Reaction Modulation Spectroscopy
J. Phys. Chem., Vol. 100, No. 45, 1996 17857
to produce a peroxy radical. In the absence of NOx, this radical
will react with other peroxy radicals along several possible
pathways, often eventually producing a substituted alkoxy
radical. For large and highly substituted hydrocarbons these
alkoxy radicals may decompose into an aldehyde and an unstable
radical, which in turn may rapidly react with oxygen to produce
a second aldehyde and the hydroperoxy radical, HO2. If NOx
were present in sufficient quantity, the alkoxy radical or an
organic nitrate would be produced, and HO2 would be rapidly
converted to OH, closing a catalytic hydrocarbon oxidation
cycle. A parallel sequence, with OH addition to the terminal
carbon, yields the identical aldehydes. We therefore expect that,
under the proper conditions, we should be able to force the
system to produce one of each of the two aldehydes for every
C3F6 removed.
sections, since the sample volumes for all species are identical;
however, companion LIF and RF measurements can provide
vital additional information, especially for atoms, which of
course do not absorb in the IR, and OH, which has a very small
integrated cross section.
To exploit the steady state concentrations in the HPFS without
losing sensitivity because of base line drifts in the FTIR, we
modulate the concentration of the intiating radical by modulating
the flow of a radical precursor in the radical source. For
instance, in the most common source of OH radicals we
dissociate H2 in a microwave-induced plasma (MIP) and then
titrate the resulting hydrogen atoms with NO2, producing OH
and NO:
H2 f H + H
H + NO2 f OH + NO
(1)
Experimental Method
All the experiments described here were conducted on our
high-pressure flow system (HPFS). The system has been
extensively described in the literature,10,11 so we shall describe
only the capabilities specific to the RMS technique.
In this case, the OH is modulated by modulating the (tiny) flow
of H2.
Spectra are taken and coadded during each phase of the
radical modulation cycle. The two “offline” spectra surrounding
an “online” spectrum are then averaged to produce a background
spectrum, and the ratio of the two gives a transmittance spectrum
showing directly any changes in absorption caused by the
modulation of the radical. This is the most important aspect of
the RMS technique: we directly record the change in abundance
of all species, including the excess reagent. There is no spectral
subtraction to be performed; the spectra may be directly analyzed
to quantify the reaction. Any number of these transmittance
spectra may be averaged to increase the signal-to-noise ratio
(S/N); the minimum noise level observed to date, on a two-day
experiment, corresponded to an optical depth of roughly 10-6
at 4 cm-1 resolution.
To optimize the S/N for the several species being measured,
we often collect spectra at more than one resolution during each
cycle. Sampling resolutions are generally matched to the
coherence length of the resolvable portion of a given spectrum.
Where rotational transitions can be resolved, this is generally
the highest resolution of the instrument. Where the spacing is
too narrow or nonexistant, 8 cm-1 is often optimal.
Reaction modulation spectroscopy (RMS) is conceptually
straightforward. We exploit several attributes of the HPFS to
achieve very high sensitivity to the disappearance of reactants
and appearance of reaction products while guaranteeing con-
servation of material in the sampling volume. We observe the
radical plume with FTIR absorption spectroscopy in a White
cell located roughly 50 cm downstream of the radical source.
Reagents other than the radical (e.g., a hydrocarbon, O2, NO,
NO2) are well mixed in the carrier flow. The carrier gas and
reagents are not recirculated in this experiment; they pass once
through the reaction zone and are pumped out of the system.
The flow out of the radical source is a tiny fraction (∼1%) of
the carrier flow, so the reagents immediately mix into the radical
plume, though the plume itself spreads only slowly toward the
flow tube wall. In the White cell, the plume is a Gaussian of
roughly 2 cm half width. Thus, essentially none of the radical
has reached the wall at 6 cm radius.
The experiment is in a steady state; all flows are controlled
with mass flow controllers, and very little surface area requires
any conditioning. Practice has shown that the system will
remain stable for days; reagent concentrations maintained by
flow controllers drift by a few percent or less per day, while
radical concentrations typically drift by roughly 10% per day.
The White cell is placed so that it examines a volume element
located roughly 100 ms into the reaction sequence (for a 500
cm/s flow). We generally tailor the conditions so that the initial
reaction will go to completion in 10 ms or less, leaving reaction
products equal in abundance to the initial radical abundance.
The excess reagent concentration will have dropped by an
amount equal to the amount of radical removed, assuming that
no secondary reactions involve the initial reagent. However,
this will generally be only a small fraction of the total excess
reagent concentration.
Data Analysis
Data are collected with and without the initating radical. The
resulting spectra are then divided to give a transmittance, which
is in turn converted into an optical depth,
τν ) ln(Iν,off/Iν,on
)
(2)
where τν is the optical depth as a function of wavenumber and
Iν is the observed spectrum. Provided that either features in
the spectrum are well resolved by the spectrometer or that the
concentrations of all absorbing species are small enough for
the actual absorption of sharp features to be small (<20%), this
observed optical depth may be simply related to the absorption
cross sections of all constituents in the path of the spectrometer.
Our goal is to identify products of the initial reaction and
succeeding fast reactions and to quantify the branching of the
overall mechanism. The sample volume has not come into
contact with the flow tube wall, and very little diffusion has
occurred (and that has little or no dependence on mass under
most experimental conditions, where turbulent diffusion domi-
nates). We therefore expect the sample volume to contain
products of the gas phase reaction exactly equal to (within a
few percent or less) the quantity of reagents consumed.
Quantification of reactants and products observed with the FT
spectrometer requires knowledge of only their relative cross
τ ) Lσ C + N
ν
(3)
∑
ν
j,ν
j
j
where L is the absorption path length, σj,ν is the cross section
of species j as a function of wavenumber, Cj is the number
densitiy of species j, and Nν is a wavenumber dependent noise
term. We may freely multiply each side of eq 3 by an arbitrary
function. We choose σi,ν, the cross section of a species i, whose
column abundance we wish to determine: