FIGURE 11.9 Schematic diagram of nondispersive infrared device (adapted from Skoog et al., 1998).

in the second step. Water must be removed either before or after collection of the sample to minimize its contribution to absorption and scattering of IR. The cryogenically trapped air sample is then transferred to a low-temperature window for infrared analysis, usually by reflection-absorption spectroscopy. The C02 matrix is used as an internal standard, and because its concentration in air is well known (see Chapter 14), the concentrations of the trapped gases can be obtained from the strength of their infrared absorptions compared to those of C02.

Table If.2 also shows the detection limits for some atmospheric gases using MI infrared spectoscopy and a f5-L air sample (Griffith and Schuster, 1987). Clearly, this technique can measure quite small concentrations, typically in the ppt range. The disadvantage is that in the configuration used to date, samples must be collected and brought back to the laboratory for analysis. As a result, it is not a "real-time" measurement, as is the case for FTIR and TDLS. In addition, the possibility of reactions during sampling and transfer onto the analysis window must be considered.

d. DOAS (UV-Visible Absorption Spectroscopy)

(1) Basis of technique Because of the relatively large absorption cross sections in the UV and visible for many gases of atmospheric interest, use of absorption spectroscopy in this region presents an obvious analytical approach. In the case of laboratory studies, measurement of the light intensity in the absence (I0) and presence (/) of the species of interest is readily applied to obtain concentrations using the Beer-Lambert law (see Chapter 3.B):

where a is the absorption cross section (cm2 molecule-1), N is the concentration (molecules cm-3), and L is the path length.

However, the fact that so many species in air absorb in this region presents a limitation in that one must be able to distinguish various species from each other as well as from background broad absorption and Rayleigh and Mie scattering of light by gases and particles. Because of these factors, UV-visible spectroscopy is, in practice, applied in air only to those species with banded structures, i.e., "fingerprints," of width ~5 nm or less. The technique used to do this is differential optical absorption spectrometry (DOAS). For reviews of DOAS, see Piatt (1994) and Plane and Smith (1995).

Figure 11.10 illustrates the basis of this technique for a species that has narrow absorption bands at wavelengths Aa and Au, superimposed on a slowly varying background. Because of Rayleigh and Mie scat-


FIGURE 11.10 Light intensities relevant to DOAS spectrometry.

tering, the "true" 70 shown by the upper dashed line, i.e., the intensity in the absence of air, cannot be measured. Scans of this spectral region do allow the broad background /,',, however, to be interpolated from the measurements of /(A). Thus, rather than measuring (/„//), the ratios (/,f'//A) and ('//u) are measured and used to obtain the concentration of the absorbing species. That is, one is measuring the differential optical absorption (D) rather than the true optical absorption (A). However, this can be used for measuring concentrations as well since the differential optical absorption also follows a Beer-Lambert relationship:

In this case, cr' is the differential optical absorption cross section for the absorption band. In practice, of course, there are many different absorbers, i, present at different concentrations Nj and absorbing at different wavelengths over the path length L.

Returning to Fig. 11.10, the relationship between I and the "true" /„ can be expressed as

/(A) =/0(A)^(A)e'-''[ECT'(A)N' + £«(A)+e«(A)]'. (C)

In Eq. (C), ^4(A) is an attenuation factor characteristic of the measurement system, and sM are the equivalent extinction coefficients due to Rayleigh and Mie scattering of gases and particles, and 0; are the total absorption cross sections of the absorbing gases, all of which are wavelength dependent. Although the Rayleigh and Mie scattering contributions are not absorption processes, their contributions to the reduction

in light intensity can be treated for DOAS measurements as if they were. The value of eR(A) is 1.3 x 1CT6 cm 1 at 300 nm for STP conditions, reducing the light intensity by about 12% in each kilometer. The value of eM(A) strongly varies with aerosol loading. Typical values at 300 nm range from f x 10~6 cm-1 for clean maritime air (without sea spray) to ~10~5 cm-1 for rural continental air. However, fog or heavy pollution can limit the application of DOAS because of the associated high values of the extinction.

The total absorption cross sections (tr,) of a single trace gas i can be broken down into a contribution from the structured portion, <r/, and one from the broadband portion that varies only slowly with wavelength, o-jB:

Substituting into Eq. (C), one obtains

Taking natural logarithms, the differential optical ab-sorbance (D') is given by

A major advantage of DOAS is its high sensitivity for species that meet the requirement of having narrow absorption bands in the UV-visible. Furthermore, because the differential optical absorption coefficients are fundamental spectroscopic properties of the molecule, the measurements need not be calibrated in the field.

(2) Analysis of spectra Different approaches to spectral analysis are described by Piatt (1994) and Plane and Smith (1995). Calibration spectra of the absorbing species must be available for fitting the DOAS spectra. These spectra are usually obtained using the same instrument and settings. However, literature spectra of the same or higher resolution can be used if they are converted to the same resolution as used in the measurements.

To quantify the measured spectra, a combination of linear and nonlinear least-squares fitting routines are used, in which the measured intensities are fit to those of scaled reference spectra while minimizing the residual absorbance. Taking the natural logarithm of Eq. (E), one obtains

This is of the form

where a • are scaling factors for each species j chosen to give the best fit to the total spectrum and 5- are the known reference absorption spectra of each of the species. It has been observed that the term P( A) in Eq. (H), which contains the components that vary slowly with wavelength, i.e., /„(A), A(X), eR(A), eM(A), and cr;u(A), can be approximated by a polynomial function of the form P(A) = Ea„ A", where n is typically ~5. Thus, In /(A) is fit using least-squares analysis with combination of a polynomial and the second term to obtain the scaling factors aj. From these scaling factors and the known path length, L, the concentration of the absorber j can be calculated. Care must be taken to ensure that the wavelengths are properly calibrated (e.g., using a low-pressure Hg lamp) and that small drifts in the spectra due to thermal drift (typically ~0.1 pixel K"1) are taken into account. In addition, changes in air pressure can cause shifts, ~0.2 pixels in going from 1000 to 750 mbar. Such problems and the details of analysis of DOAS spectra, including methods of error estimation, are discussed by Stutz and Piatt (1996).

(3) Typical apparatus Figure 11.11 is a schematic diagram of the components of a typical DOAS system. A broadband light source is needed, which, for example, can be a high-pressure Xe or incandescent quartz-iodine lamp, a broadband laser, or the sun or moon. The light traverses the air sample, either in a single-pass system or in a multipass system using an open White cell. The light strikes the entrance slit of a spectrograph which disperses the radiation. Detection as a function of wavelength of the dispersed light is carried out using a slotted-disk mechanism or, more commonly, a photodiode array (PDA) or charge-coupled device (CCD).

The use of the sun or moon as the light source allows one to measure the total column abundance, i.e., the concentration integrated through a column in the atmosphere. This approach has been used for a number of years (e.g., see Noxon (1975) for N02 measurements) and provided the first measurements of the nitrate radical in the atmosphere (Noxon et al., 1978). As discussed later in this chapter, such measurements made as a function of solar zenith angle also provide information on the vertical distributions of absorbing species. Cloud-free conditions are usually used for such measurements; as discussed by Erie et al. (1995), the presence of tropospheric clouds can dramatically increase the effective path length (by an order of


Light source

single, double or entrance slit multipass open cell

Slotted disc with PM, or PDA, or CCD

grating exit slit


Was this article helpful?

0 0

Post a comment