Oxidation Chemistry and Evolving Volatility of Organic Aerosol

Transcription

1 Oxidation Chemistry and Evolving Volatility of Organic Aerosol Neil M. Donahue Carnegie Mellon University Center for Atmospheric Particle Studies US - Nordic Workshop on Secondary Organic Aerosols 2 Jul 27 Hyytiällä, Finland 1

2 The Center for Atmospheric Particle Studies 2

3 Organic Aerosol Primary Emission POA SOA fixed yield Precursor Volatile Products Vapor Pressure Not quite state of the art, common in models... 3

4 Organic Aerosol Primary Emission SOA Precursor α 1 POA p 1 p 2 semivolatile products (Odum 2-product model) α 2 Volatile Products Vapor Pressure State of the art, in some models... 4

5 154 M. Kanakidou et Secondary - Primary Split in Models pounds re Many gas the gas ph have suffic themselve pounds ar when resi Aerosol (S that organ the aeroso oxidation organic co categories the atmosp into the ae istry. Sinc ticular sou the POA phase org and subseq 5 droplets ev Most of the world mostly POA [Kanakidou et al. ACP 25]

6 ng et al.: Hydrocarbon-like Aerosol and oxygenated Mass organicspectrometer aerosols Data 43 = C 3 H 7 = 14 = CH 2 57 = C 4 H 9 O:C ~ :1 29 = C 2 H 5 71 = C 5 H = CO (watch out for N 2...) 18 = H 2 O (% = %44) 44 = CO 2 O:C ~ 1:1 43 = CH 3 CO Mass spectra of (a) HOA and (b) OOA, colored with the estimated contribution of each element (C, H, and O) to the mass o gment. The elemental compositions of each m/z in HOA and OOA are estimated according to Table 2. Ambient organic aerosol in AMS resolve into 2 factors (these from Pittsburgh). HOA looks like diesel and has little oxygen. OOA looks highly oxidized. [Qi Zhang et al. ACP 25] aratna et al., 24); 2) the OOA spectrum is domby m/z 44 (CO + 2 ) and m/z 28 (CO+ ) and demonclose similarity in the overall pattern with those of 6

7 AMS Global Observations Most of the world 33-67% OA [Qi Zhang et al. GRL 27] 7

8 AMS OOA Cities mixed, more than 5% OA Remote sites almost all OOA [Qi Zhang et al. GRL 27] OK, so what is OOA??... HOA is convincingly POA, so... 8

9 Organic Emissions Processing Precursor Generation # Product Product Product Vapor Pressure 9

10 Continued Processing Precursor Generation # t~1 day? Product Product Product Product Product Product Product t~1 day Vapor Pressure 1

11 How Far Does it Go?? Precursor Generation # t~1 day? Product Product Product Product Product Product Product t~1 day mean particle age ~ 7 days? Vapor Pressure 11

12 Production of a Semi-Volatile Compound quantity produced Precursor α 1 O 3 Product 1 vapor condensed C* saturation quantity reacted 12

13 Semi-Volatile Mass Fraction mass fraction 1 vapor.5 condensed C* saturation quantity reacted 13

14 Partitioning of Single Component 1.9.8! (Aerosol Fraction) C * = 1. µg m! C OA (µg m!3 ) ξ i = C i C OA 14

15 Partitioning of Single Component (log X Axis) 1.9.8! (Aerosol Fraction) C * = 1. µg m!3!3!2! log 1 C OA (µg m!3 ) ξ i = C i C OA 15

16 Partitioning at Specified C OA in Solution 1.9! (Aerosol Fraction) C OA = 1. µg m!3.2.1!3!2! log 1 C * (µg m!3 ) ξ i = C i C OA 16

17 The Volatility Basis Set 1.9! (Aerosol Fraction) C OA = 1. µg m!3.2.1!3!2! log 1 C * (µg m!3 ) C i = {.1,.1, 1, 1, 1, 1, 1 4, 1 5, 1 6} µg m 3 17

18 α-pinene + Ozone!"*+/ *)6183*9288*:12;<4=. "%$ "%#& "%# "%!& "%! "%"& >14??4= et al.*@!aaab ';C61 et al.*@#""!b D168< et al.*@#""&b!e*fg=6*#""& #H*FG=6*#""&!$*FG3I*#""& ##*FG3I*#""& 6J<12K32<67*#!K17G;<*?4< L2848!86<*?4< Remote Urban (b) "!"!#!"!!!" "!"!!" #!" $ ' () *+!,*-!$. 2x SOA under remote atmospheric conditions vs. extrapolation. [Presto and Donahue, ES&T, 26] 18

19 α-pinene and the Basis Set "%$ α 6!"*+/ *)6183*9288*:12;<4=. "%#& "%# "%!& "%! "%"& I α 5 α 4 α 3 "!"!#!"!!!" "!"!!" #!" $ ' () *+!,*-!$. α 1 α 2 (mass yields α ) α i = {.4,,.5,.9,.12,.18,...} 19

20 α-pinene + Ozone Mass Balance!"+, *832:25+;4::+<34=>62?!'%!'$!'#! "'& "'% "'$ "'# 1 ppt 1-7 torr C 1 O 4 = 1.4 O OOH O O 3 + C 1 =1 1 atm 76 torr "!"!#!" "!" #!" $!" %!" & ( )* +,!-+.!/ Mass balance for nominal product demands ξ max = i α i

21 α-pinene + Ozone Product Distribution Mass Fraction ^4 1^5 1^6 C * (µg m!3 ) Products distributed over volatility space (a transformation vector) Note very small yield of nucleator, consistent with [Burkholder et al. 26] Multiply yields by mass of α-pinene consumed to get product masses. [Donahue et al. in prep] 21

22 Basis-set 11: Basis Basics Mass Fraction µg a-pinene oxidized µg aerosol ~ 5% µg 3 µg.225 µg 2.25 µg µg µg.1 µg.5.1 µg ^4 1^5 1^6 C * (µg m!3 ) Organic Mass (µg m!3 ) a) Low!!pinene (26 µ g m!3 ) C OA = 1. µg m! ^4 1^5 1^6 C * (µg m!3 ) Oxidize some amt. of precursor, say 25 µg m 3, and distribute products. Start adding from left and see which bin is roughly saturated. Partition that bin 5:5, others accordingly. Add salt to taste. Adjust accordingly. [Donahue et al. in prep] 22

23 Basis-set 11: Pop Quiz!! Mass Fraction ^4 1^5 1^6 C * (µg m!3 ) Oxidize 5 µg m 3, of α-pinene and calculate partitioning. 23

24 Basis-set 11: Answer Mass Fraction ^4 1^5 1^6 C * (µg m!3 ) Organic Mass (µg m!3 ) b) High!!pinene (468 µ g m!3 ) C OA = 1. µg m! ^4 1^5 1^6 C * (µg m!3 ) 24

25 α-pinene + Ozone Partitioning Organic Mass (µg m!3 ) a) Low!!pinene (26 µ g m!3 ) C OA = 1. µg m!3 Organic Mass (µg m!3 ) 2 b) High!!pinene (468 µ g m!3 ) C = 1. µg m!3 OA ^4 1^5 1^6 C * (µg m!3 ) ^4 1^5 1^6 C * (µg m!3 ) Partitioning changes with mass loading: x18 total loading = x1 C OA. Most of the OA compounds at 1 µg m 3 are not in the particles at 1. [Donahue et al. in prep] 25

26 α-pinene + Ozone: UV (no NO x ) α i!" (Normali5e7 Aerosol Mass Fraction) Presto et al. [25] 2 June 25 low!nox fit UV fit 1!2 1! C OA (µg m!3 ) α i = {.4,,.5,.9,.12,.18,...} UV (UV) = {.4,,.2,.8,.12,.18,...} (d) [Presto et al., ES&T, 25a] 26

27 α-pinene + Ozone: VOC:NO x!" (Normali5e7 Aerosol Mass Fraction) # $.> # $.1 low!nox fit (c) 1!2 1! C OA (µg m!3 ) α i (HO 2) = {.4,,.5,.9,.12,.18,...} NO x β = (VOC : NO x ) /1, (VOC : NO x ) < 1 α i (NO x) = {,,,,.15,.2,...} [Presto et al., ES&T, 25b] 27

28 RO 2 Fate terpene Ox NO RO2 HO2 ROONO β ROOH (+ NO2) hv (+ OH) RONO2 RO Bottom line tie SOA module into gas-phase RO 2 chemistry: {α} = β {α} low NOx + (1 β) {α} high NO x 28

29 α-pinene + Ozone T Dependence Low ( ppb) Medium ( ppb) High (33-5ppb) Saathoff et al. [24] (17 ppb) SOA Mass Fraction Temperature ( o C) Upturn in SOA between 15 and o C. [Pathak et al., JGR, 27], [Stanier et al., ACPD, 27] 29

30 α-pinene + Ozone Total Mass 3 K!" (Normalized Aerosol Mass Fraction) ! C OA (µg m!3 ) Dark green yields are guesses total is constrained. 3

31 α-pinene + Ozone Total Mass 243 K!" (Normalized Aerosol Mass Fraction) ! C OA (µg m!3 ) Products shift left by 2.5 orders of magnitude. (Note [Saathoff et al. IAC 26] saw 1 AMF at 1-2 µg m 3 and 243 K.) 31

32 α-pinene + Ozone Total Mass 35 K!" (Normalized Aerosol Mass Fraction) ! C OA (µg m!3 ) Products shift right by 2.5 orders of magnitude. 32

33 α-pinene + Ozone Thermodenuder Given time, all SOA evaporates at 7C. [An et al., Aerosol Sci., 27] 33

34 α-pinene + Ozone Thermodenuder!"+, *832:25+;4::+<34=>62?!'%!'$!'#! "'& "'% "'$ "'# transfer function (1.6 sec) transfer function (16 sec) "!"!#!" "!" #!" $!" %!" & ( )* +,!-+.!/ Denuding is a function of evaporation timescales. [Pierce et al., in prep, 27] 34

35 α-pinene + O 3 Dilution.4 Dilution Sampler Data Normalized AMF !!Yield Experiment Curve Chamber Dilution Experiments: Experiment # 1: DR = 3, then 8; Total Time: 1 hr Experiment # 2: DR = 8; Total Time: 4.8 hrs Experiment # 3: DR = 31; Total Time: 3.7 hrs ""#! 1 1 C OA ( g m -3 ) Generate ""*! $%&'()!*+!,-./(%1-2!-3!%2!&14/(5%67)!/(5%5%-2%2&!.)1'()8!8'(%2&!8%7'5%-2!-3!!4/%2)2)! high SOA and then flush 9% of chamber air. Particles shrink slowly to expected size. [Grieshop et al., GRL 27] ""D! 9:;!<1=.>-71?!@%5A!!3%5!-3!=%)78!85!3(-.!5(8%5%-27!9:;!)B/)(%.)251!<1-7%8!7%2)?C!! 35

36 Dilution of Diesel Emissions Normalized Organic Aerosol Emission Factor (a) 1, 1, 1 1 Ambient Conditions Dilution Ratio Measured Fit of data 95% PI C OA (!g m -3 ) SVOC IVOC Dilution to smbient C OA causes 67-9% evaporation of primary emissions. (b) Volatility Distribution sion Factor.8.6 Fit of Data (" bars = 1) Inferred (" bars = 1.5) (g kg-fuel -1 ) Organic Aerosol Emission Factor (g kg-fuel -1 ) [Shrivastava, Robinson, et al., ES&T [26] 36

37 1 Basis Set Representation of 2 1 Diesel 3 Emissions C OA (!g m -3 ) Normalized Emission Factor (b) Volatility Distribution Fit of Data (" bars = 1) Inferred (" bars = 1.5) SVOC IVOC 2 1 Emission Factor (g kg-fuel -1 ) C* (!g m -3 ) Emissions span volatility range, including IVOC. [Robinson, et al., Science [27] 37

38 Basis-set 71: Advanced Basis Organic Mass (µg m!3 ) b) High!!pinene (468 µ g m!3 ) C = 1. µg m!3 OA Organic Mass (µg m!3 ) 2 e) High!!pinene (468 µ g m!3 ) with wall deposition 18 C 16 OA = 1. µg m! ^4 1^5 1^6 C * (µg m!3 ) ^4 1^5 1^6 C * (µg m!3 ) Consider static chamber experiments i.e. 12 m 3 CMU chamber. Generate 1 µg m 3 SOA and then let 9% deposit to the walls (purple). What is in equilibrium? Vapors + suspended aerosol or Vapors + suspended aerosol + wall aerosol?? (work it out...) [Donahue et al. in prep] 38

39 Basis-set 71: Wall Deposition Organic Mass (µg m!3 ) f) High!!pinene (468 µ g m!3 ) after wall deposition C OA = 1. µg m!3 Organic Mass (µg m!3 ) 5 c) Medium!!pinene (94 µ g m!3 ) 45 4 C OA = 1. µg m! ^4 1^5 1^6 C * (µg m!3 ) ^4 1^5 1^6 C * (µg m!3 ) The answer is BOTH! The suspended particles still represent the composition of 1 µg m 3 SOA! Deposition fractionates the total suspension toward the volatile products. 39

40 Basis-set 72: Steady-State Chambers 5 45 c) Medium!!pinene (94 µ g m!3 ) Organic Mass (µg m!3 ) C OA = 1. µg m! ^4 1^5 1^6 C * (µg m!3 ) Now what happens in a steady-state flow-through chamber? Imagine an experiment with 1 µg m 3 suspended SOA that lasts 9 deposition lifetimes, so that C wall = 9 C suspended. What happens to the distribution as particles collect irreversibly on the walls?? 4

41 Basis-set 72: Flow-Through Wall Deposition Organic Mass (µg m!3 ) 5 c) Medium!!pinene (94 µ g m!3 ) 45 4 C OA = 1. µg m! Organic Mass (µg m!3 ) 5 d) Medium!!pinene (94 µ g m!3 ) after wall deposition 45 4 C OA = 1.2 µg m! ^4 1^5 1^6 C * (µg m!3 ) ^4 1^5 1^6 C * (µg m!3 ) The suspended particles still represent the composition of 1 µg m 3 SOA! However, the walls are covered with lower-volatility junk, total system at 1 µg m 3 Perturbing this system could have all kinds of strange consequences (it is buffered). Bottom Line: Individual particles only know about the vapors. Be very, very careful designing (and interpreting) experiments. 41

42 CONGRATULATIONS! You are now a basis-set black belt. 42

43 Fraction Difference of total mass (%) 3!2!2 Does Chamber 5 SOA = 1 OOA? 15 3 M/z (amu) 3 43 C B- B ! Difference!2! M/z (amu) 1!1 And now for the start of a lot of bad news... Chambers don t make OOA AMS spectrum from limonene + O 3 is not OOA. [J. Zhang et al. JPhysChem. 26] α-pinene + O 3 just as bad in fact no volatile precursors make OOA in chambers. 43

44 Where Are We? The existing picture nonvolatile POA and semi-volatile SOA is wrong. SOA (as OOA) has relatively low volatility. POA (as HOA) is quite volatile. At the same time, chamber SOA is not oxidized enough in the AMS. The hypothesized solution is chemical aging: Most mass, even low volatility emissions, is in vapor phase. Gas-phase OH rate constant : τ 6 hr. 44

45 Implications: Vapors!"+, *832:25+;4::+<34=>62?!'%!'$!'#! "'& "'% "'$ "'# " Vapor Phase Condensed Phase!"!#!" "!" #!" $!" %!" & ( )* +,!-+.!/ The mass not seen in the particles is in the gas phase, very low vapor pressure. 45

46 Limonene + Ozone Mass Balance!"+, *832:25+;4::+<34=>62?!'%!'$!'#! "'& "'% "'$ C +,943D E:>6.4>89+F434.8>83: <6>+F434.8>83: G38=H3:23+I254>656>J <H55+K4:6:!:8>+=H3I8 #!F6?8?8+K4:6:!:8>+L6> C 9 O 7 = 1.6 O HOO OOH O O O 3 + C 1 =1 "!"!#!" "!" #!" $!" %!" & ( )* +,!-+.!/ D-limonene + O 3 makes more SOA than α-pinene (2nd generation of oxidation). [J. Zhang et al., JPC, 26] 46

47 Implementing Basis Set in PMCAMx July 21 Biogenic SOA Old constant yield New Basis set yields Multi-generation aging New basis set parameters cause most SOA to evaporate at ambient COA. Adding aging (gas-phase OH oxidation) can generate lots of SOA. Aging parameters are not yet known (these are a reasonable guess). [Lane et al. in prep] 47

48 Photooxidation of Diesel Emissions Oxidized fraction looks a lot like OOA. 48

49 Implementation in PMCAMx (A) Traditional Model (B) Semivolatile POA (C) Semivolatile POA + Aging (D) SVOC + IVOC + Aging 1 OA (µg m -3 ) Oxidation of vapors produces regional organic aerosol. Robinson et al. Science [27] 49

50 SOA in PMCAMx (A) Fraction SOA Traditional (B) Fraction SOA Revised 1 (C) Revised / Traditional OA Ratio The aerosol is now 9% SOA in the region and 6% SOA in cities, just like AMS Urban / Regional OA Ratio (D) Traditional Revised Balt NYC Pgh Phil Robinson et al. Science [27] 5

51 Conclusions Picture log Emissions Fresh POA condensed at ambient COA POA condensed at source anthropogenic biogenic log Concentration HOA Aged Het Aging 1:1 Traditional SOA 1:1 New SOA :1 OOA OOA Decomposition : C*(µg m 3 )