Nitrate radical quantum yield from peroxyacetyl nitrate photolysis

Article abstract of DOI:10.1021/jp045529n

Peroxyacetyl nitrate (PAN, CH₃C(O)OONO₂) is a ubiquitous pollutant that is primarily destroyed by either thermal or photochemical mechanisms. We have investigated the photochemical destruction of PAN using a combination of laser pulsed photolysis and cavity ring-down spectroscopic detection of the NO₃ photoproduct. We find that the nitrate radical quantum yield from the 289 nm photolysis of PAN is φ_NO3^PAN = 0.31 ± 0.08 (±2σ). The quantum yield is determined relative to that of dinitrogen pentoxide, which is assumed to be unity, under identical experimental conditions. The instrument design and experimental procedure are discussed as well as auxiliary experiments performed to further characterize the performance of the optical cavity and photolysis system.

Full text of DOI:10.1021/jp045529n

2552
J. Phys. Chem. A 2005, 109, 2552-2558
Nitrate Radical Quantum Yield from Peroxyacetyl Nitrate Photolysis
Bradley A. Flowers,† Mark E. Angerhofer, William R. Simpson,* Tomoki Nakayama,‡ and
Yutaka Matsumi‡
Department of Chemistry and Geophysical Institute, UniVersity of Alaska, Fairbanks,
Fairbanks, Alaska 99775-6160
ReceiVed: October 1, 2004; In Final Form: January 7, 2005
Peroxyacetyl nitrate (PAN, CH3C(O)OONO2) is a ubiquitous pollutant that is primarily destroyed by either
thermal or photochemical mechanisms. We have investigated the photochemical destruction of PAN using a
combination of laser pulsed photolysis and cavity ring-down spectroscopic detection of the NO3 photoproduct.
PAN
NO3
We find that the nitrate radical quantum yield from the 289 nm photolysis of PAN is Φ
) 0.31 ( 0.08
((2σ). The quantum yield is determined relative to that of dinitrogen pentoxide, which is assumed to be
unity, under identical experimental conditions. The instrument design and experimental procedure are discussed
as well as auxiliary experiments performed to further characterize the performance of the optical cavity and
photolysis system.
Introduction
The absorption cross section for PAN has been measured.3,11
The work of Talukdar et al. demonstrates a 40% decrease in
Polluted environments contain NO2 and partially oxidized
hydrocarbons that react to form peroxyacetyl nitrate (PAN,
CH3C(O)OONO2). PAN remains stable enough at low temper-
ature to transport NO2 to remote regions of the globe.1,2 PAN
is only slowly removed by OH radical oxidation and dry and
wet deposition.1,3 Due to these properties, PAN contributes
significantly to the long-range transport of NO2. The thermal
stability of PAN is primarily reponsible for its accumulation in
the arctic during winter. Thermolysis via
σ
PAN at 320 nm between 298 and 250 K.3 Due to the instability
of the acetoxy radical produced in (3), photolysis in this channel
leads to irreversible loss of PAN. Subsequently, NO3 is
photolyzed to NO2 or NO by visible light; therefore (3) directly
converts PAN to NOx and is an irreversible loss from the PAN
reservoir. For PAN photolysis resulting in (2), the photoproducts
remain stable and can recombine to re-form PAN, thereby
extending the effective lifetime of the PAN reservoir. Thus, the
PAN photolysis channels (2) and (3) affect the PAN reservoir
very differently.
CH3C(O)O2NO2 + M h CH3C(O)O2 + NO2 + M (1)
Laser-induced fluorescence (LIF) detection of NO2 and NO3
photoproducts following 248 nm photolysis of PAN measured
has been studied extensively as the predominant ground-level
PAN loss process.1,4 Further experiments have indicated that
the thermal decomposition of PAN proceeds dominantly via (1)
despite near isoenergy of the O-O and O-N single bonds in
PAN.4,5 Process (1) is strongly temperature dependent,6 leading
to a PAN lifetime with respect to thermal decomposition of 1
month at 255 K. Recombination via (-1) extends the effective
thermal lifetime. Field measurements indicate that PAN is the
predominant active nitrogen species (NOy) during the winter to
spring transition at high latitudes.7 Typical arctic temperatures
are sufficiently low enough to preclude thermal decomposition
as the sole viable PAN loss mechanism throughout the winter
and spring at ground level and at high altitude.7,8
the quantum yield of NO2 (ΦPAN) to be 0.70 ( 0.109 and ΦPNAON
NO2
3
to be 0.30 ( 0.10.10 Recently, the nitrate radical quantum yield
from PAN photolysis was investigated again at 248 nm and
also at 308 nm detecting NO3 using laser absorption spectros-
copy.12 In both studies ΦNPAON was determined relative to ΦNO
3
3
from N2O5 photolysis. The 248 nm results of the Harwood et
al. study are slightly different from the previously reported
PAN
NO3
results of Mazely et al. Φ
(248 nm) ) 0.19 ( 0.04.
Harwood et al. indicate that the previous work of Mazely et al.
N2O5
NO3
used an incorrect value for Φ
at 248 nm and that after
correction, the agreement between the studies is acceptable,
indicating approximately 20% NO3 production. Harwood et al.
also measured the photolysis quantum yield for NO3 at 308 nm
The second loss process for atmospheric PAN is photolysis.
Many possible photoproducts may be formed by UV photolysis
as reviewed by Mazely, Friedl, and Sander.9,10 Two major
decomposition channels dominate for UV photolysis,
PAN
NO3
and found Φ (308 nm) ) 0.41 ( 0.10. The results of these
studies9,10,12 suggest that NO2 and NO3 (and their respective
organic radicals) are the major direct photoproducts of PAN
photolysis at or above 248 nm. It also appears that the nitrate
radical quantum yield from PAN is wavelength dependent.
CH3C(O)O2NO2 + hν f CH3C(O)O2 + NO2
CH3C(O)O2NO2 + hν f CH3C(O)O + NO3.
(2)
(3)
While the results so far reported9,10,12 indicate that NO2 is the
PAN
NO3
dominant direct photoproduct in PAN photolysis, Φ
cannot
be discounted due to the differences in resulting atmospheric
chemistry of the two photolysis channels.
* To whom correspondence should be addressed. E-mail: ffwrs@uaf.edu.
† Permanent Address: Department of Chemistry and Biochemistry, The
University of Texas at Austin, Austin, TX 78712.
In the present work, PAN photolysis was studied at a
photolysis wavelength of 289 nm using cavity ring-down
absorption spectroscopy (CRDS) to detect nitrate radicals.13 The
‡ Present address: Solar-Terresterial Environment Laboratory and Gradu-
ate School of Science, Nagoya University, Toyokawa, Japan 442-8507.
10.1021/jp045529n CCC: $30.25 © 2005 American Chemical Society
Published on Web 02/26/2005

Nitrate Radical Quantum Yield
J. Phys. Chem. A, Vol. 109, No. 11, 2005 2553
Figure 1. Pulsed CRDS photolysis-flow experimental schematic. Not drawn to scale.
photolysis wavelength was chosen to coincide with the begin-
ning of the solar flux available for atmospheric photochemisty.
The description of the instrument and experimental technique
is given along with analysis of these results at 289 nm in the
following sections.
laser is scanned from 655 to 669 nm in 0.5 nm steps. The ring-
down signal is detected by an avalanche photodiode detector
(Hamamatsu C5460) and read into a computer by use of
LabVIEW virtual instruments (National Instruments).
The main beam (95%) from the Nd:YAG laser is directed to
a second dye laser (Sirah Cobra Stretch). The 289 nm UV light
is produced by frequency doubling the 578 nm output of
Rhodamine 6G dye laser using a KDP doubling crystal. The
pump and probe beams cross in the center of the flow chamber
at a 2.1° angle. In this crossing geometry the CRDS beam (e1
mm2) probes the interior of the photolysis beam (3 mm2).
Typical photolysis energies of 1 mJ are used. Maximum output
energy of the photolysis laser is 3.3 mJ, and the energy is
attenuated by rotatable half-wave plate/polarizer optics20 placed
in the beam path before the photolysis beam entered the flow
chamber. With these optics, the photolysis beam could be
attenuated between 0 and 3.0 mJ.
Initially, the baseline ring-down time was found to decay due
to heating of the CRDS mirrors from the photolysis beam. The
CRDS mirrors are coated with Ta2O5 layers that absorb in the
UV, causing this heating. To decrease the UV heating problem,
the chamber windows were replaced with antireflective coated
windows that had e0.05% UV reflection. The decreased
reflectivity caused less scattered UV to be present in the chamber
and minimized the UV heating effect.18 To remove residual
effects due to UV mirror heating, a rapid background subtraction
procedure is used. Both photolysis and the CRDS beams are
driven at the 10 Hz repetition of the Nd:YAG laser. Between
the main chamber and the UV photolysis laser is a shutter timed
to allow five UV pulses through and to block five pulses hereby
establishing photolysis on and off states of the experiment. We
then observe the change in ring down time between photolysis
on and off states to observe photochemistry specifically. The 1
Hz shutter frequency proved fast enough that no heating was
observable in the difference between on and off states. By
maintaining total pressures of 10 Torr and a flow rate of 0.1
SLPM in the main flow chamber, we were able to completely
flush the photoproducts between successive laser shots.18
A purge flow of N2 is used between the main flow and mirrors
to minimize contamination of the CRDS mirrors. Because of
this purge gas, the mirror-to-mirror length (l ) 50 cm) of the
ring-down cavity is not completely filled by precursor gases.
Experiment
Samples of N2O5 and peroxyacetyl nitrate (PAN) are pho-
tolyzed, producing NO3 radicals that are detected by cavity ring-
down absorption measurements. The relatively low number
density of NO3 due to weak absorption of PAN at 289 nm is
overcome using the high sensitivity of CRDS pulsed photolysis-
flow experiments described below. Using CRDS to detect
photochemical reaction products has been reported previously
by several research groups.14,15 Figure 1 is a schematic drawing
of the experiment. The flow chamber is constructed of a 100
cm long, 7.6 cm outer diameter stainless steel tube with a 50
cm long, 3.8 cm outer diameter set of arms crossing its center
at 30 °. A mechanical vacuum pump (Edwards E2M18) is used
to evacuate and flow samples in the main chamber. The sample
flow rate was sufficient to replace completely the precursor gases
every laser shot. Total system pressure is measured by a
capacitance manometer (MKS Baratron Type 622) connected
via 1.3 cm outer diameter Ultratorr fitting directly atop the center
crossing point.
The photolysis windows and cavity mirrors are held to the
ends of the crossing volume with 6.9 cm diameter bellow flanges
held in adjustable mounts to allow alignment of the ring-down
cavity. The ring-down cavity is formed by two high-reflectivity
mirrors (Research Electro Optics). The reflectivity of these
mirrors is 0.99995 and optimal alignment of the 50 cm long
cavity produced empty cavity ring down times of 37 µsec. To
produce the CRDS probe beam, 5% of a pulsed Nd:YAG laser
beam (Spectra Physics Quanta Ray Lab 150) pumps a dye laser
(Dakota Technologies Northern Lights) filled with DCM dye.
The dye laser typically produces 1.0 mJ per pulse. The CRDS
beam passes through focusing optics and an iris to condition
the beam in an attempt to excite primarily low order cavity
modes. Selection of low-order cavity modes decreases the noise
in the CRDS measurement. To detect the strong (B˜ r X˜)
electronic absorption band of NO3 centered at 662 nm, the dye

2554 J. Phys. Chem. A, Vol. 109, No. 11, 2005
Flowers et al.
Separate experiments using water vapor showed that ap-
proximately 15 cm along the cavity axis were filled by flowing
precursor gases. In the pump-probe experiments, only the
coincident region between the photolysis and CRDS laser beams
is sampled. We estimate this crossing length to be ∼7 cm and
is entirely within the flowing precursor gas. However, the precise
value of this crossing length is not important because the PAN
quantum yield is measured relative to a reference molecule,
N2O5, as described later.
by the report of Wangberg et al., and no error is assumed for
these cross sections in this work either. Although HNO3
contamination was sometimes present in the N2O5 samples, the
NO3 quantum yield from HNO3 photolysis is very low (<0.002),26
thus nitric acid did not contribute to photolytic NO3 significantly
in these experiments.
When using the N2O5 sample, the carrier gas is seeded with
∼1014 molecules cm-3 of NO2. The added NO2 decreases the
thermal NO3 background in the N2O5 sample by forcing the
chemical equilibrium NO2 + NO3 h N2O5 toward N2O5.19,24
Because the NO2 is present throughout the photolysis modulation
cycle, its absorption is removed by differencing between the
absorption measured in photolysis on and off states. A small
fraction of the NO2 is photolyzed to produce NO, which reacts
Peroxyacetyl nitrate samples are synthesized by the technique
of Gaffney et al.17 and stored at 220 K. To supply PAN to the
photolysis experiment, a flow of N2 carrier gas is bubbled
through PAN sample in tridecane solvent placed in a ethylene
glycol/water bath chilled to 269K. The typical number density
of PAN is (3-30) × 1014 molecules cm-3. Dinitrogen pentoxide
is synthesized in a home-built glass rack using a method similar
to Davidson et al..23 Briefly, N2O5 is trapped as a product of
the reaction between NO2 and excess O3. After several freeze/
thaw/react/trap cycles the NO2 charge introduced to the glass
rack is completely oxidized to N2O5 and stored at 220 K. The
N2O5 sample was stored in dry ice/trichloroethylene slush bath
during use. Typical concentrations of (1-5) × 1014 molecules
cm-3 N2O5 are supplied for photolysis experiments. The purity
of both samples was established by FTIR absorption and used
without further purification. Both samples are supplied to the
vacuum chamber through PFA Teflon tubing from their respec-
tive baths. The loss of either sample to the surface of the tubing
and vacuum chamber was tested by switching the flow order
to have the FTIR quantification either before or after the sample
chamber. This test showed negligible loss of precursor gases.18
rapidly with NO3 (k
) 2.6 × 10-11 cm3 molecule-1 s-1
)
289
and thus could present a photolytic impurity. However, only
∼1011 molecules cm-3 of NO are formed, leading to a lifetime
of NO3 of ∼0.4 s., which is longer than both the ring-down
time scale and the flow flushing time scale. Therefore, addition
of NO2 has only the desired effect of suppression of thermal
NO3 background.
In the relative quantum yield experiments, a scan of the
detection range was performed during photolysis of N2O5, then
the photolysis chamber was isolated from the N2O5 flow and
evacuated. After evacuation, the carrier flow was switched to
the PAN sample then that flow was introduced to the photolysis
chamber at the same conditions. We then switched back to
flowing the N2O5 sample to bracket the PAN photolysis
measurements. The CRDS detection range was scanned at 4
nm per minute. This scan rate covers the detection range in 4
min, so 4 FTIR scans at 70 s each were taken concurrently to
determine the precursor concentration as a function of time
during the scan. The concentration of N2O5 supplied to the
photolysis experiment was controlled by the temperature of the
bath. It was found to be straightforward to maintain relatively
constant concentrations of N2O5 over the course of the experi-
ment. Despite our best efforts, the PAN concentration decreases
rapidly over the duration of the experiment. The final concentra-
tion of PAN was typically found to be half the concentration at
the beginning of the spectral scan. Much of this decay affects
the wings of the NO3 spectrum and is thus not critical for
accurate determination of the amount of photolytic NO3, but
we sought a systematic method to remove this precursor
concentration decay.
Precursor concentrations are measured upstream of the
photolysis chamber by an FTIR absorption spectrometer (Perkin-
Elmer Paragon 1000, 4.0 cm-1resolution). The precursor gases
passed through a 17.3 cm length flow cell with CaF2 windows
housed permanently in the FTIR sample chamber. Gas concen-
trations were determined by fitting the FTIR spectrum over the
region 1000-2000 cm-1 to a linear combination of spectra of
the precursor gas and other possible impurities. A nonlinear
Levenberg-Marquart algorithm was used in the fitting (Wave-
metrics Igor Pro). In this manner, both the precursor gas
concentration and possible impurities were quantified. For PAN
quantification, the spectra of PAN, acetyl nitrate, acetic acid,
nitric acid, as well as a linear slope and offset were used. All
reference spectra were measured on our FTIR spectrometer. For
quantification of PAN, the integrated cross sections of Tsalkani
and Toupance16 were used to scale the standard reference
spectrum. The maximum percent error on these cross sections
is reported to be 2% for the 1163.5 cm-1 peak. The cross
sections of Gaffney et al.17 are similar except for the 1740.5
cm-1 peak, as described in Tsalkani and Toupance.16 The
reported errors for the cross sections are significantly less than
the instrumental precision in the number densities measurement
(∼ 10-20%), thus the cross section error is not propagated
through the number density calculations. In all cases we find
the concentration of selected impurities negligible. Also, no
impurity in the PAN sample has an NO3 photoproduct at the
photolysis wavelength. For N2O5 quantification, the same
method was used, but the reference spectra of N2O5, NO2, H2O
and nitric acid were used in addition to the linear slope and
offset. The N2O5 reference spectrum was scaled by using the
cross section of Wangberg et al. at 1245 cm-1(σN O ) 8.56 ×
During each experiment, the four determinations of precursor
concentration were fitted to an exponential or linear decay
function. This decay function was then normalized to unity at
the time when the NO3 product scan was at the 662 nm
absorption peak. We then divided each of the four FTIR
determinations by the normalized decay function to give a
detrended precursor concentration. The average and standard
deviations of these detrended precursor concentrations are shown
in Table 1. The difference in absorption from the CRDS
measurements are also divided by the same normalized precursor
decay function, determined by FTIR, resulting in a detrended
product absorption spectrum. This product spectrum is then fitted
to a literature NO3 spectrum for quantification of the amount
of NO3 produced in the experiment. The error on the NO3
measurement comes from the error estimate in the nonlinear
least-squares fitting procedure. Photolysis pulse energy was
monitored throughout each run to ensure that a constant
photolysis energy was maintained for the experiment.
2
5
10-19 cm2 molecule -1).19 The NO2 reference spectrum quan-
tification was also taken from Wangberg et al. at 1602
cm-1(σNO ) 4.71 × 1019 cm2 molecule-1). No error is given
2

Nitrate Radical Quantum Yield
J. Phys. Chem. A, Vol. 109, No. 11, 2005 2555
TABLE 1: Number Densities and Quantum Yields for
PAN/N2O5 Photolysis Experiments at 289 nm
PAN
NO3
expt no.
[N2O5]a
[PAN]a
NO3 ratio
Φ
1
2
3
4
5
6
1.30 ( 0.10 6.49 ( 0.89
2.37 ( 0.02 27.7 ( 0.3
1.31 ( 0.10 3.37 ( 0.19
0.84 ( 0.02 2.76 ( 0.02
0.14 ( 0.01
0.23 ( 0.01
0.08 ( 0.01
0.08 ( 0.01
0.35 ( 0.06
0.24 ( 0.01
0.39 ( 0.06
0.30 ( 0.04
5.43 ( 0.51 5.81 ( 1.50 0.024 ( 0.002 0.28 ( 0.08
4.44 ( 0.42 4.28 ( 0.55 0.024 ( 0.001 0.31 ( 0.05
a Number densities are reported per 1014 molecules cm-3. The
uncertainties reported for the number densities are 2σ. The NO3 ratio
error is described in the text.
Results and Discussion
The photolytic NO3 concentration is determined from the
difference of absorption between photolysis on and off states.
For both states the measured ring-down time, (τ), is converted
to a loss rate per distance, r ) (cτ)-1, which is proportional to
the optical absorption. The difference in loss rate, ∆r, between
photolysis on and off values is related to the photolytic change
in [NO3] number density by
Figure 2. Ring-down signal decay profiles for photolysis off (dark)
and photolysis on (empty) states during N2O5photolysis.
1 1
1
l
∆r )
-
) ∆[NO3]σNO .
3
(4)
(
)
c τON τBL ls
The absorption cross section σNO used in this analysis is taken
3
from the 2003 JPL reference.26 Many researchers have measured
this cross section and have found different values, indicating
significant uncertainty in the actual peak cross section of NO3
absorption, however the actual peak value is unimportant for
the present experiment because only relative amount of NO3
formed between the PAN and N2O5 photolysis experiments
enters into the final analysis. The cavity length scaling factor
(l/ls) describes the fraction of the total cavity length, l, divided
by the length that is filled by absorber, lS. These values were
described in the Experimental Section and are not critical
because of the use of a reference molecule. The ring-down time
of the photolysis off state is τBL and the ring-down time for the
photolysis on state is τON. Typical total NO3 number densities
of ∼109 molecules cm-3 NO3 were detected. Previous work on
this system has indicated that the detection limit for NO3 is
Figure 3. Temporal profile of the cavity loss rate measured in the
presence of NO3 (empty circles) and the empty cavity loss rate (dark
circles). Individual data are the result of the photolytic differencing
procedure described in the text.
A very careful examination of the ring-down decays shown
in Figure 2 shows a slightly nonexponential decay behavior in
both photolytic and nonphotolytic signals. To establish that the
slight deviation of the ring-down waveforms from exponential
behavior is not caused by temporal changes in the photolytic
radical number density observation, the temporal profile of the
cavity loss rate r(t) was determined. Figure 3 shows the
calculated instantaneous loss rates (r) for the cavity in the
presence of NO3 (open circles) and absence of sample (dark
circles). The change in loss rate with time is calculated as the
difference between slopes of the intensity decay during modu-
lated photolysis with and without N2O5 present in the cavity.
This kinetic analysis was presented in detail in the work of
Brown et al.15 termed SKaR (simultaneous kinetics and ring-
down). In our case, however, we are interested in ensuring that
there is no NO3 loss on the time scale of the ring down event.
The data points shown Figure 3 are differences between r values
for photolysis states via
approximately 2 × 107 molecules cm-3 18
.
Quantitative measurement of the NO3 quantum yield from
PAN photolysis requires that the photolytic NO3 is detected
more rapidly than is lost due to motion away from the photolysis
volume or lost by chemical reaction. Therefore, we studied the
temporal behavior of the ring-down decay profiles to ensure
no loss of NO3 on the relevant time scale, which is on the order
of the ring-down time. Figure 2 shows the ring-down decay
profile for N2O5 photolysis with NO3 detection at 662 nm. The
filled circles show the baseline decay profile, in the absence of
photolysis, while the open circles show the photolytic decay
profile. The open circles decay more rapidly due to increased
absorbance of the photoinduced NO3. The flat plateau at early
time in both profiles is due to a detector cutoff at 9 V and is
ignored in the analysis routine. We intentionally align the system
to have some cutoff intensity to remove rapidly decaying high-
order mode cavity excitations from the ring-down profile. These
higher-order modes would cause nonexponential ring-down
decay profiles and cause increased noise in the CRDS measure-
ment. Inspection of both profiles shows nearly exponential
decays before the signal decays into noise. The observation of
an exponential decay for the photolytic (open) data indicates
that the concentration of NO3 is constant over the decay profile.
d(ln I(t))
dt
d(ln I(t))
dt
1
c
∆r )
-
(5)
ON
OFF
[
]
The dark circles in Figure 3 are the difference between loss
rates with and without photolysis in the absence of sample. The

2556 J. Phys. Chem. A, Vol. 109, No. 11, 2005
Flowers et al.
Figure 4. Plot of the natural logarithm of the photolytic difference in
optical loss, ∆r, vs the natural logarithm of the photolysis power. The
photolysis power was varied between 0 and 2.0 mJ to measure the
power dependence for N2O5 photolysis. The slope of the line is 0.9 (
0.1, indicating a linear relationship between power and signal for N2O5
photolysis up to 2.0 mJ.
zero value indicates that the rapid photolysis modulation
removes any effect of mirror heating. The open circles are the
difference in loss rates observed during N2O5 photolysis. By
inspection of the data up to ∼40 µs, the loss rate of the cavity
in the presence of NO3 varies by less than 10%. Taken together,
the data shown in Figures 2 and 3 show that the photolytic [NO3]
is constant on the time scale of the ring-down event; i.e, the
bulk flow rate of sample into the vacuum chamber is not
removing NO3 on the experimental time scale. An additional
concern is the possibility of the photolysis event depositing
thermal energy in the NO3 fragments such that the hot fragments
fly out of the CRDS probe region while they are being detected.
At longer ring down times there would effectively be less NO3
to detect and the open circles in Figures 2 and 3 would approach
the dark circles. Clearly we do not observe this flyout effect.
Torabi27 reported a relaxation time for vibrationally hot NO3
produced from N2O5 photolysis of 7 µs in 50 Torr of He. Our
conditions are at lower pressure, 10 Torr, but with a more
efficient collisional quencher, nitrogen gas, thus one might
expect a similar time scale for collisional quenching. We see
no evidence of changes in the concentration of NO3 early in
the ring-down decay profile, thus any affect of this vibrational
relaxation is not evident in our experiment. Additionally, the
un-analyzed signal saturation in the beginning of the ring-down
decay (∼10 µs) may preclude observation of effects from this
collisional relaxation. Thus, we conclude that cavity ring-down
spectroscopy of the NO3 B˜ r X˜ band quantitatively measures
the product from these photolysis reactions.
Another requirement for determining a photolysis quantum
yield for one species relative to another is that both species
must be undergoing one-photon photolysis reactions. In this
case, N2O5 has a larger absorption cross section at 289 nm than
does PAN. Therefore, we sought to ensure that we are observing
a one-photon photolysis process for N2O5. If linear photochem-
istry is observed for N2O5 photolysis, it will be so for the weaker
absorbing PAN photolysis as well. A power dependence study
for N2O5 photolysis was performed detecting NO3 at 662 nm
using CRDS. By using the polarizer/wave plate optic, the
photolysis power used for N2O5 photolysis was varied from 0
to 2.0 mJ by 0.2 mJ increments. The results from this study are
shown in Figure 4, which plots the natural logarithm of the
Figure 5. Ring down times τON and τBL recorded for photolysis on
(dashed) and off (solid) states during PAN (a) and N2O5 (b) photolysis.
The general decrease in τBL is due to the transmission of the CRDS
mirrors. The dip in the solid line at 662 nm in part b is the detection
of thermal NO3 from N2O5 thermolysis. This background signals are
eliminated through the differencing procedure described in the text.
To minimize this background in practice the flow of N2O5 was seeded
with NO2. This NO2 shows some small absorption features, most notably
at 656 nm.
photolysis power vs the natural logarithm of the NO3 signal
observed. The NO3 signal observed at each power datum was
normalized by the absorbance of N2O5 recorded upstream by
the FTIR spectrometer. The points in Figure 4 are the result of
averaging four measurements at each photolysis power. The
slope of the log-log plot is the order of the power dependence,
N, and should be 1 for a one-photon process. The slope of the
data in Figure 4 indicates that N ) 0.9 ( 0.1, therefore we are
confident that for photolysis powers used in our experiment we
are observing a one-photon process for N2O5 and subsequently
PAN photolysis up to 2.0 mJ per pulse.
Raw data from N2O5 and PAN photolysis are shown in Figure
5, parts a and b, respectively. The general decrease in τBL vs
wavelength is characteristic of the reflectivity of the mirrors
used in the cavity, which decreases at longer wavelengths. In
Figure 5b, the additional decrease in τBL at 662 nm without
photolysis is due to absorption of background NO3 from N2O5
thermal decomposition. This spectral feature is further evidenced
by the return of τBL to its wavelength transmission near 665
nm. The thermal NO3 is removed from the photolytic NO3
number density by the differencing procedure in (4). The CRDS

Nitrate Radical Quantum Yield
J. Phys. Chem. A, Vol. 109, No. 11, 2005 2557
Figure 6. CRDS difference spectra recorded during N2O5 photolysis
at 289 nm. The open circles are the difference between optical loss (r)
observed in the presence and absence of photolysis. The solid line is
a reference spectrum of NO3 taken from the JPL compendium.26
Figure 7. CRDS difference spectra recorded during PAN photolysis
at 289 nm.
laser wavelength scan is compared to a reference spectrum of
NO3 to eliminate the possibility that additional absorbers
contribute to the NO3 absorption measurement.
The nitrate radical quantum yield from PAN relative to N2O5
is determined by
σ
N˜
N2O5
[NO3]PAN
[NO3]N O
N2O5
PAN
NO3
N2O5
NO3
Φ
) Φ
×
×
5
×
.
(6)
σPAN N˜PAN
2
Here σx is the UV absorption cross section of either PAN or
N2O5, [NO3]x is the nitrate radical concentration detected from
either PAN or N2O5 photolysis, and N˜x is the precursor
concentration of PAN or N2O5 measured upstream of the
photolysis chamber via FTIR. The NO3 quantum yield from
N2O5 at 289 nm is assumed to be unity. The accepted value for
Figure 8. Nitrate radical quantum yields from PAN photolysis reported
at several wavelengths. The data of Mazely et al. at 248 nm is denoted
by the open box and has been adjusted as described in the text. The
data of Harwood et al. at 248 and 308 nm is shown as O. The point at
289 nm is the result of the current study and is shown as 4.
N2O5
Φ
has been the topic of some discussion, however the
NO3
available results indicate that at wavelengths longer than 248
N2O5
NO3
nm, Φ
approaches unity rapidly.25,26 The 248 nm ΦNN2OO5
)
3
0.8 ( 0.1, while at 308 nm, the quantum yield is unity.25 Simple
linear interpolation between these two values indicates a 289
largest contribution to the overall error. Another error is the
noise for the photolysis product ratio involved in the relative
quantum yield expression (6). These error estimates were
propagated for each individual experiment to give a 2σ error
estimate for each individual experiment, as listed in the right-
hand column of Table 1. To combine these error estimates, we
average them and find an 2σ estimate of 0.05. In addition, each
of the six experiments gave a different numerical result, which
we took into account by calculating the standard deviation of
the individual results. This experiment-to-experiment variation
gave a 2σ estimate of 0.05. We combine these two error
estimates in quadrature (root-mean-square) to give an experi-
mental precision of 0.07 (2σ). In addition to these experimental
errors, we also want to consider the systematic errors. We found
that the FTIR cross section error estimates were much smaller
than the experimental precision, so we ignored these small
errors. Because the quantum yield of the reference gas, N2O5
is not known at 289 nm, we consider a 10% error for this value.
Addition of this 10% error in quadrature to the experimental
N2O5
NO3
nm Φ
) 0.96, which is within the 248 nm experimental
N2O5
NO3
error of unity. Therefore, we assume that the 289 nm Φ
)
1.0 and propagate 10% error for this assumption. Because the
photolysis measurements are done under identical experimental
conditions for both PAN and the N2O5 reference, the crossing
volume length from (4) cancels. The precursor detrended
photolytic difference spectra for nitrate radical from PAN and
N2O5 photolysis are fitted to a NO3 reference spectrum and
shown in Figures 6 and 7. The open circles represent the
difference between NO3 absorption in the dark and light states
of the experiment. The solid line is a reference spectrum of
nitrate radical taken from the JPL reference.26 Only the relative
NO3 number density is of interest, making the absolute value
used for σNO3 unimportant. From the agreement between the
difference and reference spectra we feel assured that the
photoproduct is nitrate radical.
The measurement of NO3 quantum yield for PAN photolysis
was performed in six independent experiments. The precursor
number densities (N˜N O , N˜PAN) and photoproduct NO3 ratios
errors gives an overall error estimate of 0.08 at the 2σ level.
2
5
PAN
NO3
([NO3]PAN/[NO3]N O ) are shown in Table 1. Taking the average
Therefore, our overall result is Φ
) 0.31 ( 0.08 (( 2σ).
2
5
PAN
NO3
of the six experiments, we find the quantum yield (Φ ) to be
Our result agrees qualitatively with previously reported results
at 248 and 308 nm,10,12 although using a different detection
method for NO3 and the wavelength is close to but not the same
as the other measurements. Figure 8 plots the result of this work
0.31. There are several uncertainties that need to be combined
in the final reported error estimate. One error takes into account
the noise (2σ) in the precursor number density, which is the

2558 J. Phys. Chem. A, Vol. 109, No. 11, 2005
Flowers et al.
along with the values reported in Harwood et al. at 248 and
308 nm and the adjusted results from Mazely et al. at 248 nm.
in acquiring some of the data described in this paper. We would
also like to acknowledge helpful discussions from Dr. Ruth
Shear.
The adjustment made to the results of Mazely et al. was to use
N2O5
NO3
the value for Φ
of 0.80 as reported in Harwood et al.25 The
References and Notes
results in Figure 8 are all derived from PAN/N2O5photolysis
experiments using different detection methods for nitrate radical.
The combination of our and past work implies that as photolysis
energy increases, the NO3 quantum yield for PAN photolysis
decreases, which is similar to the trend for N2O5. As the
photolysis energy increases, more photoprocesses within NO3*
become possible leading to NO3 fluorescence and direct
dissociation into NO2 and/or NO.25,28,29 For atmospheric pho-
tochemistry, the wavelength dependence between 290 and 320
nm is of critical interest. The trend in NO3 quantum yield for
PAN is less clear in this atmospherically relevant region. Our
(1) Roberts, J. M. Atmos. EnViron. 1990, 24A, 243.
(2) Moxim, W. J.; Levy, H., II; Kasibhatla, P. S. J. Geophys. Res. 1996,
101, 12621.
(3) Talukdar, R.; Burkholder, J.; Schmolter, A.; Roberts, J.; Wilson,
R.; Ravishankara, A. J. Geophys. Res. 1995, 100, 14163.
(4) Orlando, J.; Tyndall, G.; Calvert, J. Atmos. EnViron. 1992, 26A,
3111.
(5) Miller, C. E.; Lynton, J. I.; Keevil, D. M.; Francisco, J. S. J. Phys.
Chem. A. 1999, 103, 11451.
(6) Bridier, I.; Caralp. F.; Loriat, H.; Lesclaux, R.; Veyret, B.; Becker,
K.; Reimer, A.; Zabel, F. J. Phys. Chem. 1991, 95, 3594 and references
cited within.
(7) Honrath, R. E.; Jaffe, D. A. J. Geophys. Res. 1992, 97, 20615 and
Beine, H. J.; Jaffe, D. A.; Blake, D. R.; Atlas, E.; Harris, J. J. Geophys.
Res. 1996, 101, 12613.
(8) Stroud, C.; Madronich, S.; Atlas, E.; Ridley, B.; Flocke, F.;
Weinheimer, A.; Talbot, B.; Fried, A.; Wert, B.; Shetter, R.; Lefer, B.;
Coffey, M.; Heikes, B.; Blake, D. Atmos. EnViron. 2003, 37, 3351.
(9) Mazely, T.; Freidl, R.; Sander, S. J. Phys. Chem. 1995, 99, 8612.
(10) Mazely, T.; Freidl, R.; Sander, S. J. Phys. Chem. A. 1997, 101,
7090.
(11) Senum, G.; Lee, Y.; Gaffney, J. J. Phys. Chem. 1984, 88, 1269.
(12) Harwood, M.; Roberts, J.; Frost, G.; Ravishankara, A.; Burkholder,
J. J. Phys. Chem. A. 2003, 107, 1148.
result at 289 nm suggests a lower limit of 0.31 and an increasing
PAN
NO3
trend for Φ
relevant in atmospheric chemistry models.
PAN
However, there is only one measurement at 308 nm for Φ
NO3
to confirm this trend.
Conclusions
PAN
NO3
We report a photolysis quantum yield (Φ ) relative to that
of N2O5 at 289 nm of 0.31 ( 0.08 ((2σ). This result is derived
from six separate measurements using pump-probe CRDS
experiments at the University of Alaska. Several experimental
checks were performed to ensure the accuracy of the experi-
mental results from this instrument including a detailed char-
acterization of the performance of the ring-down cavity. These
tests included establishing the proper flow conditions for the
sample gases and timing of the laser shots to ensure that the
photoproducts are removed from the detection volume between
each laser shot. Also, a power dependence study of the reference
gas N2O5 was performed. This check ensured that we are
observing a one-photon process. Furthermore, we have inves-
tigated the temporal profiles of the cavity loss rates in the
presence and absence of photoproducts to ensure there is no
systematic NO3 loss, or gain, on the time scale of the ring-
down event due to flow conditions, cavity alignment, or
secondary chemistry.
(13) Cavity ring-down absorption spectoscopy has been reviewed
extensively in recent years and only described briefly here. The reader is
directed to the review article Wheeler, M.; Neuman, S.; Orr-Ewing, A.;
Ashfold, M. J. Chem. Soc., Faraday Trans. 1998, 94, 337 for further detail.
(14) Chen, Y.; Zhu, L. J. Phys. Chem. A. 2003, 107, 4643 and references
cited within.
(15) Brown, S.; Ravishankara, A. R.; Stark, H. J. Phys. Chem. A. 2000,
104, 7044.
(16) Tsalkani, N.; Toupance, G. Atmos. EnViron. 1989, 23, 1849.
(17) Gaffney, J.; Fajer, R.; Senum, G. Atmos. EnViron. 1984, 18, 215.
(18) Angerhofer, M. Photodissociation and Quantum Yield Determina-
tion of Peroxyacetyl Nitrate. M.S. Thesis, University of Alaska, Fairbanks,
AK, 2003.
(19) Wanberg, I.; Etzkorn, T.; Barnes, I.; Platt, U.; Becker, K. J. Phys.
Chem. A. 1997, 101, 9694.
(20) Cooper, J.; Zare, R. J. Chem. Phys. 1968, 48, 942.
(21) Wayne, R.; Barnes, I.; Biggs, P.; Burrows, J.; Canosa-Mas, C.;
Hjorth, J.; Le Bras, G.; Moortgat, G.; Perner, D.; Poulet, G.; Restelli, G.;
Sidebottom, H. Atmos. EnViron. 1991, 25A (1), 1.
(22) King, M.; Dick, E.; Simpson, W. Atmos. EnViron. 2000, 34, 685.
(23) Davidson, D.; Viggiano, A.; Howard, C.; Dotan, L.; Fehsenfeld,
F.; Albritton, D.; Ferguson, E. J. Chem. Phys. 1978, 68, 2085.
(24) Wanberg, I.; Etzkorn, T.; Barnes, I., Platt, U.; and Becker, K.; J.
Phys. Chem. A., 101, 9694.
(25) Cantrell, C.; Davidson, J.; McDaniel, A.; Shetter, R.; Calvert, J. J.
Chem. Phys. 1988, 88, 4997.
(26) Harwood, M.; Burkholder, J.; Ravishankara, A. J. Phys. Chem. A.
1998, 102, 1309 and references cited within.
(27) Sander, S. P.; Freidl, R. R.; Golden, D. M.; Kurylo, M. J.; Huie,
R. E.; Orkin, V. L.; Moortgat, G. K.; Ravishankara, A. R.; Kolb, C. E.;
Molina, M. J.; Finlayson-Pitts, B. J. Chemical Kinetics and Photochemical
Data for Use in Atmospheric Studies; Evaluation 14, Publication 02-25;
NASA/Jet Propulsion Laboratory: Pasadena, CA, 2003, and references cited
within.
Figure 8 plots the results of this work along with that of the
PAN
NO3
previous studies of Φ
using different experimental tech-
niques. We find our result at 289 nm to be in qualitative
agreement with the previously reported trend in results for
PAN
NO3
Φ
between 248 and 308 nm. However, uncertainty in the
trend at longer wavelengths indicates that consistent experiments
at more wavelengths within the UV-B range would be most
useful to determine whether there is a wavelength dependence
for the nitrate radical quantum yield from PAN photolysis
relevant to atmospheric photochemistry.
Acknowledgment. B.A.F. thanks John F. Stanton of The
University of Texas for support during the initial portion of this
collaboration. JFS is supported by the National Science Founda-
tion and the Robert A. Welch Foundation. W.R.S. and M.E.A.
are also supported by the National Science Foundation (CHE-
0094038). We acknowledge the assistance of Sarah Kvasnicka
(28) Torabi, A. An Investigation of the Kinetics and Excited-State
Dynamics of the Nitrate Free Radical. Ph.D. Thesis, Georgia Institute of
Technology 1985.
(29) Oh, D.; Sisk, W.; Young, A.; Johnston, H. S. J. Chem. Phys. 1986,
85, 154.
(30) Johnston, H. S.; Davis, H. F.; Lee, Y. T. J. Phys. Chem. 1996,
100, 4713.