THE ASTROPHYSICAL JOURNAL, 515 : 868È872, 1999 April 20 ( 1999. The American Astronomical Society. All rights reserved. Printed in U.S.A.
DETECTION OF THE METHYL RADICAL ON NEPTUNE1 B. BE ZARD DESPA, Observatoire de Paris, FÈ92195 Meudon, France
P. N. ROMANI NASA/Goddard Space Flight Center, Code 693.2, Greenbelt, MD 20771
H. FEUCHTGRUBER Max-Planck Institut fur Extraterrestrische Physik, D-85748 Garching, Germany
AND T. ENCRENAZ DESPA, Observatoire de Paris, F-92195 Meudon, France Received 1998 September 30 ; accepted 1998 November 30
ABSTRACT We report the Ðrst detection of the methyl radical (CH ) in the upper atmosphere of Neptune. Obser3 vations with the Short-Wavelength Spectrometer of the Infrared Space Observatory (ISO) satellite at a resolving power of 2200 revealed several emission features from the l Q-branch of CH around 16.50 2 1.6`1.2 ] 1013 molecules 3 km. The column density of methyl radicals above the 0.2 mbar level is cm~2. ~0.9 Results are compared with predictions of photochemical models. The CH abundance is mostly sensitive 3 rate at low pressure. Using to the eddy mixing proÐle and to the poorly known methyl recombination Slagle et al.Ïs expression for this rate, the present observations imply an eddy mixing coefficient at the methane homopause in the range of 2È9 ] 106 cm2 s~1. Recombination rates higher than used in current photochemical models would lead to larger eddy mixing coefficients. Subject headings : infrared : solar system È planets and satellites : individual (Neptune) 1.
observational database for investigating the photochemistry of the giant planets.
Hydrocarbon photochemistry in the atmospheres of the giant planets is initiated by the photolysis of methane (CH ). Photodissociation produces CH, CH , and CH 4 (Mordaunt et al. 1993), mainly in the2 homopause3 radicals region. Ethane, the most abundant hydrocarbon after M methane, is formed by the reaction 2CH ] C H . 2 6 3 Acetylene and heavier hydrocarbons are produced through more complex reaction schemes (e.g., Romani et al. 1993). These compounds slowly di†use downward to the deep troposphere, where they are eventually recycled back to methane. Methane, ethane, and acetylene have been observed in NeptuneÏs stratosphere from ground-based and V oyager infrared spectroscopy for over 15 yr. In this paper we report the Ðrst detection of the methyl radical (CH ), a direct 3 product of methane photolysis. The l rovibrational band, 2 centered at 16.50 km, was detected in speciÐc observations conducted with the Short-Wavelength Spectrometer (SWS) of the Infrared Space Observatory (ISO). A subsequent search for CH in Saturn also proved successful ; results 3 presented in Bezard et al. (1998). have recently been Methyl is the Ðrst radical ever detected on the outer planets. Produced by the photodissociation of methane, it is thought to be the sole photochemical source of ethane. For the Ðrst time, ISO observations give us access to a key intermediate in the hydrocarbon photochemistry. They can help us to understand the formation of ethane, which constitutes, along with methane and acetylene, the bulk of the
Observations of Neptune were recorded on 1997 November 12 UT using the grating mode of the SWS (AOT 02) aboard the ISO. The spectrum covers the 16.44È16.56 km range at a resolving power of 2200. The total observing time was 64.5 minutes. Descriptions of the ISO satellite and the SWS spectrometer can be found in Kessler et al. (1996) and de Graauw et al. (1996), respectively. The aperture ([email protected]
@ ] [email protected]
@ at half-maximum signal) was centered on Neptune, whose equatorial diameter was 2A. 2. The subEarth latitude on Neptune was [30¡. The accuracy of the absolute Ñux scale is estimated to B20%, and the uncertainty on the continuum level, arising from dark current subtraction, was estimated to B0.4 Jy. Raw data were rebinned to a resolution of 4000 ; the rms noise level is then about 0.04 Jy. The observed spectrum is shown in Figure 1. The most obvious feature in the observed spectrum is a 2 Jy emission at 16.497 km, corresponding to the superposition of the Q(1, 1), Q(2, 2), Q(3, 3), and Q(4, 4) transitions from the l band of CH . Weaker lines due to the Q(7, 7) 2 Q(6, 6) (16.474 3 km), Q(5, 5) (16.486 km), Q(2, 1) (16.458 km), (16.531km), and Q(3, 2) (16.551 km) transitions are also clearly detected. 3.
RADIATIVE TRANSFER ANALYSIS
Synthetic spectra were generated from a line-by-line radiative transfer program that includes the H -He collision2 the l band induced absorption and molecular opacity from 2 used of CH . The line compilation and spectral parameters 3 to model CH opacity are described in Bezard et al. (1998). 3 The mean stratospheric molecular weight adopted in our modeling is 2.39, corresponding to a He mixing ratio of 0.19 (Conrath et al. 1991).
1 Based on observations with ISO, an ESA project with instruments funded by ESA Member States (especially the PI countries : France, Germany, the Netherlands, and the United Kingdom) with the participation of NASA and ISAS. The SWS instrument is a joint project of the SRON and the MPE.
DETECTION OF METHYL RADICAL ON NEPTUNE
occultation data (Yelle et al. 1993). We assign an uncertainty of ^50 K to the 0.1 kbar temperature, in line with the numerous stellar occultations reported by Roques et al. (1994). The composite proÐle, as well as extreme (““ cool ÏÏ and ““ warm ÏÏ) proÐles used in the error analysis, are shown in Figure 2. The CH emission is optically thin ; it is thus roughly 3 proportional to the CH density, but depends on the tem3 perature levels where methyl is present. Its mixing ratio is predicted to peak around the methane homopause, the level to which methane is well mixed in the atmosphere. The location of this level depends on the eddy mixing coefficient (K) proÐle. To derive constraints on the column abundance, we considered vertical proÐles derived from photochemical calculations (see ° 4). We tested distributions generated with an eddy mixing coefficient at the methane homopause (K ) h varying between 106 and 5 ] 107 cm2 s~1. This spread is based on the range of K \ 2È10 ] 106 cm2 s~1, derived in the 0.2 kbar region, i.e., slightly below the methane homopause, from the analysis of the UVS occultation light curves (Bishop, Romani, & Atreya 1998 ; Yelle et al. 1993). We considered two types of K proÐles. In the Ðrst class (type A), K is inversely proportional to the atmospheric number density to the [0.6 power (Romani et al. 1993). Figure 2 shows a subset of CH proÐles that were generated with this 3 type of proÐle. Distribution 1 is our nominal case (case A of FIG. 1.ÈObserved spectrum of the CH l band on Neptune and syn3 2 shown in Fig. 2. thetic spectra based on the atmospheric models
Spectral radiances were calculated for disk-averaged conditions and converted to Ñuxes using a value of 3.87 arcsec2 for NeptuneÏs solid angle. 3.1. Atmospheric Model The disk-averaged stratospheric temperature proÐle used in this analysis is based on the following observations. At pressures greater than D0.5 mbar, the proÐle is constrained by the ISO observations of the S(0) and S(1) quadrupolar H lines at 28.3 and 17.1 km (Feuchtgruber et al. 1999). It is 2 extended to 1 kbar using ISO observations of the l band of CH in the 7 to 8 km range and the analysis of 4the 1985 4 August 20 stellar occultation (Hubbard et al. 1987). The latter, rescaled to a molecular weight of 2.39, indicates a temperature of B142 K at the 0.39 mbar level. Weak methane lines in the P and R branches of the l band, 4 probing the 0.02È0.2 mbar range, allow a determination the CH mole fraction in this region (Bezard 1998). Emission from4 the peak of the Q branch at 7.66 km indicates a temperature of 170 ^ 3 K near 3 kbar. The complete analysis of the l methane band as observed by ISO will be presented in a 4forthcoming publication. Our temperature proÐle is quasi-isothermal (165È180 K) in the 0.3È20 kbar interval, and satisÐes constraints from V oyager ultraviolet spectrometer (UVS) measurements, which yield a mean temperature of 170È190 K between 0.3 and 50 kbar (Yelle et al. 1993). Above the 0.3 kbar level, temperature steeply increases to reach 250 K at 0.1 kbar, and higher, a thermospheric temperature of 550 K. This nominal proÐle yields an atmospheric density equal to 1.5 ] 1012 cm~3 at 710 km above the 1 bar level, in agreement with the determination of the H density at this altitude from the UVS 2
FIG. 2.ÈTemperature proÐle and CH vertical distributions used in the 3 analysis of the ISO spectrum. Dotted curves around the nominal temperature proÐle mark the ““ cool ÏÏ and ““ warm ÏÏ models representing lower and upper limits. Curves labeled 1È3 correspond to CH mixing-ratio 3 ; proÐles 1 proÐles generated from photochemical calculations (see text) and 3 assume K \ 1 ] 107 cm2 s~1, while proÐle 2 assumes K \ 5 ] 106 h cm2 s~1. ProÐlesh 1 and 2 were generated with the CH ÈCH combination 3 3 rate given by Slagle et al. (1988), while proÐle 3 is based on the formulation of Macpherson, Pilling, & Smith (1983).
BEŠZARD ET AL.
Romani et al. 1993), and corresponds to K \ 1 ] 107 cm2 h s~1, while proÐle 2 is based on a half lower value (K \ 5 h 1, ] 106 cm2 s~1). ProÐle 3 has the same K proÐle as case but includes a larger CH self-recombination rate (see ° 4). 3 The second class of K proÐles (type B) have a stagnant stratosphere (K \ 2 ] 103 cm2 s~1 at p [ 2 mbar), with a rapid transition to a region of rapid mixing (K D 5 ] 107 cm2 s~1 in the 0.5È0.001 mbar range), decreasing at higher altitudes, as described in Romani et al. (1993 ; case B) and Bishop et al. (1998). 3.2. Determination of the CH Column Density 3 To determine the CH abundance implied by the obser3 vations, we multiplied the trial proÐles by a constant factor that allowed us to match the intensity of the observed features. We found column densities for these rescaled proÐles in the range 1.1È2.0 ] 1013 cm~2 at the 0.2 mbar level (Table 1). The CH abundance below this level (less than 3 10% of the integrated column density) in the lower, colder stratosphere does not contribute signiÐcantly to the observed emission. Spectra generated with our nominal temperature proÐle and the CH photochemical distributions (1È3) are shown in Figure3 1. Most of the emission originates from the 0.1 to 3 kbar pressure range for proÐles 1 and 3, and from the 0.3 to 4 kbar pressure range using proÐle 2. The inferred column abundances are subject to uncertainties in the Ñux calibration of the ISO spectra (^20%) and in the strength of the l band (^30% ; Wormhoudt & 2 McCurdy 1989). Another error source derives from uncertainties in the thermal proÐle. Calculations with the warm and cool proÐles displayed in Figure 2 lead to the error bars listed in Table 1. Combining the various error bars, we conclude that the CH column density is in the range 0.7È2.8 ] 1013 mol3 cm~2. ecules 4.
COMPARISON WITH PHOTOCHEMICAL MODELS
Photochemical calculations were carried out with a onedimensional model most recently described in Bishop et al. (1998). Chemical reactions and kinetic rates included in the modeling are listed in Table 3 of Bishop, Romani, & Atreya, with minor updates since then, except for the CH recombination reaction discussed below. Photolysis rates3are calculated for disk-averaged conditions, and account for both solar irradiance and the Lya sky glow from the local inter-
stellar medium. Solar minimum conditions, representative of the time of the ISO observations, were used, because the chemical lifetime of CH radicals was found to be much 3 We used a CH mole fraction shorter than the solar cycle. 4 equal to 1.4 ] 10~3 in the lower stratosphere, as favored by observations of the l band of methane by ISO (Bezard 4 1998). The model incorporates a downward Ñux of atomic hydrogen at the upper boundary equal to 4 ] 107 cm~2 s~1, representing H production from solar EUV. Methyl radicals are formed from methane photolysis hv either directly via the CH ] CH ] H pathway, or indi4 3 rectly by the reaction of CH (a8 1A ) (another direct pho2 1 tolysis product) with H (Mordaunt et al. 1993 ; Bishop et 2 al. 1998). Peak production occurs near the methane homopause. Ethyl radicals (C H ) reacting with H atoms are an 2 5 additional signiÐcant source of methyl. The dominant methyl loss processes are the recombiM nation reaction to form ethane 2CH ] C H and H addi2 6 M 3 tion to reform methane CH ] H ] CH . Since these 4 3 reactions are three-body reactions, their kinetic rate coefficients increase with atmospheric pressure. There is also only one laboratory measurement of the CH ] CH reaction 3 rate in the temperature range relevant to 3the methane photolysis region on Neptune (150È300 K) (Walter et al. 1990), and none for the other reaction. The methyl abundance is thus most sensitive to (1) the altitude of methane photolysis, which depends primarily on the eddy di†usion coefficient proÐle, and (2) the adopted rate coefficients for the CH combination reactions. This is 3 similar to the situation on Saturn (see Bezard et al. 1998). It is less sensitive to other model parameters, such as the methane mixing ratio in the lower stratosphere, the downward Ñux of atomic H, or the assumed stratospheric temperature. For instance, varying the CH mixing ratio between 0.5 and 2.5 ] 10~3 leads to at most4a ^20% variation in the CH column abundance compared to our nominal case (1.43 ] 10~3). This is because methane photolysis is primarily a photon-limited process, the atmosphere remaining optically thick in the UV for the range of CH mixing ratios considered. In4 our ““ nominal ÏÏ model, the eddy mixing coefficient at the methane homopause (K ), located near 0.06 kbar, is 1 ] 107 cm2 s~1, decreasing ath deeper levels as n~0.6 (case A of Romani et al. 1993). The CH proÐle predicted by this 3 model (distribution 1) overestimates the observed CH emission by 50% (Fig. 1), slightly beyond our spectral mod-3
TABLE 1 RETRIEVAL OF THE CH COLUMN DENSITY 3 Photochemical Model Case A . . . . . . . . . . . . . . . . . . . K \ 1 ] 106 . . . . . . . . . h K \ 5 ] 106 (2) . . . . . . h K \ 1 ] 107 (1) . . . . . . h K \ 1 ] 107 (3) . . . . . . h K \ 5 ] 107 . . . . . . . . . h Case B . . . . . . . . . . . . . . . . . . .
Model CH Column Densitya 3
Rescaled CH Column Densitya,b 3
1.1 ] 1013 2.0 ] 1013 2.5 ] 1013 0.9 ] 1013 3.8 ] 1013 4.2 ] 1013
2.0 ] 1013 1.8 ] 1013 1.6 ] 1013 1.5 ] 1013 1.2 ] 1013 1.1 ] 1013
a In molecules cm~2, at the 0.2 mbar level. b Needed to reproduce the CH emission intensity. 3
Uncertainty from T ProÐle (%) ^7 ^15 ^19 ^21 ^30 ^35
No. 2, 1999
DETECTION OF METHYL RADICAL ON NEPTUNE
eling uncertainties. The CH emission may be reduced by 3 decreasing K , e.g., distribution 2 with K \ 5 ] 106 cm2 h h s~1. The homopause is then located at deeper levels (0.1 kbar), and the resulting increase in the rate of the CH 3 three-body loss reactions leads to a 25% decrease in the methyl column abundance and reduces the emission intensity by 35%. Acceptable values for K range from 2 to h 9 ] 106 cm2 s~1 considering all sources of uncertainty. This model, however, may be in conÑict with the UVS observations of CH . Analyzing the solar egress occultation 4 (applicable to the ISO observations), Bishop et al. (1998) determined a methane-mixing ratio in the lower stratosphere of the order of 1 ] 10~4 and an eddy mixing coefficient of D1.7 ] 107 cm2 s~1 at the 0.2 kbar level. At this level, our preferred range for K is signiÐcantly lower : 2.5È 4 ] 106 cm2 s~1. Modeling the UVS data with a larger CH mixing ratio, as used in the present analysis, will 4 reduce the derived eddy mixing coefficient, but may provide a worse Ðt to these observations. Another analysis of the same data set by Yelle et al. (1993) leads to K \ 2È3 ] 106 cm2 s~1 at 550 km (0.6 kbar), with values between 5 and 8 ] 106 cm2 s~1 at the 0.2 kbar pressure level, still slightly larger than our inferred range. Eddy mixing proÐles exhibiting a stagnant lower stratosphere surmounted by a broad region with vigorous mixing (K \ 5 ] 107 cm2 s~1 in the 1È500 kbar interval) are more difficult to reconcile with the ISO observations. The CH proÐle generated with the case B K proÐle of Romani et al.3 (1993) yields an emission feature 4 times too strong. To bring the methyl emission into agreement with the observations with this type of proÐle, it is necessary to restrict the region where vigorous mixing applies to pressures higher than D5 kbar. In the above models, the kinetic rate constant for the (M) reaction 2CH ] C H comes from Slagle et al. (1988) in 2 6 (k ), and from Hessler (1997) in the 3 the low-pressure regime high-pressure regime (k ).0 Note that Bishop et al. (1998) = used data from the compilation of Baulch et al. (1992). Recent papers (Hessler 1997 ; Robertson 1995) agree well on k in the temperature region of 200È300 K. Unfortunately, = model is most sensitive to the choice of k where there is the a wide range of values (compare Baulch et al.01992 ; Slagle et al. 1988 ; and Macpherson, Pilling, & Smith 1983). For example, at 180 K, the expression adopted by Slagle et al. (1988) leads to k \ 5.4 ] 10~26 cm6 s~1, while that of 0 is 2 times slower, and that from MacpBaulch et al. (1992) herson et al. (1983) is 12 times faster. Using k from Macpherson et al. (1983) and the nominal K proÐle, 0the photochemical model yields a CH proÐle 3 times (Fig. 2, distribution 3) with a column abundance 2.5 lower than that in distribution 1. The emission feature is now 40% smaller than observed (Fig. 1). We conclude that a value of k intermediate between the Slagle et al. (1988) and 0 et al. (1983) low-temperature rates would Macpherson allow us to reproduce the CH emission without modifying 3 that on Saturn, use of the the eddy mixing proÐle. Note Macpherson et al. rate allows to reproduce the CH emis3 s~1) sion with the ““ nominal ÏÏ value of K (6 ] 107 cm2 h (Bezard et al. 1998). Such a change in the CH ] CH rate would not be 3 3calculations with the sufficient to reconcile photochemical ISO observations when K proÐles exhibiting vigorous mixing in the 1-500 kbar region are used. More speciÐcally, the CH emission generated with K from case B of Romani 3
et al. (1993) is twice too large when the rates from Macpherson et al. (1983) are used for the CH recombination, still outside the modeling uncertainties. 3 The model methyl abundance is less sensitive to the (M) assumed rates for the CH ] H ] CH reaction. The 3 4 ““ nominal ÏÏ model includes the formulation of Brouard, Macpherson, & Pilling (1989) in the low-pressure regime and the value found by Seakins et al. (1997) for the highpressure regime. Moving to the expression given in the compilation of Baulch et al. (1992) only reduces the CH column 3 density and emission intensity by D25%.
ISO observations of the CH l band on Neptune imply a 3 2 CH column density equal to 1.6`1.2 ] 1013 molecules 3 ~0.9 cm~2 at the 0.2 mbar level. This is about half the abundance determined in SaturnÏs stratosphere (Bezard et al. 1998), although the error bars allow for a Saturn/Neptune abundance ratio between approximately 1 and 6. On the two planets, current photochemical models tend to overpredict the methyl abundance when a value for K is used based upon other hydrocarbon observations (e.g., hdistribution 1). This discrepancy could be due to the poorly known chemical rates of the CH -CH combination, which a†ect the loss 3 3and thus their column density. A rate of methyl radicals CH -CH combination rate of about 1.7 ] 10~25 cm6 s~1 3 K,33 times larger than extrapolated from the expresat 180 sion of Slagle et al. (1988), would be required to reconcile the nominal photochemical model (case A of Romani et al. 1993) with the Neptune CH observations. On the other hand, if the3 Slagle et al. (1988) formulation for the CH self-recombination is valid (or not too wrong) 3 at low temperatures, then the eddy di†usion coefficient proÐle likely needs to be reduced in the homopause region. The CH emission implies a homopause value of 4`5 ] 106 cm2 s~13when the chemistry is left unchanged. This~2value is somewhat smaller than the range derived by Bishop et al. (1998) from the analysis of the V oyager UVS solaroccultation data, which is based, however, on a smaller CH 4 abundance than used here. It might also happen that with improved laboratory measurements of the CH self recombination rate, the photochemical models fail3 to explain both the Saturn and Neptune observations. In that case, the problem in the models may well lie in the source of methyl radicals, methane photolysis, which is still not well understood (see Mordaunt et al. 1993 ; Heck, Zare, & Chandler 1996). As outlined by Bezard et al. (1998), observations of the CH radical are a diagnostic tool to constrain the location 3 homopause, i.e., the height to which methane is mixed of the in the atmospheres of the outer planets. This location, depending on the strength of the atmospheric eddy mixing at high altitudes, is difficult to determine from other observations. However, it requires a precise knowledge of the CH recombination rates. In addition, it is important to 3 understand the sources and sinks of CH , since it is the sole 3 source of C H , the main product of methane photolysis. 2 6 For the Ðrst time, ISO has given us insight into the formation of this key compound. In conclusion, we can only renew our call for laboratory measurements of the CH recombination rates at the low temperatures appropriate to3 the upper atmospheres of the giant planets (140È200 K).
BEŠZARD ET AL.
REFERENCES Baulch, D. L., et al. 1992, J. Phys. Chem. Ref. Data, 21, 411 Macpherson, M. T., Pilling, M. J., & Smith, M. J. C. 1983, Chem. Phys. Bezard, B. 1998, Ann. Geophys., 16 (Suppl. III), C1037 Lett., 94, 430 Bezard, B., Feuchtgruber, H., Moses, J. I., & Encrenaz, T. 1998, A&A, 334, Mordaunt, D. H., et al. 1993, J. Chem. Phys., 98, 2054 L41 Robertson, S. H., Pilling, M. J., Baulch, D. L., & Green, J. B. 1995, J. Phys. Bishop, J., Romani, P. N., & Atreya, S. K. 1998, Planet. Space Sci., 46, 1 Chem., 99, 13452 Brouard, M., Macpherson, M. T., & Pilling, M. J. 1989, J. Phys. Chem., 93, Romani, P. N., Bishop, J., Bezard, B., & Atreya, S. 1993, Icarus, 106, 442 4047 Roques, F., et al. 1994, A&A, 288, 985 Conrath, B. J., Gautier, D., Lindal, G. F., Samuelson, R. E., & Sha†er, W. Seakins, P. W., et al. 1997, J. Phys. Chem., 101, 9974 A. 1991, J. Geophys. Res., 96, 18907 Slagle, I. R., Gutman, D., Davies, J. W., & Pilling, M. J. 1988, J. Phys. de Graauw, Th., et al. 1996, A&A, 315, L49 Chem., 92, 2455 Feuchtgruber, H., Lellouch, E., Bezard, B., Encrenaz, Th., de Graauw, Th., Walter, H.-H., Grotheer, J. W., Davies, J. W., Pilling, M. J., & Wagner, A. & Davies, G. R. 1999, A&A, 341, L17 F. 1990, Twenty-Third Symp. (International) on Combustion Heck, A. J. R., Zare, R. N., & Chandler, D. W. 1996, J. Chem. Phys., 104, (Pittsburgh : Combustion Institute), 107 4019 Wormhoud, J., & McCurdy, K. E. 1989, Chem. Phys. Lett., 156, 47 Hessler, J. P. 1997, Int. J. Chem. Kinet., 29, 803 Yelle, R. V., Herbert, F., Sandel, B. R., Vervack, R. J., & Wentzel, T. M. Hubbard, W. B., et al. 1987, Icarus, 72, 635 1993, Icarus, 104, 38 Kessler, M. F., et al. 1996, A&A, 315, L27