Chapter 2. Very Short-Lived Halogen and Sulfur Substances1

2Lead Authors: Malcolm Ko (USA) and Gilles Poulet (France)3

4Co-authors5Donald Blake (USA)6Olivier Boucher (France)7James Burkholder (USA)8Mian Chin (USA)9Tony Cox (UK)10Christian George (France)11Hans Graf (Germany)12James Holton (USA)13Daniel Jacob (USA)14Kathy Law (UK)15Mark Lawrence (Germany)16Pauline Midgley (Germany)17Paul Seakins (UK)18Dudley Shallcross (UK)19Susan Strahan (USA)20Donald Wuebbles (USA)21Yoko Yokouchi (Japan)22

23

Contributors:24

Nicola Blake (USA)25James Butler (USA)26Anne Douglass (USA)27Victor Dvortsov (USA)28Ian Folkins (Canada)29Peter Haynes (UK)30Abdelwahid Mellouki (France)31Michael Prather (USA)32Jose Rodriguez (USA)33Susan Schauffler (USA)34Theodore Shepherd (Canada)35Christiane Textor (Germany)36Claudia Timmreck (Germany)37Debra Weisenstein (USA)38

39

Lead Authors:

Malcolm Ko Atmospheric and Environmental Research Inc. USAGilles Poulet Centre Nat.de la Recherche Scientifique France

Co-authors

Donald Blake University of California, Irvine USAOlivier Boucher Centre National de la Recherche Scientifique /

Universite' des Sciences et Technologies de LilleFrance

James Burkholder NOAA Aeronomy Laboratory USAMian Chin Georgia Institute of Technology USATony Cox University of Cambridge UKChristian George Laboratoire d'Application de la Chimie à

l'Environnement (LACE)France

Hans Graf Max-Planck-Institute for Meteorology GermanyJames Holton University of Washington USADaniel Jacob Harvard University USAKathy Law University of Cambridge UKMark Lawrence Max Planck Institute for Chemistry GermanyPauline Midgley M&D Consulting GermanyPaul Seakins University of Leeds UKDudley Shallcross University of Bristol UKSusan Strahan NASA Goddard Space Flight Center USADonald Wuebbles University of Illinois USAYoko Yokouchi National Institute for Environmental Studies Japan

Contributors:

Nicola Blake University of New Hampshire USAJames Butler NOAA CMDL USAAnne Douglass NASA Goddard Space Flight Center USAVictor Dvortsov ?? USAIan Folkins Dalhousie University CanadaPeter Haynes University of Cambridge UKAbdelwahid Mellouki LCSR CNRS FranceMichael Prather University of California, Irvine USAJose Rodriguez University of Miami USASusan Schauffler Natl Ctr Atmospheric Research USATheodore Shepherd University of Toronto CanadaChristiane Textor Max-Planck-Institut für Meteorologie GermanyClaudia Timmreck Max Planck Institute for Meteorology GermanyDebra Weisenstein Atmospheric and Environmental Research Inc. USA

VSL very short-livedBL boundary layerUT upper troposphereFT free troposphereLS lower stratosphereODP ozone depletion potentialSGI source gas injectionPGI product gas injectionTTL tropical tropopause layerSTT secondary tropical tropopauseTTT thermal tropical tropopauseSTE stratosphere troposphere exchangeEESC equivalent effective stratospheric chlorineCLP chlorine loading potentialSTREAM stratosphere troposphere experiment by aircraft measurementMOZAIC measurement of ozone and water vapor by Airbus in-service aircraftSTRAT Stratospheric Tracers of Atmospheric TransportCARIBIC Civil Aircraft for Regular Investigation of the atmosphere Based on anInstrument ContainerPEM Pacific Exploratory MissionTrace-P Transport and Atmospheric Chemistry near the Equator-Pacific

1

2Chapter Summary3

42.1 INTRODUCTION5

62.2. DYNAMICAL AND CHEMICAL CHARACTERISTICS OF THE UPPER7

TROPOSPHERE AND THE TROPICAL TROPOPAUSE LAYER82.2.1. Moist Convection And The Tropical Tropopause Layer92.2.2 Stratosphere-Troposphere Exchange10

2.2.2.1 Exchange across the tropical tropopause112.2.2.2 Exhange across the extra-tropical tropopause12

2.2.3 Chemistry in the UT and TTL132.2.4 Chemical Tracer Observations for Model Evaluation14

152.3 ATMOSPHERIC CHEMISTRY OF HALOGENATED VERY SHORT LIVED16(VSL) SUBSTANCES17

2.3.1 Removal Of The Halogen Source Gases182.3.2 Production And Removal Of Degradation Products19

2.3.2.1 Degradation mechanisms for source gases202.3.2.2 Intermediate products212.3.2.3 Inorganic bromine and iodine species22

2.3.3 Degradation Mechanism For n-Propyl Bromide232.3.4 Stratospheric Iodine Chemistry24

252.4 CONTRIBUTION OF HALOGENATED VSL SUBSTANCES TO THE STRATOSPHERIC26

INORGANIC HALOGEN BUDGET272.4.1 Observed Concentrations And Sources Of Organic Chlorine, Bromine And28

Iodine Compounds292.4.1.1 Chlorine Compounds302.4.1.2 Bromine compounds312.4.1.3 Iodine compounds32

2.4.2 Estimates Of Contributions From VSL Halogen Source Gases332.4.3 Modeling Studies342.4.4 Information Needed For Future Evaluation35

362.5 ESTIMATES FOR THE IMPACT OF HALOGENATED VSL SUBSTANCES ON37

COLUMN OZONE382.5.1 Review Of Methodolgy For Calculating ODP For Long-Lived Source Gases392.5.2 Modified Approach For estimating ODP for Halogenated VSL Substances40

2.5.2.1 Methods for estimating inorganic halogen loading412.5.2.2Estimates for effects on stratospheric ozone422.5.2.3 Effects on ozone in the free troposphere43

2.5.3 Existing Modeling Studies On ODP For n-PB442.5.4 Information Required For Evaluating Ozone Impact From VSL Source45

Gases46

12.6 VSL SULFUROUS SPECIES AND STRATOSPHERIC AEROSOL2

2.6.1 Introduction32.6.2 Chemistry Of The VSL Sulphurous Species42.6.3 Sources For The Background (Non-Volcanic) Stratospheric Sulfate Layer5

2.6.3.1 Direct input from explosive volcanic eruptions62.6.3.2 VSL precursors emitted in the troposphere7

2.6.4 How Can We Predict Future Changes In The Stratospheric Sulfate Layer?89

10

p. 1

Chapter Summary1

2

Definition of Very-short Lived Substances3

4

• Very short-lived (VSL) substances have chemical lifetimes comparable to tropospheric5

transport timescales, with the result that the steady state mixing ratio of the substance in6

the troposphere depends on where and when (time of the year) it is released. In practice,7

this happens for species with atmospheric lifetimes of a few months or less.8

9

10

Ozone Depletion Potential (ODP) for Halogen Source Gases11

12

• The ozone depletion potential of a halogen source gas X ( ODP X x te e( , , )r

) is defined as13

the time integrated effect on stratospheric ozone caused by emission of X at location rxe14

and time te relative to the time integrated effect on stratospheric ozone from the same15

mass emission of CFC-11 at the same location and time.16

17

• In the general case, ODP X x te e( , , )r

consists of two terms: ODP X x tSGI e e( , , )r

and18

ODP X x tPGI e e( , , )r

. ODP X x tSGI e e( , , )r

(Source Gas Injection term) and ODP X x tPGI e e( , , )r

19

(Product Gas Injection term) represent the effects caused by inorganic halogen atoms that20

are transported to the stratosphere in the form of the source gas and in the form of21

degradation products respectively.22

23

• For long-lived source gases, the value for ODP X x tSGI e e( , , )r

is independent of rxe and24

te , and ODP X x tPGI e e( , , )r

is negligible in comparison to ODP X x tSGI e e( , , )r

. Thus, a single25

value of ODP can be obtained based on the atmospheric lifetime of the source gas26

independent of location and time of emission. These are the traditional single value ODP27

reported in the literature.28

29

p. 2

• For VSL source gases, the ODP X x tSGI e e( , , )r

value depends on the location and time of1

emission. The ODP X x tPGI e e( , , )r

value depends also on the properties of the degradation2

products.3

4

• If degradation of the VSL source gases gives rise to significant inorganic halogen5

concentration in the upper troposphere (UT), this may lead to ozone depletion in this6

region. The question then arises if this depletion should be added to the ODP value,7

which traditionally refers only to depletion in the stratosphere. It also points to the need8

to evaluate the role of halogen chemistry in the UT.9

10

Transport from the Boundary Layer to the Stratosphere11

12

• The most efficient route for irreversible transport of VSL source gases and their13

degradation products from the surface to the stratosphere is in the tropics, because the14

vertical transport times from the surface to the upper troposphere are short, and air that15

enters the stratosphere through the tropical tropopause remains there for more than a year.16

17

• In the tropics, the transition from tropospheric to stratospheric air takes place in a18

"tropical tropopause layer" (TTL) between the secondary tropical tropopause (STT, 11-1319

km) and the thermal tropical tropopause (TTT, 16-17 km). Transport of VSL source20

gases and degradation products from the surface to the UT and the lower part of the TTL21

is primarily through convection. From the TTL, a small fraction of the air enters the22

stratosphere via slow ascent through the tropical tropopause or via isentropic transport23

into the extra-tropical lower stratosphere.24

25

• Current estimates indicate that air at the base of the TTL is replaced by convection from26

the tropical boundary layer on a time scale of 10-30 days. Thus, a significant fraction of27

the emitted VSL source gases can be expected to reach the TTL in regions of deep28

convection.29

30

p. 3

• In the extratropics, exchange across the tropopause maintains a ‘transition layer’ of 1-21

km with chemical characteristics intermediate between that of the UT and LS (lower2

stratosphere). Frontal lifting can transport air on time scales of a few days to the UT or3

even occasionally into the LS. Through such mechanisms, VSL source gases are4

expected to enter the ‘transition layer’.5

6

7

Chemistry for VSL Halogen8

9• Inorganic chlorine, bromine, and iodine have potential to destroy stratospheric ozone.10

For the same concentration in the stratosphere, iodine is most efficient in removing11

ozone, followed by bromine and chlorine. Laboratory data on iodine stratospheric12

chemistry has led to downward revision of the iodine efficiency in depleting ozone in the13

stratosphere. The estimated efficiency factor relative to chlorine (~ 150-300) is still14

higher than that of bromine (~ 45).15

16

17• The main uncertainties in estimating the impact of VSL halogen source gases on18

stratospheric ozone lie in the physical and dynamical processes transporting halogen into19

the stratosphere and the behavior of their intermediate degradation products; uncertainties20

in the gas phase chemistry of the source gases are minor.21

22

• The products expected from many of the bromine and iodine VSL source gases23

considered in this chapter are all rapidly converted to inorganic halogen. The inorganic24

halogen multiphase chemistry for bromine and iodine is not very well known.25

Heterogeneous processes which recycle reservoir species to reactive inorganic bromine26

and iodine may significantly increase the efficiency of halogen injection (via the PGI27

pathway) into the stratosphere.28

29

• The HOx-related chemistry in the upper troposphere is very critical for both the30

destruction/production of ozone and for the delivery of VSL source gases to the31

stratosphere. The understanding of this chemistry requires a better knowledge of the32

p. 4

sources and processes for the oxygenated organics (carbonyl compounds, peroxides,…)1

and for the NOx and NOy species.2

3

4

VSL Halogen Source gases in the present day atmosphere5

6

• The observed concentrations of VSL halogen source gases are typically a few ppt for7

bromine and iodine VSL source gases, and up to twenty ppt for VSL chlorine source8

gases. The majority of these gases are of natural origin.9

10

• Two model studies simulated the atmospheric distribution of bromoform (CHBr3)11

assuming a simplified uniform (space and time) ocean source. The calculated12

atmospheric distributions indicate that an average surface mixing of 1.5 ppt at the surface13

could maintain about 1 ppt of inorganic bromine in the stratosphere. The simulation14

shows that one half to three fourths of the bromine enter the stratosphere in the form of15

inorganic products. These results can be used as a guideline for estimating contributions16

from other source gases that have local photochemical lifetimes of about 10 days, no17

long-lived intermediate degradation product, and relatively uniform sources over the18

globe.19

20

• Very short-lived bromine and iodine source gases with surface concentrations a few ppt21

could have large effects on the current inorganic bromine and odd iodine budget since the22

stratospheric concentrations of inorganic bromine and iodine are about 20 ppt and less23

than 1 ppt respectively. The transfer of inorganic bromine from the UT may account for24

the current discrepancy in the stratospheric inorganic bromine budget (as discussed in this25

chapter and in chapter 4).26

27

28

Modeling Studies to determine values for ODP of n-PB29

30

p. 5

• n-Propyl Bromide (n-PB : CH3CH2CH2Br) is removed by reaction with OH with local1

photochemical lifetimes in the tropical troposphere of about 10 days. Laboratory data2

indicate that its degradation products, and particularly bromoacetone, have lifetimes3

shorter than a day.4

5

• Given the complexity of atmospheric processes, three-dimensional numerical models6

are the preferred tool to evaluate the ODP for VSL source gases. Results for n-PB from7

three separate model studies that computed ODP n PB x te e( , , )−r

were assessed. These8

results can be used as guideline for other gases with similar local photochemical lifetimes9

(~10 days) and degradation schemes.10

11

12

• Two of the three studies provided values only for ODP n PB x tSGI e e( , , )−r

. The third13

study computed contributions from ODP n PB x tPGI e e( , , )−r

as well. In the latter study, the14

values of ODP n PB x te e( , , )−r

are 0.1 for tropical emission and 0.03 for emissions15

restricted to Northern mid-latitudes, with about two thirds of the effect from16

ODP n PB x tPGI e e( , , )−r

.17

18

19

VSL Sulfur Species and Stratospheric Sulfate20

21

• A source of 0.06-0.15 Tg(S)/yr is needed to maintain the observed burden of the22

background stratospheric sulfate layer during periods with low volcanic activity. Current23

model estimates indicate that degradation of OCS in the stratosphere can account for 20-24

50% of the needed source, with the balance from transport of sulfur VSL gases25

(CH3SCH3, H2S, SO2) and sulfate formed in the troposphere.26

27

• Our current best estimates for the surface sources of sulfur are 10-30 Tg(S)/yr for DMS,28

1-2 Tg(S)/yr for OCS, 0.5-1.5 Tg(S)/yr for CS2, 6-9 Tg(S)/yr for H2S, and 67-10029

p. 6

Tg(S)/yr for SO2. Emission from aircraft engines provides a source of 0.06 Tg(S)/yr in1

the upper troposphere and lower stratosphere.2

3

• Model simulations show that a few tenths of one percent of the SO2 emitted at the4

ground reaches the stratosphere. This is consistent with estimates for VSL halogen5

source gases. For DMS, multiphase oxidation may occur and reduce its possible transfer6

to the UT.7

8

• Previous climate calculations assumed that the change in stratospheric sulfate loading is9

proportional to changes in total SO2 emission. Changes in the spatial distribution of the10

emission can have a large effect. Depending on how much of the sulfur gas is converted11

to small sulfate particles in the UT, the change in surface area per unit mass of sulfur12

transferred to the stratosphere could differ.13

1415

Model evaluations1617

• In order to evaluate the contribution of VSL source gases to halogen and sulfur loading,18

global 3-D model simulations are required. Simulating aerosol microphysics in the19

troposphere with very short time constants in global scale models presents a large20

challenge. Significant uncertainties exist in the treatment of dynamical and physical loss21

processes in such models.22

23

• Further evaluation of model performance against observations is necessary, particularly24

relating to rapid transport processes and treatments of cloud scavenging. Various tracers25

can be used to test specific components: Radon-222 and short-lived hydrocarbons for26

continental convection; methyl iodide, dimethylsulfide, and ozone for marine convection;27

lead and beryllium isotopes, and soluble gases for precipitation scavenging; and ozone28

for stratosphere-troposphere exchange.29

p. 7

2.1 INTRODUCTION12

An important component of previous Ozone Assessment Reports is the discussion3

of the observed concentrations, budgets, trends and atmospheric lifetimes of source gases4

that affect stratospheric ozone. These gases are mostly emitted at the surface. The5

source gas and its degradation products are transported to the stratosphere where they6

could affect the ozone abundance. Examples of these include: odd nitrogen species from7

nitrous oxide; chlorine, bromine and iodine radicals from organic halogen-containing8

chemicals; and methane reacting to produce water vapor in the stratosphere. Sulfur spe-9

cies found in the troposphere (e.g. carbonyle sulfide, OCS; carbon disulfide, CS2; hydro-10

gen sulfide, H2S; dimethyle sulfide, CH3SCH3; sulfur dioxide,SO2) provide the source of11

sulfur to maintain the stratospheric sulfate layer. The sulfate aerosol particles provide12

sites for heterogeneous reactions that control the partitioning of the nitrogen and halogen13

radicals. One aim of the Ozone Assessment Reports is to establish the tools that enable14

one to predict how changes in emissions of the various source gases would affect ozone.15

In the case of the halogen species, it is done in terms of the equivalent chlorine concen-16

tration estimated from emissions and lifetimes. A similar connection could be made for17

the sulfur source gases if one could predict how the stratospheric sulfate may change in18

response to changes in emissions. In this chapter, we restrict the discussion to halocar-19

bons and the sulfurous gases. Other species with similar lifetimes are affected by many20

of the same processes discussed in this chapter. One example is oxides of nitrogen re-21

leased at the ground (from combustion and biomass burning) and produced in the free22

troposphere (by lightning or aircraft engines). They are not included in this chapter.23

24

The purpose of this chapter is to discuss how very short-lived (VSL, to be defined25

below) substances released at the surface or produced in situ in the troposphere could af-26

fect ozone in the stratosphere. Because the subject is relatively new, this chapter will de-27

viate from the standard approach of providing a detailed review and assessment of the28

information in the published literature. The current state of our understanding does not29

provide concrete evaluations to assess the merit of the approaches and make recommen-30

dations for specific values. Instead, the chapter will focus on outlining viable approaches31

p. 8

with emphasis on how observations may be used in future evaluations. By making cer-1

tain simplifying assumptions, order of magnitude estimates for some of the parameters2

will be provided so that policy makers can start to consider how best to approach the3

problem. The emphasis will be on the issues needed to be resolved and the observations4

needed for evaluation.5

6

In the discussion, it is often convenient to distinguish among the different forms7

in which the halogen atoms and sulfur atoms are transported to the stratosphere. Halogen8

atoms bound in the source gases and the intermediate degradation products will be re-9

ferred to as organic halogen. The final products will be referred to as inorganic halogen10

which include, in the case of chlorine, the reservoir species (HCl, ClONO2, HOCl) and11

the radical species (Cl, ClO). For sulfur, an analogous approach would identify carbon-12

yle sulfide (OCS), carbon disulfide (CS2), hydrogen sulfide (H2S) and dimethyl sulfide13

(DMS: CH3SCH3) as source gases and sulfur dioxide (SO2) and various forms of sulfate14

(SO4) in gaseous, liquid, or solid form as the final product. As we shall see in section 2.6,15

this turns out not to be desirable because SO2 and SO4 are also released at the surface.16

We will refer to a sulfur species as sulfate or sulfate precursor in the discussion.17

18

A major goal in the study of halogen source gases has been to track the life cycle19

of a source gas so that one can evaluate how its emission may affect the inorganic halo-20

gen budget in the present-day stratosphere. This is accomplished by combining the21

amount of the source gas emitted at the surface with the percentage of the emitted halo-22

gen atoms that is delivered to the stratosphere. It is important to provide values for the23

latter as it allows one to define the relative ozone depletion potential of an individual24

source gas. Traditional methods for evaluating the contribution of organic halogen25

source gases with lifetimes of 1 year or longer have been reviewed and assessed in previ-26

ous Ozone Assessment Reports. However, such methods cannot be applied to VSL spe-27

cies. To understand why, we examine the processes that deliver the final products to the28

stratosphere. For ease of discussion, we will present the arguments in the context of the29

halogen species. It should be noted that many of the ideas are equally applicable to the30

sulfur species.31

p. 9

1

Figure 2-1 illustrates that there are two pathways for delivering the final products2

to the stratosphere. For the SGI (source gas injection) pathway, the source gas is trans-3

ported the stratosphere and then reacts to release the halogen atoms. For a given emis-4

sion location and time, its efficiency depends mostly on the properties of the source gas5

(essentially its removal rate by chemical or physical processes in the troposphere as com-6

pared to the transport rate from the ground to the stratosphere). The PGI (product gas7

injection) pathway involves transport of intermediate or final products produced in the8

troposphere. Its efficiency depends also on the properties of the degradation products.9

10

To put things in perspective, we provide some rough numerical estimates for the11

two pathways for long-lived source gases that have uniform mixing ratio in the tropo-12

sphere. Long-lived source gases that are inert in the troposphere (most chlorofluorocar-13

bons) are recycled between the troposphere and stratosphere until they are all removed14

via photochemical reactions in the stratosphere. Thus, 100% of the final product is deliv-15

ered to the stratosphere via the SGI pathway. For long-lived source gases that are re-16

moved efficiently in the troposphere (most hydrogenated chloroflurocarbons), at least17

90% of the mass burden is found in the troposphere at steady state. If we assume that the18

local lifetimes in the troposphere and stratosphere are similar, this would imply that 90%19

of the final product is released in the troposphere. Thus, only 10% is delivered to the20

stratosphere via the SGI pathway. If one assumes that than less than 1% of the degrada-21

tion products (in the form of intermediate degradation products or inorganic halogen) re-22

leased in the troposphere is transported to the stratosphere, then the supply from the PGI23

pathway (0.009 of emission) is a factor 10 smaller than the SGI pathway (0.1 of emis-24

sion). This is why the contribution from the PGI pathway is usually ignored in consider-25

ing the long-lived source gases.26

27

We now explain the criteria for classifying a substance as VSL as distinct from28

those discussed in Chapter 1 of this report. The established method for estimating inor-29

ganic halogen loading from the SGI pathway depends on the observed fact that the mix-30

ing ratio of the long-lived source gas is essentially uniform in the troposphere. [Note that31

p. 10

the key issue is the difference in mixing ratio between the boundary layer (BL) and the1

upper troposphere (UT). Small inter-hemispheric difference ( ~20%) in the BL will not2

affect the accuracy]. In these cases, the fraction of the emitted halogen atoms entering3

the stratosphere is independent of the location and time of emission. The inorganic halo-4

gen loading in the stratosphere can be estimated from the steady state burden of the5

source gas which is related to its atmospheric lifetime. Details of the transport processes6

in the troposphere play no role. As the local lifetime of the source gas gets shorter, its7

mixing ratio will no longer be uniform in the troposphere. The method used for long-8

lived species is no longer accurate when the mixing ratio in the UT is less than half of the9

mixing ratio in the BL. Survey of chemicals in the atmosphere indicates that this occurs10

when the local lifetime (inverse of the local removal rate) is 0.5 year or shorter. This will11

serves as the criteria for identifying VSL species.12

13

The fraction of the emitted halogen atoms that are transported to the stratosphere14

via the PGI pathway depends on the properties of the intermediate and final products as15

well. Consider first the situation where the intermediate products are so short-lived that,16

for practical purposes, the inorganic halogen species can be treated as being produced by17

the initial reaction experienced by the source gas. Many of the inorganic halogen species18

are water-soluble or will react in solution to form water-soluble products. They are re-19

moved efficiently in the lower troposphere with local residence time of around 10 days.20

In order for the PGI pathway to provide an effective source to the stratosphere, the prod-21

ucts should be produced at least as fast as the washout rate (Ko et al., 1997). Thus, the22

PGI pathway will be important if the local photochemical lifetime of the source gas is of23

the order of 10 days or shorter.24

25

The properties of the intermediate products could change the estimate in a signifi-26

cant way. If the intermediate products are long-lived in the troposphere (i.e. with photo-27

chemical lifetime longer than the source gas or the washout lifetime of the inorganic28

halogen), they could serve as a vehicle for transporting halogen atoms to the stratosphere29

and deliver much more of the degradation products via the PGI pathway. An example of30

an intermediate product having lifetime longer than the source gas is phosgene (COCl2)31

p. 11

formed from degradation of C2HCl3. At the same time, if the local photochemical life-1

time of the intermediate products in the stratosphere is longer than its residence time in2

the stratosphere, they may be removed from the stratosphere before they release their3

halogen atoms. This will effectively decrease the efficiency of the SGI pathway (see4

Kindler et al., 1995).5

6

Many of the same issues apply to the VSL sulfate precursor gases and how their7

fates in the troposphere would affect the stratospheric aerosol layer. In addition to source8

gases released at the surface, there is a direct way of injecting sulfur species into the9

stratosphere through explosive volcanic eruptions. In this chapter, we will briefly review10

the effects of volcanoes building on work from the previous Ozone Assessment Report11

(Godin et al., 1999). The major emphasis will be on the role of the VSL precursor gases.12

13

The structure of this chapter is as follows. Section 2.2 discusses the transport14

mechanisms in the troposphere and cross tropopause exchange, and the chemical envi-15

ronment in the upper troposphere. Section 2.3 describes the removal mechanisms of the16

VSL halogen source gases and their degradation products. It also includes a brief discus-17

sion of the ozone removal efficiency of the iodine radicals in the stratospheric. Section18

2.4 discusses how VSL halogen source gases may contribute to the inorganic halogen19

budget in the current stratosphere. The observations for these source gases are reviewed20

with emphasis on our understanding of the budgets. Section 2.5 focuses on the issue of21

evaluating the inorganic halogen loading estimates and ozone depletion estimates of VSL22

halogen source gases. In addition to natural species, the discussion includes industrial23

species and the information that is needed on emission inventories to estimate their im-24

pact. Finally, section 2.6 discusses the effects of sulfur-containing VSL gases and how25

they affect future trends in the stratospheric sulfate layer.26

27

p. 12

2.2. DYNAMICAL AND CHEMICAL CHARACTERISTICS OF THE1

UPPER TROPOSPHERE AND THE TROPICAL TROPOPAUSE2

LAYER3

4The critical parameters for understanding the processes that control fluxes of key5

trace gases from the troposphere to the stratosphere are the mixing ratios of the trace gases6

in the upper troposphere (UT) and in the tropical tropopause layer (TTL) and the transport7

rates across the tropopause. We employ the following definitions in this chapter: the UT is8

the region from about 8-12 km, which is capped by the tropopause in the extratropics, and9

which overlaps with the lowest part (~1 km) of the TTL in the tropics. The TTL is10

bounded above by the thermal tropical tropopause (TTT, defined here as the cold point of11

the sounding) at about 16-17 km, and bounded below by a secondary tropical tropopause12

(STT) at 11-13 km. The secondary tropical tropopause (STT) is related to large scale13

circulation in that it corresponds to the maximum Hadley-cell outflow (Highwood and14

Hoskins, 1998). At these levels, convection is no longer able to maintain a moist adiabatic15

temperature lapse rate owing to the rapid decreases with height of the cumulus mass flux,16

and IR radiative cooling. Within the TTL there is a gradual transition from tropospheric to17

stratospheric dynamical and chemical characteristics. However, the rapid decrease with18

height of the convective outflow, and the rapid transition from clear-sky radiative cooling19

to radiative heating, combine to make the transition at the TTT fairly sharp (see e.g.20

Folkins et al., 1999; Hartmann et al., 2001, Jensen et al., 2001; Thuburn and Craig, 2000).21

The TTL is a relatively new concept, and its nature and regional variations are still under22

investigation. The concentrations of trace gases in the UT and TTL are controlled by23

transport, physical and chemical removal processes in the troposphere and lower24

stratosphere, and the location and timing of emissions.25

26

We consider two broad classes of transport mechanisms, illustrated in Figure 2-2.27

The first is deep moist convection, which transports air rapidly from the boundary layer28

(BL) to the UT, with a fraction detraining well into the TTL and some directly into the29

stratosphere. The second is large-scale stratosphere-troposphere exchange (STE), which30

transports air from the TTL upward into the stratospheric overworld, and laterally into the31

p. 13

extratropical lower stratosphere. In addition, this section includes a discussion of1

extratropical STE, the chemical environment in the UT and TTL, and a survey of the2

database of observations that are needed to critically evaluate the results of models.3

4

The numerical simulation of very short-lived (VSL) substances is difficult because5

of the wide range of scales on which relevant processes occur, ranging from microphysical6

(uptake of gases in cloud droplets) to synoptic and global (advection). Global models are7

hindered by their coarse spatial and temporal resolution (which are typically at best of the8

order of 100 km in the horizontal, 0.5 km in the vertical in the free troposphere and 0.1 km9

in the BL, and 10 minutes in time), which results in the need for "parameterizations" of10

important sub-grid scale processes. Regional models are able to resolve small-scale11

processes better, but their limited spatial domain requires the specification of boundary12

conditions, which makes it difficult to extrapolate to global scale estimates. In one13

comparison of simulated results from global tropospheric chemistry transport models with14

O3 data from the Measurement of Ozone and Water Vapor by Airbus In-service Aircraft15

(MOZAIC) program, it was found that the models generally smeared out observed16

gradients, that they performed less well in regions where meteorological analyses, used as17

input, are expected to be less reliable, and that higher resolution seemed more important18

than an extensive hydrocarbon chemistry scheme for the quality of the simulated upper19

tropospheric O3 (Law et al., 2000).20

21

22

2.2.1. Moist Convection And The Tropical Tropopause Layer23

24

Moist convection is characterized by rapid updrafts (driven by latent heat release from25

condensing water) with sustained velocities of up to several m/s over regions of the order26

of a kilometer in diameter. The updrafts are balanced by downdrafts (largely driven by27

evaporating precipitation) with a similar spatial extent, and by large scale subsidence over a28

much larger area where clear sky radiative cooling occurs. In the UT (particularly the29

tropics), where moist convective adjustment holds, trace gases are strongly influenced by30

rapid upward transport within convective clouds, slow descent outside convectively active31

p. 14

regions, and nearly all air parcels have had recent contact with the tropical boundary layer.1

Convective clouds are often associated with the formation of precipitation (rain, snow, and2

graupel) and anvil cirrus clouds. This leads to efficient scavenging of soluble gases by3

precipitation. 4

5

Convective events are highly stochastic. An individual cloud can transport air parcels6

(thermals) from the BL to the UT in less than an hour. However, the actual concentration of7

the source gases and degradation products in the UT and TTL depends on how quickly they8

are evacuated from the BL, how often thermals reach and penetrate the UT and the TTL,9

and how much old air from the mid-troposphere (which has much smaller concentrations of10

the source gas) is entrained into the convective cores.11

12

An estimate for the mean evacuation time for the BL in the tropics can be obtained13

using relatively well-defined values for the Hadley cell circulation and the clear sky14

radiative cooling. The observed annual mean mass flux from the boundary layer in the15

Hadley circulation between 15° N and 15° S is about 1011 kg/s (Hartmann, 1994). This,16

however, is the net mass flux arising from the difference between the upward flux in17

convection and the clear sky subsidence. The clear sky radiative cooling rate in the tropical18

lower troposphere is ~1.6 K/day (e. g., Hartmann et al., 2001). Since convective19

disturbances cover less than 10% of the tropics, the above cooling rate implies a downward20

mass flux between 15° N and 15° S of about 2.0x1011 kg/s (assuming descent along a dry21

adiabat). Thus the upward flux from the boundary layer in convection must be about22

3.0x1011 kg/s. Assuming that the BL has a 100 hPa depth, then the mass between 15° N and23

15° S is about 1.3x1017 kg. The upward flux from the BL then implies an evacuation time24

of about 5 days if one assumes that the material is taken from the whole BL rather than just25

from the top of the BL. Based on a 3D general circulation model simulation, Prather and26

Jacob (1997) estimated a mean evacuation rate of the 130 hPa deep BL in the tropics of27

18%/day (see Figure 2-3), which implies an e-folding removal time of 5.5 days, consistent28

with the above estimate. This estimate is at the low end of the 5-10 day estimate made in29

Kley et al. (1996) based on data from a different general circulation model, and is slightly30

p. 15

longer than the estimate of 4 days made by Cotton et al. (1995). The actual evacuation rate1

will vary regionally, depending on the strength and frequency of convective events.2

3

The model results in Figure 2-3 show that the magnitude of the mass flux across the4

STT is about half that of the mass flux out the BL. As indicated above, the latter mass flux5

implies a replacement time for the BL of about 5 days. Since the TTL and BL have similar6

depths of about 100 hPa, the average replacement time for the whole TTL by air crossing7

the STT should be of the order of 10 days. However, the replacement time is not uniform8

with altitude. Much of the detrainment in the UT and TTL occurs near the STT, and within9

the TTL the mass flux decreases rapidly with height, but the precise height dependence has10

not yet been reliably determined. The local replacement time (defined as the replacement11

time of a layer of air by detrainment from convective clouds) can be estimated by assuming12

that the upward mass flux from convection is balanced by the clear sky subsidence.13

Folkins et al. (2002) used this method to obtain replacement times of 5 days near the STT,14

and 100 days at 16 km. An alternative estimate can be obtained by using measured profiles15

of CO and O3 (Dessler, 2002), yielding 10-30 days for the lower TTL and 50-90 days for16

the upper TTL. It is difficult to quantify the uncertainty in these estimates at present, but it17

is likely that global models with different contemporary convection parameterizations18

could produce numbers that differ by at least a factor of 2.19

20

The replacement time can be used to estimate how much additional chemical21

transformation has occurred for the VSL species in air parcels since they have left the BL.22

The actual concentration of the species will depend also on the entrainment of old air as the23

convective column ascends the free troposphere. If the BL air has been diluted by 50%24

during the ascent, the replacement time for the TTL by BL air will be a factor of 2 larger25

than the estimates cited above. Cloud resolving model simulations (Lu et al., 2000) indicate26

that only a very small amount of tracers that are produced in situ in the MT are entrained27

into the updrafts, but instead are forced aside and downwards due to rising thermals. The28

wealth of literature on studies using cloud resolving models (e.g., Lu et al., 2000,29

Redelsperger et al., 2000, Bechtold et al., 2000), reveals the immense complexity of30

transport in cumulus clouds, and the difficulties of representing this in large-scale models.31

p. 16

1

The general transport characteristics of convection have also been studied using2

idealized tracers in a 3D global chemistry-transport model, with a fixed uniform mixing3

ratio at the surface and a specified local decay rate (Crutzen and Lawrence, 2000). It was4

found that an insoluble gas with a lifetime of 30 days was 40-50% as abundant in the5

tropical UT as at the surface, while with a lifetime of 10 days the tropical UT abundance6

was 25-30% of the surface mixing ratio (see Figure 2-4). Lower values were computed in7

the UT outside the tropics: 25-35% and 10-15% for the 30- and 10-day lifetime tracers,8

respectively. As above, these estimates depend strongly on the respective model9

formulations, and thus could differ significantly from the results obtained with other10

models. Such modeling studies, unfortunately, are rarely repeated once the first ones have11

been published.12

13

The situation is even more complex for soluble gases because of the variable14

relationship between updraft mass flux and the amount of precipitation. Crutzen and15

Lawrence (2000) also examined this issue using idealized soluble tracers. Compared to the16

calculated concentration of an insoluble gas, they found that about 15%, 50%, and 85% of17

gases with Henry's Law constants of 103, 104, and 105 M/atm, respectively, were scavenged18

by precipitation before they reached the UT (see Figure 2-5). This result was nearly19

independent of season, geographical location, and the decay rate applied to the tracers, but20

depended strongly on the assumptions made about whether tracers were rejected from,21

retained in, or additionally scavenged by ice, snow, and graupel. Similar studies using22

cloud resolving models (Barth et al., 2001; Yin et al., 2001) have also indicated that it is23

important to consider kinetic limitations to trace gas uptake in precipitation, since assuming24

Henry's Law equilibrium, as is generally done for the long time steps of global models, will25

lead to an overestimate of the amount of gases which can be taken up into precipitation,26

and thus to the scavenging loss of highly soluble gases. A promising approach for future27

studies is the use of observations of peroxides with various solubilities together with28

insoluble tracers like CO to test parameterizations of convective scavenging (e.g., Mari et29

al., 2000). At present it is difficult to apply these results to inorganic halogenated species30

p. 17

due to a lack of observations, combined with the fact that they are also produced in situ in1

the UT and can have highly variable mixing ratios at the ground.2

3

Although this section has focused on the role of deep convection, it is important to4

note that other processes can also contribute significantly to the transport of air parcels5

from the BL into the mid and upper troposphere, especially in the extratropics. In6

particular, recent attention has been given to forced uplift associated with synoptic-scale7

frontal systems, which may be the primary mode of vertical transport in many regions of8

the extratropics (Stohl, 2001; Zahn, 2001; Kowol-Santen et al., 2001). Since frontal9

lifting is also associated with clouds and precipitation, similar uncertainties for the10

scavenging of soluble gases as discussed above will apply. The issue of frontal lifting is11

likely to see considerable progress prior to the next assessment. Other processes, such as12

variations in the BL height in association with non-zero large scale vertical velocities, as13

well as topographical effects, can also contribute to the evacuation of the BL.14

15

16

2.2.2 Stratosphere-Troposphere Exchange17

18

Most air arriving in the lower stratosphere passes through the tropical tropopause19

(e.g. Holton et al., 1995), although net transport of chemical constituents across the sub-20

tropical and extra-tropical tropopauses is also known to take place. The precise21

mechanisms of the cross-tropopause transport and their consequences for short-lived22

substances are still controversial. Furthermore, it is also important to know the form of23

halogenated gases in the UT, since soluble gases are subject to removal from the UT by24

precipitation (as discussed in the previous section), as well as by slower sedimentation of25

cirrus ice (Lawrence and Crutzen, 1998), both of which can occur at rates that can compete26

with the residence time of gases in the TTL. In this section, we discuss how one may27

obtain estimates of the flux across the tropopause by examining the mixing ratios of the28

species near the tropopause and the mass flux across the tropopause.29

30

2.2.2.1 Exchange Across the Tropical Tropopause31

p. 18

1

The transition from air having compositions typical of the troposphere to those2

typical of the stratosphere does not occur abruptly across the tropical tropopause. Above3

the level where the clear-sky mass flux is mostly upwards, air parcels are rising slowly into4

the stratosphere, driven by the Brewer-Dobson circulation (speed of order 30-50 m per5

day). However, some deep convection penetrates deep into the TTL, and a small fraction of6

the air leaving the boundary layer (about 0.1%, based on Gettelman et al., 2002) is directly7

lofted across the TT. Thus, a small fraction of the air encountered directly above and below8

the TT may have a distinct “BL character”. At the same time, convective overshooting and9

breaking waves can mix material back downward (Fujiwara, et al., 1998; Sherwood and10

Dessler, 2001), so that not all of the air in the TTL is transported irreversibly into the11

stratosphere. Notably, the radiative properties of the TTL are sensitive to the stratospheric12

O3 distribution (Thuburn and Craig, 2000), which is in turn sensitive to the transport of13

constituents through the TTL.14

15

It is also likely that some of the air in the TTL has a stratospheric origin, entering16

via lateral mixing from the extratropical stratosphere (Figure 2-2). A 20-year climatology17

of intrusions from the extratropics into the tropics (Waugh and Polvani, 2000) showed that18

Rossby wave breaking events can result in transport of tongues of extratropical air deep19

into the tropics, and a more detailed look at four Rossby wave-breaking events (Scott et al.,20

2001) was used to “infer a potentially significant contribution to the global stratosphere-21

troposphere exhange from the wave-breaking processes”. Studies of mixing in the other22

direction also help give an indication of the timescales for two-way mixing, which can be23

comparable to convective timescales. For example, Dethof et al. (2000) reported a contour24

advection study using ECMWF (European Center for Medium Range Weather Forecasting)25

water vapor fields, in which they found that the isentropic transport of water vapor into the26

extratropical lower stratosphere is similar in magnitude to upward transport in tropical27

regions. Furthermore, the work of Strahan et al. (1998) also showed evidence of such28

isentropic transport based on the seasonal cycle of carbon dioxide simulated in a global29

chemical transport model.30

31

p. 19

Finally, according to Rosenlof and Holton (1993) the annual flux across the 1001

hPa surface (i.e., TT) in the 15° N and 15° S latitude range is 8.5x109 kg/s. Since this is2

only ~3% of the flux of air out of the tropical BL computed above, the vast majority of the3

air parcels in deep convection are circulated back down into the free troposphere to4

eventually re-enter the BL, or are transported isentropically into the extratropical lower5

stratosphere. A quantification of these fractions is not yet possible.6

7

2.2.2.2 Exchange Across the Extra-tropical Tropopause8

9

Exchange of air across the extratropical tropopause is accomplished by processes10

on a large range of temporal and spatial scales. Although the net (downward) mass11

exchange rates are controlled by global-scale processes as part of the Brewer-Dobson12

circulation (Holton et al., 1995), two-way exchange occurs on synoptic and sub-synoptic13

scales and may have important consequences for chemical species. Synoptic-scale frontal14

systems and to a lesser extent mid latitude deep convection frequently bring constituents15

from the boundary layer to the upper troposphere, leading to strong chemical contrasts16

across the tropopause. These processes may also transport trace constituents, including17

VSL substances, their degradation products and inorganic halogen into the lowermost18

stratosphere.19

20

Previous analysis of observed and modeled CO2 levels (Bolin and Bischof, 1970;21

Strahan et al., 1998) suggested that there is not a large degree of communication between22

the UT and LS in mid latitudes because of a clear separation in the seasonal cycle in these23

different regions. It was postulated that this is due to the bulk of air in the lower extra-24

tropical stratosphere originating either through the tropical stratosphere or possibly via25

lateral transport from the TTL. Analysis of aircraft CO2 data from the ER-2 (Hintsa et al.,26

1998) at 40N also indicated that most of the air in the lower stratosphere must have27

entered laterally from the sub-tropics or tropics. However, there was also evidence that28

air entered the lower stratosphere directly from the mid latitude upper troposphere in29

conjunction with mixing processes near the tropopause.30

31

p. 20

Several more recent observational studies support the existence of a transition1

layer between the extra-tropical upper troposphere and lower stratosphere (see Figure 2-2

2). The layer exhibits chemical characteristics that are intermediate between these two3

regions. For example, ozone and carbon monoxide data collected as part of the4

Stratosphere Troposphere Experiment by Aircraft Measurements (STREAM) program5

show the existence of a transition layer which varies in thickness between winter and6

summer (Hoor et al., 2001), being deeper in summer. A study of NOy and N2O7

relationships also showed that more exchange appears to occur in the summer months.8

This results in a deeper summertime transition layer with higher excess NOy9

concentrations (i.e. in excess of what can be expected from photochemical destruction of10

N2O) which may have been transported into the transition layer by deep convection11

(Fischer et al., 2000). Another study showed that frontal uplifting can be important for12

transporting trace gases (SF6, deuterated water vapour (HDO)) from the sub-tropics in the13

troposphere into the Arctic lower stratosphere during winter (Zahn, 2001).14

15

These findings are consistent with seasonal variations in exchange predicted by16

model studies based on wind fields from large-scale meteorological datasets (Chen, 1995;17

Haynes and Shuckburgh, 2000). In effect exchange at middle and high latitudes18

associated with synoptic eddies and frontal systems produces a transition layer in winter19

and summer which corresponds potential temperatures between 310K and 330K. In20

winter at lower latitudes and higher levels the subtropical jet acts as an effective barrier to21

transport across the tropopause. Trajectory analyses point to frontal uplifting being a22

more effective exchange mechanism in the fall (Seo and Bowman, 2001) as well as the23

winter (Stohl, 2001). In summer the subtropical jet is a less effective barrier and24

additional exchange results in a layer extending up to ~360K. Deep convection at mid25

latitudes may also be important but its effects are not yet well quantified.26

27

In principle, the transition layer is a region in which concentrations of VSL28

substances and their degradation products may be relatively high and in which substantial29

in-situ ozone depletion may take place. Clearly, further clarification of the mechanisms30

controlling the characteristics of the transition layer, the origin of air in this layer and its31

p. 21

seasonal evolution are needed. Also, time scales on which air parcels remain in this1

region, e.g. before returning to the troposphere are unknown. All these factors have2

implications for the lifetimes and impact of VSL substances and their degradation3

products in this region.4

56

2.2.3 Chemistry In The UT And TTL78

Modeling the transport of VSL substances into the stratosphere requires an9

accounting of photochemical losses in the troposphere, particularly in the upper10

troposphere and tropical tropopause layer, where the VSL substances may reside for11

some time before ascent to the stratosphere. This section focuses mainly on the UT.12

Currently very few chemical observations of species other than ozone are available in the13

TTL, which is of critical importance for delivery of VSL substances to the stratosphere.14

Because the TTL is dry, sources other than water vapor are likely to dominate the supply15

of HOx radicals there.16

17

Loss by photolysis in the UT can be calculated with confidence if absorption18

cross-section and quantum yield data are available, as observed UV actinic fluxes in the19

UT are typically within 15% of values computed from clear-sky radiative transfer models20

(Shetter et al., 2002). Loss by reaction with OH and possibly other radicals is more21

difficult to constrain. Work over the past five years has profoundly changed our22

understanding of HO x (OH + HO2) chemistry in the UT (Jaegle et al., 2001). Aircraft23

measurements of HO x radicals (Wennberg et al., 1998; Brune et al., 1998, 1999; Tan et24

al., 2002) and supporting photochemical models have shown that photolysis of acetone25

and of convected peroxides and aldehydes provides a large source of HOx to the UT,26

comparable to or exceeding the source from oxidation of water vapor (Chatfield and27

Crutzen, 1984; Jaegle et al., 1997, 1998a, 2000; Prather and Jacob, 1997; McKeen et al.,28

1997; Crawford et al., 1999; Muller and Brasseur, 1999; Collins et al. 1999; Wang et al.,29

2000; Ravetta et al., 2002). Current photochemical models constrained with aircraft30

observations reproduce observed HOx concentrations in the UT to within the instrument31

accuracy (± 40%), and account for 60-70% of the observed variance (Jaegle et al., 2001).32

p. 22

Large unresolved biases remain at sunrise and sunset, in cirrus clouds, and under high-1

NOx conditions (Brune et al., 1999; Faloona et al., 2000; Jaegle et al., 2000). The 40%2

uncertainty in HOx translates into errors in the computed concentrations of some VSL3

species in the UT and TTL.4

5

Our ability to represent UT radical chemistry in global models remains weak,6

despite the constraints offered by HOx and NOx measurements from aircraft. Limitations7

are not so much in our understanding of the fast HOx photochemistry as they are in8

defining the factors that control the concentrations of HOx precursors and NOx.9

Recognized uncertainties in models include: 1) the sources of UT NOx from lightning and10

from vertical transport of surface emissions, the chemical cycling between NOx and11

HNO3, and possible complications involving HNO4 chemistry (Lawrence et al., 1995;12

Folkins et al., 1998; Jaegle et al., 1998b; Keim et al., 1999; Liu et al., 1999; Folkins and13

Chatfield, 2000; Schultz et al., 2000; Staudt et al., 2002); 2) the sources and chemistry of14

oxygenated species (peroxides, acetone and other carbonyls) and their deep convective15

transport to the UT (Zuo and Deng, 1999; Jaegle et al., 2000; Jacob et al., 2002); and 3)16

low-temperature rate constants for critical reactions such as CH3O2+HO2 (Tyndall et al.,17

2001; Ravetta et al., 2001). The possible role of halogen radicals themselves in18

influencing HOx chemistry, and in turn the fate of other halogens, is another unresolved19

issue (Borrmann et al., 1996; Crawford et al., 1996; Davis et al., 1996; Solomon et al.,20

1997). These issues are particularly relevant in the presence of convective clouds, and21

studies with cloud resolving models (e.g. Wang and Prinn, 2000) are providing additional22

insights.23

24

25

2.2.4. Chemical Tracer Observations For Model Evaluation26

27

Observations of chemical tracers offer valuable diagnostic information for testing28

atmospheric transport of VSL substances in global models. Of particular interest is the29

simulation of vertical transport from the BL to the UT and across the tropopause. An ideal30

chemical tracer for testing vertical transport in models should have the following31

p. 23

properties: 1) well-defined boundary conditions (concentrations or fluxes) in the BL or in1

the LS; 2) well-defined production and loss rates within the atmospheric column; 3) an2

atmospheric lifetime such that the transport processes of interest translate into measurable3

concentration gradients; and 4) extensive observations for model evaluation statistics. No4

single tracer fully meets all these conditions, but several are sufficiently close to be helpful.5

6

Table 2-1 summarizes the utility of various chemical tracers for global model7

evaluation of the key transport processes related to VSL substances simulation. Current8

models can reproduce the vertical distributions of 222Rn, nonmethane hydrocarbons, and9

CO without evident bias, providing some confidence in the simulation of convective10

transport from the BL to the UT at least in a global mean sense (Jacob et al., 1997).11

Simulations of ozone and 7Be indicate that models often overestimate the downward12

transport of air from the LS to the UT (Bey et al., 2001; Liu et al., 2001). Aircraft13

observations of soluble gases in convective outflows suggest that the scavenging of these14

gases during deep convection from the BL to the UT can be successfully described on the15

basis of Henry’s law solubility and knowledge of gas retention upon cloud freezing (Mari16

et al., 2000). However, global models generally overestimate HNO3 in the UT (Thakur et17

al., 1999), possibly because of insufficient scavenging including neglect of cirrus18

precipitation (Lawrence and Crutzen, 1998). Overall, the suite of tracers in Table 2-119

provides important information for testing transport and scavenging processes of key20

relevance to VSL substances, and it is essential that any chemical transport model used to21

simulate VSL substances be first validated with an appropriate ensemble of these tracers.22

23

Ozone is a particularly valuable tracer of vertical transport in the UT/LS region but24

it has so far been underutilized for this purpose. It is the only gas sampled regularly25

enough in the UT (by both ozonesondes and aircraft) to provide data representative of26

actual monthly means (Logan, 1999; Thompson and Witte, 1999; Emmons et al., 2000;27

Lawrence, 2001). Ozone has a lifetime against chemical loss of months to years in the28

UT/LS, hence its vertical concentration gradient is primarily determined by transport and29

chemical production (Folkins et al., 1999). Vertical profiles of ozone across the tropopause30

region and correlations with water vapor provide an important test for evaluating the nature31

p. 24

of the tropopause and cross-tropopause mass flux in global models (Law et al., 2000). Of1

particular relevance for VSL substances, ozonesonde profiles in the tropics can be used to2

diagnose vertical transport in the TTL and ascent into the stratosphere (Folkins et al.,3

1999). One difficulty with the use of ozone as a tracer of vertical transport is that its long4

lifetime in the UT allows for horizontal propagation of signals associated with vertical5

transport, and varying chemical production along these horizontal transport trajectories6

may have a major effect on concentrations (Pickering et al., 1992; Lawrence et al., 1999;7

Martin et al., 2002).8

910

p. 25

2.3 ATMOSPHERIC CHEMISTRY OF HALOGENATED VERY1

SHORT LIVED (VSL) SUBSTANCES2

3

This section presents an assessment of the state of knowledge of the kinetics and4

mechanisms of the chemical processes that determine the local lifetimes and distributions5

of the VSL organic halogen source gases and their degradation products in the atmosphere.6

The compounds to be considered are summarized in Table 2-2. They were selected be-7

cause they have been either observed in the atmosphere or proposed for applications that8

may lead to release. Fully fluorinated VSL substances are not covered in this assessment,9

since the F-containing radicals produced in their degradation do not deplete ozone.10

11

Subsection 2.3.1 deals with the kinetic data and reaction mechanisms for the or-12

ganic source gases that determine the efficiency of the SGI (source gas injection) pathway,13

as defined in section 2.1. The efficiency of the PGI (product gas injection) pathway de-14

pends on the kinetic data and reaction mechanisms of the degradation products and the15

inorganic halogen species. These are discussed in subsection 2.3.2. Some of the species16

were assessed in the most recent WMO Ozone Assessment Report (Kurylo and Rodri-17

guez et al., 1999), and the current assessment provides an update and extends that work18

to include the inorganic halogen chemistry in the UT/LS region. The degradation mecha-19

nism for n-propyl bromide is discussed in subsection 2.3.3. Subsection 2.3.4 provides20

the state of knowledge of iodine chemistry in the stratosphere, which is used for the21

evaluation of the ozone removal efficiency of iodine.22

23

2.3.1 Removal Of The Halogen Source Gases24

25

The atmospheric removal of the VSL halogen source gases under consideration is26

primarily initiated via reaction with the OH radical and/or by UV photodissociation. Re-27

moval via reactions with Cl atoms and NO3 radicals is expected to be of only minor im-28

portance. Table 2-3 summarizes the kinetic and photochemical parameters for the VSL29

halogen source gases. For larger molecules such as n-propyl bromide (n-PB,30

p. 26

CH3CH2CH2Br), the reaction rate constant with OH depends on the site of the OH reac-1

tion (see also 2.3.3).2

3

Table 2-4 summarizes estimated values of the local lifetimes for some VSL halo-4

carbons. For the chlorine and bromine containing compounds, reaction with OH is the5

dominant loss process, apart from CHBr3 and CHBr2Cl, where photolysis contributes at6

least a comparable amount. For alkyl iodides (such as CH3I, C2H5I, n-C3H7I, i-C3H7I,7

CH2ICl, CH2IBr and CH2I2), photolysis is always the dominant loss process leading to8

prompt release of iodine atoms :9

10

RI + hν → R + I (2.3.1)11

12

Local lifetimes for these iodides can be estimated from their absorption cross-sections (Rat-13

tigan et al, 1997; Roehl et al., 1997; Mössinger et al., 1998) : (4-12) days for CH3I, (2-8)14

days for C2H5I, (1-6) days for n-C3H7I, (1-3) days for i-C3H7I, less than 1 day for CH2ICl,15

hours for CH2IBr and minutes for CH2I2. The range reflects variations with latitude, alti-16

tude and seasons. Reaction with OH may also be a significant loss process for n-C3H7I17

and i-C3H7I, but I atoms are still believed to be released promptly (Cotter et al., 2001). 18

19

Deposition of the source gases into the ocean would constitute a permanent re-20

moval if there are efficient chemical and/or biological pathways to transform the species in21

the water. Estimates for the ocean removal lifetime lead to negligible effect on the local22

photochemical lifetime of the VSL halogen source gases (Yvon-Lewis and Butler, 2002).23

24

2.3.2 Production And Removal Of Degradation Products25

26

2.3.2.1 Degradation mechanisms for source gases27

28

As stated, the atmospheric degradation of the source gas is primarily initiated via29

reaction with OH and/or by UV photodissociation. The radicals produced following the30

initiation reaction are rapidly converted into more stable products. Halogen atoms may31

p. 27

be released in the initiation reaction or during the subsequent sequence of reactions occur-1

ring in the degradation mechanism.2

3

The degradation mechanism for the halogen containing compounds parallels that4

previously reported for the HCFCs (Cox et al., 1995). However, many of the reactions5

and reaction intermediates in the degradation have not been studied or observed in the6

laboratory. In these cases, formulation of the elementary steps in the degradation mecha-7

nisms has been based on analogy with known mechanisms. Both OH reaction and UV8

photolysis loss processes will lead to the rapid formation of alkylperoxy radicals. The9

degradation of the halocarbons leads to the formation of carbonyl halides, peroxynitrates,10

hydroxyhalomethanes and haloperoxy acids. The yield of a specific degradation product11

will depend to some extent on the atmospheric conditions (temperature and pressure) as12

well as the atmospheric concentrations of NO, NO2, HO2 and RO2.13

14

A general degradation mechanism for the halomethanes is shown in Figure 2-6 and15

a summary of the degradation products derived from the various source gases is given in16

Table 2-5. Upon photolysis, the singly iodinated compounds (alkyl iodides) and the di-17

halo-iodo compounds (CH2ICl, CH2IBr and CH2I2) eliminate an I atom with unit quantum18

yield. Photolysis therefore does not lead to the formation of iodine containing organic19

degradation products, with the exception of CH2I2 which could form HIC(O) as a minor20

product. Photolysis also removes CF3I rapidly from the atmosphere with CF3 and I as21

photolysis products. The atmospheric degradation of the CF3 radical does not lead to22

significant ozone destruction as discussed in Cox et al. (1995). The photolysis of multi-23

ple-substituted halomethanes (CH2BrCl, CH2Br2, CHBr2Cl, CHBr3) will lead to the direct24

elimination of a halogen atom and the formation of a formyl halide.25

26

H-abstraction by OH is the primary loss process for several of the less photo-27

labile compounds (CH2BrCl, CH2Br2 and CH3CH2CH2Br) and of secondary importance28

for the iodoalkanes (C2H5I, CH3CHICH3 and CH3CH2CH2I). For CH2BrCl and CH2Br2,29

upon reaction with OH, it is likely that carbonyls such as HC(O)Br, C(O)BrCl and30

p. 28

C(O)Br2 will be produced and these may be potential carriers of bromine to the strato-1

sphere. For the non-methane compounds such as CH3CH2CH2Br, reaction at the differ-2

ent carbon sites will lead to the formation of different degradation products. The scheme3

of the degradation mechanism for n-PB is detailed in subsection 2.3.3.4

5

2.3.2.2 Intermediate products6

7

The kinetics for the reactions involving the intermediate products are much less8

well understood than for the reactions of the source gases. In most cases, this doesn’t9

matter since the initiation reaction is the rate determining step for the release of inorganic10

halogen atoms. When there is a competition among alternative pathways such as RO2 (R11

= alkyl or substituted alkyl group) reactions or alkoxy radical (RO) reactions, the lack of12

kinetic knowledge limits the ability to accurately predict the extent to which stable prod-13

ucts are formed.14

15

The first generation of stable products following initiation of gas phase degrada-16

tion are generally carbonyl or peroxide species. Depending on the mechanism of degrada-17

tion, halogen substitution may be retained at the α or β position. Table 2-6 summarizes18

the data that are known for the reactions of stable halogenated species generated in the19

first steps of degradation of VSL substances. Much less is known about the site-specific20

kinetics of these reactions. Generally, the presence of a halogen atom leads to an overall21

reduction in the rate coefficients compared to the hydrogenated species.22

23

Halogenated carbonyls are slightly soluble in water. Once in solution, they un-24

dergo hydrolysis which then represents the driving force for their multiphase removal25

from a given air mass. This chemical reaction may be written as:26

27

RCOX (where X = Cl, Br) + H2O → RCOOH + HX (2.3.2)28

29

p. 29

The uptake coefficient, γ, which is a convolution of all steps involved in the multiphase1

processing, is expected to be small for all compounds. Current estimates give an upper2

limit of 10-3 (Debruyn et al.,1995, George et al., 1994), which indicates that cloud removal3

may play a role if gas phase reactions are slow. It must also be emphasized that, for the4

compounds with very low reactivity, cloud evaporation may re-inject these compounds5

into the air, possibly at a different altitude, before multiphase processing can occur.6

7

2.3.2.3 Inorganic bromine and iodine species8

9

Br and I atoms released through the degradation of the source gases and their deg-10

radation products undergo a series of reactions in which they are inter-converted among11

the oxides (XO) and inorganic reservoirs: HX (hydrogen halides), XONO2 (halogen ni-12

trates) and HOX (hypohalous acids) where X = Br or I. The partitioning reactions for13

the inorganic reservoirs are common to bromine and iodine, and are illustrated in Figure 2-14

7. Bromine does not form higher oxides but IO can react with BrO or IO at a rate com-15

petitive with HO2 to form OIO (Cox et al., 1999; Rowley et al., 2001). The chemical re-16

actions of OIO are not well known. Photodissociation to O(3P) + IO is not favored ther-17

modynamically in its main visible absorption band (Misra and Marshall, 1998) and is18

slow (Ingham et al., 2000). This photolysis pathway would lead to a null cycle for19

ozone. However, photolysis to I + O2 may occur (Ashworth et al et al., 2002), leading to20

an ozone loss cycle :21

22

IO + BrO → Br + OIO23OIO + hυ → I + O224I + O3 → IO + O225Br + O 3 → BrO + O 2 26

2 O3 → 3 O22728

Thus OIO is a potentially significant atmospheric iodine species, and indeed has been29

detected in the marine boundary layer (Allan et al., 2001). Further oxidation of OIO, ei-30

ther by gas phase or multiphase processes, to IO3- is a likely sink for tropospheric iodine,31

leading to the iodate observed in precipitation (Truesdale and Jones, 1996).32

p. 30

1

The relative importance of the halogen nitrates and HOX reservoirs depends on2

the local NO2 and HO2 concentrations, and the presence of an aqueous phase where halo-3

gen nitrates may be hydrolyzed to form HOX. The inorganic reservoirs are removed from4

the troposphere by incorporation into aerosols, cloud droplets and ice particles (cirrus5

clouds in the UT). However, the incorporation of the reservoirs into aerosol may not lead6

to permanent loss. Laboratory studies (Abbatt, 1994; Fickert et al., 1999; Holmes et al.,7

2001; Mössinger and Cox, 2001) show that HOX and XONO2 species react very effi-8

ciently with Cl- and Br- on aerosols and ice particles, releasing the corresponding halogen9

or inter-halogen molecules, which all photodissociate to atomic halogens10

11

HOBr (or HOI) + Cl- + H+ → BrCl (or ICl) + H2O (2.3.3)12

HOBr (or HOI) + Br- + H+ → Br2 (or IBr) + H2O (2.3.4)13

14

Reactions (2.3.3) and (2.3.4) are very efficient in acidic solutions or on HCl or HBr doped15

surfaces. These reactions (and similar reactions involving XONO2) have been the focus of16

many studies, especially in connection with their role in stratospheric halogen activation.17

They also play a role in tropospheric chemistry, particularly in the marine environment18

where sea salt aerosols contain Cl- and Br- ions. Additional reactive halogen is produced19

autocatalycally, leading for example to the so-called “bromine explosion” observed in the20

Arctic (e.g. Barrie and Platt, 1997). To be efficient, these reactions require cold condi-21

tions (such as frozen aerosols in the arctic troposphere) and sustained levels of particles22

(such as ice or sulphuric acid aerosols in the stratosphere) that may be observed for in-23

stance under volcanic influence. Table 2-7 lists the required parameters for an assessment24

of the fate of some of the HOX and XONO2 species.25

26

Reactions similar to (2.3.3) and (2.3.4) efficiently recycle iodine and bromine into27

radical forms, thereby effectively increasing the lifetime of tropospheric halogens. This28

explains the relatively high abundance of IO in the marine boundary layer. There are also29

recent direct observations that show that BrO may be present in the upper troposphere30

p. 31

at a concentration of 0.4 – 2 ppt, which is much higher than predicted on the basis of1

known photochemical sources and heterogeneous removal rates (Fitzenberger et al.,2

2000). Such observations indicate that the recycling processes may be important, leading3

to a significant increase of the efficiency of the product gas injection (PGI) pathway in4

supplying inorganic halogen species to the stratosphere.5

6

2.3.3 Degradation Mechanism For n-Propyl Bromide (n-PB)7

8

CH3CH2CH2Br ,1-bromopropane or n-propyl bromide (n-PB), is a compound9

that has been suggested for use in a variety of applications which would lead to its release10

into the troposphere. A reliable estimate of the amount of inorganic bromine that will be11

delivered to the stratosphere from release of n-PB requires better knowledge of its degra-12

dation mechanism and a guess as to the possible future quantities produced and emitted.13

14

The degradation mechanism for n-PB is outlined in Figure 2-8. Recent laboratory15

measurements have determined the site specific OH rate coefficients over the temperature16

range 230 to 360 K (Gilles et al., 2002). The kinetic data are reported in Table 2-3. The17

reaction rate constants reported by Gilles et al. (2002) and that recommended in Sander et18

al. (2002) give estimated local lifetimes of 13 days.19

20

Bromoacetone, CH3C(O)CH2Br, was observed as a major degradation product21

following the OH reaction at the -CH2- site of the n-PB molecule (β channel in Figure 2-22

8). Burkholder et al. (2002) have recently shown that CH3C(O)CH2Br will rapidly pho-23

tolyze in the atmosphere (~ hours). The photolysis products were not determined but24

the possible bromine containing products are expected to liberate bromine within days.25

The reaction of OH at the –CH2Br group in CH3CH2CH2Br (α channel) will lead to the26

rapid elimination of Br and the formation of CH3CH2C(O)H. The reaction of OH at the27

CH3- group in CH3CH2CH2Br (γ channel) will predominantly lead to the formation of28

bromo-aldehydes, CH2BrCH2C(O)H and CH2BrC(O)H. These compounds are expected29

to degrade in the atmosphere (~ days) through reaction with OH and UV photolysis.30

p. 32

Minor reactions of the reaction intermediates can lead to the formation of peroxy-nitrates1

(containing the OONO2 group), peroxy-carbonyl-nitrates (PAN type compounds, re-2

sulting from the further oxidation of the bromo-aldehydes), and alkyl hydro-peroxides3

(containing the OOH group). The peroxy-nitrates will thermally decompose rapidly un-4

der all tropospheric conditions. The alkyl hydro-peroxides will be removed via a combi-5

nation of OH reaction, UV photolysis and multiphase processing. The rates for these6

processes are not known but are expected to lead to atmospheric lifetimes on the order of7

days. The only possible degradation products of n-PB with atmospheric lifetimes longer8

than the parent compound are the peroxy-carbonyl-nitrates, but only if they are formed,9

if they contain Br and if they are rapidly transported into the low temperature environ-10

ment of the upper troposphere. In conclusion, it is very likely that the product gas injec-11

tion pathway for n-PB is from transport of inorganic halogen rather than intermediate12

products.13

14

2.3.4 Stratospheric Iodine Chemistry15

16

Although the chemistry of chlorine and bromine in the stratosphere is reasonably17

well known, this is not the case for iodine. Chameides and Davis (1980) were the first to18

describe the atmospheric photochemistry of iodine in the marine boundary layer where19

the iodine is produced from the oceanic source of volatile alkyl iodides. If alkyl iodides or20

their degradation products could enter the stratosphere, the I atom would be rapidly con-21

verted to IO through reaction with O3. Recycling of IO to I through reactions with ClO,22

BrO and HO2 would constitute catalytic O3 removal cycles.23

24

Solomon et al. (1994a) provided the first estimate for the ozone removal efficiency25

of iodine in the stratosphere. In their study, inorganic iodine is partitioned among I, IO,26

HI, HOI and IONO2, with 90% of the iodine in the form of I and IO. The ozone removal27

cycles were identified as28

29

IO + HO2 → HOI + O230HOI + hυ → OH + I31

p. 33

I + O3 → IO + O21OH + O 3 → HO 2 + O 2 22 O3 → 3 O23

4

and5

6

IO + XO → I + X + O2 (or IX + O2, followed by IX + hν → I + X)7I + O3 → IO + O28X + O3 → XO + O29

2 O3 → 3 O21011

where X is Br or Cl. At the time the study was done, rate data were not available for the12

reactions of IO with BrO and ClO. Solomon et al. (1994a) assumed a rate of 1x10-10 cm313

molecule-1 sec-1 for both rate coefficients and estimated that the ozone removal efficiency14

for iodine is about 1000 times that of chlorine. It was suggested that the coupling of a15

low natural abundance of iodine species with increasing bromine and chlorine levels have16

contributed to the trend in ozone loss in the lowermost stratosphere.17

18

Laboratory work in the past 5 years has significantly improved the knowledge of19

the relevant kinetic data. For the IO + ClO reaction, the rate coefficient has been found to20

be much lower than estimated by Solomon et al. (1994a), by a factor between 8 at 298 K21

and 5 at temperatures of the lower stratosphere (Bedjanian et al., 1997 and Turnipseed et22

al., 1997). Furthermore, three product channels have been identified :23

24

IO + ClO → I + OClO (a)25→ I + Cl + O2 (b)26→ ICl + O2 (c)27

28

Channel (a), which leads to a null cycle for ozone depletion is the major channel,29

ka/(ka+kb+kc) = 0.55. The channels which lead to ozone loss, (b) and (c), account for a30

significant fraction of the reaction, (kb+kc)/(ka+kb+kc) = 0.45 at 298 K (Bedjanian et al.,31

1997 and Turnipseed et al., 1997).32

33

p. 34

Several studies reported rate coefficients for the IO + BrO reaction ranging from 71

x 10 –11 to 10 x 10 –11 cm3molecule-1s-1 (Gilles et al., 1997; Lazlo et al., 1997; Bedjanian et2

al., 1998; Rowley et al., 2001). The possible channels for this reaction are:3

4

IO + BrO → Br + OIO (a)5

→ I + Br + O2 (b)6

→ IBr + O2 (c)7

→ I + OBrO (d)8

9

The branching ratio for channel (a) has been measured to be > 0.65 (Bedjanian et al., 1998)10

while channels (b) and (c) account for less than 0.35 of the reaction at 298 K (Gilles et al.,11

1997 and Bedjanian et al., 1998). The OIO molecule formed in channel (a) has been un-12

ambiguously detected in this reaction (Cox et al., 1999). As discussed in section 2.3.2.3,13

photolysis of OIO would efficiently produce I + O2, and channel (a) could lead to ozone14

loss cycle. However, uncertainties in OIO chemistry do not allow any firm conclusions re-15

garding the ozone depletion efficiency of iodine in the stratosphere. The new data obtained16

for the IO + IO reaction are not discussed here, since this reaction, which also produces17

OIO, is not relevant to stratospheric iodine chemistry, but could be important in the marine18

boundary layer.19

20

Laboratory investigations of the temperature dependence of IO + HO2 reaction21

(Maguin et al. al, 1992; Canosa-Mas et al., 1999; Cronkhite et al., 1999; Knight and22

Crowley, 2001) lead to a somewhat higher rate coefficient (up to a factor 3 at strato-23

spheric temperatures) than that used in the calculations of Solomon et al (1994a). This,24

associated with the reduced effect of the reactions of IO with both ClO and BrO, will in-25

crease the relative importance of the HO2 cycle in the ozone loss due to iodine. Finally,26

in all reactions of IO with other radicals, the existence of a termolecular channel (similar to27

the ClO dimer forming channel for ClO + ClO) was not observed.28

29

Using the new laboratory data in Sander et al. (2000), the ozone removal efficiency30

of iodine relative to chlorine has been recently re-evaluated : a value of 150 - 300 (Ko, 2000)31

is used in this chapter (see section 2.5) based on estimates provided by Bedjanian et al.32

p. 35

(1998) and Ko (2000). However, these estimates, which are lower than the value of 10001

initially suggested by Solomon et al (1994a), are still very uncertain due to the lack of2

knowledge on OIO chemistry, and also on the heterogeneous reactions which could be very3

important for iodine, since all the inorganic reservoirs are reactive towards stratospheric par-4

ticles (both ice and H2SO4 surfaces). Many data are still missing for these heterogeneous5

processes, such as those needed to estimate the hydrolysis rate of IONO2. When such data6

are available, a multidimensional atmospheric model should be used to perform more rigor-7

ous calculation of the ozone removal efficiency of iodine.8

9

10

p. 36

2.4 CONTRIBUTION OF HALOGENATED VSL SUBSTANCES TO THE1

STRATOSPHERIC INORGANIC HALOGEN BUDGET2

3There are two independent ways to obtain the total local concentration of inor-4

ganic halogen species in the present-day stratosphere. The first method (inorganic5

method) is to add up the measured concentrations of each inorganic species (HCl,6

ClONO2, HOCl, ClO, Cl2O2, etc. in the case of chlorine). A variant of this method uses7

measured concentrations of a subset of the inorganic halogen species (e.g. BrO) and de-8

rives the other concentrations using a photochemical model. The second method (the or-9

ganic method) is to add the contributions from each source gas by taking the difference10

between the measured concentration of the source gas and its expected concentration in11

an air-parcel based on when it first entered the stratosphere (see, e.g., Prather and Wat-12

son, 1990; Woodbridge et al., 1995; Wamsley et al.,1998). Contributions from transport13

of very short-lived (VSL) source gases, degradation products and inorganic halogen spe-14

cies from the troposphere are usually ignored in the organic method. Agreement between15

the two methods provides some reassurance, but not irrefutable, that our understanding of16

the present-day halogen budget may be correct. For example, Woodbridge et al. (1995)17

carried out a study of the inorganic chlorine budget and concluded that Cly concentrations18

derived using the two methods are in reasonable agreement. However, the work of19

Wamsley et al. (1998) showed that derived Bry using the organic method is 30% smaller20

than the expected Bry from the measured BrO but that these differences are barely within21

calculated uncertainties. Recent analyses by Pfeilsticker et al. (2000) of a different22

dataset concluded that the two methods give values that agree within the estimated un-23

certainties (~ ±3 ppt) (see Chapter 3).24

25

The measured concentrations of VSL source gases such as CH3I and CHBr3 are26

sufficiently high near the tropical tropopause that they could make a non-negligible con-27

tribution (see Table 2-8, review by Kurylo and Rodriguez et al., 1999; Sturges et al.,28

2000). This section summarizes the observations of the VSL halogen source gases and29

provides an estimate of their contribution to the present-day halogen budget in the strato-30

sphere. Recent modeling studies are also reviewed and provide some estimates of the31

p. 37

amount of Bry entering the stratosphere from particular substances (CHBr3, n-PB) and1

their degradation products.2

32.4.1 Observed Concentrations And Sources Of Organic Chlorine, Bromine And4

Iodine Compounds5

67

The atmospheric distribution and sources of organic chlorine, bromine and iodine8

compounds were reviewed in Kurylo and Rodriguez et al. (1999). Recent measured con-9

centrations of the species in the tropical troposphere from TRACE-P, PEM Tropics A and10

B are summarized in Table 2-8 and compared with values given in Kurylo and Rodriguez11

et al. (1999). Burdens estimated using the numerator in the R.H.S. of equation (2.4.3)12

(see section 2.4.2) are given in Table 2-9. Using the estimated local li