Introduction
Short-lived free radicals play a crucial role in determining the composition
of the atmosphere. The catalytic cycle of HOx (= OH + HO2)
radicals is of central importance to tropospheric chemistry. OH acts as the
primary daytime oxidant, initiating the degradation of most trace gases and
thereby controlling their atmospheric concentrations and lifetimes. The
short lifetime of the radicals generate HOx concentrations which are
uninfluenced by transport; therefore reproducing observed HOx
concentrations can be an excellent test of any chemical model (Heard and
Pilling, 2003). However, it should be mentioned that agreement between
measured and modelled [HOx] could be fortuitous as both sources and
sinks of HOx radicals could be simultaneously under- or overestimated.
The development of detection techniques that permit the speciation of a
wider range of atmospheric components (e.g. volatile organic compounds (VOCs) and aerosols), together
with OH reactivity measurements, can help to further constrain modelling
studies and reduce the potential for the coincidental agreement. OH has
been detected by long-path differential absorption spectroscopy (DOAS) in
the field (Brauers et al., 1996; Dorn et al., 1996) and in the
SAPHIR
chamber (Schlosser et al., 2009), and chemical ionisation mass spectrometric
techniques (CIMSs) have also been used in field observations (Eisele and
Tanner, 1991; Berresheim et al., 2002; Sjostedt et al., 2007; Kukui et al.,
2008). However, a majority of field measurements have been made using
laser-induced fluorescence (LIF) spectroscopy, and intercomparisons exist which
have validated the technique against DOAS and CIMS in both chamber
(Schlosser et al., 2007, 2009; Fuchs et al., 2012) and
field environments, including aircraft-based measurements (Eisele et al.,
2001, 2003). Low concentrations and potential interferences
(Fuchs et al., 2011; Mao et al., 2012; Whalley et al., 2013) make HOx
measurements challenging. In addition, most HOx detection methods are
not absolute, and hence calibration is required.
Fluorescence assay by gas expansion, FAGE, is a low-pressure LIF technique
commonly used for the detection of OH and HO2 radicals (Heard, 2006, and references therein). The low concentrations of ambient OH
(∼ 106 molecule cm-3) require a viable measurement
technique to discriminate between laser-scattered light and small signal
levels. Originally this was attempted by exciting OH to the first
vibrational level in the A state at 282 nm (A2Σ+ (ν′=1) ←X2Πi (ν′′=0) OH transition) and
observing off-resonant fluorescence at ∼ 308 nm using an
interference filter to help discriminate against scattered laser radiation
(Davis et al., 1976). Although non-resonant LIF has been successful in
stratospheric applications (Wennberg et al., 1994), in the troposphere 282 nm photolysis of ozone (and subsequent reaction of O(1D) with water
vapour) generates an unacceptably high interfering OH signal, and
on-resonant LIF with excitation at 308 nm is used instead. Expanding the
sample through a pinhole to low pressure (∼ 1–2 Torr)
increases the fluorescence lifetime of the A state beyond the laser scatter
pulse, allowing for temporal discrimination against the resonant 308 nm
excitation pulse (A2Σ+ (ν′=0) ← X2Πi(ν′′=0) OH transition). Injection of an OH
scavenger (e.g. C3F6) allows quantification of any laser-generated
OH interference (Mao et al., 2012; Novelli et al., 2014). HO2 is
converted into OH via reaction with added NO:
HO2+NO→OH+NO2
and the resultant OH is detected in the same way. Detection of OH and
HO2 either simultaneously or in series can be achieved using the same
LIF detection axis (measurements in series; Creasey et al., 1997a), with two
separate LIF axes within the same cell (simultaneous; Stevens et al.,
1994) or with two separate detection cells (simultaneous; Whalley et al.,
2010).
LIF is a very sensitive but non-absolute detection method, and therefore
each channel of the instrument needs to be calibrated. The
vacuum-ultraviolet (VUV) photolysis of H2O vapour was originally
developed for the calibration of HOx measurement instruments in the
1990s (Aschmutat et al., 1994; Schultz et al., 1995; Heard and Pilling,
2003). Since then the methodology has become the HOx measurement
community standard. Upon the photolysis of a known H2O vapour
concentration (in synthetic air at atmospheric pressure) by a mercury (Hg)
Pen-Ray lamp at 184.9 nm, OH and HO2 are produced in unity ratio (Fuchs
et al., 2011) via Reactions (R2) and (R3) (Schultz et al., 1995):
H2O+hν⟶λ=184.9nmOH+H,H+O2+M⟶M=N2,O2HO2(+M).
The radicals are then sampled by the HOx instrument at atmospheric
pressure; the concentrations of OH and HO2 produced can be determined
using Eq. (1):
OH=HO2=H2OσH2O, 184.9nmΦOHF184.9 nmΔt,
where [H2O] is the water vapour concentration; σH2O, 184.9nm is the known
absorption cross section of H2O vapour at 184.9 nm ((7.22 ± 0.22) × 10-20 cm2 molecule-1 (Cantrell et
al., 1997; Creasey et al., 2000)); ΦOH (=ΦHO2=1) is the photodissociation quantum
yield of OH and HO2 (Fuchs et al., 2011); F184.9 nm is the photon
flux of 184.9 nm light; and Δt is the exposure time of the air to the
Hg lamp output. There are two main methodologies used for obtaining the
product F184.9 nm,Δt in Eq. (1). In the first, the
two parameters are measured separately, F184.9 nm using a calibrated
phototube and Δt using knowledge of the volumetric flow rate and
geometric parameters of the flow tube (Stevens et al., 1994). In the other,
a chemical actinometer is used to obtain the product directly, with both
O2 and N2O photolysis at 184.9 nm used to generate either O3
or NO, which is subsequently detected using commercial analysers, with good
sensitivity (Creasey et al., 1997a; Hofzumahaus et al., 1997; Heard and
Pilling, 2003; Faloona et al., 2004). There are two main methods for
delivery of the OH radicals to the FAGE inlet at atmospheric pressure,
either using a laminar or turbulent flow tube. In the laminar flow regime
there is a radial gradient in the OH concentration for which the so-called
profile factor (P) has to be quantified (Holland et al., 1995; Creasey et
al., 1997a), whereas in a turbulent flow system the radial OH
concentration is constant except very close to the walls.
Alternative calibration methods have also been developed, but typically not
deployed in the field, and examples of these will be employed in the current
study. A detailed evaluation of calibration techniques has been presented by
Dusanter et al. (2008). In some of the earliest field measurements, Hard et
al. (1995) developed a calibration method based on hydrocarbon decays. The
concentration of a hydrocarbon with a known and well-characterised rate
coefficient for reaction with OH, k (in this case 1,3,5-trimethylbenzene),
was measured as a function of time using gas chromatography, allowing
determination of all the parameters in Eq. (2) with the exception of [OH]. The
rate of loss of a hydrocarbon (HC) through reaction with OH is given by Eq. (2):
d[HC]dt=kOH[HC].
This methodology has also been applied more recently to FAGE validation
measurements in the EUPHORE chamber (Bloss et al., 2004).
For HO2 the well-defined second-order recombination rate coefficient
for Reaction (R4) can be used to determine [HO2], where for a second-order
reaction the half-life of the decay is related to the initial starting
concentration.
HO2+HO2→H2O2+O2
In a short set of experiments, Pilling et al. (2005) generated HO2 from
the photolysis of formaldehyde in the EUPHORE chamber and observed the
second-order HO2 decay with a FAGE instrument. The decays were in good
agreement with the calibrated HO2 measurements, but no systematic
studies have been undertaken using this reaction as a calibration method.
The deployment of the FAGE technique for aircraft-based measurements
(Faloona et al., 2000; Commane et al., 2010; Martinez et al., 2010) raises
two issues. First, the need to sample air from outside of the boundary layer
of the aircraft fuselage requires a significant length of flow tube before
the gas sample is interrogated by the laser beam. Secondly, the pressure in
the FAGE cell will vary as the aircraft changes altitude (e.g. 0–7 km,
1.3–2.8 mbar internal cell pressure range, from Commane et al., 2010),
altering the instrumental sensitivity (Commane et al., 2010; Martinez et
al., 2010) owing to changes, for example, in the nature of the initial
expansion into the FAGE apparatus. The current design of the flow tube
calibration method is limited to delivering the calibrated [OH] at
atmospheric pressure; however, by using different nozzle pinhole diameters
(typically 0.2–1.0 mm) it is possible to alter the pressure in the FAGE
cell over the range typically encountered during a flight. Importantly, this
method does not compensate for the changing pressure differential across the
inlet nozzle experienced during a flight and what effect this might have on
the expanding gas before it reaches the FAGE cell. The possible changes in
radical surface losses due to the change in inlet pinhole diameter are also
assumed to be negligible.
Potential systematic uncertainties around the application of calibrations
performed at atmospheric pressure to HOx data obtained whilst sampling
from different pressures (e.g. in flight) highlight the need to obtain
calibrations at relevant external pressures. Martinez et al. (2010) have
investigated the effect of external pressure on instrument sensitivity by
calibrating during flight, reporting an increase in the instrument
sensitivity to OH in the free troposphere, compared to the boundary layer.
It was not concluded whether this was an effect of the calibration source
used (conventional H2O vapour photolysis) or the instrument itself;
however the increase was not characterised by the conventional calibrations
performed on the ground before the flight.
We report here an intercomparison of HOx calibrations based on the
conventional flow tube methodology, using different inlet nozzle diameters
to vary the internal fluorescence cell pressure, with two alternative
calibration methods. Analysis of the decays of hydrocarbons was used to
determine [OH], while analysis of the kinetics of HO2 decay by
self-reaction following the photolysis of formaldehyde was used to determine
[HO2]. The studies took place in the Highly Instrumented Reactor for
Atmospheric Chemistry (HIRAC), which is a custom-built atmospheric
simulation chamber providing the unique ability to simultaneously vary
pressure and temperature whilst measuring the short-lived free radical
species OH, HO2 and NO3 (Glowacki et al., 2007a; Malkin, 2010;
Malkin et al., 2010). These features make HIRAC ideally suited to the study
of the kinetics and mechanisms of atmospherically relevant reactions and the
calibration, validation and development of atmospheric measurement
instrumentation.
HIRAC and FAGE instrumentation
HIRAC
Experiments were conducted in HIRAC, a stainless-steel chamber with a total
volume of 2.25 m3 and total internal surfaces of 13 m2 (S/V ∼ 5.8 m-1). The chamber could operate over a wide range
of pressures (10–1000 mbar), with multiple access ports used to connect
an array of instrumentation and monitoring equipment (pressure gauges,
thermocouples etc.). Further details on the construction can be found in
Glowacki et al. (2007a) and Malkin et al. (2010).
The photolysis lamps, housed in eight quartz tubes mounted radially inside
the reactive volume, were used to initiate photochemistry. The lamps were
interchangeable depending on the target molecules; lamps, with primary
emissions centred at 254 and 290 nm (GE Optica, GE55T8/HO and Philips,
TL40W/12 RS), were used for the alternative OH and HO2 calibration
methods respectively (Sects. 3.2 and 3.3). The output of the lamps was
temperature dependent outside of a narrow temperature range (∼ 35–39 ∘C), and so the housings were flushed with N2 to regulate the
temperature and remove photolabile species. A photolysis-lamp-induced
chamber temperature increase of ∼ 2 K was seen over the course
of a typical experiment (< 40 min), and was therefore considered
negligible compared to the temperature of the chamber on any given day (293 ± 5 K).
Investigations into radical gradients across the HIRAC chamber have been
conducted using direct FAGE measurements of OH produced from both photolytic
(methyl nitrite) and non-photolytic (O3+ trans-2-butene) sources using
an extended inlet (800 mm) to probe across the chamber diameter. No
significant OH radical gradient was observed until the FAGE sampling nozzle
was ∼ 200 mm from the wall and a maximum ∼ 15 % decrease (compared to the centre of the chamber) was seen when the
sampling inlet was flush with the chamber walls. Other than being close to the
walls, the lack of gradient in OH radicals from both photolytic and
non-photolytic sources provides direct evidence of the homogeneity of the
lamp radiation profile and efficacy of mixing in the chamber, whilst showing
that the standard FAGE inlet (280 mm, Sect. 2.2) samples well into the
homogeneous area.
Ozone was monitored using a UV photometric O3 analyser (Thermo Electron
Corporation 49C, detection limit (d.l.) = 1.0 ppbv at 60 s averaging). The
O3 analyser had been calibrated using a commercial ozone primary
standard (Thermo Electron Corporation 49i-PS), and an intercomparison with
the Fourier transform infrared (FTIR) spectroscopy within HIRAC was linear (Glowacki et al., 2007a). A
chemiluminescence NOx analyser (TEC 42C, d.l. = 50 pptv at 60 s
averaging) was used to determine that levels of NOx (= NO + NO2) were characteristically below the detection limit of the
apparatus.
A calibrated gas chromatography instrument with flame ionisation detector
(GC-FID, Agilent Technologies, 6890N) was used for the online detection of
reactants (Sect. 3.2) using an evacuated sampling loop into which gas from
the chamber was expanded. The GC was fitted with a CP-SIL-5 column (50 m,
0.32 mm, 5 µm) using He carrier gas and a constant oven temperature (40–75 ∘C
depending on the hydrocarbon being detected) and was able to
provide hydrocarbon measurements on a 2–6 min time resolution.
Supporting measurements of iso-butene and (CH3)3COOH were made via a
long-path FTIR absorption facility. The FTIR spectrometer (Bruker, IFS/66)
was coupled to a Chernin-type multipass cell (Glowacki et al., 2007b) and
spectral resolution was maintained at 1 cm-1 across all
experiments, using 32 co-added spectra for a 30 s time resolution.
Calibration experiments were conducted over a pressure range of 440–1000 mbar in an ultra-high-purity (UHP) 1:4 synthetic air mix of O2 (BOC,
zero-grade, > 99.999 %) and N2 (BOC, zero-grade,
> 99.998 %) to match the range of pressures from the pinhole
calibration method (Sect. 3.1). The UHP gases help maintain low H2O
vapour (< 10 ppm, verified by dew-point hygrometer measurement),
NOx (< 1 ppbv) and non-methane hydrocarbons (< 1 ppbv)
during experimental runs. Thorough mixing of reaction mixtures within HIRAC
was achieved in ≤ 70 s by four vibrationally damped, variable speed
circulation fans mounted in pairs at each end of the chamber. The chamber
was evacuated to ∼ 0.05 mbar for ∼ 60 min following each
experiment using a rotary-pump-backed roots blower (Leybold, trivac D40B and
ruvac WAU251) to ensure removal of all reactants/products. Known
concentrations of precursors were introduced to the chamber in the vapour
phase through a 0.97 L stainless-steel delivery vessel. A combined sampling
rate of ∼ 9 sLm from the chamber required a counter flow of
synthetic air maintaining the desired pressure and diluting the reactants
((4.5 ± 0.2) × 10-5 s-1). This was regulated using
two Brooks mass flow controllers (N2 and O2).
HO<inline-formula><mml:math display="inline"><mml:msub><mml:mi/><mml:mi mathvariant="normal">x</mml:mi></mml:msub></mml:math></inline-formula> detection instrument
Calibrations were conducted using both the University of Leeds aircraft- and
HIRAC-based FAGE instruments, brief operational details of which are shown
in Table 1. The two FAGE systems were very similar in design except for the
inlet length and pinhole size as highlighted in Table 1. The aircraft
instrument was used as described in Commane et al. (2010) to validate the
alternative HO2 calibration technique only. The HIRAC-based FAGE
instrument has also been described in the literature by Glowacki et al. (2007a), and hence only modifications since publication will be discussed
here.
FAGE instruments used and their respective inlet designs, laser
systems and calibration methods used.
FAGE instrument
Inlet design
Laser
Calibrations conducted
Aircraft
Pinhole diameter
0.6 mm
Photonics Industries (5 kHz PRF) Nd:YAG (DS-532-10) pumped Ti:Sa (TU-UV-308)
H2O vapour photolysis, HCHO photolysis
Length
420 mm
HIRAC
Pinhole diameter
1.0 mm
Litron, NANO-TRL-250, (200 Hz PRF) Nd:YAG pumped dye laser (Lambda Physik, LPD3000)
H2O vapour photolysis, HCHO photolysis, HC decay
Length
280 mm
Schematic showing a side-on vertical cross section of the HIRAC
FAGE OH fluorescence cell. The OH scavenger (iso-butane) was introduced
∼ 40 mm from the inlet pinhole through an 1/8 in. internal-diameter
stainless-steel tube mounted in between the OH and HO2 cells
(out of frame). The tube ran flush to the cell wall to reduce possible
scattering of laser light, and the tip was angled slightly towards the centre
of the main gas flow to improve mixing.
![]()
Figure 1 shows the cross-sectional schematic of the HIRAC FAGE instrument.
Under typical operating conditions, air was sampled at ∼ 6 sLm through
a 1.0 mm diameter pinhole nozzle and passed down the inlet (length 280 mm,
50 mm diameter) into the OH detection axis maintained at low pressure (1.8–3.85 mbar) using a high capacity rotary-backed roots blower pumping
system (Leybold, trivac D40B and ruvac WAU251). Using the same pump set, the
aircraft instrument was operated with a 420 mm long inlet and a 0.6 mm
pinhole. The long inlet was used to draw a sample away from the chamber
walls where radical losses become significant (see Sect. 2.1). Both
instruments were coupled to the HIRAC chamber using custom-made ISO-K160
flanges, ensuring the pinhole, in both cases, was kept ∼ 225 mm from the chamber walls.
Concentrations of HO2 were measured simultaneously in a second
detection axis ∼ 300 mm downstream of the OH detection axis.
High-purity NO (BOC, N2.5 nitric oxide) was added ∼ 20 mm
before the HO2 detection axis into the centre of the FAGE cell in the
direction of gas flow through 1/8 in. stainless-steel tubing at a rate of 5 sccm (Brooks 5850S), converting HO2 to OH.
Recently published material on the conversion of certain RO2 radicals
to OH upon reaction with NO in FAGE detections cells (Fuchs et al., 2011;
Whalley et al., 2013) has shown a significant enhancement of the HO2
signal in the presence of RO2 derived from certain hydrocarbons. These
effects have been thoroughly studied using a range of different hydrocarbons
for the HIRAC FAGE apparatus and will be the subject of a further
publication. The potential interferences associated with HO2
measurements in the presence of certain hydrocarbons due to the presence of
β-hydroxyperoxy radicals do not apply to either HO2 calibration
method. In addition, any interference from RO2 radicals produced during
the alternative calibration methods was experimentally demonstrated to be
negligible under the conditions of these experiments (Winiberg, 2014).
Experiments with the HIRAC FAGE instrument used a new medium pulse
repetition frequency (PRF) laser light source (= 200 Hz), with a different
light delivery method to the detection cells, compared to that described by
Glowacki et al. (2007a). The previously used JDSU Nd:YAG pumped Sirah Cobra
Stretch system (PRF = 5 kHz) focussed the frequency-doubled 308 nm output
into fibre optic cables (10 m, Oz Optics), which were then attached directly
to the FAGE cell arms via collimators (Oz Optics). Using the new Litron Nd:YAG
(NANO-TRL-50-250) pumped Lambda Physik (LPD3000) dye laser system (PRF = 200 Hz), the high laser pulse energies were found to burn the ends of the
fibre optic cables, and hence direct light delivery was applied using a
combination of mirrors, lenses and irises to direct and shape the
beam to the OH and HO2 detection regions, as shown in the top-down
schematic of the modified HIRAC FAGE instrument displayed in Fig. 2.
Top-down schematic of the FAGE instrument showing the laser beam
path (blue line) through the OH and HO2 detection cells, and the
reference cell using the Litron/LPD3000, 200 Hz PRF laser source. Q –
quartz flat; M – mirror; I – iris; and L – lens. The FAGE inlet is
extended past the edge of the mounting table for insertion into the HIRAC
chamber. The calibrated photodiode was used to normalise the fluorescence
signals to fluctuations in laser power.
![]()
The UV light exiting the dye laser was split with a quartz flat (Fig. 2,
Q1) to direct ∼ 5 % of the laser light towards the reference cell
(where OH was generated continuously from a hot wire filament in water-saturated
air), which enabled precise tuning of the laser wavelength to the
maxima of the OH Q1 (2) branch (within 98 %). The remaining light was
aligned through the OH and HO2 cells sequentially using a series of
308 nm centred turning optics (M1–M4, CVI Laser Optics, Melles Griot). A lens
was used (L1, f=100 mm) in conjunction with an iris (I2), to help
transmit the laser beam through both detection cells, avoiding collisions
with any internal surfaces. Fluctuations in laser power were accounted for
using a linear-response, UV-sensitive photodiode (UDT-555UV, Laser Components
UK) at the exit arm of the HO2 detection axis to normalise the LIF
signal. Both laser systems provided 5–7 and 2–3 mW of 308 nm
light to the OH and HO2 detection axes respectively.
The OH fluorescence was collected orthogonal to the gas flow onto
electronically gated Channeltron PhotoMultiplier tubes (CPM, Perkin Elmer,
C943P) via a series of imaging lenses and a narrow-bandpass filter (Barr
Associates, 308.8 ± 5.0 nm). A spherical concave back reflector was
positioned underneath the cell, opposite the detection optics, to optimise
light collection onto the CPM. To avoid detector saturation, the CPM was
gated (i.e. switched off) for the duration of the laser pulse using a
modified gating unit based on the original design by Creasey et al. (1997a).
Signals from the CPM were analysed using photon-counting cards (Becker and
Hickl PMS-400A).
A new OH scavenger system was installed to help discriminate between OH
sampled from the chamber and laser-generated OH in the fluorescence cells due
to the higher pulse energies associated with the 200 Hz PRF laser system
(1 × 1014 compared to 5×1012 photons pulse-1 cm-2 at 5 kHz for laser
power = 8 mW). A mixture of iso-butane (20 % in N2) was injected
∼ 40 mm inside the inlet pinhole into the central flow (Fig. 1),
through a 1/8 in. internal-diameter stainless-steel pipe at a rate of
∼ 20 sccm, reacting with the sampled OH before reaching the detection
axis. The laser-generated OH was probed within the same laser pulse (12 ns)
and hence was not suppressed by the scavenger injection. Multiple photolysis
of the same gas sample was avoided as the residence time in the laser pulse
cross section (∼ 0.5 cm2) was calculated at ∼ 0.4 ms,
compared to a laser pulse every 5 ms at 200 Hz PRF (assuming plug flow at a
6 sLm ambient sampling rate). Neither a pressure increase nor attenuation of
UV light was detected during the scavenger injection process at this flow
rate and dilution.
Calibration procedures
Conventional H<inline-formula><mml:math display="inline"><mml:msub><mml:mi/><mml:mn mathvariant="normal">2</mml:mn></mml:msub></mml:math></inline-formula>O vapour photolysis calibration
The requisite equation for calibration of FAGE by water vapour photolysis
was given as
OH=HO2=H2OσH2O, 184.9nmΦOHF184.9nmΔt
and the principles were outlined above in Sect. 1. A schematic diagram of
the H2O vapour photolysis calibration source is presented in Fig. 3,
consisting of a square cross-section flow tube (12.7 × 12.7 × 300 mm) through which 40 sLm of humidified air (BOC, BTCA 178)
was passed, resulting in a turbulent flow regime (Reynolds number ≥ 4000). The air was humidified by passing a fraction of the total air flow
through a deionised water bubbler system, and [H2O] was measured using a
dew-point hygrometer (CR4, Buck Research Instrument) prior to the flow tube.
The collimated 184.9 nm output of a mercury Pen-Ray lamp (LOT-Oriel, Hg-Ar)
was introduced to the end of the main flow tube, photolysing H2O vapour
(Reactions R2–R3). The gas output from the flow tube was directed towards the FAGE
sampling inlet, where the overfill of the FAGE sample volume from the flow
tube stopped the impingement of ambient air. A range of HOx
concentrations were produced by changing both the H2O vapour
concentration and the mercury lamp photon flux.
Schematic cross section of the H2O vapour photolysis
calibration source used in the calibration of the two FAGE instruments
(inlet for HIRAC FAGE instrument shown here, for example). The
[H2O]vapour was measured prior to entering the square
cross-section flow tube, and the concentration was controlled through a series of
three taps around the bubbler. The Hg Pen-Ray lamp was housed in a second
section of the wand, and the output was collimated through a Suprasil window
using a honeycomb arrangement of ø = 1 mm aluminium tubes. The lamp was
continuously flushed with N2 to remove potential absorbers and
photolabile species, and to help regulate temperature.
![]()
The flux of 184.9 nm light, F184.9 nm, was varied by altering the Hg
lamp supply current and was dependent on the specific mercury lamp employed
along with the lamp temperature and orientation (Hofzumahaus et al., 1997;
Creasey et al., 2000; Dusanter et al., 2008). To this end, determinations of
the flux from the specific mercury lamp used in the calibrations described
in this work were made in situ for lamp supply currents between 0.2 and 3.0 mA
using the N2O actinometry method described in detail in a number of
publications (Edwards et al., 2003; Heard and Pilling, 2003; Faloona et al.,
2004; Glowacki et al., 2007a; Whalley et al., 2007). The exposure time of
the air to the 184.9 nm light, Δt, was calculated as a function of
the known velocity of the air and the cross section of the photolysis
region.
Various cell conditions and their effect on the sensitivity to OH and
HO2 have been reported in the literature (Faloona et al., 2004;
Martinez et al., 2010; Regelin et al., 2013). Here, instrument sensitivity
as a function of internal cell pressure has been determined for the HIRAC
FAGE instrument using the 200 Hz PRF laser source only (Table 1). Different
internal cell pressures (1.8–3.8 mbar) were achieved by changing the
diameter of the FAGE inlet pinhole between 0.5 and 1.0 mm. For the aircraft
FAGE instrument, inlet pinhole diameters between 0.3 and 0.6 mm were used
giving internal cell pressures between 1.4 and 2.5 mbar.
Hydrocarbon decay method – OH calibration
Hydrocarbons (0.5–2.0 × 1013 molecule cm-3) and the
OH precursor, tert-Butyl hydroperoxide (TBHP, Sigma Aldrich ∼ 40 % in H2O, 2.0 × 1013 molecule cm-3) were
introduced to the chamber before the lamps were switched on, initiating the
decay experiment. OH was produced directly from the photolysis of TBHP at
λ= 254 nm and is, as far as we are aware, the first chamber
experiment to use TBHP photolysis as a source of NOx-free OH. Upon
illumination of the chamber, rapid photolysis led to an instantaneous peak
[OH] ∼ 107 molecule cm-3 before OH decayed away over
∼ 30 min as the TBHP was removed by photolysis, whilst OH was removed through reaction with TBHP
(kOH(296 K) = (3.58 ± 0.54) × 10-12 cm3 molecule-1 s-1;
Baasandorj et al., 2010) and the selected hydrocarbon. The alternative OH calibrations
presented here were conducted for the HIRAC-based FAGE instrument operating
at 200 Hz PRF only.
Cyclohexane (> 99 %, Fisher Scientific), n-pentane (> 99 %, Fisher Scientific) and iso-butene (99 %, Sigma Aldrich) were
employed as the hydrocarbons in this study due to their sufficiently fast
and well-known rates of reaction with OH to provide a quantifiable decay
compared to chamber dilution. The rate coefficient for OH with iso-butene has
been evaluated by IUPAC as kOH (298 K) = (51 ± 12) × 10-12 cm3 molecule-1 s-1
(IUPAC, 2007), and rate
coefficients for the reaction of OH with cyclohexane and n-pentane have been
reviewed by Calvert et al. (2008) as kOH (298 K) = (6.97 ± 1.39) and (3.96 ± 0.76) × 10-12 cm3 molecule-1 s-1
respectively (all quoted to ±2σ). Whilst alkanes are
known to have a pressure-independent rate coefficient for OH reactions, the
reactions of OH with alkenes occur predominantly by addition, a process
which is pressure dependent, with the rate coefficient increasing with
pressure up to the high-pressure limit where the addition of OH is the rate-determining
step (Pilling and Seakins, 1995). A study by Atkinson and Pitts (1975) into the reaction of various small-chain alkenes showed no pressure
dependence for propene over 33–133 mbar of argon; therefore the reaction
of OH with the larger iso-butene molecule is presumed to be pressure
independent above 133 mbar (Atkinson, 1986; Atkinson et al., 2004). To confirm this, a
relative rate study in air was conducted using isoprene as a reference
(kOH (298 K) = (1.00 ± 0.14) × 10-10 cm3 molecule-1 s-1, IUPAC, 2007). Both direct
and relative rate studies have shown that the reaction of isoprene and OH is
at the high-pressure limit above 100 Torr (Campuzano-Jost et al., 2004; Park
et al., 2004; Singh and Li, 2007). Figure 4 shows that there is no
significant pressure dependence in kOH for OH + iso-butene over the 250–1000 mbar pressure range within the uncertainty of the experiment
(∼ 25 %, ± 2σ) and that the measured rate
coefficient, kOH (298 K = (4.87 ± 0.83) × 10-11 cm3 molecule-1 s-1), is in good agreement
with the literature values ((5.07 ± 0.51) × 10-11 cm3 molecule-1 s-1 for
IUPAC; Atkinson, 2003; IUPAC, 2007).
Rate constant, kOH, for iso-butene + OH over the 250–1000 mbar pressure range measured relative to an isoprene reference in the
HIRAC chamber. An empirical linear least-squares fit to the data is shown to
emphasise lack of observed pressure dependence in the measured rate
constant. Error bars represent the standard error (±1σ) in
the associated relative rate determination of kOH, and linear regression
is weighted to account for this.
![]()
The hydrocarbon decay method relies on the loss of hydrocarbon being solely
due to reaction with OH, and hence the effects of O3 and NO3 as
reagents must be considered as both are important in the oxidation of
alkenes (Atkinson, 1994). Before photolysis, O3 and NOx were
measured to be around the instrumental detection limits (0.5 and 0.050 ppb
at 60 s averaging respectively) using commercial analysers (details given in
Sect. 2.1). Upon photolysis a slow increase in O3 and NO2 was
observed, to a maximum of ∼ 40 and ∼ 20 ppbv
respectively. The [NO3] upper limit was estimated at ∼ 0.32 pptv using a simple steady-state approximation, where NO3
production was controlled purely by O3+ NO2→ NO3
(Atkinson et al., 2004) and loss by photolysis
(j(NO2) = 1.93 ± 0.10; Glowacki et al., 2007a). Under these
conditions it was estimated that > 98 % of the loss of
iso-butene would be due to OH and not O3 or NO3 where
kO3 = (1.13 ± 0.33) × 10-17 cm3 molecule-1 s-1 and
kNO3= (3.4 ± 1.0) × 10-13 cm3 molecule-1 s-1 (Calvert et al.,
2000).
Formaldehyde photolysis – HO<inline-formula><mml:math display="inline"><mml:msub><mml:mi/><mml:mn mathvariant="bold">2</mml:mn></mml:msub></mml:math></inline-formula> calibration
Formaldehyde was produced by direct heating of paraformaldehyde powder in a
glass finger (Sigma Aldrich, 99 %) and was introduced in a flow of
nitrogen into the chamber at concentrations ∼2 × 1013 molecule cm-3 (determined manometrically). The chamber was
irradiated (lamps: Philips TL40W/12 RS), resulting in an almost instantaneous
HO2 signal. Once an approximately steady-state HO2 concentration
was achieved, the photolysis lamps were turned off and the decay of HO2
was monitored by FAGE for ∼ 120 s until near-background signals
levels were reached. The measurement of HO2 decays was repeated up to
five times before the laser wavelength was scanned to the offline position.
Therefore five individual CHO2 determinations
could be achieved from one chamber fill, with the limiting factor being the
increased complexity of the reaction mixture after repeated photolysis
cycles. After five decays, the analysis often exhibited evidence of
secondary chemistry starting to distort the HO2 signal profiles,
showing non-linearity in second-order plots. The absence of OH in these
experiments was confirmed by simultaneous measurement of OH in the OH
fluorescence cell, giving signals below the detection limit (1.6 × 106 molecule cm-3 for 60 s averaging for the 200 Hz PRF laser
system).
Formaldehyde concentrations were kept low (< 3 × 1013 molecule cm-3) to avoid removal of HO2 via reaction with HCHO,
ensuring that the loss of HO2 occurs predominately via self-reaction
and wall loss (Sect. 4.2). The HO2 calibrations were conducted for
the HIRAC-based FAGE instrument operating at 200 Hz PRF and the aircraft-based
FAGE instrument operating at 5 kHz PRF. The chamber mixing fans were
used for the majority of calibration data sets discussed here,
representative of a typical experimental homogeneous gas mixture. A series
of experiments were conducted without the mixing fans to probe the HO2
recombination and wall-loss kinetics using the aircraft-based FAGE
instrument, and these are discussed in greater detail in Sects. 4.2 and
5.3.
Decay of iso-butene as a function of time through reaction with OH
in HIRAC (750 mbar, 294 K), measured using GC-FID on a 2 min time
resolution. The data are fitted with a first-order exponential decay (purely
empirical) to allow calculation of [HC] on the same timescale as the 60 s
averaged FAGE data. Time = 0 s indicates photolysis lamp turn-on time and
uncertainties are quoted to ±1σ. Error bars are
representative of the precision in the GC-FID (∼ 2 %) and
FTIR (∼ 3 %) measurements to 1σ.
![]()
Data analysis
Hydrocarbon decay
Figure 5 shows the hydrocarbon decay for iso-butene at 750 mbar and 294 K
measured by GC-FID and FTIR. Using the Guggenheim method (Guggenheim, 1926;
Bloss et al., 2004) the pseudo-first-order rate coefficient (k′) for
the hydrocarbon removal was calculated using Eq. (3):
k′=lnHC1/HC2t2-t1,
where [HC]1 and [HC]2 are the concentrations of the hydrocarbon at
time t1 and t2 respectively. The mean [OH] between t1 and
t2 was calculated using Eq. ():
[OH]=k′-kDilkOH,
where kDil is the dilution rate of the measured [HC] due to continuous
sampling from instrumentation (e.g. FAGE). Bloss et al. (2004) found the
Guggenheim method to be most effective when smoothing the inferred [OH] over
five [HC] measurements (i.e. consider 10 measurements taken at times
t1-t10; [OH] at t5 would take [HC]1 and [HC]5, [OH] at
t6 would take [HC]2 and [HC]6 etc.). Due to the short experiment time
(20–30 min) and the 2–6 min time resolution on the GC measurements,
this smoothing was not possible. For iso-butene, FTIR measurements were taken
every 30 s, and these were typically found to be in excellent agreement with
the GC-FID-measured HC decays, as shown in Fig. 5. However, measurement of
small changes in the [HC], due to low steady-state [OH] in the chamber
(∼ 5 × 106 molecule cm-3), led to large
point-to-point variation in the inferred [OH], even after the smoothing was
applied. A solution was found by fitting the hydrocarbon decay data with an
empirical exponential function of the form y=A × e(-x/t1)+y0 as shown in Fig. 5, which allowed the accurate calculation of
[HC] at the same time resolution as the FAGE instrument (20 s averaged). A
negligible difference between inferred [OH] determined using the FTIR or
GC-FID data was observed, and hence only GC-FID-measured hydrocarbon decays
were used for direct comparison with n-pentane and cyclohexane.
Comparison of [OH] traces measured using the HIRAC FAGE instrument
(200 Hz PRF) during the photo-oxidation of n-pentane at 1000 mbar and 293 K
before (a) and after (b), correcting for laser-generated OH due to TBHP
photolysis in the OH fluorescence cell. The uncorrected and corrected FAGE
signal was converted to [OH] using COH=3.6 × 10-8 counts cm3 s-1 molecule-1 mW-1 determined using the
conventional calibration method for comparison with GC-FID data. The TBHP
(3.2 × 1013 molecule cm-3) and n-pentane (2.1 × 1013 molecule cm-3) were introduced into the chamber at t≈-500 s, and the photolysis lamps were switched on at t=0 s. The [OH]
inferred from the HC decay method is also displayed in (b). Dashed
line at y=0 given for clarity.
![]()
Displayed in Fig. 6 is a typical [OH] profile for the photo-oxidation of
n-pentane (2.1 × 1013 molecule cm-3) in HIRAC at 1000 mbar and 293 K where photolysis of TBHP was used to produce ∼ 1.3 × 107 molecule cm-3
OH at t= 0. The OH was
measured directly using the HIRAC FAGE instrument with the Litron Nd:YAG
pumped dye laser light source, operating at 200 Hz PRF. Upon introduction of
TBHP (3.2 × 1013 molecule cm-3) to the dark chamber at
t≈-500 s, an OH signal equivalent to ∼ 2.5 × 106 molecule cm-3 was observed, and was typically
< 25 % of the total detected OH signal following lamp photolysis.
The measured un-normalised OH fluorescence signal was observed to increase
quadratically with laser power, suggesting a two-photon photolysis-probe
process from the OH probe laser at 308 nm, as described by Reactions (R5)–(R7).
TBHP+hv→OH+products,OH+hv→OH(A),OH(A)→OH(X)+hv(LIF).
This phenomenon was not observed during a brief test of the HIRAC FAGE
instrument with a 5 kHz PRF laser system (JDSU Nd:YAG pumped Sirah Cobra
Stretch dye laser, as in Glowacki et al., 2007a, and Malkin et al., 2010),
most likely due to much lower laser pulse energies for the 5 kHz system (1.6 µJ pulse-1
compared to 40 µJ pulse-1 at 200 Hz PRF). Using the scavenger injection system (Sect. 2.2), the decay of the
TBHP could be accurately described (compared to simultaneous FTIR
measurements) characterising the interference signal. At a time defined by
the user, the iso-butane scavenger (20 % in N2) was injected into the
FAGE cell for ∼ 90 s at ∼ 20 sccm. Before the
chamber photolysis lamps were initiated, the OH interference signal was
measured with the scavenger off and on, and the difference in signal was
observed to be negligible (within the uncertainty of the measurement). The
OH interference profile during the hydrocarbon decay was characterised and
accounted for using 3–4 scavenger injections per experiment. An empirical
fit to the averaged signals was used to correct the measured OH signal from
TBHP laser photolysis over time, shown here in Fig. 6b compared to the
inferred [OH] from the GC-FID. The type of fitting parameter (e.g. linear or
exponential) was judged depending on the quality of data.
The calibration procedure was completed by plotting the OH signals,
normalised for laser power, measured by FAGE as a function of the calculated
OH concentrations from the hydrocarbon decays producing a calibration plot
with COH, in units of counts cm3 s-1 mW-1
molecule-1, as the gradient. A typical calibration plot is shown in
Fig. 7 – produced using the decay of cyclohexane at 1000 mbar chamber
pressure (see caption for detailed operating conditions). Uncertainties in
COH are quoted to ±2σ, and error bars represent the
standard error, and hence precision, of the measured SOH to ±1σ. Error bars were kept at ±1σ as this represented the
precision used in the analysis procedure.
Calibration from the hydrocarbon decay method for cyclohexane at
1000 mbar and 293 K chamber pressure using the HIRAC FAGE instrument with
the 200 Hz PRF laser system; inlet pressure = (3.81 ± 0.02) mbar;
laser power = (7.0 ± 0.5) mW; [H2O]vapour < 10 ppmv. Extrapolated calibration from the H2O photolysis calibration
technique for inlet pressure = (3.79 ± 0.02) mbar, laser power =
(6.0 ± 0.5) mW, [H2O]vapour= (3900 ± 20) ppmv
and [OH] = (0.5–1.5) × 109 molecule cm-3. Both
fits are weighted to errors in the x and y axes and error bars are
representative of the standard error in the measurement to ±2σ. Uncertainties quoted for the slope and intercept represent the precision
of the calibration processes to ±2σ.
![]()
Formaldehyde photolysis
Calibration of the HO2 detection cell required only the generation of
HO2 radicals in the HIRAC chamber, and a time-dependent measurement of
their subsequent recombination using the FAGE instrument once the photolysis
lamps were extinguished. Upon photolysis in air (lamps: Philips TL40W/12
RS), HCHO produced H + HCO and H2+ CO (Reaction R9) in approximately a
60:40 ratio (Reactions R8, R9). Under the conditions in HIRAC, HCO reacted with
O2 to give HO2+ CO (Reaction R10) and the H atom produced in Reaction (R8) reacted
with O2 to give HO2 (Reaction R11). The loss of HO2 was characterised
by the competing bimolecular and termolecular self-reactions (Reactions R12 + R13)
and a first-order wall-loss parameter (Reaction R14):
HCHO+hv→H+HCOorH2+COHCO+O2→HO2+COH+O2+M→HO2+MHO2+HO2→H2O2+O2HO2+HO2+M→H2O2+O2+MHO2→Loss.
Therefore the rate of loss of HO2 is given by:
d[HO2]dt=-klossHO2++kHO2+HO2[HO2]2,
where kHO2+HO2is the HO2 recombination rate coefficient;
the sum of the pressure-independent (Reaction R12) and pressure-dependent (Reaction R13) rate
coefficients as determined by IUPAC (2007). Solving analytically for
[HO2]t at a given time, t, integration of Eq. (5) becomes
1[HO2]t=1[HO2]0+2⋅kHO2+HO2kloss×eklosst-2⋅kHO2+HO2kloss.
The [HO2] in Eq. (6) is unknown but is related to the
normalised HO2 signals measured by FAGE,
SHO2, and the instrument sensitivity to
HO2, CHO2, and therefore
SHO2t=1SHO20+2⋅kHO2+HO2kloss⋅CHO2×eklosst-2⋅kHO2+HO2kloss⋅CHO2-1,
where (SHO2)t and
(SHO2)0 are the HO2 signal at time t
and t=0 respectively.
The measured decay of SHO2 using FAGE and the
fit described by Eq. (7) are displayed in Fig. 8a for a typical
experiment (aircraft FAGE instrument (5 kHz PRF), 1000 mbar, 298 K,
< 10 ppm [H2O], mixing fans on). Both kloss and
CHO2 were determined by data fitting the
SHO2 decay using Eq. (7) with a
Levenberg–Marquardt non-linear least-squares algorithm, fixing the initial
signal and kHO2+HO2. The first ∼ 100 s of data were
used, ensuring analysis after an almost complete decay of
SHO2. Fitting was improved by the inclusion of
upper and lower bounds of ±10 % for the
(SHO2)0 into the fitting routine, which
accounted for the uncertainty in the determination of
(SHO2)0 (see Sect. 5.4.3).
Normalised SHO2 decay for the HCHO
photolysis calibration method at 1000 mbar chamber pressure using the
aircraft FAGE instrument (5 kHz PRF laser system) conducted with the HIRAC
chamber mixing fans on (a) and off (b); inlet pressure = (2.53 ± 0.02) mbar; laser power = (8.25 ± 0.25) mW. Data in (a) were fitted
with Eq. (7) to give CHO2 where A=(SHO2)0, kb= kHO2+HO2,
c =CHO2, ka=kloss and z= offset, with uncertainties quoted to ±1σ. Parameters without
quoted error were fixed.
![]()
For the experimental 350–1000 mbar pressure range at 0 % H2O
vapour, kHO2+HO2 was determined between
(2.00–2.85) × 10-12 cm3 molecule-1 s-1,
according to the recommendation given by IUPAC (2007). A calibration was
conducted at [H2O]vap=7500 ppmv, to validate the calibration
method at high water vapour concentrations, representative of the
conventional H2O vapour photolysis method. The
kHO2+HO2
therefore included a correction for the HO2-H2O vapour chaperone
effect (Stone and Rowley, 2005) in accordance with the IUPAC recommendation
(Atkinson et al., 2004). The wall-loss rate, kloss, was dependent on
daily chamber conditions and was therefore determined as part of the fitting
procedure along with CHO2, typically between
0.032 and 0.073 s-1 with an uncertainty of ±10 % (2σ).
Variations in the wall-loss rates have implications for the uncertainty in
CHO2 derivation (Sect. 5.4).
Results and discussion
All results presented here were taken using the HIRAC FAGE instrument using
Litron Nd:YAG pumped dye laser light source operating at 200 Hz PRF, unless
otherwise stated. Tabulated data from the alternative calibration methods
are displayed in the Supplement (Tables S1 and S2). All
uncertainties displayed are quoted to ±2σ and all regressions
shown are empirical, unless otherwise stated.
Conventional H<inline-formula><mml:math display="inline"><mml:msub><mml:mi/><mml:mn mathvariant="bold">2</mml:mn></mml:msub></mml:math></inline-formula>O vapour photolysis calibration
C<inline-formula><mml:math display="inline"><mml:msub><mml:mi/><mml:mrow class="chem"><mml:mi mathvariant="normal">OH</mml:mi></mml:mrow></mml:msub></mml:math></inline-formula> and <inline-formula><mml:math display="inline"><mml:mrow><mml:msub><mml:mi>C</mml:mi><mml:mrow class="chem"><mml:msub><mml:mi mathvariant="normal">HO</mml:mi><mml:mn mathvariant="normal">2</mml:mn></mml:msub></mml:mrow></mml:msub></mml:mrow></mml:math></inline-formula> as a function of internal cell
pressure
The FAGE instrument sensitivity to OH (Fig. 9, HIRAC FAGE only) and
HO2 (Fig. 10, top) was determined as a function of pressure using the
H2O vapour photolysis calibration method over the inlet pressure range
of 1.8–3.8 mbar. Figures 9 and 10 display the H2O
calibration data compared to those from the respective alternative
calibration methods, the results for which are discussed in Sects. 5.2 and
5.3; error bars in both figures are representative of the total uncertainty
in the calibration (Sect. 5.4 for details). Constant laser power and
[H2O] were maintained throughout the calibration process (8 ± 1 mW and 4500 ± 600 ppmv respectively).
HIRAC FAGE instrument sensitivity to OH, COH, as a function
of internal detection cell pressure as determined by the H2O vapour
photolysis and HC decay calibration techniques using the Litron Nd:YAG
pumped dye laser operating at 200 Hz PRF. All calibrations were conducted at
laser powers between 6.0 and 9.5 mW. Different internal cell pressures (1.8–3.8 mbar) were achieved by changing the diameter of the FAGE inlet
pinhole between 0.5 and 1.0 mm. Conventional calibrations were conducted at
constant [H2O]vap (4500 ± 600 ppmv), whereas alternative
calibrations were conducted in near-dry conditions (< 15 ppmv).
HIRAC chamber pressures between 440 and 1000 mbar were used to induce
internal cell pressures between 2.1 and 3.9 mbar. Error bars indicate the
total uncertainty to ± 1σ.
![]()
The linear regressions were used to describe the sensitivity as a function
of fluorescence cell pressure for experiments conducted in HIRAC, and are a
valid description of the data inside the 1.8–3.8 mbar pressure range
only. The COH and CHO2 data sets shown here
were not conducted at the same time, but 6 months apart. This was due to the
chronological order of the development of the alternative calibration
techniques, during which time the FAGE pump set was serviced, increasing the
pumping capacity and generally lowering the internal cell pressures for each
pinhole in the COH determination.
The fit displayed a greater increase in CHO2 as
a function of pressure compared to COH, where ΔCOH= (17 ± 11) %
and ΔCHO2 = (31.6 ± 4.4) % increase between 1.3 and 3.8 mbar. Altering the pinhole
diameter could change the flow dynamics inside the instrument, reducing NO
mixing efficiency, and therefore HO2 conversion efficiency, before the
HO2 cell. The decrease in CHO2 at lower
pressure has been reproduced in a more recent calibration of the HO2
cell using the 5 kHz PRF laser source (slope = (5.14 ± 0.46) × 10-9 counts cm3 molecule-1 s-1 mW-1 mbar-1), suggesting the process was not affected by changes in laser
pulse energy.
The experimental parameters controlling the instrument sensitivity,
COH, which are dependent upon pressure, are the OH concentration in the
laser-excitation region, [OH]cell; the fluorescence quantum yield
following laser excitation to the OH A2Σ+ (v′=0)
excited state, φfl; and the fraction of the fluorescence decay
which falls within the integrating gate of the photon counter, fgate
(Creasey et al., 1997b; Faloona et al., 2004). The OH concentration in the
cell held at total density [M] and the fluorescence quantum yield are given
by Eqs. (8) and (9):
[OH]cell=χ[M],φfl=AA+kq[M],
where χ is the mixing ratio of OH impinging at the pinhole (assuming
no losses at the pinhole), A is the inverse of the radiative lifetime of OH
and kq is the rate coefficient for quenching of the excited
A2Σ+ (v′=0) (averaged appropriately over all
quenching species). Assuming that fgate=1, then the overall
pressure-dependent term for instrument sensitivity to OH can be described as the
product of Eqs. (8) and (9), shown here in Eq. (10):
φχOHcell×Φfl=MAA+kq[M].
When [M]→0, the product becomes χ[M], and COH is directly
proportional to pressure ([M]). At higher pressures when
kq[M]≫A (at 18 mbar the ratio is ∼ 10), the
product becomes ∼χA/kq and COH is independent of [M],
and thus depends only on the mixing ratio of OH.
However, FAGE is an on-resonance technique, and therefore it is not possible
to achieve the limit fgate=1, because it is necessary to gate off
the CPM during the laser pulse in order to avoid saturation of the detector.
Thus, in these experiments fgate was always < 1, and depended
non-linearly on pressure because the photon-counting gate remained the same
whilst the fraction of the total fluorescence collected within this gate
(fgate) changed as a result of changes in the total fluorescence
lifetime of the excited-state OH radicals. Thus the effective area of
integration under the fluorescence decay curve reduced with increasing
pressure such that fgate reduced non-linearly as pressure increased. For
the conditions used in these experiments fgate=0.79–0.63 (between
1.3 and 3.8 mbar). Hence the observation that COH increased linearly
over pressures between 1.3 and 3.8 mbar in this study is consistent with the
expected behaviour based purely on the balance between OH number density and
the total fluorescence collected.
FAGE instrument sensitivity to HO2,
CHO2, as a function of internal detection cell
pressure as determined by the H2O vapour and HCHO photolysis
calibration techniques using the HIRAC FAGE instrument operating at 200 Hz
PRF (a) and the aircraft FAGE instrument operating at 5 kHz PRF (b).
Conventional calibrations were conducted at constant [H2O]vap ((a) 4500 ± 600 ppmv, (b) 6000 ± 600 ppmv), whereas alternative
calibrations were conducted under low [H2O]vap (< 15 ppmv). HIRAC chamber pressures of 440–1000 mbar were used to induce
internal cell pressures of (a) 1.8–3.8 mbar (pinhole diameter 0.5–1.0 mbar) and (b) 1.42–2.48 mbar (pinhole diameter 0.3–0.6 mbar).
Error bars indicate the total uncertainty to ±1σ.
![]()
The inherent complexity that results from the multiple factors which control
the sensitivity of FAGE instruments, and which also change with a variety of
conditions (i.e. pressure, [H2O], laser power) and instrumental factors
(e.g. time take for CPM to reach maximum gain), require that FAGE instruments
are frequently and carefully calibrated.
Additional investigations into the FAGE instrument sensitivity to OH as a
function of [H2O]vap and laser power are discussed in detail in
the Supplement. The change in instrument sensitivity over
[H2O] between 200 and 4500 ppmv was observed to be within the
uncertainty of the calibration (35 % at 2σ) and was therefore
considered negligible. For this reason no correction for sensitivity to
[H2O] was applied to the date taken in the alternative calibration
method where [H2O] < 10 ppmv.
Hydrocarbon decay calibration
Figure 7 shows a direct comparison of analysed data from the decay of
cyclohexane and H2O vapour calibration method at ∼ 3.80 mbar internal cell pressure (equivalent to 1000 mbar in HIRAC) using the 1.0 mm
inlet pinhole and ∼ 7 mW laser power. The COH was
determined as (2.13 ± 0.52) × 10-8 counts s-1 molecule-1 cm3 mW-1,
within error of the traditional H2O vapour photolysis calibration
(2σ) at the same pressure ((2.62 ± 0.91) × 10-8 counts s-1 molecule-1 cm3 mW-1).
Error bars are representative of the total uncertainty at ± 1σ. Additional example calibration plots for each hydrocarbon studied are
included in the Supplement. Displayed in Fig. 9 is
COH as a function of internal cell pressure using the HC decay
calibration method, determined for iso-butene, cyclohexane and n-pentane. The HC
decay calibration method was observed to be in agreement with the H2O
vapour photolysis calibration. The average of the ratio of calibration
factors (conventional : alternative) was calculated for each alternative
calibration point across the entire pressure range,
COH(conv)/COH(alt)=1.19 ± 0.26, where COH(conv) was
determined from the fit to the H2O photolysis data.
A large variability in the COH determined using the iso-butene decay was
observed, with larger uncertainties associated with this calibration
compared to cyclohexane and n-pentane, and the reason for this remains
unclear. On average, the measured OH signals were closer to the detection
limit of the FAGE instrument when using iso-butene. Initial concentrations of
each of the hydrocarbons were 2.5 × 1013 molecule cm-3,
and hence a lower OH steady-state concentration is expected when
iso-butene is present as the kOH is an order of magnitude higher than
those for n-pentane and cyclohexane. As SOH approaches 0 counts s-1 mW-1, the SOH measurement becomes increasingly imprecise, and thus
the uncertainty in the fitting of the calibration plot increases.
A general under-prediction of COH, compared to the H2O vapour
photolysis method, was observed when calculated using the decay of
cyclohexane, COH(conv)/COH(Chex) = 1.52 ± 0.44. The exact
reason is unknown. Evaluation of the HC decay data with the kOH adjusted
at the upper limit of uncertainty recommended by Calvert et al. (2008)
(25 % (2σ), kOH=8.04 × 10-12 cm3 molecule-1 s-1) brings the two data sets into better agreement,
COH(conv)/COH(Chex) = 1.21 ± 0.22. The cyclohexane
measurements were also affected to a greater extent by the chamber dilution
due to the slower rate of reaction with OH, which contributed to 25–30 % of the total cyclohexane decay rate directly after the photolysis lamps
were initiated, compared to 5–10 % for the iso-butene experiments.
Correcting the cyclohexane data for a hypothetically enhanced chamber
dilution could explain the lower sensitivity measurements (as the decay
increases, [OH]inf increases), however the dilution rate was confirmed
prior to photolysis of TBHP in each experiment.
Formaldehyde photolysis calibration
Figure 10a shows the instrument sensitivity to HO2,
CHO2, as a function of internal cell pressure
for the newly developed formaldehyde photolysis calibration technique for
the HIRAC FAGE instrument. Each data point corresponds to the average of up
to five HO2 decay traces (Fig. 8a), and the error bars are
representative of the total calibration 1σ uncertainty (Sect. 5.4). All calibrations were completed over a 4–8 mW laser power range.
The alternative calibration was observed to be in good agreement with the
conventional H2O vapour photolysis calibration technique over the
operating internal cell pressure range of 1.8–3.8 mbar
(CHO2(conv)/CHO2(alt) = 0.96 ± 0.09) for the
Litron-based FAGE system.
The kinetics of the HO2 decay due to recombination and first-order wall
loss (Eq. 7) were confirmed by studying the HO2 decay profile with
the chamber mixing fans on and off using the University of Leeds aircraft-based
FAGE instrument. With the mixing fans off, the decay was accurately
described by the recombination kinetics only (Fig. 8b), giving
CHO2 values within error of the fans on
experiments, as shown in Fig. 10b. Good agreement between the
conventional and alternative calibration methods was also observed across
the 1.42–2.48 mbar internal cell pressure range, and the overall
correlation between conventional and alternative calibration methods was
calculated as CHO2(conv)/CHO2(alt) = 1.07 ± 0.09 for
the high-frequency aircraft-based FAGE instrument.
Calibration uncertainties
The overall uncertainty associated with the calibration methods presented
here was calculated using the sum in quadrature of the accuracy and the
precision terms of the calibration. The accuracy term accounted for any
systematic uncertainty in the calculation of [HOx] for each calibration
method, signal normalisation etc., and these are displayed in Table 2. The
precision of the calibrations was defined as the random errors associated
with each method. All uncertainties are quoted as 2σ.
The systematic uncertainties to 2σ in the various
parameters that determine the accuracy in the OH and HO2 calibration
factors for all three calibration methods. Total accuracy is taken as the
sum in quadrature of the individual uncertainties. The Online position error
is the approximate error in the maximum line intensity that is achieved when
positioning the laser wavelength at the centre of the OH transition.
H2O +hν
Hydrocarbon decay
HCHO +hv
Parameter
Uncertainty
Parameter
Uncertainty
Parameter
Uncertainty
F184.9 nm
32 %
kOH
20–25 %
kHO2+HO2
35 %
Δt
2 %
kDil
10 %
SHO2 intial
20 %
[H2O]
10 %
GC-FID
4 %
Laser power
6 %
σH2O,184.9nm
6 %
Laser power
6 %
Online position
4 %
Laser power
6 %
Online position
4 %
Online position
4 %
Error
35 %
Error
24–28 %
Error
41 %
H<inline-formula><mml:math display="inline"><mml:msub><mml:mi/><mml:mn mathvariant="normal">2</mml:mn></mml:msub></mml:math></inline-formula>O photolysis calibration
The total uncertainty in the H2O photolysis calibration method was
estimated to be ∼ 36 %. The accuracy was defined by the
uncertainty associated with each term of Eq. (1) in the determination
of [HOx] and was estimated to be ∼ 35 %. The largest
contribution to the accuracy of this calibration method came from the
determination of the calibration source flux, F184.9 nm, with a total
uncertainty of 32 %. The product of the flux and the irradiation
time from Eq. (1), F184.9 nm × Δt, was determined
using N2O actinometry which relied on the detection of trace levels of
NO (0.5–3 ppbv, Sect. 2.1) followed by evaluation of the measurements
using four rate constants each with ∼ 20 % uncertainty.
Although the actinometric method gives a direct determination of the product
F184.9 nm × Δt, in order to calculate [OH] from
Eq. (1) any differences between the total volumetric flow rate during
the actinometry experiment and the OH calibration needs to be accounted for
as they change Δt. It is therefore necessary to account for the
uncertainty in Δt, which was determined to be ∼ 2 %
using the uncertainty in the flow rates from the mass flow controllers
(∼ 1 %). For the remainder of the terms in Eq. (1)
their contributions to the accuracy in the H2O photolysis calibration
method were as follows: σH2O was taken from Cantrell et
al. (1997), with a reported total error of ±6 %; the error in [H2O] was
taken from the hygrometer instrumental uncertainty (±10 %); and
laser power was defined by the laser power meter (Molectron Powermax 500A,
±0.25 mW).
The precision was typically between 4 and 10 % for the flow tube
calibration process and was taken from the standard error in the error-weighted
fit of the calibration plot. The error bars were representative of
the standard deviation in the SOH and [HOx] for the x and y axes
respectively. The flux output of the calibration source, hygrometer and CPM
measurements were observed to have good point-to-point stability and
therefore low standard deviations.
Hydrocarbon decay calibration
The accuracy of the hydrocarbon decay method was estimated to be better than
that of the flow tube method (∼ 28 % compared to 35 %).
However, due to the large variation in the random errors that defined the
precision of the experiment, the total uncertainty for the HC decay method
was larger than the flow tube calibration method, with the total uncertainty
estimated at ∼ 45 %.
The accuracy in the calibration was intrinsic to the hydrocarbon used, being
dependent on the uncertainty in kOH and kDil. The largest
uncertainty was in kOH, taken from data reviews from the Calvert series
or IUPAC recommendations: n-pentane, ±20 % (Calvert et al., 2008);
cyclohexane, ±20 % (Calvert et al., 2008); and iso-butene, ±25 %
(IUPAC, 2007). Uncertainty in kDil was calculated from repeated
measurements of chamber dilution for each of the hydrocarbons, and induced
errors in GC-FID calibration (4 %). The precision of the experiments for
both n-pentane and cyclohexane was between 10 and 25 %, whereas
iso-butene showed much greater variation of between 13 and 69 %; possible
reasons for this have been discussed in Sect. 5.2.
Formaldehyde photolysis calibration
The total uncertainty for the HCHO photolysis calibration method has been
estimated at ∼ 46 %, which is 10 % greater than the
conventional calibration method. The accuracy of the HCHO photolysis method
was estimated as ∼ 41 %; the largest contribution to
this deriving from the uncertainty was in the HO2 recombination rate
constant (35 %), taken from the IUPAC recommendation (IUPAC, 2007).
Determining the accurate initial SHO2 (i.e.
SHO2 at t0) is hard as HO2 does not
fully reach steady state before the photolysis lamps are switched off;
therefore there is a certain amount of subjective choice in the value of
S(HO2)0, and hence the uncertainty in the initial
SHO2 was based on the standard deviation of the
mean “steady-state” HO2 signal, which gives an estimation of the 1 s point-to-point variability for a chosen t0 (∼ 20 %).
The error associated with the precision of the experiment was taken from the
error propagation of the standard error terms from the Levenberg–Marquardt
iterative fitting procedure for Eq. (9) and Fig. 8. This includes
both the error in the CHO2 and kloss
parameters. The precision for this method was in line with the conventional
flow tube calibration between 10 and 20 %.
Conclusions and outlook
The first pressure-dependent calibrations of a FAGE instrument for both OH
and HO2 have been successfully conducted using the HIRAC chamber.
Previous pressure-dependent aircraft measurements had been extracted by
assuming that the calibration factor could be determined by simply
calibrating at the required internal FAGE cell pressure. Assumptions were
therefore made that variations in radical losses on the inlet and the nature
of the expansion caused by the varying pressure differential inside and
outside the FAGE cell were insignificant. The results displayed in Figs. 9
and 10 validate the conventional calibration method with the
alternative hydrocarbon decay and HCHO photolysis methods over a range of
internal FAGE cell pressures. It should be emphasised that, strictly
speaking, this validation applies only to these particular FAGE instruments,
but this work suggests that the agreement between the different calibration
techniques will translate to FAGE instruments of similar designs. As the
calibration methods are quite different in principle, they are unlikely to
be subject to the same systematic errors. The alternative calibration
results presented here have been shown to be well within the combined
uncertainty of their respective traditional calibration method, validating
the pressure-dependent flow tube calibration technique and improving
confidence in FAGE measurements both in the field and in kinetics
experiments. Both alternative methods have also shown that calibrations
conducted under high [H2O]vap conditions (2000–4500 ppmv) can
be applied to measurements at low [H2O]vap (< 15 ppmv).
The hydrocarbon decay method has shown that the FAGE instrument can be
calibrated over a range of external pressures using different hydrocarbons.
Compared to the conventional calibration method, where [HOx] are
generated typically at > 108 molecule cm-3, the
hydrocarbon decay method is conducted at a [HOx] relevant to
chamber-based experimental measurements (∼ 107 molecule cm-3) and much closer to typical ambient OH concentrations
(∼ 106 molecule cm-3).
Currently the total error associated with the hydrocarbon decay method is
greater than that of the flow tube method (∼ 45 % vs.
36 %). The accuracy or total systematic uncertainty associated with the
alternative OH calibration method is lower than that of the flow tube
calibration method (28 % vs. 35 %), and hence an improvement in the
precision of the experiment could improve the overall uncertainty to be in
line with the flow tube method. The primary source of random error arose in
the detection of OH close to the detection limit. Increasing the steady-state
OH concentration in the chamber would allow easier detection of the
hydrocarbon decay compared to chamber dilution, as well as an OH measurement
above the detection limit. The steady-state OH concentration could be
increased by increasing the 254 nm intensity in the chamber; using new lamps
or more lamps; altering the OH precursor, e.g. O3+ alkenes or
photolysis of methyl nitrite; or by lowering the initial [HC]. The latter
would require a more sensitive hydrocarbon detection technique than GC-FID
or FTIR, which are currently available in HIRAC. One such technique is proton
transfer mass spectrometry (PTR-MS), which would reduce the uncertainty in
the hydrocarbon decay measurements by providing higher-time-resolution
measurements and allow for easier simultaneous measurement of multiple
hydrocarbons at low concentrations, effectively providing multiple
independent estimates of COH from a single experiment. Uncertainties in
the rate coefficients could also be reduced by a concerted laboratory study
including relative rate and direct flash photolysis methods, decreasing the
systematic error. With careful experimental design, errors could potentially
be reduced to closer to 10 % (Orkin et al., 2010; Carr et al., 2011;
Glowacki et al., 2012).
A full range of pressure-dependent calibrations using this method would
currently take ∼ 2 days, compared to ∼ 3 h
for the flow-tube-based calibration. However, the timescale does not limit
the suitability of the method for regular confirmation of COH obtained
from the flow tube calibration method.
The total uncertainty in the HCHO photolysis method is ∼ 46 % which is 10 % greater than that of the traditional H2O
photolysis method. The HCHO photolysis method is quick and reproducible. The
time taken to complete the analysis and the errors is comparable with the
flow tube technique. An advantage of the HCHO photolysis method is that
several runs can be completed in one fill of the chamber, compared to the HC
decay method, which requires one fill per experiment (although the proposed
use of multiple HC decays will provide multiple estimates of COH from a
single chamber fill).