DATA_SET_DESCRIPTION |
Data Set Overview
=================
Quoting from the Icarus paper:
'Samples were mixed at room temperature in [a 2 liter] volume using CH4
and N2 gases supplied in high pressure cylinders by Airgas Specialty
Gases. These were connected through separate pressure regulators and
valves to the mixing volume. The reported purities were >=99.99% for the
CH4 and >=99.9% for the N2. To these gases, we could add CH3D purchased in
a lecture bottle from Sigma Aldrich (product number 490237, with a
reported purity of >=98%). Unlike in water, proton-deuteron exchange is
negligible in methane at room temperature and below, so we did not need to
consider the formation of other isotopomers such as CH2D2 and CHD3 (Sigma
Aldrich Stable Isotope Department, personal communication 2010). After
mixing gases to the desired composition at room temperature in the mixing
volume, we opened a valve to allow the gas to flow into the empty, cold
cell, condensing it as a liquid. We froze this liquid by
reducing its temperature, maintaining a vertical thermal gradient of about
2 K across the 15 mm diameter sample by means of heaters (see Fig. 1), so
that it would freeze from the bottom upward, with the location of the
freezing front being controlled by the cell temperature. Each new mixture
requires some experimentation to find appropriate cooling rates, but rates
for freezing samples were typically in the range of -0.01 to -0.1 K/min.
It was sometimes necessary to freeze an initial polycrystalline mass and
then melt almost all of it to obtain a small seed crystal before
re-freezing slowly, in order to obtain an optical-quality ice sample.
After a suitable sample had been frozen, we removed the thermal gradient
by smoothly shifting temperature-control heating from the upper to the
lower heating element over a period of 10-20 min. Using only the
bottom heater for temperature control resulted in a thermal gradient
across the sample of just a few tenths of a Kelvin.
'Spectra were recorded with a Nicolet Nexus 670 Fourier transform infrared
(FTIR) spectrometer at a sampling interval of 0.24 cm-1, resulting in a
spectral resolution of 0.6 cm-1 (measured full width at half maximum of
unresolved lines). The spectrometer beam was focused to a few mm spot
inside the cell. We typically averaged over 100 spectral scans to improve
the signal/noise ratio. After a sample spectrum had been recorded, we
would ramp to a new temperature, at rates typically in the range of
0.1-0.5 K/min. We recorded spectra through our ice samples at every
multiple of 10 K between 40 K and the host ice melting points (90.7 K for
CH4 ice, 63.1 K for N2 ice).
'Before each ice sample was prepared we also recorded spectra through the
empty, cold cell, and did the same after each sample was eliminated.
Filled-cell spectra were divided by empty-cell spectra to remove gross
effects of lamp emission, detector sensitivity, and absorptions by the
windows and air, resulting in approximate transmission spectra T(k). These
spectra are affected by subtle, spurious slopes from a variety of sources.
Wavelength-dependent refractive index contrasts exist between ice and cell
windows leading to slightly different transmission through the ice-window
interface than through the vacuum-window interface. The room temperature
refractive index n(k) of sapphire decreases gradually with wavelength from
about 1.76 to 1.59 from 1.0 to 5.5 microns (Malitson et al., 1958;
Gervais, 1991) but methane ice shows almost no wavelength dependence in
its refractive index (except at wavelengths near 3.3 microns where
absorption is so strong that we measure no transmission whatsoever; Pearl
et al., 1991). If we knew the temperature dependent n(k) of both
sapphire and ice, and the ice-window interfaces were solely responsible
for these slopes, we could easily correct for the effect. But as the ice
and the surrounding cell contract on cooling, each with their own
distinct temperature-dependent coefficients of thermal expansion, the
sample is stressed and can fracture or pull away from the windows, opening
additional ice-vacuum interfaces that can produce wavelength-dependent
scattering that varies with temperature, with thermal history, and
with location within the sample. In addition to imparting slopes,
these effects lead to a decline in overall transmission, by as much as
a factor of three in an ice sample cooled relatively rapidly from 90 to
40 K. Slow drifts in lamp filament temperature or detector sensitivity
over the course of experiments lasting multiple days can also contribute
spurious slopes. Slopes arising from any combination of the above factors
were removed by fitting a line or low-order polynomial to continuum
regions adjacent to absorption bands to be quantified and dividing by
this function to 'straighten out' the continuum. The continuum-corrected
transmission spectra were then converted to Lambert absorption coefficient
spectra alpha(lambda) via the Beer-Lambert absorption law, rearranged
as alpha(lambda) = -ln(T(lambda))/d, where d is the path length through
the cell (d = 5.4 +- 0.1 mm for all experiments reported in this paper).
'We added a small amount of CH3D to CH4 to produce a deuterium-enriched
sample that remained dominated by normal CH4 absorptions. Addition of 0.5%
CH3D (D/H ratio of 1.25 x 10-4) produced ice having sufficiently strong
CH3D absorption to be easily measurable, but not so strong as to be
saturated in transmission though our 5.4 mm optical path length. We were
not able to directly measure the D/H ratio in our ice sample, so we
assumed zero fractionation between the gas phase and the liquid condensed
from it, and again zero fractionation between the liquid and the solid
crystallized from it.
'We subtracted the ordinary methane absorption coefficients to isolate
the contribution of the CH3D. The resulting CH3D ice absorption
coefficients were divided by the 0.5% concentration of the CH3D
(assuming the composition was unchanged by condensation and freezing)
to obtain effective absorption coefficients for CH3D in methane as
if it were pure CH3D. These values can be combined with ordinary CH4
ice absorption coefficients, scaled by their relative abundances, to
approximate absorption coefficients for arbitrary CH3D/CH4 mixing ratios.
'We performed an experiment with CH3D diluted in the hexagonal beta phase
of N2 ice. N2 ice melts at 63.1 K, so this experiment spanned a smaller
range of temperatures. Uncertainty over the CH3D concentration in the
ice presented even more of a challenge with this experiment. The CH3D/N2
ratio mixed in the gas phase in the mixing volume could not be expected
to remain unchanged in the ice, for two reasons. First, methane is much
less volatile than nitrogen. On condensing the N2-dominated gas into
the cell as a liquid at about 65 K, some of the deuterated methane could
have condensed as frost somewhere in the inlet tube rather than making
it into the cell. Second, compositional gradients appear on freezing,
as a result of the separation between liquidus and solidus curves of
the binary phase diagram of nitrogen and methane (e.g., Prokhvatilov
and Yantsevich, 1983; unfortunately, the two curves are difficult to
distinguish in their figure). Quirico and Schmitt (1997) assumed that
the integrated absorptions of CH4 bands remain unchanged on dilution
in N2 ice in order to estimate the composition of their samples. Making
the same assumption for CH3D, we used the integrated absorption of the
4.56 micron CH3D band to estimate the CH3D fraction in our mixed CH3D + N2
ice sample as 0.0024 +- 0.0002 (about a factor of two below its gas phase
abundance of 0.005). As before, we subtracted the absorption coefficients
of the host N2 ice, studied in an otherwise identical separate experiment
in our laboratory, to isolate the CH3D absorptions.
|