Absorbance Spectroscopy: Spectral Artifacts and Other Sources of Error
- 148 Downloads
Modern spectrophotometers and data acquisition software are usually reliable and easy to use, but the spectra that they produce must always be regarded critically, as there are many possible sources of error and measurement uncertainty. Some of these relate to sample purity and identity. Others relate to the quality and calibration of the instrument, experimental technique, choice of sample cells and wavelength ranges, correct use of software for data acquisition and analysis, appropriate scan parameters that optimize signal-to-noise ratios yet ensure the fidelity of the recorded spectrum, and uncertainties associated with data analysis (e.g., the use of standard curves). A good understanding of spectroscopic principles, and proper instruction and supervision in the use of spectroscopic instrumentation, is essential for reliable spectroscopic work. Other factors and artifacts may also affect the reliability of absorption, linear dichroism (LD), and circular dichroism (CD) spectra, as discussed below. These may lead to distorted spectral band shapes, deviations from the Beer–Lambert law, and unreliable LD and CD data.
A number of issues that affect absorbance data reliability but may not be apparent upon inspection of a spectrum or calibration of an instrument are discussed below.
Wavelength at which the pure solvent absorbance in a 1-cm path length cuvette is 1 (arranged in order of increasing absorbance) according to Honeywell Burdick and Jackson. An absorbance of 1 is an effective cutoff for solvents since most instruments become unreliable between absorbances of 2 and 3 and some sample absorbance must be considered
Wavelength at which A = 1 in 1-cm path length
Absorbances of selected buffer components at 1 mM in a 1-cm path length cuvette. Tris = tris(hydroxymethyl)aminomethane. (Note: TBS often denotes Tris-buffered saline, which is a mixture of 50-mM Tris and 50-mM NaCl adjusted to pH 7.6 with HCl; it is a mixture of components all of which must be taken into account.) HEPES = 2-[4-(2-hydroxyethyl)piperazin-1-yl]ethanesulfonic acid. MES = 2-(N-morpholino) ethanesulfonic acid (Kelly et al. 2005)
Na borate (pH 9.1)
TrisH2SO4 (pH 8.0)
MES/Na (pH 6.0)+
Although, in principle, a double-beam spectrophotometer can correct for a weakly absorbing solvent if it is placed in the reference beam, spectroscopic accuracy and precision deteriorate when the solvent absorbance increases, and stray light (see below) may also degrade the absorbance readings. In the worst case, both the sample and reference cells are opaque at the analytical wavelength, producing a very noisy and unusable spectrum. It is thus usually better to use air in the reference beam and subtract a buffer baseline manually as then one knows how it is affecting the spectrum.
Unless there is no alternative, absorption spectra should not be recorded for turbid samples, as the measured absorbances may be misleading. However, many biomacromolecular species are of comparable size to the wavelength of light, and as a result they scatter light. The situation is worse for linear and circular dichroism as such samples may differentially scatter the two polarizations of light used. The theory of light scattering is complicated, depending not only on the size regime of the particles but also on their shape. Thus, if at all possible, one should avoid it occurring rather than trying to correct for it. In general, scattering can be reduced either by reducing the size of the particles or collecting a high percentage of the scattered photons. Three methods that may enable scattered photons to be collected (and thus to avoid the problem of scattering) are to (1) have a wide-angle photomultiplier tube (PMT), (2) place the sample very close to the photomultiplier tube, or (3) have a collecting lens close to the sample to refocus the scattered light onto the PMT.
However, it is not always possible to remove light-scattering artifacts. In such a case, the measured spectrum is a combination of true absorbance and scattering. The scattering contribution is usually apparent as a sloping baseline outside absorbance bands. Nordh et al. (1986) showed that a simple empirical correction can often be subtracted from an observed linear dichroism spectrum to remove the sloping baseline (Meeten 1981).
However, a more complete treatment as outlined by Hulst (1981) may be required and is challenging.
The same approach may also work for other spectroscopies.
In addition to any light scattering from the sample, so-called stray light (or stray radiant energy) from anywhere in the system may cause additional artifacts especially at lower wavelengths where lamps struggle for light intensity. In such a case, a significant percentage of the light that does reach the photomultiplier detector may be stray light rather than light of the correct wavelength that has passed through the sample. However, all photons get counted as unabsorbed photons at the monochromator’s nominal wavelength, causing the detector to give a misleading indication that there is less absorption than there really is. The main problem is due to imperfections of the monochromator: long-wavelength stray light bypasses the monochromator. This limits the maximum absorbance at which an instrument will give reliable readings. If the spectrum is “true,” it will follow the Beer–Lambert Law (A = ϵCℓ). A sharp drop-off in intensity on the low wavelength side of a band is often indicative of this problem.
Absorbance flattening effects for samples where the light passes through a sample of which half has no absorbance and half has the indicated absorbance
A of clear part of the solution
A of dense part of solution
A apparent of inhomogeneous solution
A for the analogous homogeneous solution
An effect analogous to absorption flattening can be caused by any arrangement that allows a fraction of the light to bypass the sample solution or to pass through a shorter path length of sample. Unusual sample holders such as capillaries are also prone to this problem (Waldron et al. 2010). In some cases, the problem of absorption flattening can be reduced to acceptable levels simply by diluting the whole solution as shown by the lower absorbance lines in Table 3. For membrane systems, one somehow has to remove the local high concentrations, and simple dilution is often not sufficient. Mao and Wallace (1984) have outlined how this may be done by reducing membrane particle size. Unfortunately, the problem cannot always be avoided, in which case methods such as those of Gordon and Holzwarth (1971) can be used to correct measured data.
While not specifically a spectroscopic artifact, photodegradation of biomolecules in a light beam is a very common cause of a measured spectrum not being the true spectrum. Fast preliminary scans are the best way to identify whether this is a problem.