Conclusions on PH3 in Saturn

By comparing the width of the observed PH₃ J=1-0 transition in Saturn with that modeled for various abundances using a radiative transfer code, we obtain an estimate of the saturnian PH₃ mole fraction of 3.0 $\pm$ 1.0 ppm in the pressure range from $\sim$ 1 bar to 100 mbar (assuming a constant mole fraction at pressures greater than a $\sim$ 100 mbar cutoff pressure). This abundance represents an enrichment of more than 5 times the solar value of (P/H) ${}_\odot = 2.7\times 10^{-7}$ (Anders and Grevesse 1989) and is larger than the saturnian PH₃ abundance derived from previous infrared measurements at these pressures. This abundance also represents a factor of $\sim 5$ enrichment over the mole fraction derived for Jupiter using the 3-2 line. While Jupiter's smaller phosphorus abundance might be partially explained by Jupiter's greater gas retention during its formation, it seems likely that other factors, such as the convective transport rate from the deep atmosphere, may play a role. The non-detection of a narrow stratospheric emission core at the line center of the J=1-0 PH₃ rotational transition in Saturn provides a lower limit on the PH₃ cutoff pressure in Saturn's stratosphere of at least 13 mbar, consistent with a low stratospheric PH₃ mixing ratio. The depth of the absorption line also places an upper limit of 140 mbar on the PH₃ cutoff pressure, with 100 mbar providing the best fit to our observations. Although we have also detected the PH₃ J=3-2 line in Saturn, the extreme width of the line combined with its unfortunate position relative to a strong terrestrial H₂O feature have thus far precluded a quantitative analysis.

A summary of saturnian PH₃ abundances inferred by various observers is given in Table . Most infrared observations of Saturn prior to 1984 were at wavelengths of 3 and 10 $\mu$ m. These observations sampled the saturnian atmosphere at pressures of $p \sim$ 400-700 mbar and obtained PH₃ abundances between 0.8 and 2.0 ppm (summarized in Prinn et al. 1984). However, measurements by Bézard et al. (1987) at 5 $\mu$ m--a more transparent region of the spectrum--suggested that the PH₃ concentration was significantly higher at the $\sim$ 4 bar level. Similarly, Noll and Larson (1990) reported a best fit to 3 and 10 $\mu$ m infrared observations--taking into account evidence for higher abundances at the pressures sampled by 5 $\mu$ m measurements--with a PH₃ mole fraction of 1 ppm for 78 mbar <p<400 mbar and 7⁺³_-2 ppm for p>400 mbar (where the transition pressure is that of the upper cloud in their two-cloud model). However, the model of Noll and Larson was not optimized for the upper troposphere (i.e., that portion of the atmosphere above the base of the NH₃ clouds at pressures $\leq$ 1.4 bar).

Table 6.5: Published Saturn PH₃ Abundances from infrared/millimeter observations.

PH₃ Mole	Pressures	$\lambda$	Instrument	Reference
Fraction (ppm)	Probed (mbar)	( $\mu$ m)
$\mathrel{\hbox{\hbox to 0pt{% \lower.5ex\hbox{$\sim$}\hss}\raise.4ex\hbox{$>$}}}400$ 0.8 ${}^\dagger$	400-700	10	CFWS	Gillet and Forrest 1974
>0.2	400-700	10	CGS	Bregman et al. 1975
0.9	500-1000	3	FTS	Larson et al. 1980
1.6	1000-3200	5	FTS	Larson et al. 1980
>0.8	400-700	10	CGS	Tokunaga et al. 1980, 1981
1.0	400-700	10	IRIS	Hanel et al. 1981b
1.3 $\pm$ 0.8	400-700	10	IRIS	Courtin et al. 1984
0.6^+0.6_-0.3	< 100	100	CGS	Haas et al. 1985, 1986
5.0 ${}^\ddagger$	4000	5	FTS	Bézard et al. 1987
3.0	1000-3200	5	FTS	Noll 1987
7.0^+3.0_-2.0	>400	5	FTS	Noll and Larson 1990
3.0 $\pm$ 1.0	100-1000	1100	FTS	Weisstein and Serabyn 1994

^*concentrations given relative to the hydrogen abundance have been converted assuming an H₂ mole fraction of 0.963

${}^\dagger$ tentative detection

${}^\ddagger$ this best current value (Bézard, pers. comm. 1994) is slightly higher than the published 4.0 ppm

CFWS: cooled filter-wheel spectrometer

CGS: cooled grating spectrometer

FTS: Fourier transform spectrometer

IRIS: Voyager infrared interferometer spectrometer (an FTS)

Infrared observations suffer from a lack of prominent PH₃ features, making it difficult to isolate PH₃ emission from that due to other continuum sources. An exception is the 1972 cm^-1 Q-branch observed at 5 $\mu$ m in Saturn by Bézard et al. (1987). Comparison with the laboratory measurements of Tarrago et al. (1992) allowed these authors to derive a PH₃ mole fraction of $\sim$ 5 ppm in the deep saturnian atmosphere. However, the analysis of 5 $\mu$ m observations of Saturn is complicated by their sensitivity to reflected solar flux from a few hundred millibars in addition to thermal radiation from several bars. Noll and Larson used micro-windows in the 2000-2160 cm^-1 region to determine the 7 ppm lower tropospheric PH₃ mole fraction for their model. Although Noll and Larson were able to fit portions of their spectrum (which contains nearly 2000 PH₃ lines) quite well, they had difficulty matching the entire spectrum, possibly as a result of the discontinuous vertical mole fraction profile they used. Finally, infrared abundance inversions are sensitive to assumptions about cloud heights and structure. All these factors conspire to make an unambiguous estimate of the true PH₃ mixing ratio from infrared measurements very difficult, although these measurements do give convincing evidence for a drop in PH₃ abundance with increasing altitude.

Millimeter measurements avoid the complications inherent in infrared work because they are sensitive only to thermal emission from the upper troposphere (pressures < 1.4 bar; see Fig. ) and are unaffected by reflected solar flux (which is negligible at millimeter wavelengths). The presumed ``upper cloud'' at 400 mbar (Tomasko et al. 1984) can also be ignored since its opacity is negligible at wavelengths longer than infrared. The constant PH₃ mole fraction of 3.0 $\pm$ 1.0 ppm which we derive is intermediate to the 3/10 $\mu$ m and 5 $\mu$ m-derived values discussed above. However, this is not due to our sampling of pressures intermediate to the two IR bands, since the central portion of the 267 GHz PH₃ line arises from higher in the atmosphere than do the IR lines. As a result, our measurements imply a PH₃ mole fraction in the upper troposphere of Saturn which is considerably larger than the previously published values.

**Figure 6.16:** Model Saturn 1300 um PH3 Spectra Compared with Noll and Larson. Dotted line: data from Fig. . Solid line: model spectrum for the mixing ratio given by Noll and Larson (1990), 1 ppm for 78 mbar 0<p< 400 mbar, 7 ppm for p> 400 mbar. Dashed line: model spectrum for a constant PH₃ mixing ratio of 3.0 ppm for p>100 mbar (central curve from Fig. ).
$\begin{figure} \begin{center} \BoxedEPSF{sat_ph3_1-0_nollfit.ps scaled 500}\\ \end{center}\end{figure}$

In order to make a more direct comparison of our observations with the two-step PH₃ abundance profile derived by Noll and Larson (1990), we also ran atmospheric models using the profile suggested by these authors. We found that a mole fraction of 1 ppm for 78 mbar < p < 400 mbar and 7 ppm for p>400 mbar resulted in a PH₃ line width somewhat too narrow and a line depth too small to be consistent with our observations (Figure ), as confirmed by a chi-squared test (see Table ). However, the difference from our best one-step model is not large. It should in principle be possible to model the PH₃ line with various abundance profiles and to use chi-squared testing to determine the profile matching our observations most closely. For example, we could use more complicated exponential profiles such as those adopted by Tokunaga et al. (1979), Tokunaga et al. (1980), Haas et al. (1985), or Griffith et al. (1992). Unfortunately, although the central portion of the observed PH₃ absorption line can be fit fairly cleanly, the remnant ripple evident in Fig. would make it difficult to distinguish small differences in the lineshape, implying that our data do not yet warrant a parameterization beyond the first-order one-step approximation. The simple one-step PH₃ model we use is already consistent with a convective origin for tropospheric PH₃ in Saturn, since convection should result in a well-mixed, nearly constant tropospheric PH₃ mixing ratio up to the level at which PH₃ is most effectively destroyed by photodissociation, making it adequate for our analysis. Until further progress is made in the analysis of the J=3-2 line near 800 GHz, or in reducing the Winston cone-induced ripple in the 1300 $\mu$ m filter, more detailed modeling does not appear warranted for Saturn.

Finally, we address the question of time dependence. The ``Great White Spot'' equatorial atmospheric disturbance that occurred in Saturn in late September of 1990 (Westphal et al. 1992) may have resulted in an increase in the transport of PH₃ to the atmospheric levels at which we observed it. Because we have no data prior to December 1992, we are unable to compare observations before and after the event in order to address this possibility. However, our 1300 $\mu$ m observations with the FTS indicate that no substantial change has occurred in the millimeter spectrum of Saturn between December 1992 and September 1995.