Hydrocarbon Photochemistry in the Upper Atmosphere of Jupiter

Transcription

1 ICARUS 119, 1 52 (1996) ARTICLE NO Hydrocarbon Photochemistry in the Upper Atmosphere of Jupiter G. RANDALL GLADSTONE Space Sciences Department, Southwest Research Institute, 6220 Culebra Road, P.O. Drawer 28510, San Antonio, Texas randy@whistler.space.swri.edu MARK ALLEN Earth and Space Science Division, Jet Propulsion Laboratory, California Institute of Technology, 1201 East California Boulevard, Pasadena, California AND Y. L. YUNG Division of Geological and Planetary Sciences, Caltech , 1201 East California Boulevard, Pasadena, California Received April 19, 1995; revised August 29, 1995 photochemistry that provides the major source of disequilibrium The hydrocarbon photochemistry in the upper atmosphere of species in the upper atmosphere, although on Jupi- Jupiter is investigated using a one-dimensional, photochemical- ter the aurora also provides an important source. Many of diffusive, and diurnally averaged model. The important chemi- the long-lived photochemical products are optically active cal cycles and pathways among the major species are outlined and contribute to the opacity of the atmospheres at waveand a standard model for the North Equatorial Belt region is lengths 2000 Å. Although the circulation patterns of the examined in detail. It is found that several traditionally domi- upper atmospheres are not currently known, it is more nant chemical pathways among the C and C 2 species are rethan likely that many species of photochemical origin are placed in importance by cycles involving C C 4 species. The important sources and/or sinks of heat, and thus contribute pressure and altitude profiles of mixing ratios for several observto upper atmospheric dynamics. Detailed observations of able hydrocarbon species are compared with available ultraviotrace species such as atomic hydrogen and acetylene will let- and infrared-derived abundances. The results of sensitivity studies on the standard model with respect to variations in eddy provide an excellent method for examining general circula- diffusion profile, solar flux, atomic hydrogen influx, latitude, tion patterns and vertical mixing in the stratospheres of temperature, and important chemical reaction rates are presented. the giant planets, over an enormous range of pressures. Measured and calculated airglow emissions of He at Hydrocarbon photochemistry also may be an important 584 Å and H at 1216 Å are also used to provide some constraints source for aerosol formation through the production of on the range of model parameters. The relevance of the model long-chain polymers. With the Galileo spacecraft set to results to the upcoming Galileo mission is briefly discussed. arrive in the jovian system in late 1995, it seems an appro- The model is subject to considerable improvement; there is a priate time to update our understanding of hydrocarbon great need for laboratory measurements of basic reaction rates photochemistry on Jupiter, before the first in situ composiand photodissociation quantum yields, even for such simple tion measurements of the archetype giant-planet atmospecies as methylacetylene and allene. Until such laboratory measurements exist there will be considerable uncertainty in sphere begin. the understanding of the C 3 and higher hydrocarbons in the Pre-Voyager studies by Strobel (1969, 1973, 1974, 1975) atmospheres of the jovian planets Academic Press, Inc. were the first to adequately explain methane (CH 4 ) photo- chemistry in the jovian upper atmosphere, although the basic problem of methane stability was recognized much INTRODUCTION earlier by Wildt (1937), and also addressed by Cadle (1962), Hunten (1969), and McNesby (1969). Strobel demonstrated The photochemistry of methane and other hydrocarbons that photolysis of methane leads to the formation in the upper atmospheres of the jovian planets is important of long-lived disequilibrium products such as ethane in a number of ways. On each of the giant planets it is (C 2 H 6 ) and acetylene (C 2 H 2 ), and that these products are /96 $12.00 Copyright 1996 by Academic Press, Inc. All rights of reproduction in any form reserved.

2 2 GLADSTONE, ALLEN, AND YUNG surements with which to compare our models in the lower stratosphere, but almost none at higher altitudes. We note that many more measurements exist than we have summarized here (cf. Atreya 1986); the entries in Table I were selected for their relevance to our models in terms of the latitude and pressure range sampled by the observers. MODEL The photochemical model calculations presented in this paper were obtained using a one-dimensional photochemical-diffusive computer program (Allen et al. 1981, Yung et al. 1984). The program uses a finite difference method to solve the continuity equation for each species i n i t i z P i L i, (1) transported by eddy mixing down into the deep atmosphere where they undergo pyrolysis back to methane. Methane is mixed upward to complete the cycle. In a later study, Yung and Strobel (1980) used observations of C 2 H 6 and C 2 H 2 abundances and the Ly brightness to determine vertical mixing properties near the tropopause and near the homopause. Gladstone (1983) continued this line of study, using revised rate constants and an improved acetylene photochemistry (Okabe 1981, 1983a), and included the concept of photosensitized dissociation of methane by C 2 H 2 that was developed by Allen et al. (1980) for the atmosphere of Titan. Since this last work was completed, new data on the state of the upper atmosphere of Jupiter have been obtained and many reaction rates involving hydrocarbons have been updated. With the exception of a study by Ashihara (1983), all previous studies of hydrocarbon photochemistry on the jovian planets have been limited to C and C 2 (i.e., containing two carbon atoms) species, with the possible addi- tion of some polyyne species (polyynes are HUCICU UCICUH polymers that are sometimes incorrectly re- ferred to as polyacetylenes). Although the major observed hydrocarbon species are indeed of the C or C 2 variety, there is every reason to expect substantial abundances of higher-order species. Several of these more complex hydrocarbons (e.g., CH 3 C 2 H, C 3 H 8, and even C 6 H 6 ) have been observed in the auroral regions on Jupiter (Kim et al. 1985). In addition, it is important to include higher hydrocarbons in the model in order to make sure that the simpler hydrocarbons are being accurately modeled, since there are often important feedback pathways in which the photochemistry of complex hydrocarbons affects the abundances of simpler species. In this study we extend the hydrocarbon photochemistry from the C and C 2 species to include most hydrocarbons up to the C 4 level. An initial investigation using our updated model focused specifically on the ratio of ethane to acetylene in the lower stratosphere of Jupiter (Allen et al. 1992). Here we present a more comprehensive investigation of the response of the upper atmosphere to variations in eddy diffusion profile, solar flux, atomic hydrogen influx, latitude, temperature, and key chemical reaction rates. We obtain density profiles for nearly all the C C 4 hydrocarbons, as well as atomic hydrogen and helium. The calculated hydrocarbon abundances are compared with some recent ultraviolet (UV) and infrared (IR) measurements. A summary of these recent measurements is presented in Table I. The dates of the observations, latitudes studied, and the techniques used are given, and the pressure ranges sampled by the measurements are estimated based on published weighting functions. It is apparent that (with the exception of UV occultations) both IR and UV measurement techniques sample the jovian stratosphere in the few to tens of mbar pressure range. Thus, we have many mea- in which n i is the number density, i is the vertical flux, P i is the chemical production rate, and L i is the chemical loss rate, all evaluated at altitude z and time t. Nonlinear chemistry is solved using Newton s method. The altitude z in all of our model atmospheres is measured from the 1-bar pressure level. In the models presented here we as- sume steady-state conditions, so that n i / t 0 and i, P i, and L i are diurnal averages at the modeled latitude. This is a reasonable assumption given the chemical lifetimes of the important species on Jupiter. The vertical flux i in (1) is determined by i (D i K) dn i dz D i K H i H a n i (2) (1 i)d i K dt T dz n i, where D i is the molecular diffusion coefficient, K is the eddy (turbulent) diffusion coefficient, H i and H a are the component and bulk atmosphere pressure scale heights, respectively, i is the thermal diffusion parameter, and T is the temperature (Chamberlain and Hunten 1987). A more accurate version of Eq. (2) would include a right- hand-side term wn i, where w is the zonally averaged vertical velocity at the latitude of the simulation (e.g., Hunten 1975). The inclusion of a vertical velocity is desirable, since it would allow the possibility for transport of long-lived species against their mixing ratio gradients, which is not possible when vertical movement is represented solely by diffusive processes. Unfortunately, the large-scale stratospheric circulation of Jupiter is not sufficiently well deter- mined at the present time for us to include the vertical velocity field in our model. Dynamical modeling by West et al. (1992) suggests that meridional circulation in the

3 JOVIAN HYDROCARBON PHOTOCHEMISTRY 3 TABLE I Relevant Trace Hydrocarbon Abundance Observations a Mixing ratio ranges, representing statistical errors in most cases (ppbv parts per billion by volume, ppmv parts per million by volume). For Maguire et al. and Kostiuk et al. total uncertainty estimates are used. b Pressure ranges with peak of contribution function in parentheses. The pressure ranges are based on weighting function FWHMs for the IR measurements (for Kim et al. the weighting functions were assumed to be the same as for the measurements of Maguire et al.) and contribution functions for the UV measurements. c IRIS Voyager IRIS experiment; UVS Voyager UVS experiment; IUE International Ultraviolet Explorer; IRFP Infrared Fabry Perot Spectrometer (IRTF observations); IRHS Infrared Heterodyne Spectrometer (KPNO and IRTF observations). d Maguire, W., R. E. Samuelson, and M. Allen, 1993, manuscript in preparation. 1 P (mbar) 100 region is driven by heating due to region or are whole disk measurements which are weighted absorption of sunlight by polar stratospheric aerosols and toward low latitudes (cf. Table I). We first establish a absorption of momentum from upward-traveling gravity standard model for the North Equatorial Belt (NEB) (at waves, with upwelling for P 30 mbar at high latitudes a latitude of 10 N) and examine it in some detail. We then and subsidence in the equatorial regions (with the reverse perform a variety of sensitivity studies on the standard situation for P 30 mbar). Intense heating from auroral model to investigate the response of the model atmospheric particle precipitation and ionospheric Joule heating proba- abundances to changes in solar flux, eddy diffusion, and bly drives a similar though more vigorous circulation pattern other important model parameters, which are either poorly at thermospheric altitudes (P 0.1 bar). The upwell- known or vary with time. As part of the sensitivity study ing velocities in the polar regions derived by West et al. we also examine models for two other latitudes; the north (1992) are as large as cm sec 1 at P 10 mbar, temperate zone (NTeZ) (at a latitude of 23 N) and a nearauroral which is comparable to the downward effective velocity region at a latitude of 60 N (POL). These sensitivwhich due to eddy diffusion at this pressure level. The abundance ity studies will allow the model results reported here to be profiles of long-lived species could thus be considerably applicable to a wider range of observations, or even utilized modified by vertical winds, and we recommend that further to constrain physical parameters as appropriate observations general circulation studies of Jupiter s upper atmosphere become available. be undertaken so that future photochemical studies may The temperature profiles used in our models are based include these effects. on Voyager IRIS results in the lower atmosphere (from 6 The focus of this study is stratospheric hydrocarbon pho- bars to 1 mbar) (W. Maguire 1993, private communication) tochemistry in the equatorial region, since most of the and are consistent with the upper atmosphere tempertochemistry available abundance measurements are for the disk center ature profile derived by the Festou et al. (1981) analysis

4 4 GLADSTONE, ALLEN, AND YUNG FIG. 1. Total density and temperature profiles as a function of pressure for the standard NEB (solid line), NTeZ (dotted line), and POL (dashed line) model atmospheres. Altitudes marked on the right side are measured from the 1-bar level and apply to the NEB model only. of the stellar occultation experiment performed by the P a T T a T a T 0 Voyager 2 UVS instrument. The temperature profiles for T Ta, T T the three latitudes we are considering are shown in Fig /H0 (3b) as a function of hydrostatic pressure. The altitude scale on the right-hand side of Fig. 1 is appropriate for the NEB where P a mbar, T a 425 K, T 1100 K, temperature profile alone (the altitude scales for the NTeZ T K, 337 km, km, H 200 km, and POL model atmospheres are only slightly comthe and H 0 30 km. This empirical expression is pivoted at pressed). Since the temperature profile in the upper atmosphere Voyager UVS occultation measurement of T a 425 of Jupiter is considerably less structured than that K at a density of n a cm 3 (Festou et al. 1981). of the Earth, we suggest that the region of atmosphere The values of 0 /H 0 and /H were chosen to obtain a between the tropopause and the homopause be referred good match to the Voyager IRIS data used at P 1 mbar to as the stratosphere. So defined, the stratosphere may be and the exospheric measurement of Festou et al. (1981). conveniently divided into three sublayers, each spanning It is important to emphasize that the models we are presabout two decades of pressure the lower ( mbar), enting here are for nonauroral regions only. Recent obser- middle ( mbar), and upper ( 0.1 bar 0.01 mbar) vations of auroral H3 and H 2 Lyman and Werner band stratosphere. These sublayers provide a useful context for emissions (e.g., Drossart et al. 1993, Trafton et al. 1994) outlining the details of the photochemistry in later sections. suggest that the upper stratosphere and thermosphere in The temperature is assumed to be fairly constant at 200 the auroral regions is much hotter than in the nonauroral K throughout the upper stratosphere, but above the homo- atmosphere, as was suggested by the calculations of Waite pause region at 0.1 bar ( 400 km) the temperature et al. (1983). increases steeply into the thermosphere and eventually The acceleration due to gravity at the 1-bar level for the approaches an exospheric temperature of 1100 K (Festou NEB, NTeZ, and POL regions is 2335, 2388, and 2616 cm et al. 1981). In extrapolating from the IRIS-derived temper- sec 2, respectively, as determined from Anderson (1976). ature profiles at pressures 1 mbar to the exospheric tem- Since our calculations apply to the region from 6 bars down perature we utilize a double-sided Bates-type profile to 1 nbar (i.e., from well below the tropopause at 100 (Bates 1959) having the form mbar to well above the homopause at 0.1 bar), the variation of gravity with altitude must be properly accounted for, as well as the centrifugal effect due to rotation P(T) P a T a T T T T Ta (3a) T T a /H (at the equator the gravitational acceleration is reduced

5 JOVIAN HYDROCARBON PHOTOCHEMISTRY 5 by 220 cm sec 2 due to the spin of Jupiter from what its carbon bond-forming reactions is quite small (see Table value would otherwise be), in order to obtain the correct IV), and the abundances of C 5 and higher hydrocarbons variation of altitude with pressure. The mean mass of the are likewise expected to be small, so that their effect on actual atmosphere changes from 2.25 amu below the the C C 4 system is probably of minor importance. We homopause to 1.98 amu at the upper boundary of our have deliberately omitted reactions involving elements model atmospheres, mostly due to the diffusive separation other than C and H (such as N, O, and P, for example), of both helium and atomic hydrogen above the homopause. since in this particular study we are primarily interested At very large altitudes the mean mass approaches unity, in studying the distribution of the major products of but this occurs at number densities 10 4 cm 3, far above methane photolysis in the stratosphere. Ammonia (NH 3 ), the region we are considering. For convenience, the mean water (H 2 O), and phosphine (PH 3 ), the main N, O, and mass of the atmosphere was held constant at 2.25 amu P bearing species in the jovian atmosphere, are greatly when calculating the pressure/altitude relation. The error depleted in the stratosphere due to both condensation introduced in the altitude scale of our figures by the ap- and photolysis in the region of the tropopause (Atreya proximation of constant mean mass increases from 0 at et al. 1977, Strobel 1985). One of the main products of and below the homopause region to 6% at the upper coupled hydrocarbon/ammonia photochemistry, methylamine boundary of the model atmospheres (i.e., where our models (CH 3 NH 2 ), is dissociated by photons up to 2500 indicate 600 km the actual altitude for that pressure is Å in wavelength, and one of its products is hydrogen closer to 636 km). cyanide (HCN). HCN was detected on Jupiter (Tokunaga Equation (1) is solved for each of the 39 species con- et al. 1981) but a more recent search has called this tained in our model (listed in Table II), with the exception result into question (Bézard et al. 1994). The coupled of H 2, which we set equal to the density of the bulk atmo- photochemistry of NH 3 and hydrocarbons in the lower sphere minus the calculated densities of H and He. The stratosphere has been investigated by Kaye and Strobel values of P i and L i in Eq. (1) are determined by the 194 (1983a, b), who find that they can produce HCN abundances chemical reactions we have chosen. The set of 78 photodissociation of in the lower stratosphere, but that reactions involving these species is presented in the coupled chemistry has only a minor effect on the Table III, along with their rate coefficients in the lower production of ethane (C 2 H 6 ). Molecular nitrogen may also stratosphere (at 8.9 mbar) and at the upper boundary of be present in substantial amounts (1 10 ppm according to the atmosphere (at 1.0 nbar), the wavelength range in Prinn and Olaguer (1981)) throughout the upper atmo- which the photodissociation cross sections are important, sphere, but it is so stable a molecule that it is effectively and references. The photodissociation cross sections used inert in the lower and middle stratosphere. Because of in the model are averaged over 50-Å-wide bins for wave- the condensation of H 2 O deep in the troposphere, the lengths longer than 1250 Å. Below 1250 Å, irregular bin dominant form for oxygen in the lower stratosphere is sizes are used and cross sections for prominent solar emis- thought to be carbon monoxide (CO). The measured sion lines wavelengths (1216, 1026, 991, 997, and 973 Å) stratospheric mixing ratio for CO is only are included separately. The solar fluxes for the standard (Atreya 1986), however, and it is not expected to affect model are taken from Mount and Rottman (1981) for the the abundances of the major hydrocarbon species. Thus, far- to mid-ultraviolet (FUV MUV), and from the 1979 although we neglect other species, our results should be solar maximum values of Torr and Torr (1985) for the realistic throughout most of the stratosphere. In the region extreme ultraviolet (EUV). Solar maximum solar fluxes near and below the tropopause (at P 100 mbar), were chosen for the standard model so that our results however, our results may be less reliable. would be most comparable with Voyager observations, The upper boundary condition for all species except H although we placed Jupiter at its mean distance (5.2 AU) in our models was a specification of zero flux. This implies rather than at the Voyager encounter distances of about that there are no sources or sinks above the upper boundary. 5.3 AU, so that our fluxes are about 4% higher than would For atomic hydrogen there is a source due to EUV be expected during the Voyager encounters. As mentioned and soft electron dissociation of H 2. Strobel (1973) estimated above, the solar fluxes were appropriately diurnally averaged that the source due to solar EUV produced a downabove, for each latitude studied. The set of 83 bimolecular ward flux of H of cm 2 sec 1. In our standard and 33 termolecular chemical reactions used in the model model we assume a considerably larger downward flux of are listed in Table IV, along with adopted rate constants H atoms at the upper boundary, based on the study by and references. Waite et al. (1983), of cm 2 sec 1. The abundances Although we consider a fairly complete reaction set of the major disequilibrium species near the homopause for the C through C 4 hydrocarbons, the possibility of are very sensitive to the value assumed for the downward fast catalytic cycles involving C 5 or higher hydrocarbons flux of H atoms from the thermosphere, and we examine cannot be ruled out. However, the number of carbon this sensitivity in detail in a later section.

6 6 GLADSTONE, ALLEN, AND YUNG TABLE II Model Species and Heats of Formation a Heat of formation at T 298 K and P 1 bar in the ideal gas state, from Lias et al. (1988, 1994), Domalski and Hearing (1993), Chase et al. (1985), and Wagman et al. (1982). Landry et al. (1991) pointed out the importance of putting the lower boundary below the tropopause, in order to avoid boundary effects in the lower stratosphere. Thus, we have located our lower boundary at the 6-bar level (several scale heights below the tropopause at 100 mbar), where we fix the mixing ratios of H 2, He, and CH 4 at 0.89, 0.11, and , respectively (Gautier et al. 1981, 1982, Sato and Hansen 1979; Bjoraker et al. 1986). All the other species are allowed to move across the lower boundary with a downward velocity given by K 6B /H 6B, where K 6B and H 6B are the eddy diffusion coefficient and the bulk atmospheric density scale height, respectively, at the 6-bar level. This velocity is an upper limit to the true diffusion velocity which is given by w i6b K 6B 1 n i dn i dz z z 6B 1 H 6B. (4) For well-mixed species the true diffusion velocity is zero

7 JOVIAN HYDROCARBON PHOTOCHEMISTRY 7 TABLE III Photodissociation Reactions

8 8 GLADSTONE, ALLEN, AND YUNG TABLE III Continued

9 JOVIAN HYDROCARBON PHOTOCHEMISTRY 9 TABLE III Continued a Photodissociation rate constants for reaction i (denoted J i ) in units of s 1. Values are for 5 mbar and 1 nbar, respectively, at 10 latitude and are calculated utilizing diurnally averaged optical depths and the solar maximum irradiances of Mount and Rottman (1981) and Torr and Torr (1985). The adopted heliocentric distance of Jupiter is 5.2 AU and the subsolar latitude is 0. Indicated also is the wavelength range (longward of 360 Å) in which the cross sections are nonzero.

10 10 GLADSTONE, ALLEN, AND YUNG TABLE IV Chemical Reactions

11 JOVIAN HYDROCARBON PHOTOCHEMISTRY 11 TABLE IV Continued

12 12 GLADSTONE, ALLEN, AND YUNG TABLE IV Continued a Two-body rate constants for reaction i (denoted k i ) and high-pressure limiting rate constants for three-body reactions (k ) are in units of cm 3 s 1. Low-pressure limiting rate constants for three-body reactions (k 0 or k 0,i ) are in units of cm 6 s 1. b M represents any third body. and our boundary condition overestimates the transport across the lower boundary. However, for the only species for which this lower boundary condition is important (i.e., the long-lived alkanes), the lower boundary is far below the production and loss regions of the stratosphere. By Eqs. (1) and (2) we expect these long-lived hydrocarbons to have an approximately constant downward flux in the troposphere. As long as K and H are slowly varying in the

13 JOVIAN HYDROCARBON PHOTOCHEMISTRY 13 troposphere, then the densities of the alkanes should be tory, and theoretical, kinetics and photochemistry literature fairly constant in the troposphere and our simplified available at that time. These tables are the largest boundary condition will be accurate (i.e., the alkanes are compilation to date of chemical processes occurring in a not expected to be well mixed in the troposphere). The reducing planetary atmosphere. The Yung et al. (1984) primary reason for placing the lower boundary as deep as work was the starting point for the chemical model used 6 bars is to avoid boundary effects in the lower stratothe in this paper. There are clearly some differences between sphere, but since we are neglecting nonhydrocarbon (N, earlier and current reaction sets simply as a result of H 2 O, and P) chemistries that may become important in the being the dominant constituent in the jovian atmosphere, upper troposphere, the abundance profiles we derive in while N 2 is the major constituent in the atmosphere of the troposphere should be regarded with caution. For the Titan. eddy diffusion profile we use the functional form The reaction lists presented in Tables III and IV reflect the laboratory measurements that have become available K(n) K H (n H /n) P P T (5a) since the Yung et al. (1984) model was prepared. Unfortunately, most of the relevant experimental research has been K T P P T, (5b) motivated by the importance of hydrocarbon chemistry in where P T represents the pressure of the tropopause (i.e., combustion studies. Consequently, the majority of pub- 100 mbar). In the formulation given by Eqs. (5) it is alloware lished results are for systems at higher temperatures than able for K to be discontinuous across the tropopause. In found in the stratosphere of Jupiter. For many recombi- the standard NEB model we use K H cm 2 sec 1, nation reactions only the high pressure limiting rate con- K T cm 2 sec 1, n H cm 3, and stants are available and the buffer gases are never H 2 as The value of K H was chosen to be consistent with the is relevant for simulating jovian chemistry. The identities result derived from the Voyager UVS stellar occultation by of product species are rarely determined along with the Atreya et al. (1981). Since n cm 3 at the measurements of kinetic rate constants nor are quantum tropopause, we have K cm 2 sec 1 at the tropoported yields for formation of neutral product species often repause in the standard model. The diffusion time constant when photodissociation cross sections are pub- just below and just above the tropopause are d lished. Fortunately, authors of these works sometimes offer H 2 /K 131 and 27 years, respectively. The value of K T estimates of the probable products based on various chemiwas taken from Allen et al. (1992), and is consistent with cal arguments. The foregoing qualifying statements have the results of Conrath and Gierasch (1984). Though the been presented as a way of indicating the overall uncertain- tropospheric eddy diffusion coefficient is expected to be ties in the reaction model utilized in this paper. much larger at higher pressures (e.g., cm 2 sec 1 The chemical model presented in this paper consists of at pressures 6 bars) according to Gierasch and Conrath reactions we have identified, based on available reaction (1985), the calculated stratospheric abundances of the rate constants, as being important for understanding the long-lived hydrocarbons will still be accurate. As the actual hydrocarbon chemistry in the jovian atmosphere. Other profile for eddy diffusion is essentially unknown for Jupiin hydrocarbon reactions are known, but are not included ter, we will examine the sensitivity of the model to changes our model since current understanding of their rate in K H, K T, and. coefficients indicates these reactions are unimportant in For molecular diffusion coefficients we use the exof the jovian stratosphere. Several key reviews/compilations pression laboratory photolysis and kinetic measurements pro- vided the starting point for updating the reaction set beyond Yung et al. (1984), including Calverts and Pitts (1966), D i (n) A H2 T n 0T (S H2 1) 0 n M H2/M i, (6) Okabe (1978), Allara and Shaw (1980), Laufer et al. (1983), Tsang and Hampson (1986), and Tsang (1988); these works where A H cm 2 sec 1 K S H2, SH2 1.75, are referenced in the tables where appropriate. n 0 T cm 3 K, and M H amu. Tables III and IV consist of several types of chemical This expression gives results that are within a factor of reactions: (a) photodissociation (AB h three for the diffusion coefficients of He, H, and CH A B), (b) inser- 4 in H tion (A BH AB H), (c) hydrogen abstraction 2 (Mason and Marrero 1970), and provides a physically reasonable estimate for the diffusion coefficients of the (A BH AH B), (d) combination (A B M AB) higher hydrocarbons in H 2. and disproportionation (AB CD AC BD), (e) exchange and transfer (A BC AB C), and (f) REACTION RATE COEFFICIENTS cracking and hydrogen scavenging (A H AH, followed by AH H A H 2 or C D). Yung et al. The tables of reactions in the Yung et al. (1984) Titan (1984) present an extensive discussion of these reaction atmosphere model provide a major review of the labora- categories and their importance in the hydrocarbon chem-

14 14 GLADSTONE, ALLEN, AND YUNG istry of planetary atmospheres, so a similar commentary be uncertain for several reasons. First, many values may will not be repeated here. However, additional comments be systematically underestimated because the input cross for the Tables III and IV entries are needed and comprise sections have been measured and/or theoretically calculated the remainder of this section. for only a limited range of wavelengths. For example, In selecting the reactions for Tables III and IV the focus the CH 3 cross sections are known only in a narrow range was on the photochemical formation of C 2,C 3, and C 4 around 1500 Å and in an isolated band near 2150 Å. Photo- compounds in general, the aromatic compound C 6 H 6, and dissociation at other, particularly shorter, wavelengths is polyynes through C 8 H 2. Formation of larger molecules was expected, but has not been included in these calculations. not considered for this work. Consequently, for reactions In some cases in which the cross sections appeared to be with probable products larger than the chosen limits monotonically decreasing with increasing wavelength, for R151, R163, R184, R185, R192, R193, R194 no specific example 1,2-C 4 H 6, we extrapolated the cross sections to product species were identified. This does represent a minor longer wavelengths and diminishing values to minimize the inconsistency in conservation of carbon in the model, amount by which the photodissociation rate coefficients but the column-integrated rates of these reactions sum to were underestimated. A second source of uncertainty is 1% of the upward flux of CH 4 from the interior. In addi- the unavailability of measurements quantifying the temperature-dependence tion, many more forms of C 3 H y and C 4 H y exist than are of cross-section values. Wu et al. explicitly considered in the Table II species list and the (1989) demonstrated the importance of this issue for hydrocarbon reaction set of Tables III and IV. For example, we have species when they found that the C 2 H 2 cross deliberately excluded all cyclic isomers from the model sections at 155 K could vary by as much as 40% from the (with the exception of benzene, the results for which will 295-K values, a typical temperature for the measurements be discussed in a separate paper) since we expect them which supply the majority of our cross-section data. Finally, to be less abundant than noncyclic isomers in the jovian and most importantly, a significant source of uncertainty stratosphere. We have chosen to model the net formation in calculating the photodissociation rate coefficient for any of representative alkanes, alkenes, and alkynes for each specific reaction in Table III is the lack of available, de- carbon number. We feel that the choices made preserve tailed information on the branching ratios (probabilities the relative abundances of total C 3 and C 4 species, and even of producing a particular set of products upon absorbing the relative abundances by general chemical classification. a photon) as a function of wavelength. There often are For a number of reactions, measurements of reaction large variations in measured branching ratios for a particurate constants have been reported in the literature at sev- lar reaction at different wavelengths (for example, see Table eral specific temperatures (or temperature ranges). In most 3A in Yung et al. (1984)). For the calculations in this instances the published data show that the rate constants paper, at wavelengths between two wavelengths for which depend on temperature. In the absence of published rate measurements were available, the branching ratio was set constant expressions as a function of temperature, we fitted equal to the measured value at the shorter wavelength. a simple Arrhenius expression of the form k(t) Ae B/T Branching ratios shortward (longward) of the lowest (high- to the literature values to estimate the reaction rate con- est) wavelength at which data exist were set equal to the stants for the relatively low temperatures in the jovian measured values at the nearest wavelength. To further stratosphere. References to the various measurements at understand the uncertainties in any particular photodissociation specific temperatures are provided in Table IV. rate coefficient value, the reader should refer to To better simulate the observed vertical variation in the the references listed in Table III, which summarize the abundances of C 2 H 2 and C 2 H 6 at near-equatorial jovian published sources of information. latitudes, Allen et al. (1992) suggest enhancing the model We now present additional comments for several specific rate of hydrogenation of C 2 H 2 to C 2 H 6. Reasonable reactions. (1) At the longer wavelengths important for the agreement between calculations and measurements was photodissociation of C 2 H 2 (R8 and R9), the cross-sections achieved (1) by increasing k 164 and k 0,86 and decreasing are sensitive to temperature (Wu et al. 1989). For our k,82 within their published uncertainties, (2) by decreasing calculations, we utilized the 155-K values, available between k 0,82 and k 85 within estimated uncertainties at low temperatures, 1530 and 2034 Å, from Wu et al. (1989) and Chen (3) by increasing k 154, k 155, and k 158 as suggested by et al. (1991). (2) In our study we assumed that the product recently published measurements for k 154 and k 156, and (4) channels and associated branching ratios in the photodisso- by increasing the low-temperature quantum yield for ciation of allene (CH 2 CCH 2 ) were similar to its isomer 1475 Å for formation of C 2 H from photolysis of C 2 H 2 (R8). methylacetylene (CH 3 C 2 H). A recent paper by Jackson et These adjusted values are used in our model and are listed al. (1991) confirms our choice of product channels, although in Tables III and IV. The reader is referred to Allen et al. the branching ratios are slightly different (0.81 and (1992) for further details for the R23 and R24 yields, respectively, as opposed The calculated photodissociation rate coefficients may to the values 0.73 and 0.27 chosen by us). The C 3 H 4 isomers

15 JOVIAN HYDROCARBON PHOTOCHEMISTRY 15 methylacetylene and allene are of considerable importance has a significant activation barrier at low temperatures, we to the hydrocarbon photochemistry of Jupiter and we implore reduced the rate constant k 179 for the comparable reaction the laboratory community to publish further studies of these molecules. (3) In the absence of published data, C 4 H H 2 C 4 H 2 H R179 we adopted similar product channel and rate coefficients for C 6 H 2 and C 8 H 2 as for C 4 H 2, for which measurements by a factor of 10 from the value for k 154. Also following exist. We expect that the C 4 H 2 photodissociation rate coef- the model of Tanzawa and Gardiner (1980) we estimated ficient is a lower limit to the values for C 6 H 2 and C 8 H 2 that all reactions leading to C 6 H 2 had equal rate constants since the spectra of polyynes (C 2n H 2 ) are known to have and all reactions forming C 8 H 2 had rate constants a factor absorption at increasingly longer wavelength as n increases of 40 smaller than the C 6 H 2 -forming reactions. Finally, (Kloster-Jensen et al. 1974). we further assumed that all reactions leading to polyyne The Anderson et al. (1987) measurement of k 120,CH product species larger than C 8 H 2 had rate constants equal insertion into CH 4, at 290-K was 60% smaller than the to those reactions forming C 8 H K value derived from the temperature-dependent rate For a number of the abstraction reactions in Table IV coefficient expression reported earlier by Berman and Lin for which no laboratory measurements are available, we (1983). Since the temperature-dependent formulation pro- estimated the rate coefficients assuming that similarities vides some ability to extrapolate reported laboratory re- in molecular structures of reactants, and similarities in results to lower temperatures, we adopted the Berman and action exothermicities, suggest similarities in reaction ki- Lin (1983) exponential factor and scaled the preexponen- netics. Reactions of hydrogen atoms with species such as tial factor by 0.6. C 3 H 5 and C 4 H 3, which contain the vinyl (C 2 H 3 ) moiety, Cole et al. (1984) modeled the synthesis of aromatic were assigned rate constants equal to that for species in flames. The reaction C 4 H 5 C 2 H 2 C 6 H 6 H R187 H C 2 H 3 C 2 H 2 H 2. R85 By a similar logic, H-atom abstraction by CH 3 from an was identified as the rate-controlling step in the formation embedded vinyl moiety in, for example, C 3 H 5 and C 4 H 5 of benzene (Rn refers to the corresponding reaction numwas assumed to be equivalent to ber in Tables III or IV). In their work, the suggested value for k 187 yielded calculated C 6 H 6 formation rates qualitatively similar to, but systematically smaller than, their measurements. CH 3 C 2 H 3 CH 4 C 2 H 2. R136 Hence, the value for k 187 in our model (see Having roughly similar reaction exothermicities, H-atom Table IV) is 5.6 times larger than the Cole et al. (1984) abstraction from large alkanes by C 2 H also was assumed value. This factor is what is necessary to bring the simulato be equally fast, so the rate constants for C 2 H C 4 H tions of the laboratory results into quantitative agreement 10 (R161) was set equal to the measured value for C 2 H with the measurements. C 3 H 8 (R159). There is unfortunately little published on the rate coeffi- The rate coefficients for combination reactions were incients involving polyyne species larger than C 2 H and C 2 H 2. terpolated between the low-pressure, three-body values A key work is that of Tanzawa and Gardiner (1980), who (k 0 ) and high-pressure, two-body limiting values (k ) with modeled laboratory measurements of high temperatures a simple expression that leads to the limiting values at low C 2 H 2 pyrolysis. Their model included several reactions inand high densities, volving C 2n H and C 2n H 2 species. In their work, the reactions k(t, M) k 0(T) k (T) C k (T) k 0 (T)M, 2 H C 2 H 2 C 4 H 2 H R156 (7) C 4 H C 2 H 2 C 6 H 2 H R181 where M is the total atmosphere density and k(t, M) isin had equal rate constants and no activation barriers. We units of centimeter 6 second 1. An alternative interpolation adopted this result for our model. In addition, we assumed formula is presented in the NASA compilation of reactions that all hydrogen abstraction reactions forming C and their rate expressions recommended for modeling ter- 2n H 2 from C restrial stratospheric chemistry (see most recently DeMore 2n H, where n 2, had rate constants equal to the compaet al. 1992): rable reactions where n 1 and no activation energies. However, since k(t, M) k 0(T)k (T) 2 k (T) k 0 (T)M 0.6(1 log (k0 (T)M/k (T))) 1 C 2 H H 2 C 2 H 2 H R154 (8)

16 16 GLADSTONE, ALLEN, AND YUNG This formula, based on the work of Troe (1977), provides ture experimental data can be compared with the temperature-independent a more realistic interpolation in the transition region. The values use of Eq. (8) rather than Eq. (7) resulted in minimal changes in the calculated abundances: differences of k 0, cm 6 sec 1 (11) 25% in the major photochemical species concentrations and typically equally small differences for the minor and k, cm 3 sec 1 (12) trace photochemical species (although at specific pressure adopted by Yung et al. (1984) from the low-temperature levels changes of a factor of 2 may occur for trace theoretical estimate of Laufer et al. (1983) and other earlier species). laboratory measurements. At the cold jovian stratopause, Laufer et al. (1983) and Yung et al. (1984) discussed the k 0,135 from Table IV is a factor 100 smaller than k 0,135 from application of theoretical and semiempirical techniques Macpherson et al. (1985) and factor of 1000 smaller than to estimating recombination reaction rate constants not the Yung et al. (1984) value. At higher temperatures, the otherwise measured in laboratory experiments. They noted standard model value and that of Macpherson et al. (1985) the correlation between large numbers of atoms in the come within 50% of each other, and both are smaller transitory, excited reaction intermediate and fast rate coefthan the Yung et al. (1984) value by a factor of 250. ficients. On the basis of these considerations, Laufer et al. However, the k,135 values from all three sources are in (1983) suggest a scaling between the rate constants for very good agreement with each other, with the standard several hydrocarbon reactions. The estimation of recombiand Macpherson et al. (1985) values almost identical, and nation rate constants in Table IV was guided by, when not smaller than the Yung et al. (1984) value above 1 mbar directly derived from, this work. by 30%. Consequently the effective rate coefficients for Laufer et al. (1983) computed three-body rate constants reaction R135 from all three sources are very similar for for several hydrocarbon combination reactions for the tempressures between 0.05 and 1 mbar, the main stratospheric peratures 100, 200, and 300 K. For the purposes of interpoformation zone for C 2 H 6, and diverge at lower pressures lating between these specific temperatures, Arrhenius exas expected from the differences in the k 0,135 values (with pressions were fit to the combination of the 100- and 200- little impact on the column C 2 H 6 abundance). K results and to the combination of the 200- and 300-K The high-pressure rate constant for the reaction results. For each of these reactions, we list in Table IV both derived expressions. For the low-pressure rate constant for the recombination reaction H C 2 H 2 M(k 0,84 ), Yung et al. (1984) derived H C 3 H 6 M C 3 H 7 R104 their expression (see also our Table IV) from the laboraby was estimated according to the scaling arguments advanced tory measurements reported by Payne and Stief (1976). Laufer et al. (1983) and Yung et al. (1984). Two measure- The published data suggest a lower limit of cm 6 ments of the temperature dependence of k,104 appeared in sec 1. This was adopted in our calculations. the same year. Watanabe et al. (1982) report the expression The temperature-dependent rate coefficient expressions k, e 811/T cm 3 sec 1, while Harris and Pitts for methyl radical recombination-forming ethane, k 0,135 (1982) find k, e 785/T. For the purposes and k,135, listed in Table IV were derived by Slagle et al. of our calculations, the Arrhenius expression appearing in (1988) from laboratory experiments in the temperature Table IV was constructed from an average of the two range K and a range of gas densities between published values at 300 K (differing by 25% from each and molecules cm 3. Since the key other) and an average of the two exponential factors (difformation zone for C 2 H 6 in the jovian atmosphere is in fering only by 5%). the middle and upper stratosphere at pressures less than In combination reactions, there are often two possible 1 mbar (temperature 180 K and total atmospheric denwith product channels that result in the formation of species sities less than molecules cm 3 ), the use of these carbon numbers larger than those of either reactant. published expressions requires extrapolating far beyond The relative reaction rates leading to these two channels, the measurement range from which the expressions were simple combination versus insertion, exchange and trans- derived (at least in the case of temperature). Alternative fer, or cracking, reflect the efficiency in deexciting the expressions from the K laboratory data of Macsuch reactive intermediate complex before it decomposes. In pherson et al. (1985) are cases of competition between product channels, labo- k 0, e 576/T cm 6 sec 1 (9) ratory measurements often only establish the rate of forma- tion of the sum of the two channels and the relative rates k, e 137/T cm 3 sec 1. (10) of each channel. Thus, for the coupled reactions The rate coefficient expressions based on high-tempera- CH 3 C 2 H 3 C 3 H 5 H R137

17 JOVIAN HYDROCARBON PHOTOCHEMISTRY 17 are exchange reactions of one type or another. These are CH 3 C 2 H 3 M C 3 H 6 R138 important in determining the steady-state concentrations Tsang and Hampson (1986) report the values of the major stable disequilibrium species in the model: C 2 H 2, C 2 H 4, C 2 H 6, CH 3 C 2 H, CH 2 CCH 2, C 3 H 6, C 3 H 8, k 137 k, cm 3 sec 1 (13) C 4 H 2,C 4 H 4, 1-C 4 H 6, 1,2-C 4 H 6, 1,3-C 4 H 6,C 4 H 8,C 4 H 10, C 6 H 2, and C 8 H 2. Most of the column net production of k C 2 C 4 hydrocarbons is balanced by the loss of C 2 H 6 and 137 /k, T 8.52 e 1248/T. (14) other alkanes by transport across the tropopause. A small For fraction is lost to the production of C 5 or higher hydrocarbons and polyyne compounds. These are ultimately lost as C 2 H 3 C 2 H 5 CH 3 C 3 H 5 R170 aerosols transported to the troposphere. In the following discussion, NEB results are taken as representative of global hydrocarbon photochemistry at Jupiter. C 2 H 3 C 2 H 5 M C 4 H 8, R171 The penetration of solar radiation into the upper atmosphere of Jupiter depends strongly on wavelength. EUV Tsang and Hampson (1986) recommend photons (wavelengths 1100 Å), responsible for the dissociation of H 2 above the homopause, are absorbed at very k 170 k, cm 3 sec 1 (15) high altitudes in the model atmosphere, at pressures less k 170 /k, T e 3289/T. (16) than 10 4 mbar. Near the homopause in the upper strato- sphere (P 0.03 mbar) CH 4 is dissociated by photons having wavelengths 1500 Å, while in the lower strato- Laufer et al. (1983) suggested that both k 0,138 and k 0,171 sphere (P 30 mbar), C2 H 2,C 2 H 4, and polyynes are should be approximately 10 times k 0,135. For dissociated by photons having wavelengths 1500 Å. Figure 2 shows the total photoabsorption rate coefficients for H CH 3 C 2 H M CH 3 C 2 H 2 R94 the major species as a function of pressure and altitude in the standard model atmosphere. The radiation field for H CH wavelengths 2000 Å is somewhat unrealistic below the 3 C 2 H M C 3 H 5, R95 tropopause, since Rayleigh and aerosol scattering, and absorption Wagner and Zellner (1972a) provide the branching ratio by NH 3, have not been included in the model (the Rayleigh scattering optical depth at 2000 Å is about k,94 /k, e 554/T (17) 1 at the tropopause). Visconti (1981) performed some calculations of the effect of multiple scattering (due to while Whytock et al. (1976) provide both Rayleigh and aerosol scattering) on the UV radiation field in the jovian atmosphere. He found that the absorption k,94 k, e 1233/T cm 3 sec 1. (18) rate of NH 3 in the pressure range mbar was up to 10 larger when multiple scattering was included than when only attenuated solar radiation was considered In each of these cases, rate expressions for each product (at a solar zenith angle of 60 ), while at pressures 150 channel were deduced by simple algebraic manipulations. mbar the enhancement diminished to the order of 20% or less. Thus, we have confidence in our calculated photoabsorption PHOTOCHEMISTRY rates for the region of the atmosphere above the tropopause at 100 mbar. Hydrocarbon photochemistry in the upper atmosphere Finally, we note that although dissociation usually folof Jupiter is initiated by the photodissociation of CH 4. lows absorption of ultraviolet photons, some species, such The carbon-containing fragments of this process have very as C 2 H 2, have surprisingly low quantum yields for dissociashort chemical lifetimes and undergo a series of reactions tion. It is quite likely that the excited species (molecules that lead to either the recycling of CH 4 or the synthesis of such as C 2 H 2 that have absorbed a UV photon but have C 2 C 4 hydrocarbons. Another important set of reactions not dissociated) play an important (though currently unknown) that results in the synthesis of higher hydrocarbons are role in the chemistry of the upper atmosphere, the polyyne-forming reactions. Miscellaneous other bond- perhaps by facilitating some other, usually slow, reactions. forming reactions also exist. Carbon carbon bonds are primarily broken by photolysis, although cracking by C and C 2 Chemistry atomic hydrogen is also an effective way of breaking C C Nearly all previous studies of hydrocarbon photochemistry bonds. A large number of the reactions listed in Table IV on Jupiter have concentrated on the C and C 2 species