Carryover of Sampling Errors and Other Problems in Far-Infrared to Far-Ultraviolet Spectra to Associated Applications

The thin, smooth curves representing spectroscopic data suggest a high degree of accuracy. Yet, experimental uncertainties do exist, as in any measurement. Overlooked problems in data collection, processing, and interpretation have repercussions for applications in mineral physics, planetary science, and astronomy. Random errors (i.e., noise) are fairly obvious, and are not discussed here. The concern is subtle and overlooked errors that arise in acquisition, processing, and interpretation of spectral data. These types of errors are systematic, not random. This chapter identifies various systematic errors and problems that the author encountered in her efforts to provide absolute values of absorbance or reflectivity. Re-occurring issues in data collection include underestimating the importance of surface polish and not accounting for peak profiles depending on sample thickness relative to band strengths. Processing of emission spectra is problematic. Common instrumental problems are briefly described. Optical spectroscopy is the name generally attached to the visible region which we probe with our eyes, which are convenient built-in spectrometers, but can also include the infrared (IR) region wherein the type of vibrational mode known as “optical” is detected. Because some applications require very high frequency (ν) data, this chapter concerns ν from ~10 to 106 wavenumbers, which is equivalent to wavelengths (λ) of ~106 to 10 nm or of ~1000 to 0.01 μm). The X-ray region is included due to the extreme breadths of metal-oxygen charge-transfer bands of minerals which peak in the ultraviolet (UV). The author points out errors in her own results as well those of others. Mistakes provide opportunity for learning! Correct methodologies are discussed along with measurements needed to improve constraints on spectral parameters and hence to make interpretations more definitive. Ideal conditions are difficult to achieve, so another goal is enable the reader to recognize what is “sufficiently accurate” and/or “representative” …