Measurement of peroxyacetl nitrate in Denver

Material Information

Measurement of peroxyacetl nitrate in Denver
Landa, Christine Robin
Publication Date:
Physical Description:
ix, 89 leaves : illustrations, map ; 29 cm


Subjects / Keywords:
Peroxyacetyl nitrate -- Measurement ( lcsh )
Air -- Pollution -- Colorado -- Denver ( lcsh )
bibliography ( marcgt )
theses ( marcgt )
non-fiction ( marcgt )


Includes bibliographical references.
General Note:
Submitted in partial fulfillment of the requirements for the degree, Master of Science, Department of Chemistry
Statement of Responsibility:
by Christine Robin Landa.

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:
26882473 ( OCLC )
LD1190.L46 1992m .L36 ( lcc )

Full Text
Christine Robin Landa
B.S., Cal Poly State University, 1988
A thesis submitted to the
Faculty of Graduate School of the
University of Colorado at Denver
in partial fulfillment
of the requirements for the degree of
Master of Science

This thesis for the Master of Science
degree by
Christine Robin Land a
has been approved for the
Department of Chemistry
John Lahning

Landa, Christine Robin (M.S., Chemistry)
Measurement of Peroxyacetyl Nitrate in Denver
Thesis directed by Professor Larry G. Anderson
Peroxyacetyl nitrate (PAN) is a photochemical oxidant formed in
the atmosphere when hydrocarbons (HC) and NOx (NO + NO2) are
emitted in the air and exposed to the UV radiation coming from the
sun. PAN is a known eye irritant, a mutagen, and a phytotoxic
compound which has lead to severe crop damage in Los Angeles.
Although PAN is typically formed in urban regions, the organic
nitrates can be transported to rural and remote areas and can be an
indicator of photochemical smog.
There is a large body of measurements for PAN involving
urban, rural, and remote atmospheres but most of the studies do
not include the measurement of PAN in conjunction with other
reactive nitrogen oxides. Also, most of these studies do not involve
measurements of PAN in colder climates, such as experienced in
Denver during the winter. Previous PAN measurements mostly
involved short term intense studies carried out in one location and
do not contain concurrent measurements of NOx and O3. There is
virtually no information regarding the spatial and seasonal

variation of PAN. To better understand the reactive nitrogen cycle
locally and globally more information on PAN is necessary.
This research project involves the synthesis,
calibration, sampling, analysis, and data interpretation of PAN in
the Denver area. The samples were acquired at locations where the
concentrations of NOx and O3, and meteorological parameters are
This abstract accurately represents the content of the candidate's
thesis. I recommend its publication.

I INTRODUCTION.................................1
H CHEMISTRY OF PAN......................... 7
Formation Processes......................7
Loss Processes..........................10
Background on PAN Detection........... 13
Detection Systems.......................14
m EXPERIMENTAL METHOD...................... 25
Preparation of PAN Standard.............25
Calibration of PAN Standard........... 29
IV INSTRUMENTATION...........................40
Packed Column Method....................41
Capillary Column...................... 48
Instrumental Determination..............51
Sampling Techniques.................... 57
PAN Concentration vs Date...............66
PAN Concentration vs Date 12-3 pm.......66
PAN Concentration vs Temperature........67
PAN Concentration vs Ozone............ 69

PAN Concentration vs NO.................70
PAN Concentration vs time 30 April 91...72
PAN Standard........................... 74
VI CONCLUSION........................... 77
Future Work.............................81
REFERENCES....................................... 83
APPENDIX A.........................................88

1. Movement of electrons during the pulse applied by the ECD..16
2. Electron capture detector.................................17
3; HNu photoionization detector..............................2i
4. FTIR of PAN standard...........................<..........30
5. IPN/PAN solution........................................ 32
6. Calibration curves:
a) no internal standard...............................34
b) internal standard..................................34
7. Ion Chromatogram of PAN standard..........................38
8. Schematic drawing of the overall analytical system........41
9. Example chromatogram with air peak........................44
10. Ambient air sample chromatogram..........................48
11. Capillary chromatogram of ambient air....................51
12. Detector temperature determination.......................53
13. Historical bar graph of [N02]/[N0] at sampling sites.....58-59
14. Map of sampling site.....................................61
15. PAN concentration vs date............................67
16. PAN concentration vs date between 12-3 pm................68
17. PAN concentration vs temperature....................... 69
18. PAN concentration vs ozone............................. 70
19. PAN concentration vs ozone on 30 April 91................71
20. PAN concentration vs NO ppb..............................72
21. PAN concentration vs time on 30 April 91.................73

22. Singh & Salas PAN diurnal variation of PAN

1. Rate constants and lifetimes for the thermal decomposition of
PAN at various temperatures.........................11
2. Ionization potential of organic compounds...............20
3. List of compounds and manufacturers.....................27
4. Results of the FTIR.....................................30
5. PAN concentration over time........................... 38
6. Packed column systems...................................42
7. Analytical columns of varying composition...............43
8. Summary of retention times..............................47
9. Summary of capillary columns............................50
10. Rate Constant of PAN standard at -20C.................75

In the mid 1940's photochemical air pollution was first
identified in Los Angeles, California. In the decade to follow,
several research groups led by Haagen-Smit, Middleton, and Blacet
(Pitts, 1969) established that this smog was caused by the action of
sunlight on hydrocarbons and oxides of nitrogen present in the Los
Angeles atmosphere. This photochemical air pollution consists of a
complex mixture of gaseous pollutants and aerosols. Among the
gaseous compounds are oxidizing species such as ozone, O3;
nitrogen dioxide, NO2; and organic nitrates (defined as organic
compounds containing the covalently bonded-ONC>2 groups). Of
the organic nitrates, peroxyacetylnitrate (PAN) is the most
abundant. PAN has the following formula:
Ozone, nitrogen dioxide, and organic nitrates are grouped
together and called "photochemical oxidants." Ozone and
peroxyacetylnitrate are formed by chemical or photochemical

reactions of the primary pollutants after they have been released
into the atmosphere and have been exposed to sunlight.
While much of the initial attention has been given to O3 as
an important oxidizing agent, PAN was given little attention and
was rarely included in routine measurements. PAN, and other
peroxyacyl nitrates, are of concern because of their extreme
reactivity with biological materials. PAN can react with the
sulfhydryl groups of enzymes and with low molecular weight
amino acids (Mudd, 1975). These reactions help to explain the eye-
irritation properties of PAN and PAN's toxic effects on plants.
Because PAN is a strong oxidizing agent, damage to crops occurred
in Southern California and in other countries around the world
leading to severe economic losses of crops (Oshim et al.,1974). PAN
has also been suggested as a possible etiological agent in the
increasing incidence of skin cancer (Lovelock and Penkett, 1974).
Lovelock even reported that the sweet smell Of line dried clothes in
urban regions could be attributed to PAN.
Since PAN's precursors (HC's and NOx) are typically formed
around urban regions, the organic nitrates can be transported to
rural and remote areas where NO2 can be released into the
atmosphere (reaction -17) and participate in ozone formation
(reactions 8 & 9). PAN can be used as an indicator of photochemical
smog and serve as a reservoir for NOx (NO + NO2). To be more

specific NOx is part of the total reactive nitrogen, NOy (NOx +
HNO3 + organic nitrates + NO3 + N2O5 +...). In the troposphere,
nitrogen oxides (NOy) contained in the organic form may be much
more abundant than in the inorganic form (Singh et al., 1981). This
organic nitrate of reactive nitrogen is in chemical equilibrium with
inorganic NO2 and acts as a reservoir of inorganic nitrogen oxides.
The following sequence of reactions represents the
interrelationships found between the organic nitrates and other
reactive nitrogen species.
2 NO +O2 -> 2NO2 (1)
RO2 + NO > RO + NO2 (2)
HO2 + NO -> OH + NO2 (3)
NO + O3 -* NO2 + O2 (4)
NO + OH HONO (5)
NO + RO RONO (6)
NO + NO3 -+ 2N02 (7)
NO2 + hv NO + 0(3p) (8)
0(3p) + 02 03 (9)
NO2 +OH > HNO3 (10)
NO2 +O3 NO3 +O2 (11)
> NO + 202 (12)
NO2 + NO3 N2O5 (13,-13)
-* NO + NO2 + O2
NO2 +HO2 HO2NO2 (14)
NO2 + RO2 O RO2NO2 (15, -15)

N02 + R0 RON 02 (16,-16)
CH3C(0)00 + NOz ** CH3C(0)00N02 (17,-17)
The main path for the removal of NOy from the atmosphere
is via reaction 10, and the subsequent wet and dry deposition of
NO3- and HNO3. The remaining inorganic species, reactions 5,11,
13, and 14, are photolytically and / or thermally unstable and
therefore do not represent reservoir species, which are involved in
other important atmospheric chemistry processes. The reactions in
italics represent the temporary reservoirs for NO2, including the
organic nitrates.
There is a large body of measurements for PAN. These
studies have involved measurements in urban, rural, and remote
atmospheres. Most of these studies have not included
measurement of organic nitrates in conjunction with other reactive
nitrogen oxides. To better understand the role of organic nitrates in
the reactive nitrogen cycle, concurrent measurements of NOy are
necessary when monitoring PAN. Also, most of these studies do
not involve measurements of PAN in colder climates (i.e.
subfreezing temperatures). PAN measurements have involved
short term intense studies carried out in one location, and most of
these studies do not include concurrent measurements of NOx and
O3. There is virtually no information available regarding spatial

and seasonal variation of PAN. To better understand the reactive
nitrogen cycle locally and globally, more information on PAN is
This research project involves the synthesis, calibration,
sampling, analysis, and data interpretation of PAN in the Denver
area. These samples were acquired at locations where NOX/ O3, and
other meteorological parameters are available.
From the onset of studying organic nitrates, most of the
compounds have been misnamed and the nomenclature is
ambiguous. The IUPAC system gives no definite name for the
substitution of -OONO2. However, one can construct the term
"nitrooxy" for a covalent bond -ONO2 and "dioxy" for the 0-0
linkage hence the term "nitrodioxy". The term "peroxy" is
substituted for the term dioxy which results with the term
"nitroperoxy". The organic nitrates are often named by linking two
different group names. Such is the case with PAN. PAN is an
anhydride of two different acids, commonly known as mixed
anhydrides. According to IUPAC rule NO. C-491.3 (Martinez, 1980)
mixed anhydrides are named by giving the specific part of each acid
in alphabetical order or order of complexity, followed by the term
"anhydride". In 1980, Martinez renamed "peroxyacetyl nitrate"

(PAN) as ethaneperoxoic nitric anhydride (EPNA), since it is a
mixed acid anhydride. Unfortunately, this leads to the acronym,
EPNA, which could explain why the proper name has not been
adopted. A proposed compromise recognizes the anhydride
structure of the compound, but retains the "PAN", is with the name
peroxyacetic nitric anhydride, which some researchers have started
to use.

The chemistry of organic nitrates, PAN, involves a number
of formation processes, reactions 18-25, and loss processes, reactions
28-35. These processes will be separated into two sections,
formation processes and loss processes which are discussed in the
subsequent sections.
Formation Processes
The sources of organic nitrates can be divided into two
categories: (1) primary emission from anthropogenic sources such as
combustion and chemical process, (2) secondary formation from
atmospheric photo-oxidation of organic compounds in the presence
of NOx.
Category 1 represents the emission of RONO2 from
specialized industrial processes such as the explosives industry
(Thompson et al., 1979) but this is probably an extremely minor
source and not normally considered in the formation of PAN.
Category 2 organic nitrates are formed by secondary reactions
through the atmospheric oxidation of organic compounds initiated
by O3, OH, and NO3. The nitrate formation mechanisms operating

in all of these processes are often minor pathways and therefore
difficult to quantify. The main formation process of PAN is a
product of reactions among hydrocarbons and nitrogen oxides in
polluted urban atmospheres (Pitts, 1969). Some of these non-
methane hydrocarbons, such as ethane and propane, have been
reported in unpolluted atmospheres. The atmospheric oxidation of
ethane has been suggested as a significant PAN source in the
troposphere and lower stratosphere (Singh et al., 1985). From these
studies, it has been determined that the oxidation of ethane (C2H6)
is initiated when a OH radical abstracts a hydrogen atom. Below is
an example of the oxidation of ethane (Singh and Hanst, 1981).
CH3CH2 + O2 ->
CH3CH2O + O2
CH3CHO + NO3 -*
CH3C(0) + O2 -*
CH3C(0)00 + NO2 ->
CH3CH2 + H2O (18)
CH3CH2OO (19)
CH3CH2O + NO2 (20)
CH3CHO + HO2 (21)
CH3C(0) + H2O (22)
CH3C(0) + HNO3 (23)
CH3C(0)00 (24)
CH3C(0)00N02 (PAN) (25)
Besides the above reactions for the formation of acetyl
radicals, acetyl radicals can be formed from the oxidation of
aromatic hydrocarbons and through the use of high oxygen fuels.
Also, the hydrocarbon oxidation processes leading to the formation

of PAN, oxidizes NO to NO2. This NO2 can photolyze forming O3.
Hence, in urban areas there is often a strong positive correlation
between O3 and PAN formation.
High Oxygen Fuel
Oxygenated fuels such as ethanol and MTBE ( methyl tertiary
butyl ether) blends have been receiving increased attention as both
automotive and industrial fuels (Grosjean and Miguel, 1988).
Denver has been using MTBE blended fuels during the winter
months in the hopes of decreasing carbon monoxide emissions.
Direct emissions from these fuels include MTBE and ethanol, the
unburned ether or alcohol and the aldehydes, formaldehyde and
acetaldehyde. As demonstrated earlier via reaction 22, acetaldehyde
can go on to form acetyl radical and hence PAN. Beside the direct
emissions of acetaldehyde, acetaldehyde can be formed by the
oxidation of alcohols. Below is an example of the oxidation of
CH3CH2OH + OH - CH3CHOH + H2O (26)
CH3CHOH+O2 -* HO2 + CH3CHO (27)
Hence, oxygenated fuels can increase the concentration of PAN .
This is another reason why PAN concentrations should be
monitored in the ambient air especially in cities adopting an

oxygenated fuel program. Even though the formation process of
PAN is interesting, the decomposition of PAN is of greater
atmospheric importance.
Loss Processes
The lifetime of PAN is determined by the O-N bond cleavage
which is the reverse of reaction 25. This reaction is the rate limiting
step in the decomposition of PAN and is thermally controlled.
Thus, the lifetime of PAN will be a function of the ambient
temperature. Below is the decomposition reaction chemistry of
CH3C(0)00 + NO2
CH3C(0)00 + NO
CH3C(0)0 ->
CH3 + 02 -
CH3O + 02
HO2 + NO
NET: CH3C(0)00N02 + 3NO ->
CH3C(0)00 + NO2 (28)
CH3C(0)00N02 (29)
CH3C(0)0 + NO2 (30)
CH3 + CO2 (31)
CH3OO (32)
CH3O + NO2 (33)
HCHO + HO2 (34)
OH + NO2 (35)
4N02 + CO2 + HCHO +OH
By inspection of the decomposition process of PAN, the
peroxyacetyl radical formed can either reform PAN via reaction 29
or if sufficient NO is present in the air, the peroxyacetyl radical can
oxidize NO to NO2. Once reaction 30 has occurred, the product

radical can then thermally decompose via reaction 31, preventing
PAN from being reformed. Hence, the decomposition of PAN
depends on temperature and the ratio of NO2/NO.
To determine the half-life of PAN one can use the Arrhenius
= 2.52 x 1016 e_13,543/r,s_1
Table 1 summarizes the estimated half lives of PAN as a function of
temperature (Tuazon et al., 1991). From this table, PAN has a
significant half-life at 0 oC. Thus, at low temperatures, PAN can
exhibit lifetimes greater than a day and, therefore, can be expected to
accumulate in appreciable concentrations in colder climates
(Grosjean, 1979).
Table 1: Rate constants and lifetimes for the thermal decomposition
of PAN at various temperatures
Temperature (C) k(s-!) 1/k
-30 1.6xl08 2.0 yr
-20 1.3 x 10-7 88 days
-10 1.0 xlO"6 10 days
0 3.4x10-6 3.4 days
10 2.0 xlO5 13.7 hrs
20 1.1 x 10-4 2.6 hr
30 5.0x10-4 33 min
40 2.1 x IQ"3 7.8 min

As mentioned earlier there is a correlation between the
NO2/NO concentration and the decomposition of PAN. In the
presence of NO and NO2, the decomposition of PAN is governed by
reactions 28-30. Thus the rate of PAN decomposition can be
-d[PAN] ^[PANfNO]
dt ^[NO]* ^[NOJ w
-d-ln[PAN] kMj.NO]
dt k30[NO] + k^[NO2\ W
Where k29 and k30 are the bimolecular rate constants.
Rearrangment of equation II leads to the following expression:
1 *[NOi\
^28 ^28^30
By inspection of equation HI, the relationship between [N02]/[N0]
and PAN decomposition is more apparent.
One more important process of PAN's decomposition should
be mentioned again. This point is the correlation between PAN
concentration and ozone formation. The overall net equation for
PAN decomposition reveals 4 moles of NO2 are produced. Once
NO2 is formed during the day, NO2 can photolyze and form O3.

N02 + hv -* NO + 0(3P)
0(3P) + 02 - 03
If PAN concentration is high, then ozone can be high and vice-
versa. Again this demonstrates the importance of monitoring PAN
concurrently with ozone.
PAN was first detected in the atmosphere as part of a series of
compounds called compound "X". In 1956, Stephens and co-
workers partially characterized one of the homologs in this
compound "X" series with long-path cell infrared spectrometry .
Long path IR was not a practical detector for PAN because of its
limited sensitivity. Following long path infrared spectrometry, the
gas chromatography with flame ionization detector (FID) was
employed. However, since atmospheric samples contain numerous
hydrocarbons, aldehydes, and ketones; PAN could not be
distinguished from these peaks. GC with the electron capture
detector (GC/ECD) seemed to be a viable alternative and has become
the predominant detector used in the measurements of organic
nitrates. ECD's are selective and sensitive for compounds that can
capture electrons and can measure compounds in the part per
trillion (ppt) level. More recent research has been involved with a
Background on PAN Detection

new analytical system such as the Luminox detector, based on
luminol chemiluminescence, and NOx detector, based on NO/O3
chemiluminescence (Drummond et al., 1988). These systems have
the advantage of being specific for nitrogen compounds and might
require less handling of ambient samples. Even though GC systems
are highly sensitive and selective, some problems with GC
techniques result from sample handling.
This project will employ a GC/ECD, GC/PID. These different
systems will be discussed in sections to follow.
Detection Systems
Electron Capture Detector
Theory. The electron capture detector uses a source of beta
rays generated from energetic electrons from nickel 63 or tritium
imbedded in titanium. These radioactive particles ionize the carrier
gas via reaction 36. One beta ray will generate millions of electrons
and have high mobility in the electric field set up by a polarizing
voltage applied to the cell. The electrons migrate to the positive
electrode. The nitrogen molecules are neutralized at the negative
electrode. This leads to a standing current through the detector. An
electron absorbing compound such as PAN can absorb some of these
electrons, reaction 37.
P + nN2 -* nN2+ + ne- (36)

PAN + e- -* PAN- + energy
PAN- + N2 + - PAN* + N2
The negative PAN iort is much less mobile than the free electron so
it can recombine with a positive nitrogen ion, reaction 38. The
presence of PAN thus leads to a decrease in the standing current
through the detector. The energy rich neutral PAN molecules
then decomposes into neutral species.
The early design of these detectors involved a constant DC
voltage apllies to the cell. Later designs improved the linear range
of the detector by applying a voltage pulse and maintained by a
constant frequency or constant current in the cell. This is
accomplished by the cell current which is measured and compared
to a reference current. The pulse interval is then adjusted to
maintain constant cell current. Hence, the pulse rate is converted to
a voltage, linearly related to the amount of electron-capturing
material in the cell. Recent work has demonstrated that a constant
frequency ECD is more sensitive to organic nitrates than a constant
current ECD (Buhr, 1990).
The voltage pulse applied to the electrodes cause the non-
captured electrons to be periodically collected. Figure 1 shows the
movement of electrons during the voltage pulse applied by the

SH 0- 0- &
#Rv8:v H \/ N2-B-
ipAN 1
----B-il Nl'
3- = beta particle
N2 = carrier gas
e- = low velocity secondary electrons
e-PAN = PAN with captured electron
Figure 1: Movement of electrons during the pulse applied by the
ECD Operation. Figure 2 is a diagram of a commercially
available ECD, where the anode and the cathode are in a "pin and
cup" geometry. The radioactive source is Ni63 plated on the walls of
the cup. The temperature limit for this detector is approximately
400C. The carrier gas can be nitrogen or 5% methane in argon.
(Methane helps to produce more low energy electrons through
energy-reducing collisions in the cell.) The choice of carrier gas is
important because nitrogen carrier gas in some cases produced a
greater signal (more sensitive) than argon/methane. However,
nitrogen also results in greater noise, so sensitivity is not

significantly changed. Flow and temperature have been shown to
have significant effect on the ECD response.
Figure 2: Electron capture detector (ECD)
Photoionization Detector (PIP')
Theory. Photoionization is one of the processes by which an
atom or molecule can absorb energy. It is a result of an electron
transition from one of the discrete, low energy levels (usually the

ground state) to the higher energy continuum of the ion, and the
energy required is about 5-20 eV. For the photon excitation, this
corresponds to the vacuum ultraviolet region of the spectrum.
Reaction 39 demonstrates this process.
R + hv -> R+ + e- (39)
R = ionizable species
hv = photon energy > ionization potential of R
The photoionization detector (PID) detects organic and some
inorganic species in the effluent of a gas chromatograph, in the
picogram to microgram range. The PID is equipped with a sealed
UV light source that emits photons which pass through a UV
transmitting window into an ionization chamber where photons
are absorbed by the eluted species. Those species with an ionization
potential less than the energy of the UV source are ionized. A
positively biased high voltage electrode accelerates the resulting
ions to a collection electrode. The current due to ion flow is
measured by the electrometer and is proportional to the
Quenching is the neutralization of ionized species that results in a
decrease of the measured ion current (signal). The following
equations indicate the mechanisms of photoionization and

R + hv * R+ + e- (39)
R+ + e- R (40)
PAN + e- PAN- (41)
PAN- + R+ > PAN + R (42)
R = ionizable species with an ionization potential < hv (i.e.: sample;
background; column bleed; carrier gas impurities).
hv = photon with an energy > ionization potential of R.
PAN = electron capturing species
The quenching caused by electrons (reaction 40) is inefficient
and therefore minimal. Quenching via electron capturing species
(reaction 41 and 42) is much more efficient as a result of the larger
diameter of the negative ions (i.e, PAN-) and thus results in a much
larger decrease in signal.
Selection of the lamp for analysis is based upon selectivity
and sensitivity requirements. Maximum sensistivity is achieved
with the 10.2 eV lamp which is 10-20 times more sensitive than any
of the other lamps and is used for most analytical applications.
However ,' molecules with ionization potential (IP) up to 0.3 eV
higher than a given lamp energy can be ionized and detected,
however, with less efficiency than molecules with IP's at less than
the lamp energy (HNu, 1986).

The most common lamp energy, is 10.2 eV, and compounds
having greater ionization potentials would give minimal response.
Table 2 lists some examples of ionization potentials.
Table 2: Ionization potential of organic compounds.
Compounds || IP (eV)
n2 15.6
ch4 13.0
h2o 12.6
CH3CH3 10.2
n-propyl nitrate 11.07
CCI4 11.42
PIP Operation. The PID consists of a UV lamp and an
ionization chamber. The lamp is filled with a carrier gas and
produces an emission line characteristic of the gas when excited.
The radiation passes through a metal fluoride window and into the
ionization chamber or cell. Here the sample absorbs the radiation
and ionizes. Electrodes collect the ions, and the current measured is
proportional to the sample concentration. Figure 3 is a diagram of a
commercial photoionization detector. The lamp and window are
selected based on the ionization potential of the compound to
measure, and the selectivity and sensitivity required.

Ionization Chamber
Figure 3: HNu photoionization detector
NO 03 Chemiluminescent Analyzer
Theory. PAN has also been measured using commercially
available analyzers. These analyzers are based on
chemiluminescent methods in which NO reacts with ozone to
oxidize NO to NO2 in an excited state. The near instantaneous
transition of excited NO2 is accompanied by photon emission (hv)
between 600 and 2500 nm by the following reactions.
NO + O3 -> NO2* + O2 (43)
NO2 * -* NO2+ hv (44)

The photo emission is converted into an electrical output by a
photomultiplier tube and associated electronics.
Ambient air enters the analyzer and passes through a
chemical convertor either comprised of a molybdenum or
gold/carbon monoxide. First the incoming air bypasses this
convertor and measures only NO. Then a solenoid valve directs
the air flow through the convertor and reduces NO2 and organic
nitrates to NO.
Few researchers have used the NOx analyzers for field
samples, most have used the NOx analyzers to calibrate a PAN
standard. Since the NOx analyzer converts most organic nitrates to
NO2, an analytical column needs to be used to separate PAN from
other organic nitrates. This could be one of the reasons more
researchers have not used NOx analyzers for ambient sampling.
For field sampling of PAN, Joos et al. (1986) and Tsani-Bazaca
et al. (1988) have employed the NOx analyzer in conjunction with
an GC/ECD system. Using the principles described earlier, a sample
containing PAN is first isolated from the byproducts produced in
the synthesis of PAN or any other possible interfering compounds
from ambient air. This separation is done by a packed column of
the same material used in a GC/ECD system. These samples are
then passed through the NOx analyzers and recorded on a strip
chart. Fahey et al. (1986) have described the same type of procedure

with one major difference. No column is used to separate the PAN
peak from any other possible interference such as methyl nitrate,
ethyl nitrate, and nitric acid. This could lead to a higher
concentration of PAN than is actually in the standard. The
advantages to using the NOx analyzer for calculating the
concentration of the PAN standard is that PAN dilutions are not
necessary, e.g. 5 uL injection into a 50 L Teflon bag. This could
minimize PAN decomposition and obtain a more accurate PAN
concentration. Problems arise since a GC column is still needed for
PAN separation. The other problem with the NOx analyzer is that
the detection limit is 1 ppb. If PAN concentration are below or close
to lppb, PAN would not be detected or there would be more error in
the determined PAN concentration
Luminox Detector
Theory. Luminol (5-amino-2,3-dihydro-l,4-
phthalazinedione) has been under study since 1928 and is known to
chemiluminescence with a number of oxidizing agents. In 1980,
Maeda described the development of an NO2 detector based upon
the chemiluminescent reaction of NO2 with luminol. Using this
recent development, Burkhardt and coworkers (1987) used the
luminol system for the measurements of NO2 and incorporated the
instrument to measure PAN. Again, a column had to be used to

separate the PAN peak from other nitrogen containing compounds.
This method offers several advantages over the above-described
GC/ECD method for PAN and the NO + O3 chemilumininescent
method for NO2, including selectivity, portabilility, and the absence
of compressed gas cylinders. Potential disadvantages of the system
include temperature dependence and other sources of sensitivity
drift of the luminol-based detector and nonlinearity at low
concentration of NO2 and PAN.
The system used to measure PAN by Burkhardt (1987) and
coworkers was a commercially available Scintrex LMA-3 Luminox
instrument. The air used to zero the instrument was ambient air
passed over a ferrous sulfate trap. This trap converted NO2 to NO
and removes PAN from the ambient air. Any NO2 or PAN that did
pass through the trap was interpreted as part of the base-line signal
for the chromatograph. The study developed by Burkhardt and
coworkers (1987) only looked at PAN in standards and did not
include ambient PAN concentrations. The more recent Scintrex
LPA-4 uses a GC column to acheive interference free detection of
PAN. From the column the sample is first fed into a converter
where it is converted to N02, then directed into the Luminox
detector. Overall, the luminol-based chemiluminescence detector
can be used as a monitoring system for atmospheric concentrations
of PAN and NO2.

Preparation of PAN Standard
Since PAN can not be purchased in the United States,
synthesis of PAN is required. In 1965, Stephens and coworkers
described the preparation of pure PAN vapor via three different
methods :
(1) Dilute mixtures of symmetrical olefin with either nitric oxide or
nitrogen dioxide in dry oxygen or air.
(2) Photolysis of dilute alkyl nitrite in dry oxygen.
(3) The dark reaction of the appropriate aldehyde with NO2 and O3
at low concentrations in oxygen.
All three of the above techniques produced PAN as well as
byproducts. These byproducts were separated via GC/ECD
(Stephens et al., 1965). In 1969, Stephens reported an explosion that
was a result of PAN synthesized in the gas phase. Hence, to avoid
further problems research was underway for a safer method of PAN
In 1982, Nielsen, Hansen, and Thomsen developed a simple
procedure for the nitration of peracetic acid, followed by extraction

with hexane and storing at or below -10C (Nielsen et al., 1982).
Even this method suffers from the fact that the PAN produced is
inevitably contaminated with reagents, side products, or
decomposition products of PAN. Following this work, Gaffney,
Frajer, and Senum (1984) modified Neilsen method which did not
require any purification of the reaction mixture since byproducts
were not formed. Gaffney (1984) and coworkers substituted a heavy
lipid solvent (n-tridecane) in the extraction steps which enabled the
separation of liquid PAN from the solvent and other reagents,
allowing yields of greater than 98% PAN (Gaffney et al., 1984). This
method was followed, with a few modification, for the synthesis of
PAN. Table 3 lists the compounds used in the synthesis of PAN.
Reaction 45 is the mechanism for the synthesis of PAN, and the
following paragraphs will outline the method of synthesis used in
our work.
CH3C(0)00H -* CH3C(0)00N02 + byproducts (45)
Before the reaction is started all glassware should be cleaned
and baked overnight to ensure no contamination. Also, all the
glassware and water used in the reaction is placed in the freezer
until needed. A round bottom flask containing 25 mL of n-
tridecane is suspended in an acetone/ice bath. After the n-tridecane

has reached a temperature of 0 C, 2.5 mL of 32 % (by weight)
peracetic acid and 2.0 mL of concentrated sulfuric acid are added to
the solution.
Table 3: List of compounds and manufacturers
Chemical || Company
32% peracetic acid (by weight) Aldrich
98% n-tridecane Aldrich
18 megaohm/cm H20 Our Lab
69.0-71.0% nitric acid EM Science
96.3% sulfuric acid J.T.Baker
99.5% Anhydrous Magnesium Sulfate J.T.Baker
Again, the solution is allowed to react for 5 minutes to ensure that
the temperature is 0 C. Slowly, 0.5 mL concentrated nitric acid is
added to the cooled solution, maintaining the solution below 5 C .
After all the nitric acid has been added, the solution is mixed for 15
minutes. Less reaction time results in insufficient PAN formation
and longer reaction time increases the possibility of other products
being formed. The organic layer of the reaction mixture is decanted
into a 125 mL separatory funnel containing ice cold water. The
funnel is shaken gently and the aqueous layer is removed. This
procedure is done twice. The purpose of this is to remove
unreacted nitric acid and any byproducts from the n-tridecane layer.

After the last washing, a small amount of anhydrous MgS04 is
added to the separatory funnel to remove any entrained water. The
solution is then decanted and stored in a closed container at -20 C.
At -20 C, the PAN solution is reasonably stable. Using table 1, the
lifetime of PAN is approximately 3 months. In a period of two
months, the PAN standard showed 25% PAN decomposition. The
reason for not showing more decomposition has to do with the
PAN standard being in the liquid solution. Since PAN is at a low
temperature and there are no sources of NO, the peroxyacetyl
radical will be more likely to reform PAN than further decompose
to form methyl nitrate and CO2
The thermal decomposition of PAN is known to produce two
major products, methyl nitrate and carbon dioxide. Work done by
Senum, Frajer, and Gaffney (1986) demonstrated that PAN
decomposes via a unimolecular mechanism. Hence, this
mechanism could explain why the PAN standard does not
decompose as readily as suggested by the lifetime calculated in table
1. The above unimolecular decomposition process could assist in
verifing the synthesis of PAN. Once PAN was thought to be
synthesized, the standard could be placed at room temperature. If
another peak appears on the chromatogram, then a compound was
synthesized that exhibits the same thermal characteristics as PAN

was synthesized. To date, the only known source of methyl nitrate
is via PAN decomposition in ambient air.
Calibration of PAN Standard
Once the PAN standard was synthesized, the structure and
concentration needed to be determined. Initially the long path FTIR
was going to be used for both structure and concentration
determination. Using Beer's law, concentration could be
determined via preexisting coefficients and a spectra would be
obtained in the process. Since the long path FTIR was not
functioning we were able to use a GC/FTIR at the Boulder campus.
These spectra were compared to published papers (Holdren and
Spicer, 1984 ; Bruckmann and Willner, 1983). The results are listed
in Table 4 and figure 4 is an example of the GC/FTIR.
With the GC/FTIR, concentration could not be determined.
The IR beam passes through a flow cell coated in gold. This causes
the light to reflect back and forth in the flow cell, hence the absolute
path length could not be determined. With the first problem
solved, the question of concentration needed to be answered. Most
researchers have used NOx, IC, and IR to determine the
concentration of their PAN standard. Since none of these methods
were available, another simple method was used initially.

Table 4: Results of the FTTR
Researcher v(cm-l) researcher v(cm-l) Assignment
Gaffney 1840 Our Work 1840 C=0 stretch
1799 1799 N02 asym stretch
1301 1302 N02 sym.stretch
1164 1162 C-0 stretch
929 929 0-0 stretch
794 792 NO scissors

A compound that had a similar structure to PAN, isopropyl
nitrate (IPN), was used as an internal standard via the following
method. Since the purity, and also the concentration, of the
isopropyl nitrate was known, dilutions were made of the stock
solution and injected into the GC/ECD setup for liquid injection.
Once a reasonable area was determined for IPN, PAN was injected
into the solution. Since PAN decomposes at elevated temperature
the GC parameters were varied until an optimal PAN area was
determined. Once, this was determined, the liquid PAN solution
was added to the solution until the areas of the PAN and IPN were
within 90% of each other. Figure 5 is an example of the
chromatograph obtained of IPN/PAN solution.
One important assumption with this method was that the ECD had
the same response to IPN as with PAN. With this in mind, the
concentration of PAN in the liquid phase was determined. Since
the PAN concentration was known in the liquid phase, PAN
concentration could be determined in the gas phase using the ideal
gas law. The following is an example of how these calculations
were performed.
PAN in the liquid form= 70.4 ppm
P=0.825 atm, V=50 L, R= 8.20575 L atmK-lmol-1, T= 293 K, F.W. of
tridecane= 184 g/mole, density = lg/ml
n moles of air =(50L)(0.825 atm) =1.67 moles of air
24.4 L x atm/ mole

Injecting 5 uL of PAN solution into a 50 L bag:
(5.00.10-3 mL) (1.0 g/mL) (mole /184 g)= 2.717-10-5 moles of
(2.717-10-5 moles of solution) (70.4 parts PAN in solution)/ 1,000,000
parts of solution =2.89.10-9 moles of PAN
(2.89.10-9 moles of PAN) / 1.67 moles of air= 1.69.10-9
= .00169 ppm= 1.7 ppb of PAN
To calibrate the ECD, a fifty-liter Teflon bag was inflated while
PAN was injected into the bag via a gas tight syringe. By varying

the amount of PAN injected into the bag, the concentration of PAN
will vary. Five different concentrations of PAN were injected and a
five point calibration curve was determined. From this curve the
area of an ambient sample is entered into the computer and the
PAN concentration is determined. Figure 6 is an example of the
calibration curve. Initially, no internal standard was used when
calibrating the PAN standard in the gas phase. But later the
possiblity of an infernal standard was explored. Again IPN was used
as an internal standard. Three different methods were explored:
1) Adding PAN to a solution of IPN and varying the injection of
PAN/IPN solution. One injection technique.
2) Making a solution of IPN and pulling up 5 uL of solution into a
gas tight syringe. With the syringe containing IPN, pull up the
PAN solution while making sure there are no air bubbles. The
same amount of IPN is used and the PAN solution is varied;
again, a one injection technique.
3) Injection of the IPN and PAN solution separately: The IPN
solution volume, was 5uL, and the PAN solution volume was
varied from 1-5 uL; a two injection technique.
A one injection technique was tried first to minimize error in
the injection process. First 5 injections of the same concentration of
the IPN solution were run to see the variation of the injection.

Concentration (ppb) Concentration (ppb)
y = 7.6055e-2 + 1 3B78e-6x FT2 = 0.990
(Figure 6a)
y = 1.78360-2 + 1.1738e-6x RA2 = 0.935
(Figure 6b)
Figure 6: Calibration curves, a) no an internal
b) internal standard

This was also done with the PAN standard. Method 3 was
experimentally determined to be the best method. Methods 1 & 2
showed a decrease in the IPN and PAN response and the injection
repeatability was poor. Since the injections of IPN and PAN were
without problem before the solutions were added together a
reasonable explanation for the discrepancy would be an interaction
between PAN and IPN in the liquid phase. Hence, method 3 was
used for the calibration of the GC/ECD. With the use of IPN the
calibration curve did not show any significant change (figure 6b)
hence, no internal standard was used for the calibration of PAN in
the gas phase. Moreover, ambient samples did not have an internal
or external standard because the volume of each sample varied.
After the first year of calibrating the PAN liquid standard as
described above, an ion chromatograph (IC) became available. The
hydrolysis of PAN in basic solution had been observed to be rapid by
Stephens (1967), Nicksic et al. (1967), and Helmig et al. (1984). The
products of the hydrolysis were found to be acetate and nitrite ions,
and molecular oxygen, consistent with the following overall
CH3C(0)OONC>2 + 20H- CH3COO- + NO2- + O2 + H2O
Hence the PAN concentration could be quantified using either the
acetate or the nitrite peak. For this project, the nitrite peak was used
for quantitation since the acetate peak could come from any

remaining glacial acetic acid or peracetic acid. The hydrolysis was
carried out via the following method:
1) A 4.2 mM Na2CC>3 (pH= 10.7) solution and a 1.4 mM NaHCC>3
solution were made.
2) PAN stock solution was thawed and 20 uL of this solution was
added to 5 mL of the 4.2 mM Na2CC>3.
3) The solution was shaken for five minutes.
4) 5 mL of 1.4 mM NaHCC>3 was added and the mixture was
shaken for another 5 minutes.
5) The IC was carried out on a Dionex 4500i, equipped with a
conductivity detector set at 3 uS, a 50 uL loop, the flow was set at
2 mL/min. Column; AS4A 4mm x 250mm; suppressor column,
25 mN H2SO4. Eluent, NaHCC>3 1.7 mM, & 1.8 mM Na2CC>3,
Regenerant 2.5N H2SO4
6) All work except for the IC runs was done at 4C to minimize
PAN decomposition.
7) To determine the accuracy of the IC procedure three hydrolysis
were carried out on each PAN solution and each solution was
measured three times. The % relative standard deviation was
8) 11 July 91 STD=688 ppm, 13 July 91 STD=2543 ppm
In calculating the concentration of the PAN solution, the nitrite and
nitrate areas were added together. Nitrite can oxidize to nitrate if

the samples are not analyzed soon after the hydrolysis takes place.
Because the samples were not analyzed for at least two hours and
up to two days after the hydrolysis, the nitrite and nitrate areas were
added together. The only source of the nitrite and nitrate peak was
PAN since each of the starting materials were hydrolyzed by the
above method. Chloride was found in tridecane, and acetate was
found in the peracetic acid/ glacial acetic acid. Figure 7 is an
example of an IC.
With a better method of calibrating the PAN liquid solution,
the values obtained were compared with the initial method of
calibrating the PAN standard the Values differed by 11%, with the
IC's values being higher. An explanation for the GC/ECD being
lower is the PAN standard could decompose in the injection port,
oven, or ECD detector.
The IC was performed periodically to determine the
concentration of the PAN standard. Since PAN will decompose
over a period of time, a new concentration needed to be
determined. Table 5 has the concentrations of PAN determined by
the IC and PAN/IPN method over a period of months.
The % of PAN decomposition between 11 July 91 and 19 Sept
91 using the IC method was 19%; the IPN method showed 29% PAN

111 i i i i i i i 111 i
4.00 6.00
Figure 7: Ion Chromatogram of PAN standard
Table 5 : PAN concentration over time
| Date || IC || IPN
11 July91 688 ppm 612 ppm
19 Sept91 545 ppm | 434 ppm
Also, the GC/ECD calibration curve performed on 11 July 91 was
compared to that performed on 19 Sept 91. The areas were
compared for each of the uL injection ranging from 1-5 uL. The %
difference in the areas ranged from 22%- 35%. The average value
was 26%. Hence, as shown by three different methods the PAN
standard decreased by approximately 25%. Another interesting
comparision of the PAN standard concentration was to compare the
IC values and the IPN values on the 11 July 91 and the 19 Sept 91.
The % difference on the 11 July 91 was 11% and the % difference on

the 19 Sept 91 was 22%. These different methods of analyzing the
PAN standard concentration give validity to the concentration of
the PAN standard and also give a range for the PAN standard.

Measurements of PAN have been accomplished by different
techniques. Most researchers have used GC/ECD with varying types
of packed columns and GC parameters. A review of the literature
reveals that packed columns are usually used with the stationary
phases 5-10% Carbowax 400,10% Carbowax 600, and 4.8% QF-1 +
0.18% diglycerol. These columns are usually made of glass, stainless
steel, and more recently Teflon. Because PAN is thermally labile,
low injection port temperatures, oven temperatures, and detector
temperatures are used, which vary from 17 C in oven temperature
to 80 C in the detector. Table 6 list column types and GC/ECD
parameters used by researchers in the last decade. More recent work
has involved the use of capillary systems and will be discussed in
sections to follow. Two separate instrumental techniques where
explored in this study: GC/ECD, and GC/PID. Also, a variety of
columns were compared to determine which was best for our
particular GC system.

Six port
Sampling Valve
Teflon Ambient
Sampling Bag
Figure 8: Schematic drawing of the overall analytical system. (Gas
chromatographic, detection, and chromatographic integrator)
Packed Column Method
As described in the previous section, packed columns varied
by length, stationary phase, liquid phase, and material of the
column. From examination of these different types of columns, the
column of final selection was 10% Carbowax 600 on 80/100 mesh
Chromasorb WHP. This type of column was selected after different
columns were assembled and experimentally tested. These
columns are summarized in table 7.
The retention times (RT) are defined as the time it takes for
the compound to eluted. Besides changing the length and column
phase, oven temperature and flow were varied.

Table 6: Packed column systems
Researcher Column type Oven/ECD temp.
Joos, L. (1986) 60 cm x 1.5mm i.d.5%CH4/ 95% Ar, 40ml/min) 4.8% QF-1 + 0.13% diglyerol on Chromosorb G-AW-DMCS, 80/100 mesh 350C/ N/A
Grosjean (1989) 1/8", 5ft., teflon (N2 carrier) 10% Carbowax 400 on Chromsorb P- DMCS, 60/80 mesh 30C/60C
Holdren & Spicer,. (1984) 1/8"(CH4/Ar carrier,70 ml/min) Carbowax 600 on Gas Chrom Z, 60/80 mesh 35C/60C
Lonneman (1982) 90 cm x 3.2mm o.d.,glass 10% Carbowax 600 on Gas Chrosorb Z, 80/100 mesh 23-25C/23- 25C
Peake, E. (1988) lm x 1.5mm od., glass(N2 carrier) 5% Carbowax 600 on Chromsorb W N/A
Roberts, J (1989) 60 cm x 1.5mm i.d., teflon (40ml/min) 10% Carbowax 600 on Chromsorb WHP 33OC/60OC
Singh &. Salas (1983) 36cm x 3mm i.d,teflon (5%CH4 / Ar, 30ml/min) 10% Carbowax 600 on Supelcoport, 80/100 mesh 33C/64C
Temple (1983) 30 cm x 3mm i.d. teflon (N2 / 70 ml/min) 5% Carbowax E400 on Chromsorb G- DMCS 23-25C/ N/A
Tsani- Bazaca.E. (1988) 60 cm x 1.5mm i.d.5%CH4/Ar, 30ml/min) 4.8% QF-1 + 0.18% diglyerol on Chromosorb G-AW-DMCS, 80/100 mesh 30C/30C
Rudolph (1987) 1.3 m x 2mm i.d. (N2, 50ml/min) 5% PEG on Chromsorb WHP, 80/100 mesh N/A
Our Work 63 cm x 0.8 mm i.d. (N2, 45ml/min) 10% Carbowax 600 on Chromsorb WHP, 80/100 mesh 26C/65C
The main problem with all of the above columns was the PAN
peak eluted too close to the air peak. The air peak lasted for

approximately 2 minutes and has a sizeable tail. By increasing the
length of the column, the PAN peak would elute later but at the
cost of PAN possibly decomposing through the column. This was
detected by the decrease PAN areas as the columns were lengthened.
The two different Alltech columns were purchased with the
expectation of better resolution of the PAN peak compared to the
hand packed columns.
Table 7: Analytical columns of varying composition
%liqiud phase | length || RT
5% Carbowax 400 45" 6.36
5% Carbowax 400 68" 7.12
10%Carbowax 400 62" 9.69
5% Carbowax 600 50" 6.49
10% Carbowax 600 62" 9.16
Alltech* 10% carbowax 600 62" 10.46
Alltech** 10% carbowax 600 62" 10.12
* Alltech column was purchased already packed.
** Alltech stationary support was purchased and the column hand
With both types of columns the resolution of the PAN peak was
worse and the detection limit for PAN also decreased. The amount
of PAN that could be integrated and quantified on the columns we
packed were not observed on either of the Alltech columns. Hence,

the best packing material for the analytical column was the process
we used for coating the solid support, Chromasorb WHP, with the
liquid phase, Carbowax 600, (described below). One other fact
should be mentioned. The type of Teflon material used for the
column also made a difference in the GC/ECD response to PAN.
The best type of Teflon was 1/8" o.d. PTFE tubing. Other Teflon had
a smaller Teflon diameter for the wall of the tubing and this
difference could account for the increase in the background signal of
the ECD detector. Since the background signal was higher some
sensitivity could be lost. An example of this air peak and tailing is
shown in figure 9.

Initially, the liquid phase was coated onto the solid support
using cyclohexane. The liquid phase was added to cyclohexane and
a known volume of solid support is added to the solution. After
allowing this mixture to rotate on a rotary evaporator for 24 hours,
the solvent was removed. Since the solvent and liquid phase had
similar boiling points, it was possible that some of the liquid phase
could have been removed. Consequently, dichloromethane was
replaced as a solvent. Since the ECD responds to the elution of
chlorinated hydrocarbons, column conditioning was required. This
will be discussed in following sections. After dichloromethane had
been removed, the Teflon columns were ready to be packed.
Silianized glass wool was inserted into one end of the column and
connected to an aspirator. The aspirator was turned on and the
column packing was pulled down into the Teflon column. While
the column was being packed, it was "tapped" to insure the column
was packed tightly. After the column was packed, glass wool was
inserted into the other end of the column and connected to a
nitrogen cylinder. The cylinder was then opened and the pressure
increased to 60 psi. This was to insure that the glass wool plug
would stay in place and, to for the column to be tightly packed. If
the column needed to be repacked, it would be done by the above

Once the column is packed, it was connected to the GC, flow
was initiated, but the column is not connected to the ECD. A 1/8"
stainless steel cap was placed over the ECD and the oven
temperature was increased to 100 oC. The column was allowed to
remain at this temperature overnight, hence conditioning the
column before being connected to the ECD. The following day the
oven was turned off and allowed to come to ambient temperature.
At this time, the column was connected to the ECD and the system
was checked to ascertain that carrier gas was flowing through the
column. Once the flow is confirmed, the ECD was turned on and
the temperature increased in intervals of 20 C until it reached the
operating temperature. Then the oven was turned on. The reason
for the careful procedure was that compounds could condense on
the ECD during these processes, causing a decrease in ECD's
sensitivity to target compounds.
The GC\ECD used in our lab is an HP 5890A equipped with a
constant current EC Ni 63 detector. The carrier gas used was high
purity nitrogen further purified by passing through a Supleco
air/gas purifying system. A 5 mL sample was directly injected onto
the packed column using a Valeo stainless steel six port injection
loop. Different size stainless and Teflon sampling loops Were tested
to determine if the stainless steel sampling loop decomposed PAN.
No PAN degradation was noticed and increasing the sample loop

did increase the PAN area, but at the expense of increasing the size
of the air peak. The sampling loop is controlled by the GC's purge
on/off control. In the purge mode the sample is pulled through the
sampling loop and not onto the column. When the purge is turned
off, the sample is moved onto the column. Different purge times
were explored, a time of 2.00 minutes was determined to be the
most appropriate time. The run can be started on the GC or on the
integrator. Once the run is started, there is a 2.00 minute void time
which can be subtracted from the retention time of PAN. This was
explained in the previous section. Figure 10 is an example of an
ambient air sample.
Since other peaks were found in ambient air, known
chlorinated hydrocarbons standards were injected into the GC/ECD
to help determine the identity of these peaks. Table 8 is a summary
of retention times for the Carbowax column.
Table 8: Summary of rentention times
compound || R/T
CC14 3.57
CC13CH3 6.04
CCl2=CCl2 6.04
PAN 9.13
H20 14.99

Figure 10: Ambient air sample
Capillary Column
As reported earlier, packed columns have been used for the
separation of PAN for the past 25 years. In the last decade, capillary
columns have be introduced and seem to be a logical choice for the
measurements of PAN due to increased resolution and potentially
shorter retention times relative to packed column system. Capillary
columns also have an advantage of being more sensitive as a result
of the narrow peaks and low flow rates in the capillary system.
Roberts (1989) tested three different types of capillary columns
which are as follows : 30 m x 0.32 mm i.d. fused silica coated with 1-

urn cross-linked methyl silicone (J&W Scientific, DB-1), 30 m x 0.32
mm i.d. fused silica coated with 1-um cross-linked 95% methyl
silicone 5% phenylsilicone (J &W Scientific, DB-5), and 15 m x 0.53
mm fused silica coated with 1-um cross-linked carbowax (J&W
Scientific, Durawax). Roberts (1989) compared these capillary
columns to a packed column and GC parameters that follow: 10%
carbowax 400 on Chromosorb WHP, oven temperature at 30 C and
a flow of 40ml/min was used. For the capillary columns, column
temperatures were 27-33 C with the flow rates of 1 mL/min for the
DB-1 and DB-5 columns and 2 mL/min for the Durawax column.
Besides the work by Roberts (1989), capillaries columns have been
used by Buhr and coworkers (1990). The columns used and the
measured retentions times of PAN and related species are
summarized in Table 9.
The above results demonstrate that PAN can be separated by
the use of capillary columns. By inspection of the chromatograms
of the DB-1 of Robert's work, and the DB-1 of Buhr's, it is apparent
that there are problems inherent to their methods. In Robert's
work, the problem arose with the separation of methylene chloride
from methyl nitrate otherwise the DB-1 column resolved the
halogenated compounds from the organic nitrates. With Buhr's
DB-1 column, PAN and chloroform were difficult to separate.
Moreover, methyl nitrate was not reported. Hence, methyl nitrate

might have co-eluted with air. Since the only source of methyl
nitrate is PAN, methyl nitrate could be used to monitor the
decomposition of PAN. Figure 11 is an example of a capillary
Table 9: Summary of capillary columns
Stationary phase, thickness Column Retention of PAN/ Methyl nitrate Reference
10%carbowax 400 ChromasorbWHP packed 18.7/5.68 Roberts, 1989
DB-1,1 um .32 mm x 30 m 5.11/1.19 Roberts, 1989
DB-5, i um .32 mm x 30m 6.17/1.37 Robert, 1989
DB-wax, 1.5 um .53 mm x 15 m 6.17/1.37 Roberts, 1989
DB-1,1.5 um .53 mm x 30 m 3.27/na Buhr, 1990
DB-17,1 um .53 mm x 15 m na/na Buhr, 1990
DB-Wax, 1.5 um .53 mm x 15 m na/na Buhr, 1990
10%carbowax 600 ChromasorbWHP packed 7.80/na this work
The conclusion of this review of the different capillary
systems is that packed columns still should be used in conjunction
with capillary columns because of the problems with capillary
columns, as mentioned above. The capillary columns present
themselves as viable alternatives, but more research is needed. In
this research project, we were unable explore the different capillary
systems because our GC/ECD system was an older model, additional

Hewlett Packard parts were not readily available to convert the
system over to a capillary system. Future work should explored the
use of capillary columns.
Figure 11: Capillary chromatogram of ambient air (Buhr, 1990)
Instrumental Determination
The ECD and PID theories have been described earlier in the
paper. The following section will describe the parameters for the
ECD and PID.
Since many different injection, oven, and detector
temperatures had been used (table 7), the best set of these
parameters needed to be determined for our instrument. The first

parameter to evaluate was the detector temperature. With the
injector and oven turned off, the detector temperature was elevated
to 120 C. With the initial temperature, 5 different PAN injections
were performed. The detector temperature was reduced in intervals
of 10 C and 5 points were determined as described above. The
same concentrations and volume, of PAN solution
was used for all the detector temperatures. Figure 12 is a plot of area
of PAN standard vs. detector temperature. Since the concentration
of the PAN standard was not known, the changes in the peak area
of the PAN standard were used to monitor instead of the change in
concentration of PAN. As expected, inspection of the curve
demonstrates that PAN does decompose with elevated detector
temperature. From this experiment, a detector temperature of 65C
was chosen. Lower detector temperature did not increase
sensitivity, and with the ECD at low temperature, the background
signal would increase over a period of a week.
Even with the detector temperature set at 65 C, the background
signal measured by the ECD did increase over a period of a week.
This is to be expected because of the air being injected on the ECD.
This problem was solved by increasing the ECD temperature to 120.
C. The oven is then increased to 50 C and the instrument is left
for a 24 hour period (i.e. "baking- out" the system). After this period
the oven is turned off and the ECD is brought back to 65 C.

Periodically throughout the year, the column is replaced by an
empty stainless steel column and the ECD temperature is increased
to 300 C, oven temperature to 250 C. This is also left for a 24 hour
period for a more thorough ,rbaking-out" the system. The type of
column, rate of flow, and oven temperature were discussed in
previous sections.
y = 2.75046+7 + 1.42350+5X 2872.5xA2 RA2 = 0.955
PIP Evaluation
As discussed earlier in this paper, GC/ECD, NOx analyzers,
and more recently Luminox detectors have been used to measure

PAN in standards and in ambient air samples. PID shows promise
as a detector for PAN for a variety of reason : 1) the PID will not
respond to air, 2) the PID does not contain any radioactive material,
3) the PID does not required a special license be obtained to clean the
detector, and 4) the PID has not been explored as a detector for PAN.
Initially a PID (Model PI-51-01, HNU System ) was mounted
externally to the main GC oven, on top of the GC itself. The
connection between the ECD and PID was covered with external
heating tape in order to prevent any condensation of the GC
effluent after their exit from the ECD. For detection in series (i.e.,
PID connected after the ECD, it was necessary to connect a piece of
teflon from the exit of the ECD to in inlet of the PID. Since their
was no information on the ionization potential of PAN, a
compound of similar structure, n-propyl nitrate, was found to have
an ionization potential of 11.07. With this value in mind a lamp of
10.9 eV was chosen first. A lamp with this high of an ionization
potential will ionize any compound with an IP less than 10.9 eV.
The 10.9 eV will ionize compounds with a 0.3 eV higher than the
lamp but with less efficiency. Hence, the lamp we had might not
have worked. The flow rates were checked to ensure that no leaks
were present between the detectors, and that connecting the PID did
not create any back pressure problems with the ECD. The flow from
the ECD was 40 mL/min and the flow through the PID was

approximately the same. Once the flow was deemed adequate
through the PID, the PID detector was turned on and its
temperature raised to 120 C. The 10.9 eV lamp has a maximum
temperature of 125 C and the detector was increased to 120 C and
allowed to condition overnight.
The following day the detector was turned down to 100C
and the lamp intensity was turned up to a setting of 1/2. The lamp
was left on and not used until the baseline was stable. Then a PAN
standard of approximately 1 ppb was injected into the GC/ECD/PID.
With no response on the PID the lamp intensity was increased to
3/4 of the lamp intensity. The manual does not recommend using
any higher lamp intensity on the 10.9 eV lamp because the lifetime
of the lamp will significantly decrease. With still no response on
the PID, the lamp temperature was decreased in intervals of 10C.
At each temperature, there was no response on the PID. With the
PID detector temperature left at 65C an ambient sample was
injected into the GC/ECD/PID. Their was a small response on the
PID but not at the retention time of the PAN. The peak was in the
range of retention times of chlorinated hydrocarbons.
The PID was then disconnected from the ECD and installed in
another GC ( Varian). A column of the same material found in the
GC/ECD was packed and installed in the Varian GC and connected
to the PID. All the conditions used on the GC/ECD were selected as

a starting point for the GC/PID. The flow was set at 40 mL/min,
lamp intensity at 1/2, and detector temperature at 65 C. With still
no response, a 10 mL injection loop was installed on the GC/PID.
The PID still gave no response to PAN. With the detector
temperature lamp intensity, flows, and oven temperature varied,
the PID gave no response. The next step was to inject isopropyl
nitrate to see if the PID would respond. The IPN was treated like
the PAN sample, but still no response was observed. To ensure that
the PID was in fact working, a sample of air containing 250 ppb of
tetrachloroethene was injected into the GC through the 10 ml
sample loop. The PID had a significant response to this compound
and dilutions were made and injected. The lowest level injected,
which could be integrated, was 5 ppb. Hence, the PID was
functioning. The column was then connected to the liquid
injection port and a 51.2 ppb liquid standard of isopropyl nitrate was
injected into the GC/PID. The PID did respond and it took the rest
of the day for the signal to return to the baseline. The above
procedure was performed on the PAN standard and the same
results were obtained. Hence, the PID did respond to PAN but the
PID did not have the sensitivity needed for atmospheric samples. A
lamp of lower energy, 10.2 eV, was used next. The system using this
lamp is more sensitive than when using other lamps measured at
lower levels. Since the ionization potential for PAN is not known,

a lamp which would enhance system sensitivity should be explored.
The above procedure was used, but the results were the same. Even
though both lamps were functioning, both lamps could have lost
some effetiveness to the age of the lamps. The PID might be of use
in larger urban areas, such as Los Angeles, were the PAN
concentration are much higher.
Sampling Techniques
To determine the location of a sampling site, different factors
need to be determined. A site with documented meteorological, O3,
and NOx data was needed. Also, historical data on NO, NOx, and
O3 would be useful in determining a site were PAN might be
measured. With the help of the Colorado Department of Health,
historical data from their different monitoring locations was
obtained. The Welby site historically has more O3 than the CAMP
site. In each month the NO2/NO ratio is higher at Welby than at
CAMP (figure 13). As discussed earlier in this paper, more NO2
promotes PAN formation, and less NO also decreases PAN's
destruction. Since the Welby site had more historical data than the
Englewood site, the Welby site was chosen for ambient PAN
Welby is located approximately 12 miles from downtown
Denver in the northeast direction along the Platte river. Figure 14
is a map of the Denver area showing the sampling site location. It

takes approximately 12 minutes to drive to the site and the same
time to return to the CU-Denver campus. It does take
approximately 11 minutes to take the sample from the car to the
instrument in the North Classroom Building. This became a
problem when the North Classroom Building was closed because
the sampling box did not fit through the door in an upright
position. Since the box is filled with dry ice, turning the box on its
side could cause the dry ice to fall out or stress the sides of the box.
This problem has been taken care of since the GC/ECD has been
moved to an ambient air monitoring location on the CU-Denver

g Jun-90
5 Jul-90
[N02] / [NO]
H-i ^ ls> ls>
O) o o In o o In
figure 13 (cont'd)
[N02] / [NO]
o 1-^ to to W
o U1 o CJ1 o U1 o
o o o o o o o
-1 -1- -1 H

o re 3 a
< a*


This situation was more favorable since the box did not need
to be removed from the car and no time was wasted in carrying the
ambient sample to the North Classroom Building.
With the sampling location determined the next factor to
investigate was when should the samples be taken (i.e., time of day).
By inspection of the hourly historical data, and speculation, the
time between 1:30-2:30 p.m were determined to be the best time of
day to sample from PAN. Other times were sampled to look at the
diurnal variation of PAN.
Once the sampling site was determined, sampling equipment
needed to be determined. Three different types of air sampling bags
were tested. The first was as 50 L air bag with an on/ off valve. The
composition of the bag was unknown. This bag showed PAN
decomposition and was not easy to clean. After a PAN standard was
injected into the bag the sample was injected into the GC/ECD. Two
peaks were detected, one was PAN and the other was thought to be
methyl nitrate. Once the contents of the bag were analyzed, it was
evacuated and "clean" air was used to fill the bag. This sample was
analyzed On the GC/ECD and still showed a PAN peak. The bag
needed to be filled six times before it was clean (i.e., no peaks except
for air). This process took up to four hours, and was therefore time
consuming. The next bag tried was a 50 L Teflon bag containing a

1/4" swagelock fitting. A PAN sample was injected into the teflon
bag and analyzed on the GC/ECD.
Figure 14: Map of sampling site

Only one peak was detected and the peak corresponded to PAN.
After one cleaning of the bag, the bag contained no carryover peaks
from the first run. Hence, the Teflon bag was determined to be used
for sampling. The other bag, Tedlar, responded the same as the
Teflon bag but came in the film thickness of 2 mil and was to thin
for outdoor sampling. The bag used for most of the sampling was a
Nutech Teflon double sealed air sampling bag with a maximum
volume of 30 L.
The next item to look at was a pump. Most pumps have the
ambient air pulled through the pump and then into the
instrument. With the nature of sampling that was going to be
conducted this sort of pump would not be suitable because PAN
would decompose through a pump. Consequently a bellow's pump
was used which consists mostly of Teflon parts. To insure that PAN
was not lost or decomposed through the pump, a PAN sample was
injected into one Teflon bag, connected to the pump, and another
bag was connected to the outlet of the pump. The pump was started
and the second bag was filled with the PAN standard. Both bags
were analyzed on the ECD. There was a small decrease in the PAN
concentration but this was due to the amount of time the bags were
left at ambient temperatures. Also, the second bag contained one
peak on the chromatograph, PAN. Each bag was first analyzed to

ensure that both bags were clean. Then an air sample of clean air
was connected to the bellow's pump and the effluent port connected
to the other Teflon bag (#2). The pump was turned on and when
the second bag(#2) was full, the bag was disconnected, the pump
turned off, and bag # 2 was analyzed on the GC/ECD. Analysis of
bag #2 showed no peaks and that of bag #1 showed no peaks. Hence
the bellow's pump itself would not produce any contamination
while sampling. Because all of the sampling would take place
outside filters needed to be used on the inlet side of the pump. To
determine that no contamination problem existed with the filter a
similar procedure was followed as described above. The only peak
that showed up on the chromatogram was a water peak at 14
minutes. The most probable explanation for this observation was
that the connection between the bag, filter, and pump was not air
tight. The pump could be pulling air from the room. Overall, the
pump and filter did not contaminate the air bag.
Once the sampling location and equipment were determined,
the next problem was to determine how to transport the sample
with minimal sample loss. Most researchers have a portable GC at
the sampling location. Since this was not possible in this project a
Styrofoam box was built. The box was 2 inch thick Styrofoam with
the dimension of 22"x 32"xl4". The dimensions of the box were
calculated not only to fit in the air bag but also to fit the trunk of a

car. Dry ice was placed in the box and the box was then transported
to the sampling location. The temperature in the box was
monitored and the temperature was approximately -5 C. This
temperature should keep PAN stable until the sample is returned
for instrumental analysis. To insure that there is no significant loss
of PAN an ambient sample was taken, returned to lab, analyzed,
and left in the box for 5 hours. The sample was then analyzed again
and showed only a 3% loss in the area of PAN. Another ambient
sample was taken, analyzed, and left out of the dry ice box. The
sample was then reanalyzed and showed a 70% loss in the area of
PAN. Along with this loss of PAN another peak appeared which
might have been methyl nitrate. As shown earlier in this paper, the
only source of methyl nitrate is PAN. Hence, the sampling box
worked for transport of the ambient samples.

Ambient air samples were collected from 5 Nov 90 to 18 Oct
91. Historical data obtained from the Colorado Department of
Health determined the best time to collect PAN. Samples were
collected between the hours of 12:00 p.m. and 3:00 p.m and on
certain days ambient samples were collected between the hours of
8:00 a.m. to 3:00 a.m. to observe diurnal variation in the PAN
concentration. Appendix A contains all the data for date, time,
PAN area, PAN concentration, temperature, NO, O3, NO2. The
PAN concentrations ranged from 40 ppt- 525 ppt. Every ambient
sample contained PAN and was quantified. Each sample was
quantified in the following manner: a five point calibration curve
was performed before ambient samples were obtained. The
concentration and area for each PAN concentration were entered
into either Prostat or Quattro computer programs. From these
programs and linear regression was performed. Once an ambient
sample was acquired, the area of PAN from the sample was entered
into the computer and generated a y value corresponding to the
PAN concentration. Using appendix A, 7 graphs were plotted to

assist in interpreting the data. These graphs are as follows: 1) PAN
concentration vs. date, 2) PAN concentration vs. date with time 12-3
pm, 3) PAN concentration vs. temperature, 4) PAN concentration
vs. ozone, 5) PAN concentration vs. ozone on 30 April 91 6) PAN
concentration vs. NO, 7) PAN concentration vs. time on the 30
April 1990.
PAN Concentration vs. Date
The first graph (figure 15) is PAN concentration vs. date.
Since PAN's decomposition is a thermally controlled process, PAN
concentration might be higher in the winter months than in the
summer months. The plot demonstrates a scattering of the data
which is expected in environmental data. One can not draw a
conclusion from this graph because their are too few data points.
Also, most points were collected between the months of February -
May 91.
PAN Concentration vs. Date 12-3 pm
Data was extracted from appendix A which had a sampling
time between the hours of 12- 3 pm (figurel6). This was done to
take out any diurnal variation in the PAN data. Again we see a
scattering in the data. One of the highest PAN concentration (521
ppt) was observed on 6 September 1991. To better understand this

value one needs to look at the temperature and NO2/NO ratio. The
temperature was 26C which would tend to support PAN's
decomposition, but the NO2/NO ratio is 7.7. This suggest that even
with the temperature high enough to promote PAN
decomposition, the NO2/NO ratio is also high enough to promote
PAN formation and not follow the decomposition reactions 30-35.
Figure 15: PAN concentration vs. date
PAN Concentration vs. Temperature:
As stated earlier, the lifetime of PAN is a thermally
controlled processes. Hence, as the temperature increases and as

long as NO is present, the PAN concentration should decrease. By
inspection of Figure 17 one can see a trend in the data to support
0.6 t-
0.45 -
I 0.4
a 0.35 .-
0.3 ..
0.05 4*
0 ro

O) '.
<0 O

-L Ci) C7I O) CD 1
IO * A ro
-U CJ1 "x* 0) to
'X CO xx Xs, Xx.
12-3 pm
Figure 16: PAN concentration vs. date between 12-3pm
this assumption. The highest concentration of PAN was collected at
25 C. This point was explained in the above section and
demonstrated the importance of collecting NO and NO2 data at the
same time as the sampling. Since most of the data points were
acquired between the temperatures of 10 C and 25 C, ambient
PAN concentration has a lifetime between 3-14 hours. Because
there are not enough samples collected at subfreezing temperatures,
more data points are needed at higher and lower temperatures,
especially at temperatures below 0 C. Again, there are not enough

data points to draw a definite conclusion but there does seem to be a
trend that at higher temperatures, PAN concentration decreases.
Figure 17: PAN concentration vs temperature
PAN Concentration vs. Ozone
By plotting PAN concentration vs. ozone the relationship
between PAN and O3 can be explored. As mentioned earlier, PAN's
decomposition processes produces 4 N02* Even though NO2 is
primarily formed from other sources, the decomposition of PAN
increases the formation of O3. Figure 18 shows the scattering of the
data but there does seem to be a trend to support this assumption.
Another way to possibly demonstrate this relationship is to plot the
PAN concentration vs. the ozone on the 30 April 91 (Figure 20). On

30 April 91 we obtained 5 measurements that indicated a ozone
Figure 18: PAN concentration vs ozone
Thus figure 19 does demonstrate that as the ozone
concentration increases as the PAN concentration increases. One
point does not seem to fit this conclusion. This point was obtained
at 8:30 p.m.; since ozone does not form at night, this value was not
consistent with the rest of the data. As an overall assessment, more
data points are needed.
PAN Concentration vs NO
Reactions 28-30 demonstrate the relationship between NO
and the decomposition of PAN. Not only does the decomposition
of PAN depend on the ambient temperature but also on the NO
concentration. If the ambient temperature is sufficient for PAN to

dissociate to form the peroxy radical and the NO concentration is
high, PAN will futher decompose according to reactions 31-35.
Hence, as the NO concentration increases, the PAN concentration
should decrease. Figure 20 demonstrates that at high NO
concentration the PAN concentration does decrease. Also, during
the time of sampling the NOx instruments were being calibrated
and a new NOx analyzer was being installed. Because of these
reasons, most of the data points for NO and NO2 were reported as
"n/a". Again, more data points are needed to determine if this
trend does exist.

PAN Concentration vs. Time 30 April 91
Fgure 21 demonstrates the possible diurnal variation of PAN. As
expected, PAN's concentration was highest around the time of 2:00
pm. This was predicted by using the Colorado Department of
Health information on NO, NO2, and O3. Another set of
researchers have data from the same months as this work but
different year (Singh & Salas, 1983). Their daily plot of PAN
concentration vs. time of day is similar to ours. Figure 22 is a plot of
Singh & Salas PAN vs. Time. Since Singh & Salas obtained data
from 7 different locations and plotted the diurnal behavior of PAN
at 3 of these cities, one can compare the diurnal behavior of PAN to
the other cities. For all the cities, the highest concentration of PAN

was obtained at around 2:00pm. After 2:00 pm the concentration of
PAN significantly decreases in the urban cities except for Denver.
This could be due to the effects of colder temperatures at night in
Denver than in the other cities cited in their paper. The Singh &
Salas daily plot is similiar to our daily plotted on the 30 April 91.
Even though our data is one day, the comparision and observation
of the similar diurnal should be pointed out. Overall, more data
points are needed to substantiate any of these observations.
Figure 21: PAN concentration vs time April 30,1991

Figure 22: Singh & Salas diurnal variation of PAN at Denver
PAN Standards
With regards to the PAN standards, a few observations
should be mentioned. When the PAN standards were initially
synthesized, the concentrations were determined. After this time
the standards were reanalyzed to determine the concentration of
PAN. The concentration of PAN should decrease over time and
this was observed. To see if the decomposition of PAN in the liquid
phase followed first order kinetics, the rate constant was calculated
for each standard. Table 10 has the calculated rate constant from the
synthesized PAN standards and rate constant at -20 C from table 1.
By comparing the calculated rate constant to the theoretical
value, the synthesized PAN does decompose according to the
7 4

theoretical value. This could be used to estimate the concentration
of the PAN standard over time.
Table 10: Rate constant of PAN Standard at -20C
# Date synthesized time elapsed || k (s_l)
1 26May90 9 months 1.10x10-7
2 6June90 9 months 1.05xl0-7
3 3Nov90 3 months 2.20x10-7
4 lMarch91 2 months 2.99xl0-8
5 11 July 91 2.5 months 1.20 xlO-7
n/a Table 1 88 days 1.3 x 10-7
Since the only decomposition product of PAN is methyl
nitrate, methyl nitrate synthesis was attempted. Following the
method used by Grosjean (et al., 1990), methyl nitrate was
synthesized. This product was injected into the GC/ECD and a peak
was obtained. To ensure that this peak was not one of the starting
materials, each starting material was injected into the GC/ECD.
None of the starting materials matched the peak of the product. To
determine if the PAN standard in the gas phase would acquire a
peak at the same retention time, the following experiment was
performed. A PAN standard was injected into the air bag. The bag
was analyzed and left out over night and reanalyzed. The PAN area
did decrease and another peak was observed. This peak did have a
retention time comparable to that of the methyl nitrate peak.

Standard 2 was injected into the GC/ECD and contained the same
peak as the methyl nitrate and PAN bag left out over night.
Standard 2 also contained an extra peak in the liquid injection. This
supports the assumption that the peak might be methyl nitrate.
Some of the ambient air peaks contained a peak that corresponded
to methyl nitrate but some of the chlorinated hydrocarbons had the
same retention times. Another way of possibly determining
whether the peak is methyl nitrate is to compare retention times of
other researchers using similar types of columns. Ridley and
coworkers (1990) used a packed column of a similar nature to ours.
According to Ridley paper, PAN has a Rt=7.40 and methyl nitrate
Rt= 2.55. The difference in Rt is 4.85. With our column, PAN has a
Rt =9.80 and methyl nitrate" Rt= 4.75. The difference in Rt is 5.05.
These values seem to be close and points in the direction that the
peak we assume to be methyl nitrate is methyl nitate. Overall, a few
conclusions can be draw from this: 1) the concentration of methyl
nitrate in the ambient sample was too low to be measured on the
ECD, 2) The chlorinated hydrocarbons in the ambient sample have a
similar retention time to the possible methyl nitrate peak, 3) PAN
formation was more favored at the time of sampling. Last, without
any FTIR charaterization of the methyl nitrate standard, one cannot
say with confidence that the peak observed is methyl nitrate.

The main purpose of this research project was to develop a
method for PAN measurements in Denver. The method was
successful. It involved synthesis, calibration, sampling, and data
interpretation of PAN measurements. The synthesis of PAN was
accomplished by making a few modification to the Gaffney (et al,
1984) procedure. Once PAN was synthesized a spectra was obtained
by GC/FTIR and compared to previous published papers. The
concentration of PAN was obtained by two different methods; 1)
IPN/PAN liquid injection, 2) Ion chromatography. The %
difference between the calculated values of the two method on the
11 July 1991 is 11%. For determining the concentration of PAN by
two totally different analytical techniques, the values are close and
gave validity to the standard concentration. The area of
investigation was to determine the best instrumental parameters
such as the type of packed column and GC parameters. The best
column was a hand packed 63 cm x 0.8 mm i.d., 10% Carbowax 600
on 80/100 mesh Chromsorb WHP. The GC/ECD conditions were as
followed: injection port and oven were kept at ambient temperature

(26 C), ECD temperature was at 65 C. At the end of each week the
oven and ECD were brought to temperature of 60 C and 120 C,
respectively. This was done to condition the ECD and column since
the background signal would increase over the period of a week
when 5 ml of air was being injected on the ECD.
The sampling site was determined with the assistance of the
Colorado Department of Health. Using the N02/N0, and O3 from
the historically data at three different sites in Denver, it was
determined the best sampling location would be Welby. Sampling
equipment and transport of the ambient sample were
experimentally tested in the laboratory before field sampling was
started. It was determined that PAN loss was minimal in the
sampling apparatus and the transport of the sample back to lab.
Once all of the sampling methodology was done ambient sampling
was started.
Forty points were obtained and the PAN concentration
ranged between, 39.5 ppt-520 ppt. Researchers Singh and Salas
measured PAN in Denver during the time 24 March-1 April 1984
(Singh & Salas, 1989). Their measurements were conducted on site
and in real time using an instrumented mobile laboratory. Hourly
measurements were conducted at each site on a 24-h per day
schedule, 7 day-per-week. While at their sampling site, they
encountered snow storms, subfreezing temperatures and substantial

ice/ snow cover on the ground. The sampling site was located at
Marion & E. 51st St. By inspection of figure 15, this site is located on
the Northeast side of Denver but within a few miles of our Welby
site location. During Singh & Salas sampling time, the PAN
concentration had a range of 644 ppt- 2039ppt. They also state that a
background of 500 ppt is always present because of PAN's stability in
colder climates and because PAN is much more stable at night, with
ground deposition being the only significant loss process. As stated
earlier our PAN concentration ranged from 39.5 525 ppt with the
mean equal to 235 ppt and a standard deviation of 0.1383.
If we look at the data points obtained during the same
months as Singh & Salas, the range is 127- 525 ppt with the mean
equal to 302 ppt. The percent different in our values compared to
Singh & Salas is 23%. For PAN measurements from different
researchers, sampling locations, and year, the PAN concentrations
are close. Our values are lower then those of Singh & Salas; there
can be many explanations for the difference. The first explanation
to consider to consider is that the Singh & Salas data were obtained
in real time, hence PAN loss was minimal because the sample had
minimal handing. Our samples had to be acquired at a sampling
location and then transported back to the laboratory for analysis.
Even though we tested the sampling apparatus for PAN stability
and determined PAN loss was minimal, some PAN could still be

lost. The next explanation to consider is temperature. Singh &
Salas described cold, snow covered ground during their sampling
time. By looking at the same sampling time in our data, the
temperature of the ambient air ranged from 10-25 C. At these
temperatures PAN would be expected to linger in the atmosphere
between 3-14 hours. Hence, even in the winter months Denver
might not sustain subfreezing temperature for PAN to accumulate
over a period of time. The last explanation to consider was the
determination of the PAN concentration at this particular time.
The PAN standard was calculated by the IPN/PAN liquid injection
method. As demonstrated earlier in the paper, the IPN method
calculated the PAN concentration ll%-22% lower than the IC
method. For most of the PAN samples the IC was nof available for
determining the PAN concentration. The ambient samples
acquired during the months of Sept-Oct 1991 were calculated with
the IC value and obtained a more true value for the PAN
concentration in ambient air. Overall, the values obtained from
Singh & Salas lend support to the validity of the PAN
concentrations we obtained in Denver.
The important aspect of this data analysis was the availability
of NO, NO2, and O3. Without these values, the PAN data would
not have as much significance. Much more information is needed

to better understand PAN's role in the photochemistry process in
Denver and how much of the NOy budget can be attributed to PAN.
Future Work
Their are many areas of this project that need to be explored
and will be outlined in the following paragraph.
The first area to address is the determination of the PAN
standard concentration. The availability of a long path FTER could
be used to obtain the spectral characteristics of the PAN standard
and by using Beer's Law, concentrations could be determined via
preexisting coefficients. Once the concentration was determined,
the GC/ECD could be calibrated directly from the long path FTIR.
Also, the use of an NOx analyzer should be explored in calculating
the PAN standard concentration. With respect to the packed
column system, better resolution could be obtained by the use of
megabore and capillary columns. Because capillary/ megabore
columns cannot handle the load capacity of packed columns,
cryogenic sampling capacities are needed for the GC. Lastly, real
time PAN measurements with hourly round the clock sampling
would be highly desirable. This would assist in determining the
spatial and diurnal variation of PAN in Denver. Also, with the
ambient sampling being directly injected into the GC/ECD would
minimize PAN loss because of less sample handling.

As a last note, the PID should be further explored as a possible
detector of PAN.

Bruckmann, P.W., H. Willner, Infrared Spectroscopic Study of
Peroxyacetyl Nitrate (PAN) and Its Decomposition Products,
Environ. Sci. Technol., 17, 352-357, 1983
Buhr, M.P., Ph.D. Thesis, University of Colorado at Boulder, May
Buhr, M. P., D.D. Parrish, R.B. Norton, F.C. Fehsenfeld, R.E.
Sievers, J.M. Roberts, Contribution of Organic Nitrates to the
Total Reactive Nitrogen Budget at a Rural Eastern U.S. Site, /.
Geophys. Res., 92, No. D7, 9809-9816, 1990
Burkhardt, M.R., N.I. Maniga, D.H. Stedman, R.J. Paur, Gas
Chromatographic Method for Measuring Nitrogen Dioxide
and Peroxyacetyl Nitrate in Air without Compressed Gas
Cylinders, Anal. Chem, 60, 816-819, 1987
Drummond, J.W., H.I. Schiff, G.I. Mackay, Atmoshperic
Measurements of PAN (Peroxyacetyl Nitrate) Using a Newly
Developed Lummox^ Instrument, 196th ACS National
Meeting, Vol. 28, No. 2, 89, 1988
Fahey, D.W., G. Hubler, D.D. Parrish, E.J. Williams; R.B. Norton, B.
A. Ridley, H.B. Singh, S.C. Liu, F.C. Fehsenfeld, Reactive
Nitrogen Species in the Troposphere: Measurements of NO,
NO2, HNO3, Particulate Nitrate (PAN),03, and Total
Reactive Odd Nitrogen (NOy) at Niwot Ridge, Colorado,/.
Geophys. Res., 91, 9781-9793, 1986
Gaffney, J.S., R. Frajer, G.I. Senum, An Improved Procedure for
Fligh Purity Gaseous Peroxyacyl Nitrate: Use of Heavy Lipid
Solvents, Atmos. Environ, 18, 215-218, 1984

Grosjean D., Nitrogenous Air Pollutants: AnnArbor Science, 137-
Grosjean D.G., A.H. Miguel, Atmospheric chemisrty of alcohol
fuels: ethanol, acetaldehyde, and PAN, Preprinted Extended
Abstract, ACS, 28, 79,1988
Grosjean D., S.S. Parmar, E.L. Williams III, Southern California Air
Quality Study: A Search for Methyl Nitrate, Atmos. Environ.,
24A, No.5,1207-1210,1990
Helmig D., J. Muller, W. Klein, Improvements in the Analysis of
Atmospheric Peroxyacetyl Nitrate (PAN), Atmos. Environ.,
23, No. 10,2187-2192,1984
HNu, The HNU Photoionization Detector Instruction Manual,
Holdren, W. H., C. W. Spicer, Field Compatible Calibration
Procedure for Peroxyacetyl Nitrate, Environ. Sci. Technol., 18,
Joos, L. F., W. F. Landolt, H. Leuenberger, Calibration of Peroxyacetyl
Nitrate Measurements with an NO\ Analyzer, Environ. Sci.
Technol., 20, 1269-1273, 1986
Lonneman, W. A., J. J. Bufalini, G. R. Namie, Calibration Procedure
for PAN Based on Its Thermal Decomposition in the
Presence of Nitric Oxide, Environ. Sci. Technol., 16, 655-660,
Lovelock, J. E., S. A. Penkett, PAN over the Atlantic and the Smell
of Clean Linen, Nature, 249, 434, 1974

Maeda, Y., Kenji Aoki, Makoto Munemori, Chemiluminescence
Method for the Determination of Nitrogen Dioxide, Anal.
Chem., 52, 307-311, 1980
Martinez, R. I., A Systematic Nomenclature for the "Peroxyacyl
Nitrates", the Functional and Structural Misnomers for
Anhydride Derivatives of Nitrogen Oxo Acids, lnt. J. Chem.
Kind., 12, 771-775, 1980
Mudd, J. B., Peroxyacyl nitrates. In Response of Plants to Air
Pollution, 97-119, Academic Press, New York, 1975
Nicksic, S.W., J. Harkins, P.K. Mueller, Some Analyses For PAN
and Studies of Its Structure, Atmos. Environ., 1, 11-18, 1967
Nielsen, T., A. M. Hansen, E. L. Thomsen, A Convenient Method
for Preparation of Pure Standards of Peroxyacetyl Nitrate for
Atmospheric Analysis, Atmos. Environ, 16, 2447-2450, 1982
Oshim, R., O. C. Taylor, E. A. Cardiff, Severe Air Pollution Episode
in South Coast Basin, Calif. Agric., 28, 12-13, 1974
Pitts Jr., J.N., Environmental Appraisal: Oxidants, Hydrocarbons,
and Oxides of Nitrogen, /. Air Pollut. Control Ass., Vol 19, No.
Peake, E., M. A. MacLean, P.F. Lester, H. S. Sandhu,
Peroxyacetylnitrate (PAN) in the Atmosphere of Edmonton,
Alberta, Canada, Atmos. Environ., 22, No.5, 973-981, 1988
Ridley, B.A., J.D. Shetter, J.G. Walega, S. Madronich, C.M. Elsworth,
F.E. Grahek, The Behavior of Some Organic Nitrates at
Boulder and Niwot Ridge, Colorado, J. Geophys. Res., 95, No.
D9., 13,949-13,961,1990

Roberts, J. Mv Capillary Gas Chromatographic Separation of Alkyl
Nitrates and Peroxycarboxylic Nitric Anhydrides, Anal.
Chem., 61, 771-772, 1989
Roberts, J. M., The Atmospheric Chemistry of Organic Nitrates,
Atmos. Environ., 24A, No. 2, 243-287, 1990
Rudolph, J., B. Vierkorn-Rudolph, F. X. Meixner, Large-Scale
Distribution of Peroxyacetylnitrate Results from the
STRATOZ HI Flights, J. Geophys. Res., 92, No. D6., 6653-6661,
Senum, G.I., R. Fayer, J.S. Gaffney, Fourier Transform Infrared
Spectroscopic Study of the Thermal Stability of Peroxyacetyl
Nitrate, J. Phys. Chem., 90, 152-156, 1986
Singh, H. B., P. L. Hanst, Peroxyacetyl Nitrate (PAN) in the
Unpolluted Atmosphere: An Important Reservoir for
Nitrogen Oxides, Geophy. Res. Let., 8, No. 8, 941-944,1981
Singh, H. B., L. J. Salas, Methodology for the Analysis of
Peroxyacetyl Nitrate (PAN) in the Unpolluted Atmosphere,
Atmos. Environ., 17, No. 8, 1507-1512, 1983
Singh, H. B., L. J. Salas, B. A. Ridley, J. D. Shetter, N. M. Donahue, F.
C. Fehsenfeld, D. W. Fahey, D. D. Parrish, E. J. Williams, S. C.
Liu, G. Hubler, P. C. Murphy, Relationship Between
Peroxyacetyl Nitrate and Nitrogen Oxides in the Clean
Troposphere, Nature, 318, 347-349, 1985
Singh, H. B., L. J. Salas, Measurements of Peroxyacetyl Nitrate
(PAN) and Peroxypropionyl Nitrate (PPN) at Selected Urban,
Rural, and Remote Sites Atmos. Environ., 237, No. 1,231-238,

Stephens, E.R., P.L. Hanst, R.C. Doerr, W.E. Scott, Recent
Development in the Study of the Organic Chemistry of the
Atmosphere, J. Air Pollut. Control Ass., 6,159, 1956
Stephens, E. R., F. R. Burleson, E. A. Cardiff, The Production of
Pure Peroxyacyl Nitrates, J. Air Pollut. Control Ass., 15, No. 3,
Stephens, E. R., The Formation of Molecular Oxygen by Hydrolysis
of Peroxyacetyl Nitrate, Atmos. Environ., 1, 19-20, 1967
Stephens, E. R., F. R. Burleson, K. M. Holtzclaw, A Damaging
Explosion Of Peroxyacetyl Nitrate, ]. Air Pollut. Control Ass.,
19, No. 4,261-264,1969
Stephens, E. R., M. A. Price, Analysis of an Important Air Pollutant
Nitrate, J. Chem. Ed., 50, 351-354, 1973
Temple, P. J., O. C. Taylor, World-Wide Ambient Measurements of
Peroxyacetyl Nitrate (PAN) and Implications for Plant Injury,
Atmos. Environ., 17, No. 8, 1583-1587, 1983
Thompson, C.R., G. Kats R. W. Lennox, Phytotoxicity of Air
Pollutants Formed by High Explosive Production, Envir. Sci.
Technol., 13, 1263-1268, 1979
Tsani-Bazaca, E., S. Glavas, H. Gusten, Peroxyacetyl Nitrate (PAN)
Concentration in Athens, Greece, Atmos. Environ., 22, No.10,
Tuazon, E. C., W. P. L. Carter, R. Atkinson, Thermal Decompostion
of Peroxyacetyl Nitrate and the Reaction of Acetyl Peroxy
Radicals with NO and NO2 over the Temperature Range 283-
313 K,/. Phys. Chem., 95, 2434-2437, 1991

Date II Time || PAN II II Area PAN (PPb) Temp. || Ozone || NO || NO2 || NO2/ C || (ppb) || (ppb) | (ppb) || NO
11/20/90 ,2:30 PM 74004 0.04001 8.33 n/a n/a n/a n/a
11/29/90 2:30 PM 387630 0.2015 6.11 n/a n/a n/a n/a
12/5/90 12:30 PM 363440 0.1901 11.66 4 n/a n/a n/a
1/30/91 2:30 PM 524310 0.2726 8.88 54 n/a n/a n/a
2/6/91 1:30 PM 311910 0.1621 8.33 n/a n/a n/a n/a
2/16/91 1:30 PM 332830 0.173 14.44 9 53 55 1.037
2/18/91 11:30 PM 69423 0.0395 -3.33 2 27 43 1.593
2/19/91 10:30 AM 163540 0.0851 5 34 0 5 n/a
2/19/91 1:30 PM 156070 0.0805 6.11 n/a n/a n/a n/a
3/5/91 12:30 PM 74936 0.0405 7.78 22 n/a n/a n/a
3/5/91 3:30 PM 151600 0.0788 3.33 31 n/a n/a n/a
3/5/91 8:30 PM 152790 0.0816 2.22 2 n/a n/a n/a
3/7/91 1:30 PM 269470 0.1401 5.55 38 n/a n/a n/a
3/12/91 2:30 PM 203660 0.2317 10 n/a n/a n/a n/a
3/14/91 2:30 PM 484490 0.3881 8.33 35 1 7 7
3/14/91 8:30 PM 378880 0.311 1.66 25 0 15 n/a
3/19/91 2:30 PM 481340 0.3858 16.67 4 0 5 n/a
3/19/91 7:30 PM 671940 0.5248 10 30 0 13 n/a
3/24/91 2:30 PM 134030 0.1325 20.55 52 0 5 n/a
3/26/91 2:30 PM 127030 0.1274 16.67 37 1 6 6
4/4/91 2:30 PM 287290 0.2442 23.33 38 10 29 2.9
4/11/91 2:30 PM 374160 0.2863 2.77 37 1 6 6
4/17/91 7:30 PM 695800 0.4608 12.22 36 0 12 n/a
4/18/91 2:30 PM 580930 0.3992 5.56 48 1 5 5
4/30/91 9:30 AM 168750 0.1795 3.88 32 18 27 1.5
4/30/91 12:30 PM 600740 0:4092 7.22 45 11 24 2.18
4/30/91 2:30 PM 685240 0.455 10 47 10 31 3.1
4/30/91 5:30 PM 578170 0.397 11.1 41 3 25 8.33
4/30/91 8:30 PM 523660 0.3674 6.11 0 33 66 2
9/6/91 12:30 PM 710190 0.5211 25.56 48 3 23 7.66
9/7/91 1:30 PM 298950 0.1558 25.56 48 0 4 1.33
9/21/91 i:io PM 112330 0.137 29.44 50 0 6 n/a
9/21/91 6:30 PM 66824 0.1094 26.67 33 0 10 n/a
9/21/91 9:30 PM 164410 0.2143 11.11 38 0 1 n/a

9/22/91 2:30 PM 152570 0.1718 14.44 49 n/a n/a n/a
9/26/91 2:30 PM 120386 0.1045 25 52 70 26 0.37143
10/12/91 1:30 PM 313680 0.2541 29.44 50 1 7 7
10/13/91 2:30 PM 360470 0.2831 17.78 55 0 2 n/a
10/18/91 2:30 PM 467500 0.3495 11.67 42 1 4 4
10/19/91 2:30 PM . 256790 0.2193 18.89 48 0 4 n/a