproduce experiments?How do anthropogenic emissions impact biogenic

“aqueous” SOA formation?

Warneck AE 2003; Ervens JGR 2004; Lim EST 2005

Previous research:Cloud chemistry/parcel model predictions

Overall Goal:Improve the simulation of secondary organic aerosol (SOA)

formation through atmospheric aqueous chemistry

Approach:• Conduct aqueous experiments at cloud concentrations and

HNO3 (glyoxal/ methylglyoxal + OH)

• Validate/refine aqueous chemical mechanisms; update the cloud chemistry model

• Add in-cloud SOA formation to CMAQ

• Begin to explore the magnitude of in-cloud SOA formation and role of NOx /HNO3 in SOA formation from isoprene through cloud processing

Improved Prediction of In-Cloud Biogenic SOA

Roles of NOx in Aqueous SOA from Isoprene:

1. Gas phase formation of atmospheric oxidants

2. Gas phase formation of water soluble carbonyls

3. Aqueous NO3 catalysis, organic nitrogen products?

Project Experiments

Modeling (collaborators)

This Project

On-line ESI MS; IC-ESI-MS; FT-ICR MS; ESI-MS-MS, UV or IC for organic acids, DOC for mass balance, H2 O2

Experiments 10 ‐

3000 M ORG H2

O2

+hν ∙OH   (10‐12 M)2‐5  pH± HNO3

, NH4

ORG: glyoxal, methylglyoxal

Controls ORG+Prod+UV 

± HNO3

, NH4

ORG+Prod+H2

O2

± HNO3

, NH4UV+H2

O2

Aqueous-Phase Reactions with Product Analysis

Goal: Validate/Refine Aqueous Chemistry Model

Precursor/products modeled in reaction vessel: C

once

ntra

tion

(M

)

Time (min)

Methylglyoxal + OH radical

methylglyoxal

pyruvic acid

acetic acid

glyoxylic acid, formic acid, formaldehyde

oxalic acid

Methylglyoxal

Acetic acid

Glyoxylicacid

Pyruvic acid

formic acid

H2O2

·OH

Oxalic acid

Formaldehyde

Tan AE, 2010

Dilute aqueous chemistry model reproduces oxalic,  pyruvic acid and total organic carbon at 30 M

0 50 100 150 2000

5

10

15

20

25

30

35

0 50 100 150 2000

50

100

150

200

250

0 50 100 150 2000

200

400

600

800

1000

(a)

30 M glyoxal

(b)

300 M glyoxal

Oxa

lic A

cid

Con

cent

ratio

n (

M)

Model Prediction No H2SO4

280 M H2SO4

840 M H2SO4

(c)

3000 M glyoxal

Time (min)

0

2

4

6

8

10

0

20

40

60

80

100

0 100 200 300 4000

100

200

300

400

without H2SO4

280 M H2SO4

840 M H2SO4

model prediction

Oxa

lic a

cid

(M

)

Time (min)

30 M methylglyoxal + 0.15 mM H2O2

300 M methylglyoxal+ 1.5 mM H2O2

MethylglyoxalGlyoxal

Cloud relevant ~30 M

300 M

3000 M

Oxa

lic A

cid

(M

)

Oxa

lic A

cid

(M

)

Time (min) Time (min)

Tan EST 2009; AE 2010

Methylglyoxal (3000 M) + OH radical (10-12 M)

Oxalate formation verified with IC ESI-MSMethylglyoxal

Acetic acid

Glyoxylic acid

Pyruvic acid

formic acid

H2 O2

·OH

Oxalic acid

Formaldehyde

Oxalic acid (pk I, m/z- 89)

Pyruvic acid

40

30

20

10

00 5 10

Abu

ndan

ce

93

J

I

H

contaminant 1

CA+B

JI

H

contaminant 1

C

117

89

11997

92

87

14175 A+BConductivity (S)

Tim

e (m

in)

1000

2000

3000

1x105

2x105

2000

4000

1x104

2x104

1x105

2x105

0 100 200 300

3x104

6x104

m/z-

IC ESI-MS

Acetic acid

Oxalic acid

Tim

e (m

in)

m/z-

(180 min)

Tan AE, 2010

40

35

30

25

20

15

10

5

00 10 20 30 40 50

157143131

8789

89

191

C

A

unknown 1

C

Aunknown 1

Conductivity (S)

Tim

e (m

in)

0

2x104

4x104

6x104

8x104

0

1x105

2x105

3x105

abun

danc

e

0 100 200 3000

2x104

4x104

6x104

8x104

m/z-

40

35

30

25

20

15

10

5

00 20 40 60 80

113 199

191B

A+B

A

Conductivity (S)

Tim

e (m

in)

0 100 200 3000

2x104

4x104

6x104

m/z-

Methylglyoxal (3000 M) + UV

Methylglyoxal experiment with IC ESI-MS

IC ESI-MS IC ESI-MS

Methylglyoxal (3000 M) + H2 O2

Acetic acidPyruvic acid

Oxalic acid

Tim

e (m

in)

m/z-

m/z-

Oxalic acid forms only in the presence of OH radical

No oxalic acid forms in control experimentsNo oxalic acid

forms in control experiments

(180 min)

Tan AE, 2010

Little effect on oxalate (slighly faster decay with nitric acid)No change in nitrate or sulfate throughout the reactionNo discernable change in IC ESI-MSMight still form enough organic-N to observe by FT-ICR MS

Addition of HNO3

(1.7 mM) or (NH4

)2

SO4

(0.84 mM)to glyoxal (1 mM) and OH radical (10‐12

M)

Kirkland, in prep

0 10 20 30 40 50 600.3

0.4

0.5

0.6

0.7

0.8

0.0 0.1 0.2 0.3 0.40.3

0.4

0.5

0.6

0.7

0.8

0.0 0.1 0.2 0.3 0.43

4

5

6

7

8

9

10

0 10 20 30 40 50 60

Y c [%

]

3

4

5

6

7

8

9

10

[min h-1]

[min h-1]

Y c [%

]

Y c [%

]Y c

[%]

LWCav [g m-3]

LWCav

[g m-3]

(a) VOC/NOx = 5 r2 = 0.72

(b) VOC/NOx = 5 r2 = 0.45

(d) VOC/NOx = 100 r2 = 0.55

(c) VOC/NOx = 100 r2 = 0.69

0.1 1 10 10005

1015202530354045

Trajectory 1Trajectory 2Trajectory 3

VOC/NOx

Y c [%

]

Impact of NOx on SOA from isoprene through cloud proc.

Note: higher SOA yields with higher NOx because more gas phase production of water soluble carbonyls

Ervens GRL 2008Ervens ACP 2010

Yield (%) = mass C in SOAmass isoprene C

Feingold microphys. cloud model

•multiple cycles •stratocumulus•partitioning of wsoc•gas+aq chem•Ervens aq chem •altered Gly, PA chem

Cloud contact time

VOC/NOx

Yiel

d c(%

)

Yiel

d c(%

)

Yiel

d c(%

)

Model deviates from measurements at higher  concentrations – what happens in wet aerosols?

0 50 100 150 2000

5

10

15

20

25

30

35

0 50 100 150 2000

50

100

150

200

250

0 50 100 150 2000

200

400

600

800

1000

(a)

30 M glyoxal

(b)

300 M glyoxal

Oxa

lic A

cid

Con

cent

ratio

n (

M)

Model Prediction No H2SO4

280 M H2SO4

840 M H2SO4

(c)

3000 M glyoxal

Time (min)

0

2

4

6

8

10

0

20

40

60

80

100

0 100 200 300 4000

100

200

300

400

without H2SO4

280 M H2SO4

840 M H2SO4

model prediction

Oxa

lic a

cid

(M

)

Time (min)

30 M methylglyoxal + 0.15 mM H2O2

300 M methylglyoxal+ 1.5 mM H2O2

MethylglyoxalGlyoxal

Cloud relevant ~30 M

Aerosol relevant ~1-10 M

Oxa

lic A

cid

(M

)

Oxa

lic A

cid

(M

)

Time (min) Time (min)

300 M

3000 M

(a) ESI-MS 69 min. neg mode

(b) ESI FT-ICR MS69 min. neg mode

Peak Location300 350 400 450 500

Peak

Hei

ght

0

20

40

m/z50 100 150 200 250 300 350 400 450 500 550 6001000

Ion

abun

danc

e

0

2000

4000

6000

8000

10000

C14H17O12377.070380

C17H13O11393.044360

C16H19O14435.075940

C13H15O12363.054800

C23H13O20465.177920

C12H13O20476.99925

m/z0 100 200 300

Ion

abun

danc

e

0100000200000300000400000500000

73

87

89

PA

GA

•High MW products

•O/C ~ 1

•Not seen in std mix (not ESI artifact)

•Not in controls (·OH involved)

Ultra high resolution FT-ICR MS (9.4 T) provides exact elemental comp. > m/z 300

OA

Methylglyoxal (1mM) + ∙OH

*Electrospray ionization Fourier transform-ion cyclotron resonance mass spectrometry

Altieri AE 2008

40

35

30

25

20

15

10

5

00 10 20

Abu

ndan

ce

unknown 2

221 263249

235

89

119177

133177

175161

87

75

177

103

117

177

131

unknown 3

I

H

F

E

unknown 2

A+B+C

unknown 3

IH

FE

A+B+C

Conductivity (S)

Tim

e (m

in)

02x104

4x104

0

3x105

6x105

03x105

6x105

05x104

1x105

0

1x104

2x104

01x105

2x105

0 100 200 3000

3x104

6x104

m/z-

Chemistry at Higher Concentrations―Methylglyoxal

Higher-MW ions

Oxalic acid (peak I, m/z- 89)

RT of Succinic acid (m/z- 117)

RT of Malonic acid (m/z- 103)

Formation of higher C# products

Tan AE, 2010

0 100 200 300 4000

100

200

300

400

Oxa

lic a

cid

(M

)

Time (min)

Chemistry at Higher Concentrations―Methylglyoxal

Larger acids are important products at higher concentrations.

Formation of higher C# products suggests radical-

radical reactions0

100

200

300

peak F quantified as malonic acid

peak E quantified as succinic acid

Con

cent

ratio

n (

M)

without H2SO4

280 M H2SO4

840 M H2SO4

0 100 200 300 4000

50

100

150

200

Con

cent

ratio

n (

M)

Time (min)(Tan AE 2010)

OO 2 H2O OHHO

OHHOH+

OH

H2O

OHHO

OHHO

O2OHHO

OHHO

OHO

OHHO

OO

RO2

OHHO

OHHOO

HO

HO+ O

OH

OH

O2 HO2

O

HO

glyoxal(hydrated)

1 glyoxylic acid(hydrated)

2 formic acid

+ HO2

OHO

OHHO

O2OHO

OHHOOO

OO

OHHO3 oxalic acid

+ HO2

OHO

OHHOO

OHO

OHHOO +

OO

OHO

O

HO+ CO2

CO2

CO2

glyoxal

decomposition

+ O2

+ O2

OH

H2O

RO2decomposition

decomposition

decomposition

O2 HO2

OH

H2O

O2 HO2

decomposition

R*1

R*3

R*2

R*4

R*4

4(hydrated)

4

4

4

4(hydrated)

Aqueous Mechanism - Glyoxal Y Lim ACPD 2010

Y Lim ACPD 2010

Mass o

xalat

e

formed

per m

ass

glyox

al rea

cted

Mass high MW

products formed

per mass glyoxal

reacted

Model produces oxalate at cloud‐relevant  concentrations and oligomers at aerosol‐relevant 

glyoxal concentrations (OH radical = 10‐12

M)

Cloud wet aerosol

Chemistry in wet aerosols –

complex

•Photolysis and photooxidation reactions

•Reactions of organics with ammonium and amines to form organic-nitrogen compounds

•Reactions of organics with sulfate to form organic-sulfates

•Acid and ammonium catalyzed oligomerization

Investigators include:Anastasio, Claeys, Cordova, de Haan, Flagan, Galloway, Grgic, Guzman, Hoffmann, Jimenez, Keutsch, Liu, Maenhaut, Michaud, McNeill, Monod, Noziere, Sareen, Seinfeld, Shapiro, Sun, Tolbert, Volkamer, Wortham, Yasmeen, Zhang

SOA formation  from glyoxal 

increases with LWC 

SOA formation  faster with OH  radical (light)

Volkamer ACP 2009

(Seed: ammonium bisulfate + humic acid salt; acetylene; hydrogen peroxide; light/dark) Dark reaction

(Carlton EST 2008)

Improved model

performance;

ICARTT

SOA through cloud processing now in CMAQ (yields)

“Aqueous” comparable in magnitude to “smog chamber” SOA(Chen ACP 2007; Fu JGR 2008; Carlton EST 2008; Fu Atmos.Environ. 2009)

Dept of Environmental SciencesOverview

(Lim et al., ACPD 2010)

•Predictive models capture organic aerosol poorlyaqueous SOA precursors different (Volkamer ACP 2009)improved with “aqueous” SOA (Carlton EST 2008)

•Atmospheric O/C > smog chamber O/C (Aiken EST 2008)

atmos. OOA-LV = 0.8 – 1.0 atmos. OOA-SV = 0.5 – 0.6sm. chamber SOA = 0.3 – 0.5aqueous SOA = 1-2

•Sometimes liquid water is more accessible than OM e.g., high humidities of eastern US

•Atlanta - SOA is correlated with liquid waterAtlanta (Hennigan GRL 2008; Hennigan ACP 2009)

•Anthropogenic pollutants generate biogenic SOAhigher yields for aqueous SOA from isoprene at high NOx (Ervens GRL 2008)

Secondary Organic Aerosol         

Past/present Students, PostdocsHo Jin LimKatie AltieriYi TanMark PerriYong Bin LimDiana Ortiz SupportMary Moore U.S. EPA – STARAnjuli RamosPhyllis KuoJeff Kirkland

Acknowledgements