Organic Aerosol Components VAPOrganic Aerosol Components VAP for

Aerosol Mass Spectrometers

Qi Zhang

OA Components VAP• Goal: To represent the enormously complex atmospheric

OA system as lumped descriptions of a limited number of t th t b l t d t di ti tcomponents that may be related to distinct sources,

physicochemical properties, and atmospheric processes.

• Methodology : Multivariate statistical analysis of DOE ACRF and IOP aerosol mass spectrometer data.

• Assumption: An AMS organic aerosol data matrix is comprised of the linear combination of OA components p pwith constant profiles that have varying contributions across the dataset.

Aerosol Mass Spectrometer (AMS) SystemsLong term measurements (3 systems) –Aerosol Chemical Species

MonitorACSM

(mini-AMS)

Long term measurements (3 systems) Tropical Western Pacific in Darwin, AustraliaThe Southern Great Plains, OklahomaACRF MAOS mobile facilityMonitor

Quadrupole AMS

(mini AMS)

Q AMS

ACRF MAOS mobile facility

Intensive focused studies (IOP)Quadrupole AMSCompact Time-of-Flight AMS

Q-AMSc-ToF-AMS

Intensive, focused studies (IOP)• CARES: 3 HR-ToF-AMS (T0, T1, G1)• BNL-IOP

GVEX t l t 3

High Resolution Time-of-HR-ToF-AMS• GVEX: at least 3 • ASP previous G1 studies: NEAQS (2002),

MASE (2005), MAX-Mex (2006), MAX-Tex Flight AMS(2006), CHAPS (2007), and VOCALS

(2008)

3. 0

2. 5

2. 0

1. 5

Mas

s C

once

ntra

tion

(µg

m-3

)

Am m on ium 4 .8 ug /m 3Nitra te 5 .8Su lpha te 9 .4Organ ics 13.4Ch lo ride 0.15

M ex ic o C i ty 2/200 2

43 44

High time resolutionAMS Data

1. 0

0. 5

0. 0

Nitr

ate

Equi

vale

nt M

141201 0080604 020m /z (D alt ons )

SO+SO2

+

S+

SO3+

HSO3+

H2SO4+

57ACSM(mini-AMS)

High time resolution

1. Quantitative composition(mini AMS)

Q AMS15 - 30 min.

Q-AMSc-ToF-AMS

2. Size‐resolved composition

m/z 43C2H3O+HR-ToF-AMS

CHNO+C2H5N+

C3H7+

CH3N2+

2-5 min. (ground)5-30 s (airborne)

3. Elemental comp. (CnHmNpOzSw)

AMS Organic Mass Spectrum

ORGt×i

msmeasured = ca· msa + cb· msb + cc· msc + …

0.30

ORGt×i

t = 50 – 300 / dayi 200 1000

0 15

0.20

0.25

Inte

nsity

All

i = 200 - 1000 t×i = 1E4 – 3E5 /day

0 00

0.05

0.10

0.15

150200

250

Sign

al All

observations

0.00

050

100150

23:20:00 23:30:00

23:40:00 23:50:00 00:00:00 00:10:00 00:20:00

m/zSampling timet0t1

t2t3

......

multivariate

OA1, OA2, OA3, …analysis

Multivariate analysis methods:• Positive Matrix Factorization (PMF)Positive Matrix Factorization (PMF)• Tracer-based linear decomposition• Spectra based linear decomposition• Spectra-based linear decomposition

(chemical mass balance-style)Literature (applications to AMS data):• Zhang, Q. et al. (2005), Deconvolution and quantification of hydrocarbon-like

and oxygenated organic aerosols based on aerosol mass spectrometry, ES&T,and oxygenated organic aerosols based on aerosol mass spectrometry, ES&T, 39(13), 4938-4952

• Ulbrich, I., M. et al. (2009), Interpretation of Organic Components from Positive Matrix Factorization of Aerosol Mass Spectrometric Data, Atmos. Chem. Phys., 9, p y2891-2918.

• Ng, N. L. et al. (2010), Development of Real-Time Methods for the Estimation of Organic Components for Aerosol Mass Spectrometer Data, ES&T (submitted).

Modelers are using the OA component data derived from AMS field data.

9

6

Sulfate

8

6

(a) LV-OOA (low volatility, oxygenated OA) Sulfate 18

12

H/C=1.29; O/C=0.63; N/C=0.011OM/OC=1.96

CxHy+ CxHyO1

+ CxHyO2

+

+ + +

LV-OOAOA Components – Urban Case

10

6

3

0

e (µg/m3)

4

3

Nitra

0.2

Chlo

4

2

0(b) SV-OOA (semivolatile, oxygated OA)

m3 ) Nitrate

Chloride10

8

6

0H/C=1.40; O/C=0.38; N/C=0.004OM/OC=1.63

HyO1 +CxHyNp

+ CxHyOzNp

+

SV-OOA

5

0

6

3

2

1

ate (µg/m3)

C3

0.1

0.0

oride (µg/m3)

mpo

nent

s (µ

g/m

(c) NOA (nitrogen-enriched OA) C3H8N+ 12

6420

H/C=1.50; O/C=0.37; N/C=0.053OM/OC 1 69na

l

NOASeparation of different types of POA and SOA

9

6

4

2

0

0.2

0.1

0.0

3 H8 N

+ (µg/m3)

tion

of O

A c

om

( ) ( g )

+

9630

OM/OC=1.69

% o

f Tot

al S

ig

Op ypLow Volatility Oxygenated (LV-OOA) ~ Aged, regional SOA

Semi-volatile Oxygenated (SV-OOA) ~ Fresher semi-volatile SOA9

6

3

0.04

0.02

C6 H

10 O+(µg/m

3(d) COA (cooking-emission related OA)

Con

cent

rat

C6H10O+ 8

6420

H/C=1.58; O/C=0.18; N/C=0.004OM/OC=1.38

COASemi-volatile Oxygenated (SV-OOA) ~ Fresher, semi-volatile SOA

Nitrogen-enriched (NOA) ~ Amine type, likely SOA

H d b lik (HOA) T ffi b i POA09

6

3

0.00

3)

150

100

50

NO

x (ppbv

6

4

2

EC

(µg/m3

(e) HOA (hydrocarbon-like OA) ECNOx

010

8642

H/C=1.83; O/C=0.06; N/C=0.005OM/OC=1.24

HOAHydrocarbon-like (HOA) ~ Traffic, combustion POA

Cooking related (COA) ~ POA from cooking emissions

7/14 7/16 7/18 7/20 7/22 7/24 7/26 7/28 7/30 8/1 8/3

Date & Time (EST)

0 0

v)

0

3)

20092009120110100908070605040302010

m/z (amu)

20

Sun, Zhang et al., ACPD, 2010

OA Components – Rural / Remote Case

Chebogue Pt.Chebogue Pt.

2004 ICARTT

Nova Scotiabiogenically influenced

anthropogenically influenced.2004 ICARTT anthropogenically influenced.Zhang et al., in prep.

Separation of biogenic and anthropogenic SOASeparation of biogenic and anthropogenic SOA

CARES may be a good study case

Products:Ti i f OA (OA OA OA )

What Do We Get?

• Time series of OA comp. conc. (OAi; OAi ≈ OA mass)ACSM: 15 - 30 min.ToF-AMS: 2-5 min (surface) 15-30 s (airborne)

• Mass spectra of OA comp. distinct chemical properties, e.g., O/C, N/C, H/C, OM/OC

ToF AMS: 2 5 min (surface), 15 30 s (airborne)

g , , , ,• Typical OA comp.: HOA, OOA (SV-, LV-), BBOA, COA,

NOA...AAccuracy: • Propagated uncertainty for AMS data: < 30%, mainly due

to the collection efficiency (CE) correctionto the collection efficiency (CE) correction.• Uncertainties in OA components:

Can be evaluated: e.g., running the PMF algorithm from g , g g fdifferent random starting points; bootstrapping analysis (Ulbrich et al., 2009, ACP).

• Addressing the lifecycle processes of atmospheric OA

Applications of the OA Component VAP• Addressing the lifecycle processes of atmospheric OA.

• Temporal variations• Correlations with tracer compounds• Correlations with tracer compounds • size distributions…, integrated analysis

• Add i th di ti ti f OA• Addressing the radiative properties of OA. • Surrogates (e.g., O/C) for hygroscopicity and/or

volatilityvolatility.• Correlations and intercomparison with aerosol optical

and radiative measurementsand radiative measurements. • Validation and evaluations of models• Model development• Model development

1008404UrbanU b D i d

Global Distribution of SOA and POA840-4Urban DownwindRemote

12.3Pinnacle Park, NY

11.6

New York(Winter)

12.2

City, NY (Summer)

5.2

Manches(Winter)

14.3

ter, UK (Summer) 3.0

Edinburgh, UK16.3

TaunusGermany

4.2

MainzGermany 25.5

(W

in

5.3Chelmsford, UK

2.0

HyytialaFinland

Zurich,nter)

9.6

Switzerland

(Summ

14.7

burg

h, P

Adm

er)

2.3 JungfraujochSw

itzerland

7.0

Vanc

ouve

rC

anad

aPi

ttsh

16 2

4 4

2.1

Stor

m P

eak,

CO

V

80.0

Beijing

China

16.2 Tokyo,(W

inter)

13.2

Japa 11.07.910.72.98.59.52.826.812.819.1A

4.4

Bou

lder

, CO

an (Summ

er)

FukueJapan

OkinawaJapan

Cheju IslandKorea

ChebogueCanada

Off Coast New England, US

Thompson Farm, NH

Duke ForestNC

MexicoCity

Houston, TX

Riv

ersi

de, C

A

Jimenez et al., 2009

Global Distribution of SOA and POA

16

12

42%

Avg. OA = 7.3 µg/m3

18% 82%

Avg. OA = 4.9 µg/m3

10% 90%

Avg. OA = 2.7 µg/m3

Oxygenated OA

SemivolatileLow volatility

8

4

0

58%

nc. (g

m-3

)

Hydrocarbon-like OA16

12

8

4

Urban

Urban

Rural/RemoteDownwind

Mas

s C

on Hydrocarbon-like OA

Other, BB Fuel combustion

4

0

eijin

g, C

NTo

kyo,

JP

Sum

mer

)To

kyo,

JP

(Win

ter)

burg

h, P

A

exic

o C

ity

rsid

e, C

A

usto

n, T

X

ulde

r, C

Ok

City

, NY

mer

)k

City

, NY

nter

)ve

r, C

AN

urgh

, UK

hest

er,U

Km

mer

)es

ter,

UK

inte

r)M

ainz

, DE

uric

h, C

HS

umm

er)

uric

h, C

H(W

inte

r)st

NE

, US

rch

1, U

K

rch

2, U

K

unus

, DE

ue IS

, JP

wa

IS, J

P

S, K

orea

Pea

k, C

O

ores

t, N

C

Farm

, NH

PS

P, N

Y

otia

, CA

N

ytia

la, F

N

ujoc

h, C

H

OOA (l l SOA) d i OOA % i i h di f b

Be T (S T

Pitt

sb Me

Riv

er

Hou Bou

New

Yor

k(S

umN

ew Y

ork

(Win

Van

couv

Edi

nbM

anch

(Sum

Man

che

(Wi M Zu (S Zu (

Off

Coa

s

Tor

Tor

Tau

Fuku

Oki

naw

Che

ju I

Sto

rm P

Duk

e Fo

Thom

pson

F

Nov

a S

co Hyy

Jung

frau

• OOA (largely SOA) dominance, OOA % increase with distance from urban• HOA (largely urban POA) less significant at rural/remote, due to dilution

Zhang et al., 2009

Factor Analysis of AMS Organic Mass Spectramsmeasured = ca· msa + cb· msb + cc· msc + …

3.0

m-3

)

Ammonium 4 8 ug/m3

t = # of time stepsi = # of m/z p = # of comp.

ORGt×i =  Ct×p × MSp×i + Et×i=  tsa × msa + tsb × msb + tsc × msc + … 

a a b b c c

43 442.5

2 0entra

tion

(µg

m Ammonium 4.8 ug/m3Nitrate 5.8Sulphate 9.4Organics 13.4Chl id 0 15

m/z (i) Component a Component cComponent b

SO+

SO2+

2.0

1.5

t Mas

s C

once Chloride 0.15

Mexico City 2/2002xtsa msa xtsc mscxtsb msb

Multivariate SO+

S+

SO3+

HSO3+

H2SO4+

57

1.0

0.5ate

Equi

vale

n

Time  (t)

. . . + +  …+

Multivariate Analysis

0.0

Nitr

a

1412010080604020

.

m/z (Daltons)t ~ 1,000si ~ 300