- Sample cell
Fig. 3.2 Principles of measurement of absorption coefficient of non-scattering and scattering samples. The instrument is assumed to be of the double-beam type, in which a rotating mirror is used to direct the monochromatic beam alternately through the sample and the blank cells. (a) Operation of spectrophotometer in normal mode for samples with negligible scattering. (b) Use of integrating sphere for scattering samples.
and the sample cell are taken to be proportional to I0 and I, respectively, and so the logarithm of the ratio of these intensities is displayed as the absorbance of the sample.
When used in their normal mode (Fig. 3.2a), however, spectrophotometers have light sensors with a collection angle of only a few degrees. Thus any light scattered by the sample outside this collection angle will not be detected. Such spectrophotometers do not, therefore, measure true absorbance: indeed if their collection angle is very small, in effect they measure attenuance (absorption plus scattering, see §1.4). If, as is often the case, the sample measured consists of a clear coloured solution with negligible scattering relative to absorption, then scattering losses are trivial and the absorbance displayed by the instrument will not differ significantly from the true absorbance. If, however, the pigmented material is particulate, then a substantial proportion of the light transmitted by the sample may be scattered outside the collection angle of the photo-multiplier, the value of I registered by the instrument will be too low, and the absorbance displayed will be higher than the true absorbance.
This problem may be solved by so arranging the geometry of the instrument that nearly all the scattered light is collected and measured by the photomultiplier. The simplest arrangement is to have the cells placed close to a wide photomultiplier so that the photons scattered up to quite wide angles are detected. A better solution is to place each cell at an entrance window in an integrating sphere (Fig. 3.2b): this is a spherical cavity, coated with white, diffusely reflecting material inside. By multiple reflection, a diffuse light field is set up within the sphere from all the light that enters it, whether scattered or not. A photomultiplier at another window receives a portion of the flux from this field, which may be taken to be proportional to I or I0 in accordance with whether the beam is passing through the sample, or the blank cell. With this system virtually all the light scattered in a forward direction, and this will generally mean nearly all the scattered light, is collected.
Another procedure, which can be used with some normal spectropho-tometers, is to place a layer of scattering material, such as opal glass, behind the blank and the sample cell.1211 This ensures that the blank beam and the sample beam are both highly scattered after passing through the cells. As a result, the photomultiplier, despite its small collection angle, harvests about the same proportion of the light transmitted through the sample cell, as of the light transmitted through the blank cell, and so the registered values of intensity are indeed proportional to the true values of I and I0, and thus can give rise to an approximately correct value for absorbance.
With all these procedures for measuring the absorbance of particulate materials, it is important that the suspension should not be so concentrated that, as a result of multiple scattering, there is a significant increase in the pathlength that the photons traverse in passing through the sample cell. This would have the effect of increasing the absorbance of the suspension and lead to an estimate of absorption coefficient that would be too high. Even with an integrating sphere, some of the light is scattered by the sample at angles too great for collection: a procedure for correction for this error has been described.701
An instrument based on completely different principles to any of the above is the integrating cavity absorption meter,357,418,419 (ICAM). The water sample is contained within a cavity made of a translucent, and diffusely reflecting, material. Photons are introduced into the cavity uniformly from all round, and a completely diffuse light field is set up within it, the intensity of which can be measured. The anticipated advantages are, first, that since the light field is already highly diffuse, additional diffuseness caused by scattering will have little effect; second, because the photons undergo many multiple reflections from one part of the inner wall to another, the effective pathlength within the instrument is very long, thus solving the pathlength problem. A prototype instrument gave very satisfactory results.419 In an alternative version of the ICAM -the point source integrating cavity absorption meter (PSICAM) - the cavity is spherical, and a uniform light field is achieved by introducing the photons from a centrally placed point light source.715,783 In this case also prototype instruments have given excellent results.1150
Most seawater, and many freshwater, samples have absorption coefficients (above those of pure water) too low to measure accurately in laboratory spectrophotometers An approach pioneered by Yentsch (1960), applicable to the particulate fraction is to pass the water through a filter - usually glass fibre - until enough particulate matter has accumulated, and then measure the spectrum of the material on the filter itself without resuspension, using a moist filter as a blank. To correct for the additional attenuation due to scattering by the particles on the filter the value of apparent absorbance at some selected wavelength (715-750 nm) in the near infrared, where it is assumed that true absorption is negligible, is subtracted from absorbance at all other wavelengths.
Implicitly it is assumed that the substantial attenuation due to scattering by the filter itself is the same for both the sample and blank filters. This assumption is not necessarily correct. The particles collected on the sample filter, as well as adding to total scattering may change its angular distribution and thus alter the proportion of transmitted light captured by the photomultiplier. To reduce this error, Tassan and Ferrari (1995) have developed a variation on the procedure - the 'transmittance-reflectance' method - in which, as well as measuring absorbance in the usual manner, spectral reflectance is also measured and used to correct for backscatter-ing. This procedure has been found to reduce the variability of the measurements of absorption coefficients of particulate matter in the somewhat turbid coastal waters of the Adriatic.1346
The most fundamental problem with the filter method, however, is that, as a consequence of multiple internal reflection within the particu-late layer, there is very marked amplification of absorption, the amplification factor being commonly indicated by b. To calculate the true absorption coefficients that the particulate material has, when freely suspended in the ocean or lake, the absorption amplification factor must be determined reasonably accurately, and this is not easy. The approach has generally been to find correction factors by growing phytoplankton species in the laboratory and comparing optical densities measured on filters with those obtained on suspensions in cuvettes.245 The most commonly used algorithm in recent years makes use of a quadratic relationship, first arrived at by Mitchell (1990)
00(2)Corrected - C1 0D(1)measured + C2 0°2(1)measured (3'5)
where C1 and C2 are empirically determined coefficients. Equation 3.5 implies that the value of b, the amplification factor, varies with optical density.
Roesler (1998) points out that if, as seems plausible, a completely diffuse light field is established within the glass fibre filter, then the average cosine of the forward-transmitted photons should be 0.5, which would correspond to a pathlength amplification factor of 2.0. Her investigations suggest that much of the variability encountered in using this method arises from the fact that the optical properties of the filter pad itself are markedly affected by the volume of water filtered through it, and it is typically the case that a much larger volume is passed through the sample filter than the blank filter. With the appropriate filter preparation, the assumption that b = 2.0 gave satisfactory results in this study.
Allali et al. (1995) have developed a method that, while still making use of filtration to concentrate the particles, avoids the optical problems associated with the filter itself. In their filter-transfer-freeze method the samples are filtered through a polycarbonate filter with 0.4 mm pores, so that the particles collect on the surface. The loaded filter is then transferred, particle side down, onto 5 ml of sea water on a glass microscope slide and quickly frozen using liquid nitrogen. The filter is then peeled off, leaving the frozen layer of particle suspension attached to the slide, and replaced with a glass cover slip. After thawing, the absorption spectrum is measured using an integrating sphere.
A different experimental approach has been developed by Iturriaga and Siegel (1989), who use a monochromator in conjunction with a microscope to measure the absorption spectra of individual phytoplankton cells or other pigmented particles. If sufficient particles are measured, and their numbers in the water are known, then the true in situ absorption coefficients due to the particles can be calculated. While laborious, this does avoid the uncertainty associated with estimation of the amplification factor in the filter method.
For measurements of soluble colour, scattering problems can be largely eliminated by filtration, so that measurements over long pathlengths can be used. D'Sa et al. (1999) used a 0.5m long, 550mm internal diameter, capillary waveguide (see reflective tube principle, below) to measure the absorption spectrum of dissolved organic matter in seawater samples.
As we shall see later, the absorption coefficients of the different constituents of the aquatic medium can be determined separately. The value of the absorption coefficient of the medium as a whole, at a given wavelength, is equal to the sum of the individual absorption coefficients of all the components present. Furthermore, providing that no changes in molecular state or physical aggregation take place with changes in concentration, then the absorption coefficient due to any one component is proportional to the concentration of that component (Beer's Law).