Heard D.E. (editor) Analytical Techniques for Atmospheric Measurement

Подождите немного. Документ загружается.

328 Analytical Techniques for Atmospheric Measurement

and resulted in conversion efficiencies of about 0.50, with an instrumental uncertainty

range of 10–30%. These NO

measurements were used in combination with NO, O

ClO, and HO

measurements to make a comparison with calculated NO

values in the

sunlit lower stratosphere, with the finding of a 19% difference, which was well within the

combined uncertainties of 50–70%.

Around 2000 Ryerson et al. (2000) built an NO

converter using a high-pressure Hg arc

lamp. This design was driven by the need for high time resolution aircraft measurements,

which in turn requires a high NO

-to-NO conversion fraction during a relatively short

time in the photolysis cell. The Hg arc lamp makes this possible while at the same time

reducing the heat load, also a significant consideration. An ellipsoidal reflector is used

to direct the light toward the photolysis cell through a set of optical filters designed to

minimize sample heating and attendant interferences from thermally labile species, as well

as to restrict the light to longer wavelengths. This results in the loss of 35% of the useful

lamp output, but this is compensated for by increased specificity due to reduced photolysis

of potential interferent species (HNO

,HO

; see Figure 7.7). Interferences

from PAN and HNO

were measured in the laboratory and found to be negligible. The

long pass filter (350 nm) was found to significantly reduce the UV artifact described

earlier. This combined with the apparently high water solubility of the contaminant

species serves to confirm earlier suggestions that HNO

contamination of the cell walls

plays a role in the artifact signal. The cell pressure is maintained at 250 torr (typically

flown on a mid-altitude P3 aircraft) for a cell residence time of 0.8 s and 1/e response

time of 0.65 s. The conversion factor of NO

to NO is 0.7. Achieving such high conversion

at relatively short residence times has enabled high resolution aircraft measurements to

be made in atmospheric features with small spatial scales, such as plumes from power

plants and petrochemical facilities (Ryerson et al., 2003).

7.4.4 The measurement of total reactive nitrogen (NO

)

The lack of converter specificity for NO

or any other particular species is turned to

an advantage in the measurement of total reactive nitrogen, or NO

, which includes

NO, NO

,NO

, HONO, HNO

,HO

, ClONO

, PANs, and other organic

nitrates. For an NO

measurement it is desired that all reactive nitrogen species be

converted to NO, but not other nitrogen-containing species (e.g. NH

O, HCN). As a

practical matter, it is required that near-unity conversion be achieved for all NO

species

contributing significantly to the total NO

amount, and that there be only minimal

conversion of potential interferents, as determined by the product of the conversion

efficiency and abundance for a particular species.

Several early studies used a variety of converters that were originally intended for use

as NO

-to-NO converters, but which suffered interferences from the conversion of other

nitrogen-containing species as well. So, these converters were pursued as means to measure

HNO

and other nitrogen species. Joseph & Spicer (1978) used a dual channel instrument,

each channel having a molybdenum converter in line, and placed a nylon filter in line for one

of them to scrub HNO

, so that the difference signal would yield the HNO

mixing ratio.

Kelly and Stedman (1979) used the same principle to measure HNO

, but instead with a

converter consisting of a quartz tube at 350



C packed with Pyrex beads. Helas & Warneck

Chemical Methods 329

(1981) used the difference in signals from molybdenum and ferrous sulfate converters to

infer a measure of what they termed ‘excess NO

’, or non-NO

compounds, which they

recognized might include N

,HO

,CH

, and PAN, although the conversion

efficiencies for these species were not demonstrated. The molybdenum converter found

application in commercial instruments, as flown by Dickerson (1984) to measure total

reactive nitrogen in the free troposphere, where convection was demonstrated to have a

significant impact on reactive nitrogen mixing ratios. One difficulty observed with the

molybdenum converter is a ‘memory effect’ in which, after exposure to elevated levels of

reactive nitrogen, the background signal remains high, introducing uncertainty into the

measurement. Controlled, stepped heating has been used to thermally dissociate PANs,

organic nitrates, and HNO

to NO

for detection by LIF (Chapter 4).

Bollinger et al. (1983) and Fahey et al. (1985) introduced the use of gold catalysts

for the measurement of NO

via the reduction of NO

species to NO by CO. Bollinger

et al. (1983) employed a gold-coated quartz tube and tested the conversion efficiency

of NO

, HNO

, and n-propyl nitrate (NPN), selected as being representative of organic

and inorganic reactive nitrogen compounds. Conversion was measured as a function of

converter temperature and CO mixing ratio. Under appropriate conditions, complete

conversion was found for all three species. Fahey et al. (1985) extended this work by

testing the conversion of a thin-walled solid gold tube, which was found to be more

reliable than the coated quartz tube used earlier. Nickel and stainless steel converters were

also tested, but were not as good. NO

species chosen for tests of conversion efficiency

and linearity were NO

, HNO

, and PAN. For a converter at 300



C, the conversion

efficiencies of all species exceeded 90%, and the conversion was linear for mixing ratios

in the range 0.1–50 ppbv. Temperatures higher than 300



C were not used in order to

minimize potential interferences, such as from N

In addition, the non-NO

compounds NH

, HCN, N

O, CH

, and various chlorine

and sulfur compounds were tested as interferents. NH

and HCN were found to be the

principal interferents, at least in dry, synthetic air. Conversions of 2–8% and 2–20% were

found for NH

and HCN, respectively. The addition of 0.5–2.5% H

O was found to

reduce these conversion efficiencies to about 2%. Using laboratory air at 20% relative

humidity resulted in conversion efficiencies of less than 1%. In contrast with this,

later work by Kliner et al. (1997) found near-complete conversion of HCN. On the

other hand, Weinheimer et al. (1998) measured HCN conversion efficiencies for three

different converters sampling ambient air from a DC-8 aircraft, with results for two of

the converters being consistent with the results of Fahey et al. (1985), and one, not

consistent. Conversion in humidified ambient air for two converters was small, at about

5%. However, the third converter showed 30% conversion. Taken all together, the results

from the various studies show that outwardly identical converters can experience different

interferences from HCN, perhaps due to their own particular histories (air sampled,

cleaning history, temperature history). However, these effects are minimized by ambient

humidity, and irrespective of humidity, conversion is smaller in ambient air than in

synthetic air. This result points to the value of testing each converter for its own specific

interference from HCN in order to verify that ambient levels of NH

and HCN are

generally small enough that interferences will usually be negligible in field measurements.

This also points to the general value in testing converters for the conversion efficiency of

several component species, as well as interferents. Fahey et al. (1986) used the then new

330 Analytical Techniques for Atmospheric Measurement

gold-converter technique in conjunction with measurements of many of the component

species at a ground site in the Colorado mountains to show that PAN was often the

dominant NO

species. They also found that correlations of NO

with O

gave evidence

for the production of O

in polluted air masses and showed that some NO

species are

subject to surface loss.

The molybdenum and gold converters have been compared in side-by-side tests, with

mixed results. Fehsenfeld et al. (1987) found that the two gave similar estimates of NO

ambient air over a wide range of mixing ratios, ranging from 0.4 ppbv in clean continental

air to about 100 ppbv in very polluted conditions. In this test, the molybdenum converter

did not suffer from memory effects noted above. However, it did suffer significant

degradation in conversion efficiency after extended operation in polluted conditions. In a

second ground-based test, Williams et al. (1998) report generally more favorable results.

Seven converters were compared, all followed by chemiluminescence detection of NO,

but some with gold catalysts using either CO or H

as the reducing agent and some

using molybdenum converters. No significant difference between gold and molybdenum

converters was found for polluted conditions experienced, where NO

was in the range

2–100 ppbv. This limits these favorable conclusions to these high mixing ratio conditions.

The authors note that differences could become significant at lower mixing ratios, and

especially for conditions when NO

is a smaller fraction of NO

Another important condition for measurement reliability is the transmission of HNO

through inlets and sample lines, as HNO

can be a dominant component of NO

, and

it is also very sticky. That is, it is prone to adsorb onto surfaces, either reversibly or

irreversibly. Neuman et al. (1999) used a sensitive, fast-response HNO

CIMS instrument

to test various surfaces for their ability to pass HNO

. They found that teflon above



C is very good for this purpose, but that stainless steel, glass, fused silica, aluminum,

nylon, silica-steel, and silane-coated glass all take up a large fraction of the HNO

. HNO

is reversibly taken up by teflon at low temperatures, but if it is kept moderately warm

>10



C it is a suitable material for inlets and sample lines. Inlets for NO

are typically

composed of teflon and are often heated to somewhat higher temperatures of 30–70



to insure good transmission (e.g. Ridley et al., 1994).

measurements from the NASA ER-2 aircraft have proven to be a valuable

component of extensive field missions aimed at understanding polar ozone loss processes.

Fahey et al. (1990) found that correlation of NO

and N

O, which derives from N

being the source of stratospheric NO

, could be used to infer the degree of denitrification

in the winter polar stratosphere, a process critical to ozone chemistry, as the removal of

reactive nitrogen from an air mass allows chlorine species to catalytically destroy ozone.

This nitrogen loss process was later observed by Fahey et al. (2001) in the form of

large HNO

-containing particles that were precipitating in the winter arctic stratosphere,

causing denitrification and hence making ozone loss more likely (Figure 7.9).

A new and interesting use of an NO

instrument is unattended measurements aboard

civil in-service commercial aircraft. The MOZAIC (Measurements of Ozone and Water

Vapour aboard Airbus in-service Aircraft) project includes a small NO

instrument

installed on long-haul aircraft to provide regular surveys of the poorly sampled tropopause

region. These are difficult constraints under which to operate a research-grade instrument,

and especially with the long inlet line required, but with the large number of flights taken,

and with proper attention to quality control, a large quantity of valuable data are being

generated (Volz-Thomas et al., 2004).

Chemical Methods 331

Figure 7.9 Simultaneous measurements from two Au-converter/chemiluminescence NO

channels

aboard the NASA ER-2 in the Arctic stratosphere. The darker trace is for air sampled through an aft-facing

inlet and so is primarily gas-phase NO

(mostly HNO

in this case) as particles are inertially separated.

The lighter trace is for air and particles sampled through a forward-facing inlet, and the two traces closely

overlap, except at the “spikes” which are due to the presence of HNO

-containing particles, likely nitric

acid trihydrate, which evaporate and release a burst of HNO

. By means of sedimentation, these particles

irreversibly remove HNO

from the air mass and this denitriﬁcation allows chlorine to remain active for a

longer time in destroying ozone. (Figure 2c of Fahey et al., 2001, reprinted with permission from AAAS.)

7.4.5 Routine NO

monitoring

As NO

is a major urban pollutant, central to photochemical ozone production, its routine

monitoring has become an important aspect of pollution control in urban environments.

In the UK, a network of over 200 automated sites employs chemiluminescence NO

instruments, most with molybdenum converters, to allow an assessment of ambient NO

levels for the purpose of determining whether NO

concentration objectives, imposed by

government agencies, have been met (AQEG, 2004). These objectives are not to exceed

200 gm

−3

over a one-hour average, nor to exceed 40 gm

−3

as an annual average (divide

these values by 2 to get the rough equivalent in ppbv for standard temperature and

pressure). These data also allow the observation of patterns, cycles, trends, and correlations

with measurements of other species that can be tested against our understanding of the

basic processes of pollution chemistry as embodied in numerical models. In particular, it

enables the connection of emission rates to ambient concentrations, so that predictions

of future concentrations may be inferred from projections of future emissions.

Figure 7.10a shows the diurnal variation in NO

at various sites in the monitoring

network (AQEG, 2004). That road traffic is the largest source is reflected in the diurnal

cycle with a maximum during the morning rush hour and secondary maximum in the

evening, except for the most remote site, which shows little in the way of a diurnal cycle.

The monitors are also valuable for the determination of longer-term, multi-year trends.

Emissions in the UK have declined by 37% from 1990 to 2000, and this is reflected in

the decadal trend in NO

mixing ratios shown in Figure 7.10b (AQEG, 2004). Emissions

are expected to decrease another 25% by 2010 due to tighter vehicle emission standards,

so it is important to continue to monitor ambient levels to see whether their decline is

commensurate with this. The monitoring network may also be used to construct the large-

scale spatial distribution of NO

and correlate the measured concentrations to emission

estimates. Figure 7.10c shows the distribution within the UK of annual-average NO

concentrations derived from the network dataset interpreted through use of an empirical

model (Stedman et al., 2002). As the density of sites throughout the monitoring network

332 Analytical Techniques for Atmospheric Measurement

(a)

200

180

160

140

120

100

(μg m

–3

, as NO

)

London Bloomsbury (μg m

–3

, as NO

)

London Bexley (μg m

–3

, as NO

)

West London (μg m

–3

, as NO

)

Manchester Town Hall (μg m

–3

, as NO

)

Glasgow City Chambers (μg m

–3

, as NO

)

Belfast Centre (μg m

–3

, as NO

)

Port Talbot (μg m

–3

, as NO

)

Lullington Heath (μg m

–3

, as NO

)

01234567891011

Hour of Day

12 13 14 15 16 17 18 19 20 21 22 23

Figure 7.10a The diurnal cycle of NO

at eight sites in the UK, showing the morning rush hour maximum

due to vehicle trafﬁc and a secondary maximum in the evening. (Figure 6.14 of AQEG, “Nitrogen Dioxide

in the United Kingdom”, 2004, printed with permission from Crown.) (Reproduced in colour as Plate 3a

after page 264.)

225

(b)

200

175

150

125

100

Bridge Place

Annual mean NO

(μg m

–3

, as NO

)

London Bloomsbury

London Bexley

West London

Manchester Town Hall

Glasgow City Chambers

Belfast Centre

Port Talbot

Lullington Heath

1982

1983

1986

1987

1988

1989

1990

1991

1992

1993

1994

1995

1996

1997

1998

1999

2000

2001

Figure 7.10b The decadal trend in NO

at nine sites in the UK, showing the gradual decline due to

the tightening of vehicle emission standards. (Figure 6.5 of AQEG, “Nitrogen Dioxide in the United

Kingdom”, 2004, printed with permission from Crown.) (Reproduced in colour as Plate 3b after page 264.)

Chemical Methods 333

Figure 7.10c The geographical distribution of NO

in the UK, derived from monitoring-network data

interpreted with an empirical model. (Figure 3.1 of Stedman et al., 2002, printed with permission from

AEA Technology.) (Reproduced in colour as Plate 3c after page 264.)

is not always high, and in order to elucidate the dependence of measured concentrations

on NO

sources, it is useful to interpolate the network measurements and to model

empirically the contributions of local and more distant sources to the NO

measured at

a particular site. In turn, the empirical model is validated by comparison of its results

with measurements at specific sites. Maps such as that shown in Figure 7.10c can be

used to formulate statistics on the regions in which concentration objectives for NO

are

334 Analytical Techniques for Atmospheric Measurement

exceeded, as well as the fraction of the population thus affected. Furthermore, it serves

to illustrate the impact of urban areas on more rural regions.

7.5 Ozone via homogeneous chemiluminescence

7.5.1 O

+ Ethene

The first gas-phase chemiluminescence instrument was for the measurement of O

and

was based on the chemiluminescent reaction of O

with ethene. Nederbragt et al. (1965)

mixed flows of ozone-containing air and ethene directly in front of a PMT. The detection

limit was 30 ppbv, so it was not useful for routine atmospheric monitoring but was

useful for monitoring the elevated levels (1000s of ppbv) that are produced by beams

of electrons near accelerators. Warren and Babcock (1970) made further improvements,

especially with regard to portability, but, as the application was still for accelerators, they

were satisfied with a detection limit too high for atmospheric applications.

Commercial versions based on this principle were produced and used for atmospheric

monitoring, and some were modified for use in research applications, including aircraft

deployment (Routhier et al., 1980; Gregory et al., 1983; Kondo et al., 1987). The emission

spectrum for the O

–ethene reaction peaks near 430 nm (Finlayson et al., 1972), and the

products include excited OH and HCHO. The instruments consist of inlet flow system,

an ethene flow system, and a reaction chamber viewed by a PMT. The modifications to

the commercial instruments for aircraft use generally include flow control and pressure

regulation, so that instrument sensitivity does not change with altitude or cabin pressure.

Gregory et al. (1983, 1984) report a detection limit of 2 ppbv, with a 90% response in

about 2 s, and an accuracy of 5% or 5 ppbv, whichever is larger.

7.5.2 O

+NO

Just as O

can be added in excess to the sample air stream for the measurement of

NO, so can NO be added in excess for the measurement of O

. Stedman et al. (1972)

built a prototype instrument based on this principle. A typical reaction vessel was a

cylindrical tube, 4



long and 2



in diameter, viewed from one end by a PMT through

a red filter (as for NO instruments). Conical reaction vessels were also tested. A typical

sample flow was 20 cc min

−1

, with a reagent flow of 3 cc min

−1

of pure NO. The response

time was less than 1 s, excluding inlet sampling. Calibration was achieved in multiple

ways using a photolytic O

source. One is via NO titration, which adds a measured flow

of a known concentration of NO to the sample stream before it enters the instrument

to allow reaction prior to that in the reaction vessel. As NO is increased the O

signal

decreases. An extrapolation to zero O

gives the known amount of NO corresponding

to the O

level being used. Another method used an infrared absorption measurement

of O

from the photolytic source. A linear response was found for the wide range of

50 ppbv–50 ppmv. Several species were tested for interference at the 100 ppmv level: SO

,Cl

O, C

, and C

. None gave detectable chemiluminescence. The

instrument was tested in the field where it showed excellent response to atmospheric

Chemical Methods 335

variability. The chemiluminescence technique using NO is much more sensitive than that

using ethene due to the high rate of reaction of O

with NO.

The high sensitivity and fast response of the chemiluminescence detector using NO

has led to its use in making eddy-correlation flux measurements of O

to the ground, as

surface deposition is a major sink for O

. Pearson (1990) describes an instrument designed

to make such measurements from an aircraft, where frequency response requirements are

much more demanding than for an instrument fixed to the ground. A filtered frequency

response of 12 Hz was achieved. An earlier version of the instrument (Pearson & Stedman,

1980) was improved upon by doubling the sample flow to 37 L min

−1

and by increasing

the diameter of the photocathode of the red-sensitive PMT to 51 mm. The reagent flow

is 28cm

−1

of 5–10% NO. The cylindrical reaction vessel has an ID of 4.7 cm and a

length of 8 cm and is operated at 8–10 mb. The inlet flow is regulated by a stainless steel

frit in an aft-facing inlet to prevent cloud and rain water from entering. The reaction

vessel is maintained at 39



C in order to keep the sensitivity constant in what may be

expected to be a warm aircraft flying in the boundary layer. The PMT is operated in the

analog mode and is low-pass filtered at 10–12 Hz. An operational consideration when

flying the poisonous NO reagent gas is the safety requirement for a containment vessel

to house the compressed gas bottle with a line plumbed to the outside of the aircraft, so

that in the event of a leak, the NO is vented outside the aircraft.

Lenschow et al. (1981) used an earlier version of this instrument to analyze the relative

contributions of chemistry and transport to the ozone budget in the boundary layer over

range and crop land and to show that the time rate of change of the O

concentration

was several times larger than contributions from horizontal advection and the divergence

of the vertical eddy flux, thus demonstrating the dominance of local chemistry over

transport in determining the local time rate of change of the O

concentration.

Ridley et al. (1992) describe a high-sensitivity chemiluminescence instrument that is

designed to have small size and weight, suitable for use as part of a multi-instrument

package on a small aircraft and still with good frequency response. Photon counting

is employed to give a sensitivity of 2000 counts s

−1

per ppbv of O

and a detection

limit well below 0.1 ppbv for a 2 s integration. The overall measurement uncertainty at

higher mixing ratios is estimated to be 2 ppbv plus 5%. The reaction vessel is conically

shaped, with a volume of 17 cm

(eight times smaller than that of Pearson, 1990), and

is made of highly polished stainless steel coated with gold. It operates at a pressure of

about 10 torr, with a small sample flow of 180 sccm that is combined with a reagent flow

of 1.5 sccm of pure NO. The instrument is calibrated prior to flight using a standard

UV absorption instrument. Ridley et al. (1992) offer several interesting points about the

chemiluminescence measurement of O

compared to that for NO. First, tropospheric

is often several orders of magnitude larger than that of NO, so high signal-to-noise

ratio is expected. Second, the reagent NO is much easier to provide (from a bottle)

than is the reagent O

for the NO measurement (silent discharge). Third, since the

reagent NO is pure compared to 3–5% for reagent O

, smaller reaction vessel volumes

are possible as the reaction occurs more quickly and less residence time is required. And

fourth, the limiting background signal in an NO instrument due to reagent O

is not

present in the O

instrument. All these factors combine to make possible instruments

which are very fast and sensitive. The high sensitivity and low noise of this instrument

were put to test in low altitude ∼100 m aircraft measurements in the Arctic, where

336 Analytical Techniques for Atmospheric Measurement

ozone depletion events occurred in which O

mixing ratios were measured to be as

low as 0.03 ppbv due to processes not yet fully understood but likely involving bromine

chemistry.

7.6 NO

via luminol

Luminol has found application in the measurement of NO

via chemiluminescence

near 425 nm. As luminol was described earlier to be used for the chemiluminescence

measurement of O

, an obvious concern is the interference from O

to the NO

determi-

nation. Despite this potential limitation, the luminol technique initially offered some hope

of avoiding the significant interferences encountered in the NO

techniques described

earlier (Section 7.4.3) that coupled heated catalytic converters with NO chemilumines-

cence detectors. Although photolytic converters for NO

detection do not suffer serious

interference problems and are generally used instead of the catalytic converters, the

luminol technique still enjoys application in commercial NO

instruments, as well as in

some peroxy radical instruments (Section 7.9.1). It has the advantages of small size, low

power consumption, and high sensitivity.

Maeda et al. (1980) converted a gas-phase reaction vessel so that it could hold a pool

of liquid luminol solution with a constant surface area, and they varied the luminol

concentration to optimize the response to NO

. KOH was added to the solution, and its

concentration was optimized at 0.05 M. KOH was determined to facilitate the dissolution

of NO

in the luminol solution, and the OH

−

may participate in the chemiluminescent

reaction. Potential interferents were tested. No interference was found from NH

, organic

nitrate, organic nitrite, NO, and hydrocarbons. A negative interference was found with

, but this was minimized by increasing the pH with additional KOH (1.0 M). O

is a very strong interferent with a chemiluminescence signal 30 times that for NO

but that is remedied by the addition of sodium sulfite to the luminol solution, which

also removes an SO

interference. The detection limit for NO

for this prototype was

50 pptv.

Wendel et al. (1983) improved on the earlier design. Rather than use a pool of luminol,

they used a wetted strip of cellulose fiber filter paper so that the detector would be less

sensitive to movement and orientation, and so that the PMT would not view the reservoir

of liquid. This also improved the time response by eliminating the signal that comes

from the liquid volume at some time lag after the NO

encounters the surface. Other

modifications were a cotton trap to remove O

in the inlet and the use of NaOH instead of

KOH. Also, the measurement of NO was achieved by its conversion (60–70%) to NO

passing the sample over CrO

adsorbed on silica gel (in contrast with the more common

conversion of NO

to NO followed by NO detection). Also, Na

(0.01 M) was used to

reduce the sensitivity to O

and at the same time enhance the sensitivity to NO

. This also

reduces the induction period (the time period after changing the solution during which

the sensitivity to NO

increases) to one hour rather than days. It was also found that the

addition of methanol to the solution further increases the sensitivity and specificity for

. No interference was found from 20 ppbv each of HNO

,NH

, and HCN. At best,

Chemical Methods 337

the NO

response is 30 times that for O

. PAN was found to respond like NO

(indeed

a gas chromatograph in conjunction with a luminol detector was later used to measure

PAN and NO

; Burkhardt et al., 1988). The detection limit for field measurements was

reported to be 30 pptv S/N =2, with a 2 Hz response to 20 ppb changes. However, Ray

et al. (1986) subsequently report for a later version of this instrument a detection limit

of 1 pptv and a 3 Hz response.

Schiff et al. (1986) describe a commercial instrument based on luminol (Luminox

LMA-3, Unisearch/Scintrex). In contrast to the complex solution described in the Wendel

et al. (1983) paper, comparatively little information is available in the open literature

about this commercial instrument. However, it was evaluated by Fehsenfeld et al. (1990)

in a comparison with a photolysis/chemiluminescence instrument (Section 7.4.3) and

a tunable diode laser absorption spectrometer (TDLAS, Chapter 2). For high levels of

above 2 ppbv, similar results were found for all three instruments. On the other

hand, below 2 ppbv, interferences from O

and PAN were evident. However, they were

consistent enough that they could be corrected for down to 0.3 ppbv, provided there

are simultaneous measurements of O

and PAN. An O

scrubber placed on the inlet to

remove this interference worked to do that, but it did not remove PAN and it removed

about 50% of the NO

as well. Kelly et al. (1990) also assessed the Luminox instrument

for airborne measurements and determined several corrections that needed to be applied

for low mixing ratios. Schmidt et al. (1995) determined that recommended O

corrections

did not work, in contrast to the Fehsenfeld et al. (1990) study, but they, too, found that

the proprietary scrubber had the unacceptable trait of removing NO

as well. However,

they created a scrubber using natural rubber that removed most of the O

without

significantly affecting NO

In summary, the luminol detector is sensitive, lightweight, fast, and has low power

consumption. It suffers from strong interferences from O

and PAN, which, in the past,

have been corrected for with varying degrees of success. Also of concern are calibration

instabilities, nonlinearities, and zero offsets, but Kelly et al. (1990) found that with care,

errors as low as 15% could be attained.

7.7 Peroxides, HCHO, and HONO via liquid

techniques

7.7.1 H

via dissolution and chemiluminescence

Peroxides are products of atmospheric photochemistry that serve as reservoirs of odd-

hydrogen radicals and as indicators of prior photochemistry in an air mass. One of the

early techniques for the measurement of H

employed its collection from the gas phase

for detection in the liquid phase via the chemiluminescent reaction of H

with the

ubiquitous luminol reagent. Kok et al. (1978a) designed two methods for ‘scrubbing’

from the gas phase, one of them being a coil through which both the sample

air and distilled water flowed, extracting ambient H

into an aqueous solution. Once

in solution, it is mixed with a luminol solution and a solution of copper ions directly

in front of a blue-sensitive PMT to detect the 450 nm emission. The chemilumines-

cence mechanism is not fully known, but it involves the decomposition of H