Journal of Geophysical Research: Atmospheres

Relative measurements of ozone absorption cross-sections at three wavelengths in the Hartley band using a well-defined UV laser beam

Authors


Abstract

[1] New measurements of absorption cross sections of ozone are reported using a photometer with a well-defined UV laser beam at three different wavelengths in the Hartley band, by comparison with the conventional cross-section value at the mercury line wavelength of 253.65 nm. Reported measurement results and uncertainties are dependent on the value of the cross section used at 253.65 nm. Using the value determined by Hearn (1961), ozone absorption cross-section values of 9.64 × 10−18, 10.61 × 10−18, and 11.26 × 10−18 cm2 molecule−1 are reported at wavelengths of 244.06, 248.32, and 257.34 nm, respectively; whereas, use of the Malicet (1995) value leads to ozone absorption cross-section values of 9.50 × 10−18, 10.45 × 10−18, and 11.12 × 10−18 cm2 molecule−1, respectively. In all cases, the uncertainty of the results is dominated by the contribution from the cross-section value at 253.65 nm. The experimental setup described in this article not only provides a minor contribution to the entire reported measurement uncertainties but also confirms the internal consistency of previously published ozone absorption cross-section data sets and demonstrates the biases that can be introduced by changing data sets. The results are important for the accurate measurement of ozone concentrations in the troposphere by UV absorption methods and also for all measurements of ozone concentration in the atmosphere by spectroscopic methods for which the values of the absorption cross sections have been scaled to measured values in the Hartley band.

1. Introduction and Aims

[2] Ozone is one of the most important constituents of the Earth's atmosphere. While its decrease in the stratosphere remains a concern because of the consequent increase of terrestrial UV levels, its increase in the troposphere and especially at ground level has become, more recently, a major health issue because of its well-known toxicity [Hodgson, 2005]. As such, its concentration has been continuously monitored worldwide, both in the troposphere and the stratosphere by an increasing number of monitoring networks, using an increasing variety of instruments and techniques. These include Dobson and Brewer spectrophotometers, or lidar, differential optical absorption spectroscopy, and Fourier transform infrared techniques from the ground. Instruments and techniques on board satellites include the European Space Agency's Global Ozone Monitoring Experiment, Scanning Imaging Absorption Spectrometer, and the Michelson Interferometer for Passive Atmospheric Sounding as well as NASA's Total Ozone Mapping Spectrometer and Ozone Monitoring Instrument.

[3] Most of these instruments are based on spectroscopic techniques that rely on laboratory-measured ozone-absorption cross sections at various wavelengths and temperatures. A number of comparisons between instruments have raised concerns regarding the accuracy of the ozone cross section, as recognized by the Absorption Cross-Sections of Ozone (ASCO) committee, working under the umbrella of the World Meteorological Organization (WMO) and the International Ozone Commission (see http://igaco-o3.fmi.fi/ACSO/index.html).

[4] At ground level, the reference method for the measurement of ozone concentration is based on UV absorption at 253.65 nm [International Organization for Standardization, 1998]. This method implemented in the National Institute for Standards and Technology's (NIST) Standard Reference Photometer (SRP) instrument acts as the primary standard for numerous national and international ozone monitoring networks, such as the WMO Global Atmosphere Watch program [Klausen et al., 2003]. Several replicas of this instrument are maintained by the International Bureau of Weight in Measures (French acronym BIPM), one of which is the reference for international comparisons of national ozone standards coordinated by the BIPM. The ozone absorption cross-section value at the 253.65 nm wavelength, used in the SRP, is the value measured by Hearn in 1961 [Hearn, 1961]. It is also the reference value for a number of relative measurements of the ozone absorption cross section performed in the laboratory, as reviewed by Orphal [Orphal, 2003].

[5] During the first international comparison conducted by the BIPM [Viallon et al., 2006a], 23 instrument standards based on UV absorption (same measurement principle as that of the SRP) were compared together with two systems based on gas-phase titration, an independent method for ozone concentration measurements, maintained by the BIPM and by the National Institute for Environmental Studies of Japan [Tanimoto et al., 2006]. Gas-phase titration consists of mixing the ozone in air sample with nitrogen monoxide and measuring either the loss of nitrogen monoxide or the gain of the reaction product, nitrogen dioxide, to deduce the ozone concentration in the sample. The 2%–3% bias observed between the methods requires explanation and confirmation of the ozone absorption cross-section value, which represents the major uncertainty component in measurements based on UV photometry.

[6] In 2007, the BIPM started a laboratory program, which aimed to perform new measurements of the ozone absorption cross section with improved accuracy. Efforts focused on two major sources of uncertainty in the measurements: ozone purity and knowledge of the light path length. To this end, the BIPM first developed a laser ozone photometer, capable of measuring ozone concentrations in the same range as the SRP but with improved accuracy. This was achieved by using a laser instead of a mercury lamp as the light source to allow better control of the light path length.

[7] In this study, the laser ozone photometer was used in conjunction with one of the SRPs maintained by the BIPM to perform relative measurements of the ozone cross section at three different laser wavelengths in the Hartley band. First, the setup of the laser ozone photometer is described as well as its measurement uncertainty. A series of validation studies are then presented, which demonstrate the performance of the system in comparison with the SRP. For each of the three wavelengths of the laser, measurements with the two systems performed on the same ozone sample enabled a new value of the ozone cross section, relative to the reference value used within the SRP, to be deduced. Finally, agreement with previous measurements reported in the literature and under review by the ACSO committee is discussed.

2. Measurement Setup

[8] The complete measurement setup includes a commercial ozone generator that is capable of producing between a few and 1500 nmol mol−1 of ozone in air, one of the SRPs maintained by the BIPM, and the laser ozone photometer that was developed from another SRP. The operating principle of the laser photometer and the SRP are briefly described in this section. More details of the SRP's principle and its capabilities have been presented by Paur et al. [2003]. The major changes made to obtain the laser ozone photometer, which are in its optical setup, are further described. The rest of the instrument (i.e., gas cells and gas flow system) was tested again at the BIPM but only slightly modified. Finally, the performances of the laser ozone photometer were evaluated by comparison with the SRP, enabling new ozone absorption cross-section values and uncertainties to be determined.

2.1. Photometer Measurement Equation

[9] As with an SRP, the measurement of the ozone amount-of-substance fraction by the laser photometer is based on the absorption of radiation in the UV by ozonized air in the gas cells of the instrument. One particular feature of the instrument's design is the use of two gas cells to overcome the instability of the light source (laser or lamp). This is obtained in first having ozone in one cell and reference air free of ozone in the other cell, then the gases flowing through the cells are inversed to finally calculate the product of the transmittances of the two cells D. This configuration effectively doubles the light path length and cancels the noise common to the two cells, that is, the noise from the light source. The measurement equation is derived from the Beer-Lambert and ideal gas laws. The number concentration (C) of ozone is calculated from

display math

where σ is the absorption cross section of ozone at the laser wavelength under standard conditions of temperature and pressure, Lopt is the mean optical path length of the two cells, T is the measured temperature of the cells, Tstd is the standard temperature (273.15 K), P is the measured pressure of the cells, Pstd is the standard pressure (101.325 kPa), and D is the product of transmittances of two cells, with the transmittance (Tr) of one cell defined as

display math

where Iozone is the UV radiation intensity measured from the cell when containing ozonized air and Iair is the UV radiation intensity measured from the cell when containing pure air (also called reference or zero air).

[10] Using the ideal gas law, equation (1) can be recast in order to express the measurement results as the amount-of-substance fraction (x) of ozone in air:

display math

where NA is Avogadro's number (6.023 × 1023 mol−1) and R is the gas constant (8.314472 J mol−1 K−1).

[11] The formulation implemented in the SRP software is

display math

where αx is the linear absorption coefficient at standard conditions, expressed in cm−1, linked to the absorption cross section with the relation

display math

2.2. Optical Setup of the Laser Photometer

[12] The laser source used for the photometer is a frequency-doubled argon-ion laser. The argon-ion laser can be configured to emit light at either 488.1, 496.6, or 514.7 nm (wavelengths in vacuum). A leak through one of the laser cavity mirrors is sent toward a Burleigh WA-1000 Wavemeter with a resolution of 1 pm to determine the actual wavelength. To ensure traceability the wavemeter was calibrated at the Institut National de Metrologie at 633.991 and 532.245 nm, with a negligible uncertainty compared with the instrument resolution. The frequency doubling is enabled by a beta barium borate (BBO) crystal inside the laser cavity. According to laser manufacturers, the line width of the frequency-doubled light is 10 GHz. The emitted UV light can have a wavelength of 244.062, 248.323, or 257.337 nm.

[13] The optics used for the UV light were all of fused silica with coatings adapted to the three specific wavelengths or to UV light around 250 nm. The optical setup is shown in Figure 1. Light traps (LT0, LT1, and LT2) are used for detection. The light traps are each composed of three windowless Hamamatsu S1337-1010-N photodiodes. The first two were positioned at a 45° angle and the other at a 90° angle relative to the incoming beam, making the measured beam reflect off the photodiodes five times. Electronically, the photodiodes are connected in parallel. The light traps used are similar to the ones used by Lei and Fischer [1993]. The light-trap signals are typically low-pass filtered at 1 Hz.

Figure 1.

The optical setup of the UV laser ozone photometer. M, mirror; BS, beam splitter; LT, light trap; AOM, accousto-optic modulator. The spatial filter is composed of a 75 mm focal length lens, a 20 μm pinhole, and a 40 mm focal length lens.

[14] Because of walk-off in the BBO crystal, the frequency-doubled beam has a non-Gaussian shape with fringe-like effects. This can be a cause of interference effects and so to avoid this, the beam is passed through a spatial filter composed of a 75 mm focal length lens and a 20 μm pinhole. This gives the beam a Gaussian shape, and a 40 mm focal length lens collimates the beam with a small beam size. Finally, a diaphragm removes the diffraction rings. A 200 MHz acousto-optic modulator (AOM) for UV light between 244 and 266 nm is used to stabilize the power level of the light on the LT0 light trap.

[15] The beam splitters used are high-energy beam splitters (BS1) from CVI Melles Griot, with a beam-splitting coating on the front surface optimized for S-polarized light and an antireflection coating on the back. The coatings are optimized for all three wavelengths. As these beam splitters are not truly nonpolarizing, a CVI Melles Griot PGU Alpha-BBO high-power polarizer for 200–270 nm light was placed before the first BS.

[16] The light is passed through two cells to make use of the two-cell setup in the SRP. The 89.4 cm long cells are made of quartz with fused silica windows tilted at 3° to avoid multiple reflections of the beam inside the cells. During the study of measurement uncertainties performed on the SRP [Viallon et al., 2006b], a bias in the ozone concentration measurement because of reflections of the light on the gas cells' end windows was demonstrated. Consequently, the windows were tilted to a 3° angle with the vertical plane. This setup was kept in the laser photometer. The windows at each end of a cell are tilted in the same direction to maintain the same cell length. To ensure a consistent beam alignment through the cells, and hence a light path length equal to the cell length, diaphragms are mounted on the cell ends using a Teflon mount that ensures that the diaphragms are centered on the cell windows. To avoid an alignment where the beam accidentally reflects off the cell wall, the alignment is first carried out without the cells, with the diaphragms in their approximate correct position. The diaphragms also ensure a better method of determining the optical path length uncertainty if the cell length is known. The entire optical setup, including the cells, is enclosed in a thin aluminum casing to remove air turbulence and to significantly reduce temperature changes and background lighting.

2.3. Uncertainties of the Laser Photometer

[17] A revision of the measurement uncertainties and biases in the SRP was performed by the BIPM and the NIST in 2006 and can be found in the study of Viallon et al. [2006b]. Measurement uncertainties in the laser photometer are reported in this section. They are all calculated according to the Guide to the Expression of Uncertainty in Measurement [BIPM et al., 1995].

2.3.1. Temperature Measurements

[18] The temperature in the cells is measured by thermistors fixed to the exterior of both ends of each cell. All probes were calibrated on site by comparison with two reference temperature probes that have been regularly calibrated by the French National Metrology Institute, Laboratoire National de Métrologie et d'Essais. The average of the four measurements is used to calculate the ozone concentration. In the present setup, the maximum temperature difference between the two ends of one cell is 0.3 K. At both ends of the cells, the temperature difference between the two cells is lesser than 0.01 K. If it is assumed that the average temperature of the gas in any of the cells can be of any temperature in the range between the cell-end temperatures with equal probability, the statistical uncertainty of the average temperature is inline image with Δt being the maximum temperature difference between each cell end. With a temperature difference of 0.3 K, the result is 0.061 K. Neither the probe resolution uncertainty (0.003 K) nor the calibration uncertainty (0.02 K) is significant compared with this value.

2.3.2. Pressure Measurements

[19] The pressure in the cells is measured by Setra Model 270 pressure transducers connected to the tubes, which lead the gas sent through the cells away from the cells. The transducers were adjusted by comparison to a DHI barometer regularly calibrated by the BIPM Mass Department. For practical reasons and to measure the pressure in similar to how it is measured in an SRP, the pressure transducers were connected using a 57 cm long tube to the main tubes at a distance of about 30 cm from the cells.

[20] Only the pressure measured when ozone is present in a cell is used for the calculation of the ozone concentration. Tests have shown that the pressure difference measured before entering and after leaving the cell is 1 mbar, which translates into a standard uncertainty on the cell pressure of 0.58 mbar assuming that any pressure within that 1 mbar is equally probable. The pressure gauge uncertainty is 0.29 mbar giving a total pressure uncertainty of 0.64 mbar.

2.3.3. Optical Path Length

[21] The average of the two cell lengths are used to calculate the ozone concentration. Each cell length was measured just after assembly at NIST, using a caliper 101.6 cm long (40 in.), taking into account the thickness of each quartz window measured before assembly. The standard combined uncertainty on the average cell length as given by NIST is 0.4 mm. The diaphragms centered on each cell end place a limit on how much the actual optical paths in the cells can differ from the cell lengths. The diaphragms are typically open with a 3 mm diameter. Considering the 3° angled windows and an average cell length of 893.9 mm, the shortest straight optical path hitting the diaphragm openings is

display math
display math

Similarly, the longest straight optical path is

display math
display math

Assuming that any possible straight optical length in between these two lengths are equally possible, the standard uncertainty on the average cell length is (894.1 – 893.7)/(2√3√2) = 0.08 mm. This is negligible compared with the uncertainty on the cell length as given by the cell manufacturer.

[22] The combined uncertainty of the light path length in the laser ozone photometer is thus 0.4 mm or 0.045% relative. This value can be compared with the 0.3% relative uncertainty associated with the light path length in the SRP. As explained by Viallon et al. [2006b], the vertical cell's windows and the divergence of the light emitted by the mercury lamp in the SRP allows multiple reflections on the gas cell's walls, increasing the light path length by a factor that is difficult to assess by simple optical considerations.

2.3.4. Transmittance Products

[23] The electronic measurement system used is the same as the one used for the SRP, but the photodiodes and filtering are different. This contribution to the uncertainty does not change as a function of ozone concentration. Therefore, a measurement of the measured ozone concentration in the absence of ozone was performed to deduce this background noise, which is Gaussian distributed with a standard deviation of 8.5 × 10−6. By taking into account the resolution of the detection system, the total uncertainty on the double transmission because of electronics and detection is 1.2 × 10−5. This is a slight improvement over the SRP's value of 1.4 × 10−5.

2.3.5. Uncertainty Budget

[24] Table 1 shows the uncertainty budget for the laser photometer at a nominal ozone mole fraction of 1000 nmol mol−1. For the purpose of comparison with the SRP based on a mercury lamp, the uncertainty on the ozone absorption cross section, which would be dominant and highly correlated, is not taken into account here.

Table 1. Uncertainty Budget of the Laser-Based Photometer When Measuring 1000 nmol mol−1 of Ozone in Air
ParameterTypical ValueStandard UncertaintyRelative Uncertainty
  • a

    Without the absorption cross section.

Temperature, T295 K0.061 K2.1 × 10−4
Pressure, P1000 mbar0.64 mbar6.4 × 10−4
Optical length, Lopt893.9 mm0.4 mm4.5 × 10−4
Product of transmittances, D0.951.2 × 10−52.6 × 10−4
Combined relative uncertaintya  8.5 × 10−4

[25] Figure 2 shows the total uncertainty of the laser photometer in comparison with the total uncertainty of an SRP when not taking into account the uncertainty because of the cross section. This graph shows a clear improvement for the uncertainty because of the improved optical path length determination.

Figure 2.

Combined standard uncertainty u(x) (without the ozone cross-section component) as a function of measured ozone amount fraction x with the laser photometer (red line) and with the SRP (black dashed line).

2.4. Performance of the Laser Ozone Photometer When Compared With a Standard Reference Photometer

2.4.1. Measurement Scheme

[26] To better characterize the laser ozone photometer, the instrument was compared by direct comparison with the SRP (BIPM-SRP31), while measuring on the same source of ozone. This is similar to what is typically done during SRP comparisons. Ozone is generated from purified dry air with a commercial ozone generator based on the principle of photolysis of oxygen using a mercury lamp. By adjusting the lamp intensity, the generator can generate between a few and 1500 nmol mol−1 of ozone in air. A flow of pure dry air and the ozone mixture are independently sent to two separate parts of a Pyrex manifold vented to the atmosphere from which both photometers are sampled. In each system, one cell is filled with the ozone mixture and the other cell is filled with pure air. Following a measurement taken over 5 s, the gases are reversed and another 5 s measurement is made. It takes about 8 s to purge the cells. A purge time of 10 s was used for both instruments.

2.4.2. Noise Level

[27] Figure 3 shows the noise of the ozone concentration as measured by the laser photometer in comparison with the SRP31 measurement on the same source of ozone in terms of the relative Allan deviation. The Allan deviation is derived from the Allan variance or two-sample variance introduced by D. W. Allan in time and frequency metrology to characterize the stability of frequency standards [Allan, 1987]. When displayed as a function of the sampling time on a log-log plot, a system showing a white noise behavior will typically be characterized by a linear decreasing function. Here, after 200 s, the noise is dominated by the instability in ozone production. For shorter durations, the laser system appears to be slightly less noisy than the SRP. The calculated ratio between these two measurements gives the combined noise of the two systems without the common noise because of, for example, instabilities in ozone production. This is shown in Figure 4, where it is compared with a similar measurement by comparing two SRPs (BIPM-SRP27 and BIPM-SRP28).

Figure 3.

The relative Allan deviation, σR,Allan (Allan deviation divided by the mean), as a function of the averaging time calculated on the ozone concentration measured by the laser photometer (set at 248.32 nm) and by SRP31, with ozone (in air) from the same source at an amount fraction of about 1000 nmol mol−1.

Figure 4.

The relative Allan deviation σAllan of the ratio x1/x2 between the laser ozone photometer measurements and SRP31 (red line) and between two other BIPM SRPs (black dashed line) as a function of the averaging time.

[28] This shows that the combined noise of SRP31 and the laser photometer is approximately the same when comparing the two SRPs indicating that, for longer time scales (up to ∼8000 s), the laser photometer is as stable as an SRP. The results given in Figures 3 and 4 are from measurements made on a nominal ozone mole fraction of 1000 nmol mol−1, with the laser wavelength adjusted to 248.32 nm.

2.4.3. Linearity Relative to Ozone Concentration

[29] To ensure that the laser ozone photometer measurements are linear as a function of the ozone concentration (as measured with BIPM-SRP31), 10 measurements at 8 different concentrations between 0 and 1000 nmol mol−1 of ozone were performed. The laser wavelength was 244.1 nm during this exercise. At each concentration, the average of the 10 measurements is calculated. A linear fit is made from the 8 averages and the difference of each of the 8 values from the linear fit (residuals) is displayed in Figure 5. All residuals are much lesser than the uncertainties, and no particular nonlinear trend appears, demonstrating the linearity of the laser photometer.

Figure 5.

Residuals (difference between measured and fitted values) of the linear fit of averages of 10 measurements performed with the laser ozone photometer, at 8 different ozone concentrations xSRP31(O3) as measured with BIPM-SRP31.

2.4.4. Linearity Relative to Light Power

[30] By using a laser as the light source, the intensity of the light inside the cells can be higher than when using a mercury lamp as the light source. The light power density in our laser ozone photometer is typically less than 10 μW cm−2, when compared with the power density in the SRP, which is less than 1 μW cm−2. To verify that there are no significant nonlinear effects because of a possible higher light intensity, the linearity of the laser ozone photometer was evaluated, when increasing the light power. This was carried out at a laser wavelength of 257.34 nm, as this is where the laser gives its maximum power.

[31] Figure 6 shows the relative difference between the ozone concentration values measured with the SRP and the laser ozone photometer, while the light power measured by the photodiodes is increased from about 3 to 17 μW. If it is assumed that the beam completely fills the entire 3 mm diameter diaphragm holes, this corresponds to about 10 to 60 μW cm−2.

Figure 6.

Relative difference between the ozone concentration values measured with the SRP31 (x31) and the laser ozone photometer (x32) when the laser power density is varied between 10 and 60 μW cm−2 at 257.34 nm.

[32] As shown here, no particular influence of the light intensity is observed, and the deviations are smaller than the experimental standard deviation obtained with a fixed light power. This demonstrates the absence of significant nonlinear effects because of the power density of the light from the laser ozone photometer.

3. Ozone Absorption Cross-Section Values at the Three Laser Wavelengths

[33] The laser photometer can measure the ozone concentration based on the absorption of ozone at the following wavelengths: 244.06, 248.32, or 257.34 nm. The SRP, on the other hand, is based on absorption at the mercury line wavelength of 253.65 nm. Therefore, by comparing the measured values of the two systems sampling the same source of ozone, it is possible to deduce values of the ozone absorption cross section at the laser photometer wavelength, relative to a specific absorption cross-section value at 253.65 nm, using equations (4) and (5). When doing so, the uncertainty associated with the deduced ozone cross section is the combination of the measurement uncertainty of both instruments, taking into account the uncertainty in the ozone cross-section value chosen as a reference in the SRP.

[34] The standard reference value implemented within SRP measurements is the value measured by Hearn in 1961 and is equal to 1.1476 × 10−17 cm2 molecule−1, with a relative expanded uncertainty of 2.12% [Hearn, 1961] (at 95% level of confidence). Using that value as the reference absorption cross section, deduced values of the ozone absorption cross section at the three laser wavelengths are summarized in Table 2. The expanded uncertainty on the wavelengths is 0.001 nm, as provided by the wavemeter. The same table also shows the ozone cross-section values if the value measured by Brion, Daumont, and Malicet (also called BDM values), that is, 1.128 × 10−17 cm2 molecule−1, with an expanded uncertainty of 1.3% is used as the reference within the SRP [Brion et al., 1993, 1998; Malicet et al., 1995].

Table 2. Values of the Ozone Absorption Cross Section at the Three Wavelengths of the Argon-Ion Laser, When Using Two Different Reference Values of the Absorption Cross Section at the Mercury Line Wavelengtha
Laser Wavelength (and Expanded Uncertainty)/(nm)Ozone Absorption Cross Section (and Expanded Uncertainty)/(10−18 cm2 molecule−1)
Using Hearn's Value as a ReferenceUsing Malicet's Value as a Reference
  • a

    Expanded uncertainties combined with measurement uncertainties of the SRP, the laser photometer, and the uncertainty on the ozone absorption cross-section values used as a reference.

244.062 (0.001)9.64 (0.20)9.50 (0.12)
248.323 (0.001)10.61 (0.22)10.45 (0.14)
257.337 (0.001)11.26 (0.24)11.12 (0.15)

[35] When taking into account the uncertainty of the chosen cross-section value at the mercury line wavelength, this component dominates the total uncertainty on the ozone cross section at the three laser wavelengths. In this case, values deduced from the two different references agree within the uncertainties. However, it is interesting to look at the same results without the uncertainty on the reference ozone cross section and to compare with results from other groups.

[36] Figure 7 shows these values compared with other values from the literature, recently made available in electronic format by the ASCO committee. These include measurements performed by Bogumil [Bogumil et al., 2001, 2003] and Bass and Paur [Bass and Paur, 1984; Paur and Bass, 1985], who also used the Hearn value from 1961 as a reference at 253.65 nm. These groups used a similar installation to the one presented here, in the sense that they performed the light absorption measurements on a sample of ozone in air, using a UV photometer (such as the SRP), to obtain the ozone concentration. Also displayed are the measurements performed by Burrows [Burrows et al., 1999], who also performed light absorption measurements on a mixture of ozone in air, but by a different method, to obtain the ozone concentration. The value was obtained by gas-phase titration between the ozone sample and nitrogen monoxide, NO, using the absorption cross section of the resultant nitrogen dioxide, NO2, to deduce its concentration and hence, the ozone concentration. The NO2 absorption cross section itself was measured by the same group using the same instrument. Finally, Figure 7 displays measurements performed by Brion, Daumont, and Malicet [Brion et al., 1993, 1998; Malicet et al., 1995], in which the light absorption measurements were undertaken in “pure” ozone at low pressure. Pressure measurements were used to calculate the ozone concentration, taking into account its decomposition in oxygen.

Figure 7.

Ozone absorption cross-section values, σ, at (a) 244.06 nm, (b) 248.32 nm, and (c) 257.34 nm. Red crosses show the values obtained relative to the values of Hearn (BIPM1) and Malicet (BIPM2) for the ozone absorption cross section at 253.65 nm. Measurements of Burrows (B97) [Burrows et al., 1999], Malicet (BDM93) [Brion et al., 1993, 1998; Malicet et al., 1995], Bass and Paur (BP84) [Bass and Paur, 1984; Paur and Bass, 1985], and Bogumil (Bog04) [Bogumil et al., 2001, 2003] are also shown. Expanded uncertainties (at 95% confidence level) do not take into account the uncertainty of the reference absorption cross section (Hearn or Malicet).

[37] In Figure 7, the error bars attached to the laser photometer values are the combined expanded uncertainty of the laser photometer and the SRP, excluding the uncertainty because of the reference cross-section value. It is clear from Figure 7 that the laser photometer value, based on the Hearn value [Hearn, 1961], fits well with the two measurement series based on the same reference value at 253.65 nm and also the Burrows value based on gas-phase titration. Our values also agree with the data set taken from Malicet's group [Brion et al., 1993, 1998; Malicet et al., 1995], when using the same data set (BDM) as a reference at the mercury line wavelength, although there seems to be an unexplained constant positive bias of about 0.5%.

[38] It should be emphasized that although different light sources and different optical setups were used in the four experiments (including our work) where measurements were performed relative to the value measured by Hearn at the mercury line [Hearn, 1961], they all agree within 0.5%. This confirms that photometric measurements are not the major source of uncertainty in ozone absorption cross-section measurements. The main issue is to accurately determine the ozone concentration by an independent method. The 2% difference observed between the values from Malicet's group and the others is a systematic bias, which can be seen in almost the entire Hartley band (there is an unexplained jump after 260 nm where all the data converge). This can be explained by the differences in the setup procedures used to produce pure ozone in the experiments performed by Hearn in the 1960s and by Malicet in the 1990s. Absolute measurements performed with the lowest-possible uncertainty are required to help solve this discrepancy.

4. Conclusion

[39] We have developed a laser ozone photometer capable of measuring ozone concentrations based on absorption of light at three different wavelengths in the Hartley band: 244.06, 248.32, and 257.34 nm. The system was used to perform new relative measurements of the ozone absorption cross section at three available wavelengths, by taking advantage of our usual setup procedure for ozone photometer comparisons. The measured values at the three wavelengths are in agreement with other published values, when using the same reference value of the ozone absorption cross section at the mercury line wavelength (253.65 nm), even when excluding the uncertainty on this reference value in our measurement uncertainties. This confirms the accuracy of photometric measurements, independent of the type of light source used.

[40] The use of a laser as the light source, instead of the traditional mercury lamp, leads to a reduction in the uncertainty associated with the path length of the light in the instrument gas cells, which is the second major uncertainty component of SRPs after the ozone absorption cross section.

[41] A complete uncertainty budget for the laser ozone photometer is presented, as well as results of validation studies that demonstrate the suitability of the instrument for accurate measurements of the ozone mole fraction in air between nominal values ranging from 0 to 1000 nmol mol−1, which has allowed new values for absorption cross sections of ozone to be determined.

[42] The laser ozone photometer will be used in conjunction with a setup currently under construction, in which the concentration of pure ozone will be assessed by pressure and purity measurements, allowing new absolute measurements of the ozone absorption cross section to be performed at the same three argon-ion laser wavelengths.

Ancillary