Design and validation of a temperature-controlled system for determining rate constants of the nitrate radical reaction with a variety of organic pollutants

Material Information

Design and validation of a temperature-controlled system for determining rate constants of the nitrate radical reaction with a variety of organic pollutants
Chapman, Robert Warren
Publication Date:
Physical Description:
ix, 100 leaves : illustrations ; 29 cm


Subjects / Keywords:
Nitrates ( lcsh )
Air -- Pollution ( lcsh )
Atmospheric temperature ( lcsh )
Air -- Pollution ( fast )
Atmospheric temperature ( fast )
Nitrates ( fast )
bibliography ( marcgt )
theses ( marcgt )
non-fiction ( marcgt )


Includes bibliographical references (leaves 96-100).
General Note:
Submitted in partial fulfillment of the requirements for the degree of Master of Science, Department of Chemistry.
Statement of Responsibility:
by Robert Warren Chapman.

Record Information

Source Institution:
University of Colorado Denver
Holding Location:
Auraria Library
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
17995213 ( OCLC )
LD1190.L46 1987m .C43 ( lcc )


This item has the following downloads:

Full Text
Robert Warren Chapman
B.S., Rensselaer Polytechnic Institute, 1973
A thesis submitted to the
Faculty of the Graduate School of the
University of Colorado in partial fulfillment
of the requirements for the degree of
Master of Science
Department of Chemistry

This thesis for the Master of Science degree by
Robert Warren Chapman
has been approved for the
Department of

Chapman, Robert Warren (M.S., Chemistry)
Design and Validation of a Temperature-controlled
System for Determining Rate Constants of the
Nitrate Radical Reaction With a Variety of
Organic Pollutants
Thesis directed by Associate Professor Larry G.
Present studies of the nitrate radical sink
for organic pollutants in the nighttime atmosphere
that produce such environmentally important products
as HNOgt PAN, and 2-aminonaphthalene, are limited
due to problems in experimental design as well as
uncertainties in reaction rate and mechanisms.
Previous systems have used a broad range of ambient
temperatures and intermittent data sampling. This
has resulted in flawed analysis for certain reactants,
rates, and modelling of nitrate radical equilibrium
reactions in the polluted urban atmosphere. This
is primarily due to inadequate sensitivity and
responsiveness to the temperature dependent dinitrogen
pentoxide equilibrium shifts that control the nitrate
radical concentrations.
The design and validation of a new tempera-
ture-controlled system was carried out to determine
nitrate radical, dinitrogen pentoxide, and ozone

i v
reaction rate constants with a variety of important
organic molecules that may be present in the tropo-
sphere. This comparatively inexpensive and versatile
system was used to confirm some ozone and nitrate
radical kinetics. Due to some unknown problems
encountered with the stability of the nitrate radical,
the system was not suitable for examining the kinetics
of most of the reactions of interest. With further
planned modifications and instruments now available,
it is expected that the system will produce useable
data on these kinetics in the near future.

This work would not and could not have been
completed without the expertise, hard work, and many
hours that two people put in to this effort. I will
be permanently indebted to Dr. Larry Anderson, my
advisor and friend, and my wife, Helen Chapman. The
graduate students in physical chemistry, Paul, Juan,
Sheryl, Chuck, and Mike, have proven many times
that a friend in a friend indeed.
This research is based upon work supported
by the National Science Foundation under Grants
ATM-8405394 and ATM-8521192.

1. Introduction ............................... 1
2. Inorganic Nitrogen Chemistry . ..............4
3. Organic Nitrogen Chemistry ................ 12
4. Kinetic Reaction Systems .................. 26
5. Instrumentation and Validation ............ 29
6. Sample Preparation ........................ 48
7. Modelling Results...........................54
8. Experimental Results ...................... 75
9. Future Considerations.......................85

1. Temperature Dependence of K^,.................10
2. Evaluations of at Different
Temperatures ................................ 10
3. Monoterpene Rate Constants With
Nitrate Radical ............................. 19
4. B-ulb Quantification..........................41
5. Simplified Nitrate Radical - Organic
Reaction Scheme ............................. 55
6. Reaction Scheme of Nitrate -
Formaldehyde System........................64-65
7. Ozone Oxygen Atom Comparative

1. Temperature controlled Nitrate
Radical Kinetic Study System ................ 30
2. Input System . . . .........................32
3. Bulb Calibration............................42
4. Naphthalene Detection by MS and TIC..........60

1. Decay of Ozone in 0g-CH20-N0 System .... 63
2. Various Ozone Decays..........................68
3. Ozone Auto-Decay..............................69
4. Trans-2-butene NOg Reaction.................76
5. Propene NO^ Reaction

The importance of the nitrate radical
has been established in destruction of atmospheric
organic substituents in the last decade. Studies
of the formation and kinetics of the radical from
nitrogen dioxide and ozone must include an equili-
brium with dinitrogen pentoxide which forms a
reservoir from which NO^ radical is regenerated
under steady-state conditions.
This I^O^ equilibrium has been extensively
studied and is quite temperature sensitive. There
is, as yet, no consensus for an expression for the
temperature dependence or even its value at many
temperatures in the ambient environment. The
dinitrogen pentoxide decomposes to nitrogen dioxide
and a nitrate radical which is highly reactive.
This nitrate radical controls the lifetime of many
organic compounds in the ambient atmosphere even
though it is only stable at night in the absence of

light. Nitrogen dioxide and dinitrogen pentoxide
are also reactive with certain compounds. This
reactivity has interfered with relative rate studies
involving polyaromatic hydrocarbons and monoterpenes.
Earlier studies were done at "room tempera-
ture" using nitrate radical generated by a series
of injections of dinitrogen pentoxide into large
chambers. Although this design ensured the reaction
of the generated nitrate radical, it made it imposs-
ible to follow the reaction through instrumentation
or modelling. This lack of knowledge has caused
problems in understanding the mechanism and kinetics
of many organic-nitrate reactions.
A novel system was needed that could
precisely control temperature as well as the
concentrations of reactants and follow these
concentrations throughout the reaction. Such a
system was built, tested, and validated for ozone
but similar nitrate radical reaction studies gave
spurious results. Analysis of the data generated
as a result of these reactions has led to proposals
for modification of this system. These modifications,

along with a new detector for the gas chromatograph
and the presence of a fourier transform infrared
spectrograph, should allow for better confirmation
and tracking of the substituents involved in the
reaction. This will permit a better understanding
of not only the rate constants but also the mecha-
nisms of the reactions. This understanding will
insure that the possible nitrogen dioxide and dini-
trogen pentoxide competitive reactions are not
mistaken for the actual nitrate radical kinetics.

Nitrogen oxides (NO^) are produced from
many and various sources. Even with the present
uncertainties with the figures for the amounts
1 2
generated and their sources , it can be seen
that anthropogenic and natural emissions are roughly
comparable. In a polluted atmosphere containing
nitrogen oxides, nitrogen monoxide and nitrogen
dioxide are considered to be the most important
species present in significant concentrations.
Nitrogen monoxide is generally assumed to
be approximately 90% of the N0x emitted although
this obviously depends on the combustion process
involved in the emission. The other major oxide
present in polluted urban air masses is nitrogen
dioxide, which can be produced either directly in
combustion or, more commonly, through oxidations of
the monoxide. Photodissociation of the dioxide can
regenerate the monoxide and ozone during daylight

primary combustion- N2+O2 -> 2N0 (1)
combustion exhaust- 2N0+C>2 -> 2N03 (2)
polluted atmosphere- NO+Og -> NO2+O2 (3)
during daylight -
NC>2 + photon (<430 nm) -> N0+0 (^D) (4)
0g + photon (280-350 nm or 450-700 nm)
-> 02 + 0 (5)
0+02+M -> 03+M (6)
0+H20 -> 2 OH (7)
OH + CO -> H + C02 (8)
H+02+M -> H02+M (9)
H02+N0 -> OH + N02 (10)
In a highly polluted air mass, containing
large amounts of nitrogen oxides and ozone, the
dioxide concentrations, can be up to 500 parts per
billion (ppb) while in clean southern hemisphere
maritime air masses, nitrogen dioxide concentrations
are less than 1 ppb In polluted air masses, the
further reaction of ozone with nitrogen dioxide to
form the nitrate radical can occur with concentrations
from <5-430 parts per trillion (ppt)^ ^ of the
o3 + no2 -> no3 + o2 (11)
The formation of N03 is of importance only
when NO concentrations are low due to a further

reaction of the monoxide with the nitrate radical.
NO + N03 -> 2N02 (12)
A concentration of only 5 ppb of the
monoxide gives a lifetime of only 0.3 seconds for
the nitrate radical. A brief review of reactions
2-12 shows that though there are many different
pathways and intermediates, one can represent these
reactions in an abbreviated fashion as:
02 + N02 + light <=> NO + 0^ (13)
This reaction may be considered to be at steady
state at any particular moment since the equilibrium
is established quickly. Reaction 13 ignores oxygen,
water, various radicals, and other intermediates
whose concentration is very small or whose reactions
are quite slow.
As stated previously (equation 11), the
nitrate radical may be found at significant concen-
trations only in the absence of NO. In the presence
of light, the radical rapidly dissociates back to
the dioxide and triplet oxygen^.
N03 + light -> N02 + 0 (14)
In the absence of both light and NO, or in
other words, at night in a polluted urban air mass,
the nitrate radical is comparatively stable. It can

therefore, undergo further reactions leading to more
interesting and noxious compounds.
The inorganic nitrate radical reaction of
foremost interest in this study is with nitrogen
dioxide to form dinitrogen pentoxide. This is the
anhydride of nitric acid so the further reaction
with water to form the acid is evident^. It also
dissociates readily to reform the radical and the
M + N03 + N02 < => N205 + M (15)
N25 + H2 -> 2 hno3 (16)
N03 + N02 -> NO + no2 + o2 (17)
It is important to stress the major effect
light has on the lifetimes and thus the presence of
the various nitrogen oxides. Nitrogen monoxide will
have negligible dissociation due to radiation under
tropospheric conditions. The dioxide, on the other
hand, irradiated at wavelengths less than 420 nm
(equation 4), dissociates to the monoxide and triplet
atomic oxygen. This atomic oxygen is the only
significant anthropogenic source for ozone in the
troposphere. The photon flux in the atmosphere
depends on such things as time of day, cloud cover,
latitude, reflectivity of the surface (albedo), and

season. This leads to high variability and low
consistency in the rates of the flux at different
wavelengths. Measured ambient rate constants arising
-3 -1
from this photon flux run from 3-9 x 10 s giving
lifetimes for nitrogen dioxide from 2-6 minutes
under varying conditions of clear skies to overcast
8 10
and rain at temperate latitudes It is apparent
with such a short lifetime for nitrogen dioxide,
formation of the nitrate radical and dinitrogen
pentoxide can only be expected during nighttime
hours .
Equation 15 provides another steady state
involving ozone, the dioxide, the radical, and the
pentoxide. Dinitrogen pentoxide has been calculated
to exist at levels of up to 15 ppb from measured
dioxide and nitrate radical concentrations using the
equilibrium constant.
In the presence of ozone at 0.08 ppm, the
rate constant for equation 11^:
= 3.2 x 10 ^cm^ molecule ^s ^ at 298K;
gives an NO2 lifetime of 4.4 hours. The nitrate
radical formed is then free to further react with
yet another nitrogen dioxide (equation 15). At the
upper limit of observed ambient radical concentrations

of 430 ppt, this would give an NO2 lifetime of
approximately 1 minute based on k
-12 3 -1 -1
1.3 x 10 cm molecule s
More important than the initial rate of
dinitrogen pentoxide formation is the equilibrium
constant (K^,-) involved among the three reactants
involved. Although a considerable amount of work
has been done on the pentoxide equilibrium, there
is no good consensus in the values reported^ ^
(part of this problem is certainly due to the
situation that variations of IK change the equili-
brium by approximately 10% at temperatures near
298K See Tables 1 and 2). Some of the discrepancy
in the quantifications of the equilibrium would be
resolved if Sander's larger absorption cross section
is correct for NO,
The Burrows value in Table 2
_113 _1
would then be corrected from 2.2 to 2.8x10 cm mol
1 1 1 Q 1
and Kircher's value of 3.26x10 cm mol could
then be used with a smaller caveat for error as long
as the equilibrium being studied is at 298K.
Due to this equilibrium, reaction rates of
nitrate radical undergoing thermal decomposition or
reaction with water are difficult to determine. At
298K, Cantrell et al. proposed a rate for the
nitrate radical thermal decomposition:

1.07xl0-27eI1'I94/T 1978 Graham & Johnston
1.33xlO27(T/300)32en ,080/T 1982 O) Malko & Troe
9.39*10-28en-350/T 1984 (2) Kircher et b1.
8.13xl0-29e11,960/T 1985 (3) Burrows et al.
Evaluations of Kj^ at different temperatures/(xlO
3 ,-l)
cm mol
GROUP 273 297 298 299 % difference (297-299)
Graham & Johnston 68.7 2.50 2.20 1.94 28.9%
(1) Malko & Troe 54.6 2.11 1.86 1.65 27.9%
Kircher et al. 107 3.71 3.26 2.87 29.3%
Burrows et al. 86.4 2.51 2.19 1.91 31.4%
(4) % deviation from
high to low 962 76% 75% 74%
average 77 2.7 2.4 2.1 29%
(1) Graham, R.A. and H.S. Johnston, "The Photochemistry
of NO, and the Kinetics of the N90,-0, System,"
J. Phvs. Cnem.. £2, 254 (1978) 3
(2) Malko, M.W. and J. Troe, "Analysis of the Uni-
molecular Reaction N,0,+M< >N02+N0,-fl ,"
Int. J. Chem. Kinet.. 14, 399 (1982;
(3) Kircher, C.C., J.J. Margitan, and S.P. Sender,
"Pressure and Temperature Dependence of the
Reaction N0,+N0,4M > N,0,.+M," J. Phys. Chem.,
88, 4370 (1984)-3
(4) Burrows, J.P., G.S. Tyndall, and G.K. Moortgat,
"A Study of the ^2^5 EQuilibriuni Between 275
and 315 K end Determination of the Heat of
Formation of NO,." Chem. Phvs. Lett.. 119.
193 (1985) 3 ------------------

N03 > N0+02
For this equation according to Cantrell et al.
-3 -1 19
k^g = 2x10 s while Atkinson has observed an
order of magnitude slower decomposition (1.5x10 ^s-'*')
for pentoxide equilibrium thermal decomposition
which seems to be contradictory as dinitrogen
pentoxide's stability depends upon the nitrate
radicals stability. Nitrate radical has a lifetime
of less than 1 minute when the humidity is over 50% ,
yet this could easily be due to reaction of the pent-
oxide as an anhydride since the equilibrium between
nitrate and the pentoxide shifts so rapidly.
Although some of these discrepancies can be explained
using competitive side reactions the variability,
instability, and low concentrations of these inter-
mediates make for an exceedingly complex and expensive
research area for years to come. Further discussion
of the controversy as it applies for rates in
organic reactions will be found in Chapter 3.

6 5
As mentioned before, Atkinsonu and Platt
have measured nitrate radical concentrations of up
to 430 ppt in ambient nighttime northern hemisphere
continerital air masses. These concentrations show
the importance of the nitrate radical, and possibly
of dinitrogen pentoxide, as a reactant in laboratory
and ambient air systems involving nitrogen oxides,
ozone, and organic chemicals.
The investigation of the nitrate radical's
role in atmospheric chemistry has increased dramatic-
ally within the last ten years. The key role of
this radical in, the nighttime reactions with organics
and the production of nitric acid in polluted air
21 22
was established by' Niki and others . These
processes lead to part of the observed decrease of
pH in urban air pollution (acid rain) the increasing
amounts of eye irritants (e.g., PAN), and the loss
of nitrogen oxides from the nighttime air masses.

Removal of organics from the atmosphere can
proceed in three ways: deposition, photolysis, and
chemical degradation. The latter two processes lead
to increases in the amount of carbon dioxide and
other oxidation products. Since most organics that
are of interest in the vapor phase in the atmosphere,
such as the monoterpenes and the various gasoline
fractions, are small and hydrophobic, both wet and
dry deposition can be expected to be quite small.
Photolysis in the troposphere is also negligible for
stable organics due to most higher energy photons
having already been absorbed at higher altitudes.
Of the different forms of chemical attack, free
radical initiation should be the most effective on
otherwise stable molecules. Since the ratio of the
concentrations of nighttime nitrate versus daytime
hydroxyl radicals is approximately a thousand to one
(although this varies widely depending on conditions),
anytime the rate constant of nitrate reaction with
a chemical species- is 1/1,000 the rate constant of
hydroxyl activity or more, it can be inferred that
the nitrate reaction will be a major removal process
for that substituent from the polluted atmosphere.
The reaction of nitrate radicals are an
important, if not dominant, removal process of

monoterpenes leading to the comparatively low mono-
terpene concentrations found in the early morning
versus the increasing daytime levels This mecha-
nism helps to explain the cause of the serious dis-
crepancies between emitted and observed concentrations
of isoprene and other monoterpenes if the consumption
of such species were due only to reactions with ozone,
hydroxyl radicals, or photolysis. In cities such
as Denver with prevailing winds sweeping over adjacent
pine forests, the lack of a pleasant pine scent during
the morning rush hour and the presence of eye-irritat-
ing smog must, in part, be attributed to nitrate
radicals. (Of course, if acid rain were to become
bad enough to destroy the forests, the consequent
lack of atmospheric terpenes would decrease some of
Denver's smog irritation.) The nitrate radical may
also be the reactant species surmised by Andreae and
Raemdonck in an unknown removal process of dimethyl
sulfide from marine air masses especially in the
Gulf of Mexico and Atlantic Ocean.
Winer, Atkinson, and Pitts have determined
rate constants for gas-phase reactions of nitrate
radical with dimethyl sulfide and certain monoterpenes
at room temperature. From these data, they calculated

atmospheric lifetimes that, when compared to OH
radicals and ozone reactions, are the dominant loss
pathway for these organic chemicals in the atmosphere.
Surprisingly, the dominance of nitrate
radical loss rates holds true even under relatively
"clean" conditions (30 ppb ozone, 0.04 ppt OH, 10 ppt
nitrate). In moderately polluted atmospheres (200 ppb
ozone, 0.16 ppt OH, and 100 ppt nitrate), the life-
times become extremely short, on the order of 1-5
minutes. This is ten times shorter than that for
the reaction with ozone and approximately thirty times
less than the lifetime due solely to the reaction
with hydroxyl radicals. The study goes on to show
that nitrate radicals reduce monoterpene concentra-
tions just before sunrise by a factor of somewhere
between 100-150 at typical values for water vapor,
ozone and temperature. The modelling for the study
does not take into account nitrate scavenging side-
reactions with such chemicals as hydroxy-substituted
aromatics and tri-' and tetra- substituted alkenes
such as propylene glycol-1,2-dinitrate. Nighttime
concentrations of these reactive anthropogenic
organic compounds are not believed to be large
enough to lead to a significant pathway.

In view of the wide range of lifetimes for
the organics in these calculations, (>20 days to 0.9
minutes), it is somewhat surprising that these highly
temperature-sensitive reaction conditions were
studied only at "room temperature". Dinitrogen
pentoxide concentrations varied from 0.1 to 5 ppm,
nitrogen dioxide from 1-4 ppm. The monoterpenes were
approximately 1 ppm corresponding to the widely
varying emission rates from 15 to approximately
-2 -1
7700 micrograms m hours which would reflect
vegetation ranging from high forest to low desert
emission areas.
A further report by this group remedies
the temperature inadequacies by using a relative
rate technique based on the known rate for trans-2-
butene. This technique monitors the relative decay
rates of the concentrations of the alkene being
studied versus that of at least one related organic
whose nitrate radical reaction rate is reliably
known. This assumes either no, or extremely slow,
reaction of these organics with other species present
in the chamber such as ozone, nitrogen dioxide,
and dinitrogen pentoxide. However, the rate of
nitrogen dioxide reacting with conjugated dialkenes

26 27
is non-negligible . Correction factors of
between <1 20% of the total rate had to be made.
This correction factor was done again using the
"room temperature" rate constants. The range of
uncertainty in the temperature (1-2 K) can cause an
uncertainty in the equilibrium expression of almost
The fact that the plots obtained from the
comparative rate constants are excellent straight
line plots of ln([alkene]t /[alkene] ) versus
In([trans-2-butene] /[trans-2-butene]t) is neither
surprising nor "strongly supportive" since ln-ln
plots show slight deviation from straight lines as
long as the factors are related in some manner.
These plots, contrary to the authors, should not be
used to rule out reactions of alkenes with the dinitro-
gen pentoxide present in their system.
An example of this error^^, in the case
of the reaction with naphthalene, caused this group
to falsely attribute a reaction to nitrate radical
that actually was the first reaction studied where
polyaromatic hydrocarbons (PAH) were nitrated
preferentially by the pentoxide rather than the
nitrate radical The 2-nitronaphthalene formed

(8% yield) is metabolically reduced to 2-amino-
naphthalene a human carcinogen and thus, since
naphthalene is the most prevalent PAH in the ambient
atmosphere, poses a considerable health concern.
The chemistry of PAH and dinitrogen pentoxide in air
is quite different than the same reaction in either
polar or non-polar liquid solvents . The gas-
phase reactions clearly may have major importance
in understanding the high mutagenicity of smokes and
particulates in air pollution.
The reactions of monoterpenes with nitrogen
oxides are not as controversial and the rates of
reaction relative to one another have lead to some
interesting preliminary research. Of note here,
beta-pinene has a rate constant in the reaction with
-12 3 -1-1
nitrate radical of 1.4x10 cm molecule sec
which is only slightly lower than the other mono-
terpenes studied (alpha-pinene, delta-3-carene and
d-limonene) and significantly higher than the rate
-13 3 -1 -1
of 2x10 cm molecule sec expected on the basis
of isobutene (See Table 3). This rate, therefore,
cannot be easily explained based solely on the degree
of substitution or the nature of the substituents
around the double bond. The mechanism and products

of these monoterpene-nitrate radical reactions have
not yet been studied.
COMPOUND , 3 ..-1 -1. (cm mol s ) STRUCTURE
Isoprene v58
Alpha-Pinene 6.1
Beta-Pinene 2.5
Delta-3-Carene 10.6
d-Limonene 13.9 K>
1. Finlayson-Pitts, B.J. and J.N. Pitts, Atmospheric
Chemistry: Fundamentals and Experimental
Techniques, Wiley-Interscience, New York,
1986, p. 986
The products of the nitrogen oxides reacting
with a simple alkene have been studied by Akimoto
et al. using long-path FTIR. The propene-dinitrogen
pentoxide system produced nitroalkyl nitrates,
peroxyacetyl nitrates (PAN), nitroolefins, 1,2-
propanediol dinitrate (known as PGDN propylene
glycol-1,2-dinitrate), nitroxyperoxypropyl nitrate

(NPPN) and nitroxypropy1 nitrite (NPN-found only in
the absence of oxygen gas and thus not significant
in ambient air). The kinetic behavior of the latter
products suggests that they may be produced by the
nitrate radical initially adding to the propene.
The primary reaction of nitrate radicals
with organics is presumed to be an initial hydrogen
abstraction to form the product, nitric acid, and an
organic radical.
N03+RH >HN03+R (19)
34 38
However, in the case of propene, Akimoto et al.
have proposed the following initial reactions:
N0,+C,H, >CHCH-CH0N0o 3 3 6 3 2 2 (20a)
no3+c3h6>ch3ch(ono2)ch2 (20b)
These nitropropyl radicals under atmospheric condi-
tions will react with oxygen to form alkylperoxy
radicals which then oxidize nitrogen monoxide (nitric
oxide) to the dioxide and become alkoxy radicals.
In the high nitrate radical nitrogen dioxide
environment of the'artificial systems, several
heavily nitrated products would instead be expected.
Ultimate products formed would include carbon dioxide,
carbon monoxide, formaldehyde, formic acid, acetal-
dehyde, nitric acid, methanol, and water.

A similar product and pathway study for an
aldehyde-dinitrogen pentoxide system has not yet been
published. Finlayson-Pitts believes, analogous to
the hydroxyl radical-aldehyde reaction, that this
process involves preliminary hydrogen abstraction
from the aldehyde. In any case, the reaction would
be quite slow under ambient conditions since formal-
dehyde has a lifetime of weeks when reacting with
nitrate radical versus hours for the photolysis
reaction and a day or so for hydroxyl radicals.
Cantrell et al. performed the first major
kinetic study to measure product formation rates
along with reactant removal. They found significant
error when dinitrogen pentoxide decay rates were
assumed to be independent of the generation of
secondary radicals, as is typically done in modelling
studies. These radicals might react.with both the
organic molecule (in this case, formaldehyde) and the
nitrate radical, the former increasing the perceived
rate and the latter decreasing the rate.
In addition to the previous discussion (Chap-
ter 1) of a simplified reaction scheme using equations
1-17 involving inorganic nitrogen oxides, the addition
of formaldehyde (the most important oxidation product

of the organic reactions in the atmosphere) involves
countless more possible reactions. Cantrell et al.
used a total of 36 elementary reactions for their
computer modelling of this system. Their direct
experimental determination of the formaldehyde-nitrate
reaction rate constant was done spectroscopically
by following the time dependent concentration of all
the reactants and major products using in situ long
path FTIR and visible spectroscopy (DOAS). Their
rate for reaction (21) was 6.3x10 cm"3 molecule sec
N0^ + HCHO ----> Products (21)
3 6
Finlayson-Pitts disputes this in regard to
Cantrell's choice of and, using that portion of
Cantrell's data not dependent on the pentoxide equi-
librium, finds a value of 5.6x10 ^cm\olecule ^sec ^^
Absolute rate constants for certain reac-
tions of nitrate radical were determined by avoiding
the pentoxide formation altogether and producing
nitrate radical by reacting nitric acid with fluorine
atoms in fast systems.
HN03 + F ----> HF + N03 (22)
The concentration of these nitrate radicals are not
dependent on the kinetics of K.^. The radicals were
then reacted with excess trans-2-butene and isobutene.

Comparison of this rate with that obtained from
Atkinson's research gives a result compatible with
-11 3 -1
K^,. = 3.35 x 10 cm molecule Ravishankara
and Mauldin say that in the Riverside system, the
that Atkinson uses is "the most suitable value."
They further state that "it is not suitable to use
our directly measured value of:
[NOg + trans-2-butene ---> products];
with the results of Atkinson et al. to extract the
value" of K^,.. Ravishankara and Mauldin's caution
is advisable since they modeled their system using
the same reaction scheme as the Riverside group as
well as using Atkinson's values for the wall loss
for dinitrogen pentoxide for a completely different
type of system than their own.
Thus, a more fundamental problem than finding
any particular rate constant lies in the control and
definite determination of the K^,- equilibrium. For
example, the Riverside group have performed their
kinetic studies within a temperature range of 2 C
(298 1 K). A variety of groups have studied the
temperature dependence of this equilibrium (see Table
2). Evaluation of (Table 3) shows that much, if
not all, the difference in rates for a particular

reaction, such, as trans-2-butene, may be attributed
to the variance allowed in the temperature during
these reactions. Notice, for example, Cantrell and
Pitts groups differ by less than 10% for the reaction
with formaldehyde. Also, unfortunately, the evalua-
tions of the nitrate radical concentration vary by
a factor of almost two at any particular temperature
according to which evaluation of is chosen. For
the purposes of this research, an average value of
the four methods listed in Table 2 evaluated at the
appropriate temperature will be used on K^,..
A full analysis of the simple reactions with
propene or formaldehyde is not feasible. Cantrell
et al. used over fifty different reactions, of which
36 proved necessary to describe their chemistry,
four of whose rate constants were unknown and whose
products and/or reactants could not be observed .
The computer modelling of the system can be made to
fit the results since there exist sufficient degrees
of freedom. The results can be compared to theoretic-
ally predicted values as an independent check but
these types of calculation cannot be used as proofs
of the veracity of any particular equation or rate.
They were able to show that a simpler traditional

set of five chemical reactions was insufficient to
properly describe the chemistry. The chemistry of
the nitrate-propene reaction must be much more complex
since there are more carbons in propene and since
formaldehyde itself would be an expected product of
the propene system and able to undergo further
reactions. The competitive reactions involved in
the nitrate attack on propene versus formaldehyde
and their subsequent products must be fascinating
at the elementary reaction level but are still too
vast and uncertain to be handled by modern computa-
tional methods without some form of simplification,
due to the large number of competitive reactions and
numerous transient species, many of which have not
yet been studied.

There are three ways that reactions like the
ones previously discussed can be observed quantita-
tively. The first involves direct observation of
the reactant, and hopefully, product concentrations
using Differential Optical Absorption Spectrometry
(DOAS) or Fourier Transform Infrared Spectroscopy
(FTIR). DOAS observes vibrational fine structure
on electronic transitions in the visible/ultraviolet
region while FTIR observes the fundamental vibrational
structures for the more numerous molecules absorbing
in the infrared region. DOAS is more sensitive
since absorption coefficients are higher at greater
frequencies and so this is the preferred technique
for ambient measurements of such species as nitrate
radical and nitrous acid in the atmosphere. Due to
the long path length necessary for these extremely
dilute species, these techniques are good only for
large (room-sized or exterior) air masses and thus
long reaction times (on the order of hours) are
needed for diffusion and stabilization to take place.

Another technique available is the fast flow
system. Reactant gases are mixed at one end of a
long low pressure continuously evacuated tube,
activated by light or a microwave resonance cavity,
and products are analyzed at the low pressure end of
the tube. Numerous detection methods including
chemiluminescence and mass spectrometry can be used
for detection. By changing the flow characteristics
in the tube, one can measure reaction rates that are
exceedingly fast or that involve species that have
a very short lifetime. The procedure will not work
where reactions between fairly stable ions or mole-
cules are necessary or where the reaction needs to
take place at, or near, atmospheric conditions of
pressure, trace chemicals, photo-instability, etc.
Flash photolysis (FP) is related in the technique
of activation and detection by induced fluorescence
but, since there is no flow, higher pressures may be
examined. Again, only certain reactions can be
studied due to problems with photolysis of extraneous
species and for FP there are no alternate detection
systems readily useable other than photometric.
The third main technique used in analysis
of atmospheric reactions is the one used here involving

small-scale reaction chambers made of relatively inert
materials such as one of the fluorocarbon polymers.
Although they have a larger surface to volume ratio
than the larger DOAS-FTIR chambers, their smaller
volume assures better mixing of reactants at shorter
reaction times and enhanced control of temperature.
In order to better determine the actual
kinetics of the nitrate radical, it is necessary to
develop a small reaction chamber whose temperature
can be accurately controlled that will imitate the
atmosphere in all important characteristics. In the
case under study here, the critical attribute, due
to the K^,- uncertainty, is the ability to achieve
a temperature uniform to within 0.1 K in a range of
295-310 K. The most unrealistic condition in the
system will be the surface to volume ratio, which is
predominantly controlled by the small diameter tubing
used in the circulation to the gas chromatograph.

The new kinetic system being designed and
validated in this study (see Figure 1) is to be used
for nighttime nitrate radical studies and consists
of three parts: chemical inputs, the environmental
chamber, and analytical measurements of the reaction.
The system, as built, allows fairly precise concentra
tion control of a wide variety of chemicals to be
introduced in the ppm-ppb range. Its temperature
control parameters provide the possibility of future
experiments in a wide variety of applications in
atmospheric chemistry. Improved gas chromatographic
methods involving a photo-ionization detector, rather
than the present flame ionization device, should
improve sensitivities enough to allow a wide range
of experiments involving polyaromatic hydrocarbons.
GASES Chemicals to be input include .
^2 3* ^2 anc^ various organics. Oxygen and nitro-
gen are introduced into the environmental chamber


through a pair- of Matheson 306 flowmeters. These
rates are monitored for a given amount of time usually
measured by a stopwatch to give a measured amount
of gas. For preliminary experiments involving
standardizations of the gas chromatograph's operating
parameters, a bypass allowing faster, but unmonitored,
input of nitrogen to the chamber was also provided.
A High Purity Nitrogen tank from Linde- was fitted
with an Oxy-trap for the gas chromatograph while
industrial grade nitrogen and oxygen was used for the
reaction system. (See Figure 2.)
Ozone was produced by diverting the industrial-
grade oxygen after the Matheson gauge through an
ozonizer operating with a known efficiency. This
was again done for a measured amount of time. From
the knowledge of the efficiency of the ozonizer and
the amount of oxygen flow, the amount of ozone intro-
duced into the chamber is known. Since the ozonizer
is not completely efficient in converting oxygen to
ozone, it is important to include the predominating
amount of oxygen gas when determining the composition
of the atmosphere in the environmental chamber. The
ozone is flushed into the system with enough oxygen
to make up 20% of the final contents on a molar basis.


Pure organic reactants, as well as nitrogen
dioxide, were diluted prior to use, either commer-
cially or at the laboratory, to approximately 0.1%
with nitrogen. The resultant mixture was then
released through an evacuated tube into a one liter
bulb also under vacuum, whose pressure can be moni-
tored. When sufficient gas has accumulated in the
bulb, the bulb is sealed using teflon valves. Since
volume, temperature, and pressure are known, the
amount of organic, or nitrogen dioxide, sealed within
the bulb is known. This is then flushed into the
environmental chamber by a known amount of nitrogen.
Once in the bulb, NO2 and the organic reactants are
exposed only to nitrogen, glass, teflon, and stainless
steel until it enters the chamber. Similarly, ozone
is exposed only to oxygen, stainless steel and teflon
so that no premature reactions can take place.
As soon as all the chemicals are released
into the chamber in the proper sequence, reactions
take place. Sincemany organics react quickly with
ozone, the ozone is reacted completely first with
nitrogen dioxide to form the dinitrogen pentoxide
equilibrium which takes only a few minutes. Then
the organic reference compound is flushed in with

nitrogen. More nitrogen is added to match the typical
concentrations in the atmosphere. Finally, the
organic reactants are swept in with approximately
five liters of nitrogen and the control valve is
moved from "input" to "circulation." If temperature
is being precisely monitored, the nitrogen and oxygen
make-up gases should be predominantly added before
the organic compounds. The atmosphere in the
environmental chamber will thus be given time to
equilibrate inside the constant temperature vessel.
Equipment. The vessel itself, a National
Appliance Model 3518 Incubator, consists of insulated
walls, a heater, and several fans. The heater is
controlled electronically to within 0.1 K by a Yellow
Springs Model 72 proportional temperature controller
and monitored by a pair of thermocouples (Type J)
and a thermistor (Omega 200 JC1-D1) located at
varying sites within the vessels compartment. An
independently operated fan is opposite the vessel's
main fan for assistance in circulating air within
the vessel.
Plumbing. Attached to the outside wall is
the control valve which is a four position flow
controller. As discussed, one position is for chemical

input to the chamber and another is reserved for fast
nitrogen addition. The remaining two positions are
for evacuating the chamber and for circulating the
chamber's contents to the gas chromatograph.
Two small diaphragm pumps (Metal Bellows
UB-21) that can push air through the system at
approximately 85 ml per minute (giving a residence
time outside the chamber of about 5 minutes) are
installed between the chromatograph and the chamber.
This is necessary since the chamber itself has only
one port with a narrow opening. By inserting
small-diameter teflon tubing through the port for
sampling and outflow while using the gap between the
tubing and the port for circulating input, adequate
circulation can be achieved but at the cost of
restricted flow. The gas chromatograph is installed
with a six-port automatic sampling valve driven by
compressed air and controlled by the HP 5890A gas
chromatograph. This valve intermittently samples
the flow circulating from the environmental chamber
while otherwise simply returning the flow to the
Environmental Chamber. This is a teflon film
bag, approximately 50 L in volume, having a single

opening made of stainless steel approximately 1/8"
in diameter. Ideally this chamber would contain a
sample of gas under the same influences as ambient
air. The surfaces of the chamber are the most
unrealistic condition in the system thus necessitating
the choice of an inert teflon surface. Ozone's life-
time was found to be over 50 hours in this environ-
mental chamber during continuous cycling and monitor-
ing through the ozone analyzer.
The teflon wall is semi-permeable to oxygen,
water, and other small molecules but this was not
found to be a significant source of error under our
experimental conditions. Cleaning of the bag as
received was done through heating, reacting with
ozone, and repeated nitrogen flushes over a period
of approximately two weeks. This was done until no
extra peaks at a sensitivity of less than 0.1 ppm
were discernible on the chromatograph from the
straight-run nitrogen used to fill the chamber.
Ozone Analyzer An optional by-pass through
an ozone analyzer after passing through the chromato-
graph has also been installed and its measurements
validated. This part of the system can only meaning-
fully measure ozone in the absence of other species

absorbing around 254 nm, unless those species are
known to have constant concentration, the instrument
is re-zeroed for these non-standard conditions, and
they do not react with ozone. Normally this part
of the system is bypassed to avoid contaminating the
sample cell with adsorbed UV-sensitive organics.
Ozone concentration was measured using an
ISCO UA-5 Absorbance/Fluorescence monitor modified
for gas rather than liquids by using sealed sample
and reference tubes, 1.02 meters in length, obtained
from a Beckman Infrared Analyzer Model 315A. An
ultraviolet lamp set for short UV wavelengths,
filtered to provide light of approximately 254 mm,
was used to generate the ultra-violet photons used
in the analyzer. A Sander Ozonizer Type II rated
at 25 mg/hour was used to produce ozone. This was
validated for the system using the analyzer, NO
titration, and also organic reaction rate studies.
To reduce the possibility of N0x formation
in the ozonizer, it was used only with a pure oxygen
feed during the experiment. Preliminary testing
using the analyzer and NO titration showed that the
formation of ozone was fairly independent of the nitro-
gen-oxygen mix ratio and also of the flows introduced

into the instrument. The NO titration was done
following standard techniques using the one liter
bulb whose volume determination will be described
later. The NO used for this gas phase titration
with excess ozone was obtained from Linde Specialty
at a concentration of 8,400 ppm. The titration had
an overall variation of 0.2%, probably due primarily
to errors in the flow control through the Matheson
gauges. The concentration of ozone was also confirmed
using Beers Law and known absorption coefficient of
311 atm ^ cm ^ and the length of the tube.
Gas Flow Standardization. Nitrogen and
oxygen enter the system through Matheson flow con-
trollers (Model 7481 H #603, 602, and 600). Flowmeter
calibration was done against "soap-bubble" meters
at low and moderate flows and against a wet test
meter (accuracy 1%) for high flows, that had itself
been calibrated using a soap-bubble meter, for flows
exceeding 1 liter/minute. The wet test meter had
excellent reliability even at flow rates as low as
50 ml/minute. The results were plotted against
Matheson's calibration curves using similar apparatus
and gases and were found to compare quite well.
Apparently the greater friction and inertia of the
wet test meter did not cause sufficient back pressure

to interfere substantially with the calibration of
f low.
Since the experimental protocol calls only
for comparative rates, all volumes of gases will be
stated as measured and not at STP. As all measure-
ments of reactant gases were done at room conditions,
the relative concentration of the gases will not be
Bulb Volume Determination Crucial to the
many concentration determinations of reactive species
in the system is the volume of the bulb used (along
with the use of the MKS transducer 222C measuring
pressure at 0-1000 torr) to find the amount of the
reactant introduced to the environmental chamber. The
volume was determined by evacuating a system consisting
of the bulb, tubes running to it, and another
standardizing bulb of known volume. This known
volume was determined by measuring the mass of the
standardizing bulb empty and then filled with water
and also by measuring the volume of the water needed
to fill the bulb. The pressure gauge is fitted to
the experimental bulb allowing the pressure and
volume to be measured. Air is allowed next to fill
the previously evacuated tubes leading to the
standardizing bulb and the pressure is measured again
when this reaches equilibrium. Finally, the evacuated

standardizing bulb itself is opened to the tubes and
once again, the resultant pressure is determined.
This procedure was done four times and gave a volume
for the experimental bulb >of 1.183 liters (including
the tubing to the transducer).
experimental bulb was found by comparison of pressure
variation as gas was permitted to expand from a bulb
of known volume into evacuated tubing which led to
a valve entering the evacuated experimental bulb.
A known value (V^) was connected through an unknown
volume of piping (V^) to the experimental bulb (Vg).
If the pressure (P^) in is known originally and
the gas sample expands into the tubing C^) > the new
pressure (P2) can be used to find the new volume
Similarly, if that same sample of gas is now permitted
Bulb Quantification. The volume of the
to expand into the experimental bulb (V^) and the
pressure is monitored (P^), the bulb volume is given
P2(W ww
(See Table 4 and Figure 3)

Using the derived expression for V2 yields:
P3 (Vl+VUPl-P2>+V3l(26)
2 / 2
Solving this expression for in terms of the
original volume and the pressure fluctuations gives:
Vo = V1P1(P2 P3)
p p
The known volume (V^) was determined by both volume
and mass measurements of water as 497.21 ml. Pressure
was measured in the gas bulbs using a differential
transducer indicator (Pace Model CD25) fitted with
a Validyne diaphragm (Model DP 103 10 Spec. No. 1295)
with a range of 0.1-10 torr that had, in turn, been
calibrated against a dibutyl phthalate oil manometer.
____________________TABLE 4______________________
P^ = original pressure in the Experimental Bulb
P2 = pressure in Experimental Bulb and Tubing
P^ = pressure in both bulbs and tubing
Vg = volume of Experimental Bulb
V2 = volume of tubing
= volume of Standardizing Bulb = 497.2 ml
TRIAL PRESSURE (.TORR) P P P *1 z r3 Vj, (Liters)
1 8.52 6.68 2.33 1.184
2 8.52 6.66 2.33 1.184
3 8.75 6.83 2.39 1.183
The volume of the Experimental Bulb is 1.18 liters

Temperature Control. There are four tempera-
ture control devices mounted on or near the environ-
mental chamber inside the constant temperature vessel.
One thermistor is used for feedback to the proportional
temperature controller. The other thermistor and
the two thermocouples have an independent readout
to a Heathkit EUW-4 voltmeter to insure that uniform
temperatures exist throughout the vessel. These
probes were standardized against a Parr Bomb Calori-
meter Thermometer #5J1463 that itself was standardized
according to A.S.T.M. 56C Parr No. 1603 for tempera-
tures 19-35 C. All temperatures on the devices were
uniformly 0.3 K high except the thermocouple at
position 3 which was 0.4 K high.
The constant temperature vessel's original
incubator thermostat was disconnected and its heating

coil was instead actuated by a thermistor which
controlled a YSI model 72 proportional temperature
controller set at a band width of 0.1 C. This
provided a sufficiently stable and responsive heat
source for the constant temperature compartment.
The setting on the controller was accurate within
the temperature band width at which it had been set.
This configuration provided the ability to control
temperature within the environmental chamber from
slightly above ambient to 308 K.
To this point, the major sources of error
would be found in the temperature measurements
limitation to 0.1 K and the volumes of the diluent
gases (nitrogen and oxygen) due to initial surge
problems as they flow through the Matheson gauges,
input tubing, valves, and finally.into the chamber.
The gauges do not actually measure flow but rather
the pressure imbalance due to an obstruction in the
flow stream through a tube. Since the valve on the
gauge is preset to the correct steady-state opening
and the gas is pushing against atmospheric pressure,
this surge error is probably on the order of 0.5%
or less of the overall volume but could not be
accurately ascertained. It effects only the diluent,

and not the reactive gases, so should not change the
relative concentrations of reactants or rates of
reaction in the system.
Gas Chromatograph. The Hewlett-Packard 5890A
gas chromatograph fitted with the 3392A Integrator
is the key component for output. The input valves,
columns, detectors, and the integrator have all
materially effected the data availability. Ideally,
a small portion (<0.5%) of the gases in the environ-
mental chamber is circulated outside of the constant
temperature vessel through a six-port valve and
returned to the chamber. At intervals, the valve is
rotated and a portion of 1-2 milliliters flowing
through a sample loop of this gas is injected on to
the column. The resulting peaks are analyzed for
identity and concentration using the integrator.
Typical area reproducibility is within 10% so that
concentration of reactants can be followed. The
products of these reactions in the environmental
chamber when analyzed using a flame ionization
detector (FID) will not be detected or interfere
since they would have been already oxidized during
the reaction in the chamber rather than in the

' The automatic valve system works fairly well.
Compressed air from the bench is controlled by a
solenoid valve which turns a six-port valve. The
air is diverted by an electromagnetic switch mounted
on top of the GC that is controlled by the purge
buttons. PURGE ON has chamber gases flowing through
the valve, into a sample loop of variable size, and
then back to the chamber. To enhance sensitivity,
the original 1 ml. sample loop was replaced with a
2 ml. loop. When the GC is in the PURGE OFF config-
uration, air is diverted to the other side of the
purge valve, the valve closes, and the chamber air
no longer goes through the sample loop, but rather
goes straight back to chamber. Now the carrier gas
instead of flowing directly into the column goes
through the sample loop first. This pushes the
trapped sample gas caught in the sample loop onto
the column (and also compresses the sample due to
the higher pressure of the carrier gas) .
Two problems arise due to this procedure.
The first problem is maintenance of the valve.
Mechanical problems are minimal and failure of the
switch can be detected both on the chromatograph
where there will be no peaks and either by the absence

of the brief hiss or by a continual long hiss. The
most common mechanical problem is that the compressed
air will be at the wrong pressure. This means that
either the electromagnetic switch won't be strong
enough to divert the air or that the air pressure
diverted by the switch won't be strong enough to
twist the six-port valve. The pressure of the com-
pressed air is controlled both by a screw diverter
valve open to the laboratory placed just before the
air filter and by the bench air flow diverter. A
slight hiss should be heard and air flow should be
barely felt at the opening to the screw diverter.
Too high a pressure blows off the tubing between the
screw diverter and bench yet the tubing should visibly
swell at the proper pressure.
The other problem is due to the sample loop
changing pressure between purges. These purges cause
pressure surges that are detected as spurious peaks
on the chromatogram. The peaks change position and
area depending on oven temperature, carrier flow,
column length, composition and conditioning, as well
as the flow and pressure characteristics of the chamber
gases. These peaks must be characterized, and thus
eliminated from consideration before analysis of the

chamber gases can be done. Water, oxygen, pressure
and nitrogen peaks were distinguished. This meant
that, prior to beginning an actual experiment, each
gas at the proper dilution had to have an exhaustive
series of runs performed to distinguish that gas from
other peaks that might appear on the chromatograph.
The choice of a column for a particular
experiment is crucial. The six-port valve is limited
to a maximum temperature of 125 C and most molecules
to be examined will be small organics. The structural
similarities of the molecules make distinguishing
them difficult amid the surge peaks. Longer length
columns make these surge peaks more pronounced and
difficult to handle. The nature of the gases also
caused more than the expected problem since ozone
clearly could be seen to degrade columns whose sub-
strate were olefinic or aromatic. A final choice of
column for preliminary work was a three foot Poropak
Q, 80/100 mesh, stainless steel attached to the FID.
A further refinement for future experiments is a
teflon column to be attached to an electron capture
detector (ECD) for studies on product formation of
compounds such as PAN that contain nitro groups.

The gaseous alkenes were easy to dilute and
validate the resultant concentrations while 'formal-
dehyde isoprene, and naphthalene sample preparation
led to several more intricate problems. For the
gases that werent already available at the desired
concentrations in a tank, a small commercial propane
tank was evacuated. Concentrated alkene was then
allowed to enter the tank to a desired pressure and
the gas was then diluted with technical-grade nitrogen
to a final total pressure of 42.2 psi. Technical-
grade nitrogen was used for cost reasons since its
primary impurities are oxygen, carbon dioxide, water,
and argon: all are present in higher concentrations
in the atmosphere and should not interfere with the
reaction system by removing any of the reactants.
Since isoprene is so volatile and the propane
tank is initially a vacuum, a microliter syringe
filled with a known volume of the isoprene was injected

through a small diaphragm and the liquid quickly
evaporated into the injection tubing and tank.
Nitrogen was again used to finish pushing the isoprene
into the tank from the tubing and dilute the organic
reiactant to the desired final concentration.
The method of introducing naphthalene into
the reaction system would be typical for volatile
solids. A short length of tubing was sealed with
a Swagelock fitting on both ends, weighed, reopened,
and then a few small particles of naphthalene
(approximately 0.27 mg) were inserted. The tubing
was reweighed, and if the amount of material was
appropriate, the tubing would be inserted into the
circulation tubing on the side away from the gas
chromatograph past the pump. Normal circulation
would evaporate the solid either in the tubing or in
the chamber. Care should be taken to insure that
the direction of flow in the tubing is toward the
chamber rather than into the chromatograph's sample
loop to prevent deposition in the loop or column
which would lead to erroneous results for naphthalene
Part of the problem in measuring naphthalene
lies in its low vapor pressure. The vapor pressure

between 19-35C is given by the equation:
log P(ram) = -[108.30/(t+27)] + 1.115
Approximately 27 C is the coolest that the environ-
mental chamber can be maintained without risk of
heat influx due to ambient temperature or instrumental
conditions. This temperature gives a vapor pressure
of 0.1285 mm which implies that 59 milligrams of
naphthalene would saturate a 50 liter volume in the
environmental chamber. This is the most that could
be expected to sublime and excess amounts could
precipitate in corners or in the circulation tubings'
joints and bends. Thus, using a one milliliter sample
loop, the detectors will be sampling from a saturated
sample, one microgram of material. As will be dis-
cussed in the next chapter, this amount could be
studied, by GC-MS for example, but under system
conditions was not quantitative .
Formaldehyde was initially introduced in a
similar manner but the chromatogram revealed several
peaks with unexpectedly long retention times. This
was attributed to solid formaldehyde at STP actually
being a polymer (paraldehyde). This polymer as it
evaporated would not form only the simple monomer
when decomposing, but.also some dimer, trimer, etc.

In order to form pure monomer, a cold vacuum distilla-
tion was performed..
A gram, or so, of paraldehyde was placed in
a flask that had been cleaned and then rinsed with
a dilute acid (to reduce re-polymerization). The
flask is connected to another flask immersed in a
salt-ice water bath. This is further connected via
a stopcock to another collecting flask which is
immersed this time in a dry ice-acetone bath. From
here, in turn, is connected a three-way valve that
leads on one hand to a vacuum system, and on the
other, to a storage flask kept in liquid nitrogen.
Initially, diphosphorous pentoxide was also
placed in the paraldehyde flask to act as a drying
agent. This was found to be counter-productive as
it also promoted co-polymerization with formaldehyde.
The polymer formed was found to be insoluble in non-
polar organics, as well as acidic and neutral water.
It did dissolve slowly in highly alkaline solutions.
The melting point of the white, paper-like, opaque,
and odorless solid was 238 C. The polymer condensed
throughout the vacuum system and was not easily
removed even using a heat gun.
Heating the paraldehyde without the diphos-

phorous pentoxide to 40 C in a vacuum of 5 torr or
less produced the desired results. Oligomers, water,
and other thermal decomposition by-products condensed
as a solid in the salt-ice water bath and the monomer
condensed as a liquid in the dry ice-acetone bath.
Removing the acetone bath permitted a second distilla-
tion where the first 10% or so, along with the final
20%, was evacuated through the vacuum system. The
rest of the condensate was distilled over to be
collected as a solid in the liquid nitrogen-cooled
storage flask.
Connecting the storage flask to the one liter
bulb for gas measurements and then lifting the flask
briefly out of the nitrogen permitted adequate amounts
of the monomer to be introduced to the system. Care
at this point was needed to plunge the monomer back
into the nitrogen as soon as possible as the monomer
not only quickly started to vaporize but also to
re-polymerize. The polymerization reaction, once
started, released enough heat to further vaporize
the solid and the resultant pressure could build up
in excess of the system's ability to contain the
gases formed.
Purity in every case was ascertained by gas

chromatography, peak analysis for elution time and
area. Assuming equal response sensitivity for each
carbon involved in the formaldehyde polymers or
monomer, the proportion of the monomer increased
from 34% of the total to over 99%. As mentioned
before, GC-MS was often used to confirm a peak
elution time as belonging to a particular organic

Computer modelling of the reactions taking
place in the nitrate kinetic system was done using
the Gear program^ as modified by Juan Bonilla^ on
IBM-compatible equipment. This software permitted
modelling fifty different ozone, nitrogen dioxide,
and nitrate radical bi-molecular reactions using 25
constituents (see Table 5). Reactions 1 and 2 simu-
late the dinitrogen pentoxide equilibrium and reaction
3 is the initial formation of the nitrate radical
from ozone and nitrogen dioxide. Reactions 6-10
provide a more accurate simulation of the entire
N0x system. Organic-ozone reactions are Reactions
4 and 5 while the rest of the equations are used to
detail the nitrate-organic reactions. Rate constant
values are chosen to agree with Ravishankara and
Mauldin's absolute rate constant for trans-2-butene
The modelling does not take into account the
dispersion of gases on entering the system which

Reaction RAp CONSTANT cm /molec-sec
1) no3 + no2 -> n2o5 1.3E-12
2) N205 + M -> N03 + N02 + M 4.59E-2
3) 03 + N02 -> N03 + 02 3.15E-17
4) T2BE + 03 > + 2E-16
5) PE + 03 -> + 1.13E-17
6) 03 + N02 > NO + 02 9.70E-19
7) N03 + NO -> N02 + N02 1.9E-11
8) N03 + N02 -> NO + N02 4.04E-16
9) no3 + N03 -> N02 + N02 2.29E-16
10) 03 + NO -> N02 + 02 1.82E-14
ID T2BE + N03 -> BUTYLPROD + 3.5E-13
12) PE .+ N03 -> PDDN + 7E-15
13) BDE13 + N03 -> BDE13PR0D + 9.6E-14
14) M2BDE13 + N03 -> M2BDE13PR0D + 5.8E-14
15) C2BE + N03 -> CISPROD + 3.40E-13
16) ISOPRENE + N03 -> ISPROD + 5.8E-13
17) BE1 + N03 -> BE1PR0D + 9.7E-15

means that mixing constraints will slow down the
perceived rates of the actual reactions in the
environmental chamber. Temperature changes can be
considered by changing the appropriate bimolecular
rate constants.
Modelling was done for various initial
concentrations of nitrogen dioxide and ozone in order
to achieve appropriate levels of nitrate radical with
enough dinitrogen pentoxide to insure the complete
reaction of the reference (trans-2-butene) and
experimental organic compounds. Sensitivities of
the gas chromatograph limited these compounds to
initial concentrations of 1 ppm or greater if the
decay was to be followed through a lifetime or so.
Experimentation with the reaction of nitrogen dioxide
and ozone led to concentrations of 150 ppm NC^j 5 ppm
ozone, and 2.5 ppm organics. Modelling the reaction
without the organics showed completion of the dinitro-
gen pentoxide equilibrium within two minutes. When
mixing.constraintswere taken into consideration,
it was assumed that nitrate radical formation could
be considered complete, as well as destruction of
the ozone initially present, after approximately
five minutes in the system. After that time, organics

could be injected without the complications due to
ozone destruction of the initial organic concentra-
tions. Several experimental runs were made with
quick organic injection (5-15 minutes after ozone),
and other injections were done after an hour or more
of equilibration following ozone injection. This
was done so that the system, including the tubing,
and not just the environmental chamber, would have
an equilibrium concentration of dinitrogen pentoxide.
After the nitrogen dioxide-ozone system
equilibrated in the modelling, the final values for
nitrogen dioxide, nitrate radical, nitrogen monoxide,
and dinitrogen pentoxide were used as initial values
(ozone concentration was zero by this time) along
with the initial organic concentrations to begin the
next series of reactions. This modelled the decay
in the organics concentration due to nitrate radical
reaction. Once the decay curve is matched to the
experimental decay by varying the concentrations of
the trans-2-butene (or, if need be, equilibrium N0x
concentrations), a matching concentration is used
for the other organic materials. The experimental
organic's concentration having been determined,
varying the modelled rate constant for the organic-
nitrate radical reaction would give the best fit for

the constant to be determined. This value is the
experimentally determined rate constant for the
particular reaction at that particular temperature.
A change in temperature to permit calculations
for activation energy will not yet be feasible. Not
only would the rate constants for the experimental
organic and the trans-2-butene both change, but so
would also the concentration of the nitrate radical.
The absolute rate constants for isobutene and trans-
2-butene are known from Ravishankara1s work but only
at 298 K. This lack of information, combined with
the uncertainty in the dinitrogen pentoxide equili-
brium, places too many variables into the calculations
for them to be quantified without the introduction
of more extensive instrumentation such as an FTIR
or DOAS.
GC-MS. An attempt was made to provide further
information using mass spectroscopy. A Hewlett-
Packard 5890A Gas Chromatograph fitted with a Total
Ion Current (TIC) on a Mass Selective detector (MS)
was used to confirm organic identities as undiluted
reactants and during the reaction. Due to the
concentration of the gases in the environmental
chamber, the latter use of the GC-MS was not helpful

as the only constituents to be confirmed were
nitrogen and oxygen. Confirmations of trans-2-butene,
isoprene, and propene at concentrations of 0.1%
were done as reactant gases before dilution in the
chamber. Naphthalene was confirmed at ambient
equilibrium vapor pressure in a closed sample tube.
This peak for naphthalene (see Figure 4), on both
the GC-FID and GC-MS, could be distinguished by eye
but could not be sufficiently separated and yet be
narrow enough at its ambient equilibrium vapor
pressure (approximate concentration was less than
10 ppm) to be electronically distinguishable and
quantified on a consistent basis (see Chart 1) by
the integrators. However, if the GC-MS was set on
the appropriate peak, it was able to give a confirma-
tory mass spectrogram of the compound. The overall
inadequate sensitivity led to the present acquisition
of the Photo-Ionization Detector (PID).
Organic-Ozone Reactions. In order to confirm
operation of the system, the better understood
reactions of selected alkenes with ozone were analyzed
as a preliminary study. This work confirmed the
practicality of the technique as a means of measuring
rate constants as well as validating the operation

5.448 to 5.960 min. from BOBC.D
m/z abund. m/z abund. m/z abund. m/z
51.15 4 127.15 10 127.95 85 129.05
ab .

of the individual components.
The first step was a series of measurements
to determine the stability of ozone in the operating
system using the ozone analyzer. After proper
conditioning of the environmental chamber, ozone was
found to have a lifetime of between forty and sixty
hours. The ozone decay was zero-order and approxi-
mately one ppm per 250 minutes. This is too fast
a rate for the auto-destruction of ozone unless
catalyzed in some fashion by the wall. Since it does
not react significantly with either nitrogen or
oxygen, the loss of ozone must be a wall loss due
possibly to water slowly permeating through the
teflon film. The lack of diurnal variation in the
ozone loss rate suggests that the amount of incident
light in the circulation tubing did not effect ozone
concentrations. The ozone loss rate was significantly
below any threshold that might interfere with the
study and compares favorably with loss rates in
similar studies.
The ozone. Organic modelling showed that,
since the rate constants of propene and trans-2-butene
differ by more than an order of magnitude, it would
be difficult to devise a comparative rate study that

could follow both reactions using either the GC or
ozone detector since both instruments have an effect-
ive response time during circulation of approximately
ten minutes for each point of datum. Reference to
the literature showed no available reactant whose
rate constant was of an appropriate magnitude to
bridge the gap and that other studies have encountered
the same problem with both ozone and nitrate radicals.
A further study involved the addition of
nitrogen monoxide and formaldehyde to the ozone in
the environmental chamber in a mixing ratio of 10-10-1
of ozone, formaldehyde, and NO, respectively. These
proportions lead to an ozone-rich mixture of NC^-O^
reacting slowly with formaldehyde. The initial
ozone decay was second-order for eight hours followed
by the previously observed decay with a lifetime of
over forty hours (see Graph 1). Since the system
closely follows Cantrell et al. , and this reaction
series would yield some information on the nitrate
radical reactions,computer runs using rate constants
from their work were modelled (see Table 6).

1 .06

0 48 96 144 192 240
Ozone Decay in 03~CH20-N0 System

Reaction cm molec-sec
1) no2 + 3 " > N03 + 2 3 .15E-17
2) no2 + 3 - -> 2 + NO 9 .7E-19
3) no3 + no2 -> n2o5 + 8 . 61E-13
4) n2o 5 + M - > N03 + N02 + M 4 -59E-2
5) no3 + no2 -> N02 + NO 4 . 04E-16
6) n2o 5 + -> hno3 + hno3 4E-13
7) no3 + NO - > N02 + no2 1 .9E-11
8) no3 + no3 -> N02 + no2 2 .29E-16
9) 3 f NO -> no2 + 2 ' 1 .82E-14
10) no3 + ch2o -> HNO 3 + CHO 6 .3E-16
ID CHO + 2 - > ho2 + CO 5 .5E-12
i2) ho2 + ho2 "> H22 + 2 2 .74E-12
13) ho2 + 3 -> OH + 2 2E-15
14) ho2 + NO - > OH + no2 8 .279999E-12
15) ho2 + no2 -> ho2no2 + 1 .42E-12
16) ho2no, 2 + " > ho2 + no2 7 . 16E-13
17) ho2 + no2 -> HONO + 02 2, .97E-15
18) ho2 + no3 -> hno3 + 0 2 2, .5E-12
19) ho2 + ch2o -> o2ch2oh + 1 , .69E-14
20) 02CH20H + -> ho2 f CH20 1 . . 2E-11
21) 02CH20H + 02CH20H -> OCH2OH+OCH2OH 9.939999E-14

TABLE 6 (cont. )
23) o2ch2oh + ho2 -> ho2ch2oh + o2 1.76E-14
24) och2oh + o2 -> hco2h + ho2 1.49E-14
25) OH + CH20 -> H20 + CHO IE-11
26) OH + N02 -> HN03 + 1.11E-11
27) OH + NO -> HONO + 4.57E-12
28) OH + CO -> H02 + C02 2.59E-13
29) H02 + OH -> H20 + 02 1.07E-10
30) OH + 03 -> H02 + 02 6.84E-14
31) OH + H202 -> H20 + H02 1.66E-12
32) OH + HN03 -> H20 + N03 1.28E-13
33) OH + H02N02 -> H20 + N02 4.65E-12
34) HONO + HN03 -> N02 + N02 2.71E-17
35) N03 + -> 02 + NO 7.05E-23
Although close, this modelling did not
yield appropriate results and further consideration
led to the recognition that Cantrell's experiment
and modelling did not address the situation of excess
ozone present in this system. This meant that certain
nitrate reactions with ozone and formaldehyde, which
did not contribute significantly to their system and
were ignored in the modelling, had to be accounted
for in this system. A literature search failed to

yield values for the rate constants for the reactions.
Judging from similar reactions, ozone will
react too slowly with formaldehyde, water, and other
stable molecules to be of interest here. By comparison
with the reaction of oxygen atoms, an estimate can
be arrived at for the rates for ozone on nitrate
radical and dinitrogen pentoxide (see Table 7 ).
Reactant RATE 0 (ppm min ) 3 ratio 0/03
N03 1.5xl04 ? ?
N25 <4x10 7 ?
C3H6 5.4xl03 1.5x10-2 3.6xl05
C2H4 1.2xl03 2.8xl0-3 4.3xl05
no2 1.34xl04 4.75xl0-2 2.82xl05
1. NASA Panel for Data Evaluation, Chemical Kinetics
and Photochemical. Data for Use In Stratospheric
Modelling Evaluation Number 7, JPL Publication
85-37, 1985, Jet Propulsion Laboratory,
California Inst, of Technology, Pasadena, CA
In view of this chart, the following reactions
and rates were believed to be of possible signifi-
cance and were added to the model.
36) N03+C0 -> CO2+NO2 IE-18 cm3/mol*sec

6E-16 cm /mol-sec
37) no3+ch2o - -> hno3+cho
38) N03+03 -> N02+202
The ozone-dinitrogen pentoxide reaction would
have a rate of less than IE-21; thus, due to its combined
uncertainty and lack of speed, it was dropped from
consideration. The GEARII modelling cannot handle
anything other than bimolecular rate constants so
the actual data was adjusted to compensate for the
loss of ozone by unimolecular and zero-order processes.
Graph 2 shows the decay of ozone in the
formaldehyde-ozone-NO system and examines several
possibilities for influential reaction schemes that
might model this reaction. Ozone's auto-decay (see
Graph 3) in the system and the ozone chemical reac-
tions with formaldehyde and NO had fairly similar
curves so that, during the time observed, most of
the ozone loss is not due to chemistry but is instead
attributable to wall loss. The complex modelling
of the system shows that almost all of the chemical
decay was over in about a half hour, which would be
before more than a few data points could be reliably
collected. This modelling also predicts that
approximately 76% of the original ozone should be
left when reacting 10 ppm ozone with 1 ppm NO. The


_1,1700 96 192 288 384 480 576 672 768 864 ^
Ozone Auto-Decay

actual system took approximately 32 minutes to
equilibrate at 75% of its original absorption as
measured by the ozone analyzer. The small deviation
from the curve is initially due to mixing constraints
and later due to the slightly warmer temperature
(303 K versus 298 K) than the model. Possibly, the
uncertainties of the rate constants may also have
had a slight effect.
Iii summation for the ozone system reactions,
the system and the theoretical model worked well
together in determining and elucidating the reactions
being studied. Expected problems included not being
able to follow as many reactants or products as would
be necessary to fully elucidate all the reactions
being studied, not having an appropriate response
time for all the reactants that could be followed,
and the difficulty in understanding the exact role
a small change in temperature might be having on the
various rate constants. An unexpected problem was
the larger than expected mixing and system equilibra-
tion times. Most of these difficulties would compound
the problems to be encountered in the nitrate system.
Nitrate Radical Reactions. The final series
of experiments consisted of an unsuccessful search

for the rates of reaction of nitrate radicals with
any of a variety of organic species. Since confirma-
tion for the presence in the system of NC^, ozone,
and the organic species (formaldehyde, naphthalene,
trans-2-butene, etc.) and their concentration was
completed, all that was left to be performed was
forming the nitrate radicals and then introducing
the organics into the system. This was done with a
variety of different organic chemicals, a variety
of different nitrogen dioxide-ozone concentrations,
and two different temperatures.
Since earlier experiments had already con-
firmed ozone's reactivity with several of the organic
molecules to be tested in the kinetic system, gas
concentrations had to be such that ozone would be
consumed before the introduction of the organic
molecules. This was done in order to avoid competi-
tive oxidizing reactions on the same organic compounds
that would unnecessarily complicate the analysis of
the data. This was assured by always using a nitro-
gen dioxide concentration of more than twice the
ozone and waiting at least ten minutes for the
dinitrogen pentoxide equilibrium to be established.
This was the equivalent of four to five lifetimes

for the reaction. Once the organic molecules were
simultaneously introduced, timing was started for
the reaction. Chromatographs initially were taken
every six minutes for approximately two hours, and
then every half hour until the reaction was deemed
Computer modelling of the system showed ozone
concentrations dropping to zero somewhere near 100
seconds after the mixing of NO2 and ozone. The point
at which ozone concentrations reached lO-^ ppb was
selected as the starting point for organic reactions
since an equilibrium could not be truly established.
There will always be a slow radical reaction of odd
oxygen compounds to form the more stable oxygen gas.
The various nitrogen oxide concentrations would then
be input along with the original organic concentra-
tions and then the reaction would be followed for
three to six hours.
The actual experimental system took some time
to equilibrate and establish a stable trend on the
baseline as it began to do automatic sampling. An
arbitrary selection of 20-30 minutes after injection
of the organic was taken as the initiation point to
follow comparative losses in the model and the

experimental system. By establishing the signals
at that point from both the GC and the computer
model, and then finding how long it takes for the
signals to drop to 1/e of that strength, the lifetime
of the substance being studied under the experimental
conditions can be determined. Since the nitrate
radical will have the same concentration for both
the reference and experimental organic compounds,
the ratio of the rates of the two reactions can be
found by taking the inverse of the ratio of the life-
times under the experimental conditions. This method
obviates the need to establish the actual concentra-
tion of the nitrate radical which, as has already
been discussed, would be a very difficult activity
and lead to an unacceptably high error margin. If
dinitrogen pentoxide or nitrogen dioxide was also
reacting with one of the organic compounds, varying
either the temperature or the nitrogen dioxide
concentration would expose the problem. The computer
modelling was used to establish preliminary reaction
parameters and could be used later to confirm the
experimental rate constant.
This modelling showed that the original
concentration parameters of nitrogen dioxide, ozone,

and total organic compounds of 10-4-2 ppm should be
changed to 150-4-2 ppm if acceptable lifetimes for
both the reference and experimental organics were
to be achieved. Generally, the shorter lived species
was designed to last between 90-200 minutes while
the other species would have a lifetime less than
20 hours. For example, the reaction involving 1 ppm
of trans-2-butene, lppm butadiene, 150 ppm nitrogen
dioxide, and 4ppm of ozone, had 38% of the trans-2-
butene signal being present 100 minutes later while
52% of the butadiene signal was still present six
hours later according to the modelling.

Unfortunately, the prognostications in the
previous chapter had nothing to do with the actual
experimental results. Graph 4 shows five different
experiments involving trans-2-butene "decay" and
Graph 5 shows three experiments with propene
"decaying." The careful observer will soon note
that instead of decay curves, one finds growth curves.
Some of these experiments were done at higher tempera-
tures and some were done vrith different concentrations
of reactants. Three experiments were done without
any ozone so that no nitrate could have been formed.
Changing these parameters made no significant
difference to the experiment.
Two parameters did make a difference to the
characteristics of the growth curves. One parameter
was how long a time was taken between samplings and
the other was the particular organic compound being
observed. (The identity of the other organic chemical

x 1,000
a n .
4a po *
a 4, a
u ** +
+ + ^
** +** *< ** + +* V + + *
* ^ + *+-*
* _
*a 4 A* *. * * 1
A* ^ * A* 4 A
"DA 41 A 4
V+-* .% ' * .
* ^
p *r .
it, *
6 4
/ <
2 ***
L %
&/ ,vv'V'-'*'
200/4/1 (corrected to
1.25 ppm)
a 150/0/2^ (corrected
to 1.25 ppm)
4 6 8 10 12 14 16 18 20 22 24 26 28 30 32 34 36 38 40
N02-03-C5H10 150/5/1.25 ppm T=30C
unless otherwise noted


/ ** +.***** *'*£* *
* + .*.
* ** * * * V 1 v- /t +.** ** *
* *
+ *

* ^ 0%
* a # - *
e % *
a. *
Vi '* 0
0 m
2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32 34 36 38 40
Propene NOg Reaction N02-03-CH20 150/5/1.25 ppm
GRAPH 5 unless otherwise noted 7=30C

in the chamber, did not make a difference on the growth
characteristics.) Propene took about three hours
to reach its stable maximum of 2600 counts while
trans-2-butene took about 14 hours to reach a plateau
of 12,000 counts adjusted to a concentration of 1.25
ppm in both cases. The only different behavior
characteristic between the two organics exhibited
in this study is that propene responded promptly and
returned to the standard pattern when its sampling
time was adjusted to the usual half hour intervals
while trans-2-butene only started growing in faster
and never reached 12,000. (It stabilized around
9500 after about 60 hours elapsed in the experiment.)
Both curves show similar patterns. It is
fairly easy to see both the instrumental noise
generated in the system and the errors in concentra-
tions when the experiments were performed. The curves
contain further information that can also be extracted.
Since the signal is no longer zero, the compound
still exists in the chamber long after it should have
been destroyed. The shape of the curves suggest that
a reservoir of material is reacting with a limited
amount of an unknown reagent before the organic
compound can reach the detector. As the unknown

agent is itself destroyed, more material reaches the '
detector until either an equilibrium or a steady-
state of organic material is reached at the detector.
The fact that the two organic compounds take
a significantly different time to reach this point
argues against a mechanical flaw in the circulation
such as a restricted opening in the tubing or a loose
fitting permitting some sort of bleeding of the gases.
The disparity of the average signal count at equili-
brium is more than can be accepted as error or
sensitivity differences in the detector between
propene and trans-2-butene. The most likely reason
for this disparity again lies with some sort of
chemical reaction taking place. It is interesting
to note that ratio of the equilibrium signal count
(12,000/2600=4.6) is about the same as the ratio of
the time it takes to reach the equilibrium (14/3=4.7).
This piece of evidence would seem to favor a steady-
state hypothesis if it is indicative of anything more
than the perversity of nature.
If there is an unknown chemical reaction
taking place in the system, what are some of its
properties? Examination of the elution time peak
area data revealed no peak decreasing in the same

manner that th.ese graphed peaks increased so the
material must not be FID sensitive or so polar so
as not to escape the column. Since the curves for
any particular organic are so reproducible, it
would follow that this unknown chemical, after it
has been fully reacted in one experiment, is fully
regenerated during the procedure to ready the system
for the next experiment. Propene reaches its final
equilibrium earlier and at a lower value than does
trans-2-butene which implies a disturbing conclusion
in that the unknown reaction has a higher rate with
propene, which for most reactions would form a less
stable intermediate and thus have a longer reaction
time with same reactant.
Both chemicals were sometimes in the system
together, or at different concentrations actually
than the 1.25 ppm to which the signal counts were
adjusted, and neither of these factors effected the
growth of the signa-1 count. This line of reasoning
would, contrary to the previous evidence, suggest
that there was some restriction in the circulation.
If there is a problem in the plumbing, it may lie
in the chamber, the tubing, the pumps, the purge
system, or the column and detector. If the problem

is with the chamber, the graphs should have shown
a high noise factor at first gradually settling down
to an equilibrium value. While there is an equili-
brium value, the noise is less at the beginning and
there is an obvious growth curve rather than noise.
The growth curve would be better explained if there
existed a reservoir of the organic compound in the
chamber and the unknown agent was in the tubing or
pumps. This would not explain the lower values when
the purge timing was kept at the shorter cycle since
both units have the same constant flow of gases no
matter what the position of the purge valve is on the
GC. The evidence for a chemical reaction occurring
in the plumbing would lead ultimately then to one
loop of the purge valve containing significant amounts
of some regenerable compound that reacts faster with
propene than with trans-2-butene and does not ionize
significantly in an oxidizing flame. This is not
regarded as a likely explanation.
None of the evidence helps in determining
what happened to the nitrate radical. If it is as
reactive as the ozone-NO experiments suggest, the
problem may lie in the lifetime of the nitrate radical
in the system. According to the observed ozone life-

time of 40+ ho.urs and keeping in mind that the
nitrate radical generally reacts over 1,000 times
faster than ozone, the radical itself may have a
lifetime of less than three minutes. This does not
mean that the nitrate radical concentration would
drop to l/e of what the original concentration would
be during that interval since new radical would be
regenerated from dinitrogen pentoxide on a comparable
time scale. What would happen is that nitrate radical
would be at a much lower concentration than modelling
would have suggested and nitrogen dioxide concentra-
tions would be higher as the radical decomposed.
Pitts et al. at California Riverside may
have experienced the same sort of problem. Their
chamber's volume is much larger than the one used
here containing an interior fan, for example, to aid
in circulation. Their method apparently is to pulse
in quantities of dinitrogen pentoxide which would
decompose to react as the nitrate with organic
chemicals already present in the chamber. This method
would introduce a number of difficult parameters to
measure since neither an equilibrium nor a steady
state would be established and mixing in the chamber
would not be uniform by any means during the reaction.

They are further limited to just four points of data
during any reaction. For fast reactions, extra NC^
would be added to shift the equilibrium away from
NOg, but what the actual concentration of any of
these gases might be at some point in the puff of
dinitrogen pentoxide must be fairly difficult to
ascertain even with their in-situ measurement tech-
niques. This problem would suggest why their studies
have not looked closely at the dinitrogen pentoxide
equilibrium or any temperature effects. It is
unfortunate that their articles were not more explicit
in the discussions regarding experimental design so
that their results could have been treated with more
careful consideration. In any case, their studies
were done in a manner designed to get around this
problem of nitrate decay rather than trying to solve
it. A similar technique could not be adopted here
since dinitrogen pentoxide could not be kept as the
more stable solid if the temperature effects of the
reaction were to be studied.
In conclusion, some process is removing nitrate
radical from the system, perhaps in a similar manner
to the system at California Riverside. Another
presently inexplicable process is generating a removal

process for organics in the circulating system.
These are possibly related in some form of an NO^-
wall reaction that occurs primarily at the heated
purge valve. A series of tests and improvements
could and should be made to the system that can
delineate where the problems encountered might lie
and what can be done about these problems. The
problems encountered here shed further light on the
limitations of studies done elsewhere.

While the present results for the system
have to be considered disappointing, there are a
number of improvements and modifications that are
waiting to be made that should improve sensitivity
for organic compounds and nitrogen species as well
as increase the lifetime of the nitrate radical.
Mike Hailu is already bringing on line a
Photo-Ionization Detector (PID) built by hnu equipped
with the 10.9 ev lamp. This detector has already
seemed to show greatly improved sensitivity to
naphthalene compared to the FID used in this series
of experiments. The purchase of a 10.2 ev lamp
should make peak identification somewhat simpler as
this will decrease noise and response of materials
not currently of interest. The PID will also measure
nitrogen dioxide (and possibly dinitrogen pentoxide)
simultaneously with the organics which will improve
the correlation between modelling and results. The

FID and PID are now connected in series to help in
peak identification by use of differential sensitivity.
Positive results in this area have already been seen
in the case of naphthalene.
A major simplification in the future will
be computerized data acquisition and manipulation.
At present, all peaks occurring after the purge on
to the column are recorded by retention time and area.
These are matched against known surge and purge peaks
and then identified using the known time and area
characteristics for several different concentrations.
For an experimental run, not all peaks can be identi-
fied in advance and pages of peak area-elution times
have to be handwritten and then examined to search
for overall trends. This proved to be a major bottle-
neck in analysis. Since the PID does not have an
integrator, but instead uses a strip chart recorder
whose peak area-elution time data would have to be
evaluated, an automated system has become a virtual
The computer should acquire signal data from
the FID (bypassing the integrator) and the PID. After
subtracting one signal from the other, a distinctive
recognizable peak system can be established. Care

should be taken at this point since both the FID and
PID can have negative peaks under certain surge and
column conditions. A processing program to identify
peaks and their areas will treat this generated data
and store the results from each run. Data will be
output by a particular retention times peak area
as a function of run number or elapsed time from the
start of the reaction. Peak height, rather than area,
might be used successfully as peak shape has not been
found useful for identification purposes. This
computerized data acquisition will also make compari-
sons between modelling and experimentation much
easier since both sets of data will be stored on disk
and will be available for manipulation.
Neither of these improvements will be of
effective use against the main problem of probable
premature nitrate radical loss. One effort here
that might improve matters would be to reduce wall
loss. The lifetime of ozone was about fifty hours,
but if nitrate radical is a thousand times more
reactive than ozone in the system as it is with many
other reactions, that works out to be a lifetime for
nitrate radical of three minutes. It would probably
not be possible, and certainly not reproducible, to

try and introduce the organic compounds that soon
after ozone introduction in our system. Before the
expected initial equilibrium concentrations of dinitro-
gen pentoxide could be formed, the nitrate radical
would have been destroyed. The California Riverside
group apparently has the same problem that they got
around by doing up to four successive injections of
previously prepared dinitrogen pentoxide, using an
interior fan and measuring the organic concentrations
after each injection. This may be feasible if nitrate
radical were considered to be the only reactive
species present as then all of the pentoxide would
have a chance to react as a nitrate as it decomposes.
Unfortunately, Pitts et al. have already shown this
2 8
assumption to be false in the case of naphthalene
It would be very difficult to separate the dinitrogen
pentoxide rate constant from the nitrate radical
considering mixing and wall loss problems. Changing
the NC>2 concentration, by extra addition of this
particular gas before the dinitrogen pentoxide, would
possibly show which was the reactive species but mixing
problems would be difficult to model and extracting
a meaningful rate constant near impossible. Perhaps
a better method for our system would consist of a steady

controlled steam of dinitrogen pentoxide being
admitted to the chamber at a rate equal to the loss
rate of the nitrate radical. There is present in
the PID the ability to measure certain N0x concentra-
tions so that a steady state concentration of nitrate
radical may be achieved in this manner. This
modification would make any investigation of the
dinitrogen pentoxide equilibrium impossible but at
least the system would be otherwise useful.
Other teflon environmental chambers have
been used in the past so teflon may be assumed to
cause negligible loss of the nitrate radical over
substantial units of time for these studies. Perhaps
if all the fittings, wherever possible, be
replaced with teflon there would be some improvement
but this is not likely to be of significance since
those same environmental chambers also contained
stainless steel and even some brass. Perhaps a more
important modification would be.if the circulation
pumps could be replaced with improved cooler running
pumps with teflon surfaces rather than stainless
steel. The present pumps have a high surface to
volume ratio and run at temperatures well over 50C.
A check on the efficacy of this approach could be

done by filling the environmental chamber as usual
but, instead of circulating the gases through the
GC and then returning them to the chamber using pumps
on both sides of the instrument, using a single pump
to evacuate the chamber by pulling the gases through
the GC after removing the pump that is on the line
going from the environmental chamber to the instru-
ment .
The heat of the present pumps would cause
the dinitrogen pentoxide equilibrium to shift towards
the more reactive nitrate radicals. The high surface
to volume ratio of the pumps and downstream lines
could be making for higher than necessary wall loss
rates. If this proposed series of experiments indeed
indicated an effect due to the presence of nitrate
radicals, wall loss rates could be investigated by
waiting various amounts of time before injecting
organics into the chamber or by varying the time
before onset of the evacuation through the GC of the
contents of the chamber. After finishing the environ-
mental chamber's wall loss rate studies, another
series of experiments could be done that measures
wall loss rates in the tubing. This could be done
by placing a variable flow constrictor in front of

the exhaust pump so that the time spent in the tubing
could be varied to see if tubing wall loss was a
major factor in the absence of nitrate radical.
Various pumps could then be evaluated for flow versus
reactivity correlations by placing them upstream of
the GC
Another situation that was not considered
earlier to be of significance is incidental light.
While the chamber itself is enclosed and dark (there
is a small hole used for plumbing purposes that is
less than an inch in diameter and filled with tubing),
the circulation tubing leading from the chamber to
the GC and back is made of translucent teflon. This
approximately two meters of tubing has a residence
time for the gases of about six. minutes. Although
the lab has typically been kept darkened, there may
have been enough incidental light penetrating the
teflon tubing to destroy the nitrate radicals.
The efficiency of this reaction would have
to be extremely high if this were the case but there
exist certain easy and economical steps that could
be taken that might eliminate even this faint
possibility. Black felt or electrical tape could
be wound around the tubing to stop the major influx

Full Text


1.33xl0-27(T/300)0.32ell,080/T I( i.4








t ... .t





PAGE 107