In a diode array, the optical configuration is reversed from that in a conventional spectrophotometer, and the light beam passes through the sample before being dispersed by the polychromator. This gives the benefit of fast, full spectral data with no moving parts that can wear out. The perception is therefore that diode arrays are more reliable than other detectors. This is true, but the reverse-optics design requires that both the sample and optics before the dispersing element (usually a grating) are subjected to the full-spectrum radiation also encountered by these components in a conventional spectrophotometer. In a conventional spectrophotometer, the sample and optics are outside the monochromator, and optical beam deflection and/or scattering simply reduces the intensity of the light that enters the monochromator. In an array-based spectrophotometer, deflection occurs within the monochromator, causing scatter within the confined environment and an associated increase in stray light that leads to a reduction in optimum photometric range.

The design of a UV-Vis spectrophotometer is always the result of a number of considered compromises. For example, optical performance (absolute wavelength resolution) is governed by the focal length of the monochromator, which in turn dictates the physical size of the instrument. In the case of HPLC detectors the need to provide a high-stability, low signal-to-noise ratio output at high transmittance levels through a small-aperture flow cell requires that by design these systems may not have the dynamic range or wavelength accuracy of their conventional spectroscopic counterparts.

In recent years there has been a rapid increase in the availability and use of systems built around the ability of fiber optics to channel and multiplex optical systems. Although these systems have the advantages of flexibility and ease of use and they allow measurements to be performed on micro-plates, customized systems, etc., analysts must consider the following disadvantages:

Because most spectrophotometric procedures require good spectral resolution, the spectral bandwidth is of great importance. Analysts should use the narrowest slit width that will provide an adequate signal-to-noise ratio because optimum resolution is achieved when the signal-to-noise ratio is maximized. If access to a variable-bandwidth instrument is available, then the optimum setting can be defined as the largest bandwidth at which no significant reduction in peak intensity is observed. In practical terms for molecules of pharmaceutical interest in solution, a spectral bandwidth of 2 nm is considered adequate.

Stray radiation, commonly referred to as stray light, can be defined as radiant energy at wavelengths other than those indicated by the monochromator setting and all radiant energy that reaches the detector without having passed through the sample or reference solutions (see Figure 4). It may be caused by any scattered radiation from imperfections in the dispersing medium, which commonly is a grating. The use of a holographic grating substantially reduces the levels of this source of stray radiation. Higher-quality ruled gratings yield a low level of scattered radiation. In higher-performance instruments, stray radiation can be reduced by the use of double monochromators or double-pass monochromators. Stray radiation, or apparent stray radiation, also may be caused by light leaks in the system, incorrect wavelength calibration, incorrect optical alignment, reduced source output, or reduced detector response.

Determination of the optimum photometric range is fundamental to establishing the capability of a given instrument before and during method validation. The importance of this procedure can be shown by the fact that in 1945 Vandenbelt et al. (1) attempted to establish the optimum absorbance range on the then-new Beckman DU spectrophotometer. They suggested a simple approach in which the molar absorptivities of various compounds are measured at different concentrations. This approach is shown theoretically in Figure 5, where the center plateau is used to define the photometric range.

In practice, data are more variable. Figure 6 shows three typical curves.

A more rigorous method for estimating an instrument's region of maximum precision is given by Youmans and Brown (3). The error in the measurement of the concentration of a solution of transmission T can be expressed as:

where Dc is the change in concentration, is the molar absorptivity of the compound, b is the path length, and Tav and Tmin are the mean and minimum values, respectively, from a series of n measurements on the same sample.

The statistical error is (see Figure 7):

The noise levels, and hence the precision of measurement, depend primarily on detector noise type, as shown in Figure 8.

Both approaches clearly show that in addition to the expected increasing error at higher absorbance levels caused by stray light and other factors, there is a similar, if not larger possibility for increasing error at low absorbance levels caused by instrumental variances in the form of noise from the detector, source, etc. This variance often is overlooked in the desire for economy of scale, when analysts use smaller samples and shorter path lengths and therefore lower measured absorbance values.

Quantitative absorbance measurements usually are made on solutions of the substance in liquid-holding cells. The most common cell path length is 10 mm, although path lengths from 0.01 to 100 mm are commercially available (see Figure 9). For samples with low absorbance, improved sensitivity generally can be obtained by increasing the cell path length. For example, the theoretical absorbance of a solution in a 50-mm cell is greater by a factor of five compared to the same solution in a 10-mm cell. Errors in absorption readings arising from cells almost always are caused by dirty windows that can absorb a significant proportion of the incident light beam. Less frequent causes of error are an incorrect choice of cell material for the wavelength required, e.g., use of glass cells less than 320 nm, nonrepeatability of cell positioning, differences in cell window thickness, nonparallel optical windows, or impurities in cell window materials. A 10-mm cell manufactured with a ± 0.05-mm tolerance will add a 0.005 A contribution to the overall uncertainty, when measuring absorbance values at the 1-A level. Ten-mm cells with a ± 0.01-mm tolerance are commercially available. As a general guide, the cell chosen should have a transmittance value of > 75% when filled with the appropriate solvent and measured at the wavelength of interest.

Analysts can mitigate the variability introduced by cells by adhering to the following set of best practices. Carefully clean cells before use, and store them in a manner that avoids contamination when they are not in use. Cells in frequent and repeated use, e.g., flow cells, can be stored wet in high-purity water or 1% (v/v) nitric acid. Do not use cracked or scratched cells. Wipe cells carefully with a soft, clean, lint-free cloth before they are placed in the spectrophotometer. Do not use lens-cleaning paper or any other form of paper cleaning tissue because these will either scratch or allow contamination of the optical face.

It is preferable to use a single cell for all measurements of both the reference and samples. When several sample cells must be tested against one reference cell, they all should be closely matched in their characteristics. Whenever possible, measure standards and samples at the same time and under the same conditions to minimize the potential for bias. The reference cell should contain the same solvent as the sample cell and should be checked against the sample cell at all wavelengths at which measurements are required. Analysts can apply corrections for differences between cells to absorbance readings obtained from the solutions under examination. The maximum allowable correction is ± 0.01 absorbance unit. Best practices include using the same cell in the reference beam with the same face incident to the light beam, and flushing the sample cell three times with the sample solution before the final filling before measurement. The use of flow cells effectively addresses many of these cell-handling and cell-filling issues. The contents in both the sample cell and the reference cell must be free from gas bubbles and particles. If flow cells are used, analysts may be required to degas the solution(s) before measurement in order to avoid the formation of air bubbles in the cell.

A sequence of activities is used to monitor the quality of the measurement process and to ensure that the contents are representative of the sample under measurement. Drift of the spectrophotometric baseline can be detected by measuring the blank at the start and end of the sample sequence. Adequate flushing of the cell can be confirmed by alternating between measurement of the standard (high concentration) and the blank (zero concentration) at the start of the measurement sequence. The latter check is particularly important for automated systems that can generate reproducible but inaccurate results because of inadequate flushing.

Contaminated cells are the greatest single source of error in spectrophotometry. Cells should never be handled by the optical polished faces, and analysts should rinse off residual or spilled solution. If cells are properly cared for, rigorous cleaning methods rarely will be needed, especially if the cells are cleaned immediately after use. Although acid solutions will simply contaminate surfaces, alkaline solutions can etch all types of polished glass surfaces, and the degree of attack depends on the pH and contact time. When cleaning cells, analysts should ensure that the optical faces are not scratched or chipped. Commercial cleansing solutions are available and can be used, provided that they are diluted before use in accordance with the manufacturers' recommendations. After they are cleaned with such agents, the cells should be carefully and thoroughly rinsed with high-purity water. If this fails to clean the cells, then soaking them in cold, concentrated nitric or hydrochloric acid is acceptable. Flow cells are best cleaned in situ, provided that the solvent does not interact with the connecting tubing. If the cells are stored for an extended period, then they should be dried quickly after cleaning by blowing dry with a compressed gas stream in a clean, dust-free environment.

Analysts should align the cells so that the same optical face is always incident to the light beam. Most cells have type identification engraved on one face: this can be used to ensure consistent orientation. If there are no markings, a small mark can be made on one face outside the area of the light beam. The use of cells with no type markings is not recommended, but if they are used analysts should confirm their suitability for the intended application. If cells are provided with the instrument, individual cell holders should always be used in the same beam as identified, e.g., by use of appropriate markings.

If multiple cells are used, differences in the optical transparency of individual cells can introduce a systematic bias unless the latter is accounted for by a cell correction. Do not use corrections in excess of 0.01 absorbance unit. To determine if cell corrections are necessary, analysts can fill all cells with the appropriate solvent and measure the differences in absorbance at the required analytical wavelengths. These measurements are repeated after cleaning to ensure that differences in absorbance are related to the cells and are not the result of contamination. Analysts should fill and check the cells repeatedly until the results are consistent, and they should investigate any appreciable change in cell correction because it indicates contamination, damage to the cells, or incorrect adjustment of the instrument.