Raman baseline

Normal transmission IRLD can also be used to characterize polymeric fibers, although scattering can induce sloping baselines. Raman spectroscopy then becomes a convenient alternative. Rutledge et al. have recently probed the orientation in electrospun nanofibers composed of a core of Bombyx mori fibroin and an outer shell of poly (ethylene oxide) [24], The orientation values were low, less than 0.1, as is often the case in electrospun fibers. 

The use of surface-enhanced resonance Raman spectroscopy (SERRS) as an identification tool in TLC and HPLC has been investigated in detail. The chemical structures and common names of anionic dyes employed as model compounds are depicted in Fig. 3.88. RP-HPLC separations were performed in an ODS column (100 X 3 mm i.d. particla size 5 pm). The flow rate was 0.7 ml/min and dyes were detected at 500 nm. A heated nitrogen flow (200°C, 3 bar) was employed for spraying the effluent and for evaporating the solvent. Silica and alumina TLC plates were applied as deposition substrates they were moved at a speed of 2 mm/min. Solvents A and B were ammonium acetate-acetic acid buffer (pH = 4.7) containing 25 mM tributylammonium nitrate (TBAN03) and methanol, respectively. The baseline separation of anionic dyes is illustrated in Fig. 3.89. It was established that the limits of identification of the deposited dyes were 10 - 20 ng corresponding to the injected concentrations of 5 - 10 /ig/ml. It was further stated that the combined HPLC-(TLC)-SERRS technique makes possible the safe identification of anionic dyes [150],... 

Figure B3.6.1 Rayleigh and Raman bands in fluorescent spectra, as seen in scans for solvent baseline and hen egg white lysozyme (EWL) solutions (solid lines). Circles represent the spectrum of EWL with baseline subtracted. Parameters EWL A2ao = 0.05 Xex = 280 nm excitation and emission bandwidths, 2.5 nm scan rate, 100 nm/min five scans accumulated. Spectra were measured using a Perkin Elmer LS50B fluorescence spectrometer.

Apart from the Raman band, a buffer solution should give an almost flat baseline of low intensity (see Fig. B3.6.I). At the usual working concentrations of protein (A280 >0.05), buffer fluorescence constitutes only a minor contribution to the fluorescence spectrum of the protein and is accounted for by the appropriate baseline subtraction. 

Figure B3.6.4 The effect of scanning parameters on the authenticity of spectra. The Raman band from Figure B3.6.1 is shown for a single scan at 1500 nm/min (dotted line) ten scans at 1500 nm/min (dashed line) and ten scans at 100 nm/min (solid line). The times required for scanning the complete baselines for these three measurements were 0.087 min, 0.87 min, and 13 min, respectively. Measurements made using a Perkin-Elmer LS50B.

Figure 13.4 Structure of the IRE RNA with A7 and G8 indicated (Hall and Williams, 2004). Top right fluorescence excitation and emission spectra of IRE 2AP7 and IRE 2AP8. Bottom fluorescence emission intensity of each IRE as a function of temperature, 4 jiM RNAs, 30 mM NaCl, 10 mM potassium phosphate, pH 7.0, 20 °C. The buffer baseline was subtracted from each spectrum (the Raman line is at 350 nm).

In many spectroscopic techniques, it is not unusual to encounter baseline offsets from spectrum to spectrum. If present, these kinds of effects can have a profound effect on a PCA model by causing extra factors to appear. In some cases, the baseline effect may consist of a simple offset however, it is not uncommon to encounter other kinds of baselines with a structure such as a gentle upward or downward sloping line caused by instrument drift, or even a broad curved shape. For example, in Raman emission spectroscopy a small amount of fluorescence background signals can sometimes appear as broad, weak curves. 

Raman intensity of the y th point from the baseline-corrected normalized Raman... 

Figure 7-30 Raman spectra of ibuprofen as a pure powder, a white tablet, and a brown tablet. The pure powder spectrum is the same as in Fig. 7-28 however, the baseline was flattened with a five-point function. Both the tablet spectra were measured with an FT-Raman in the near-IR.

Since T2 is readily determined from time-domain CARS with high accuracy (<2%), a combined analysis of frequency- and time-domain data was proposed and demonstrated (45), plotting the spontaneous Raman data in normalized frequency units, Aa> x T2 (note abscissa scale of Fig. 7b). In this way the bandshape only depends on the ratio rc/T2, and only this ratio has to be deduced from the wings of the Raman band. With respect to the experimental uncertainties (ordinate value of the baseline, overlap with neighboring combination tones), the approach is more reliable than the determination of two quantities, rc and T2, from the spectroscopic data. 

The spectral shape of the vibrational band differs from a Lorentzian. The deviation is evident for the wings, a factor of —20 below the maximum, and necessitates an accurate determination of neighboring bands and the baseline level. The isotropic Raman linewidth Sv (FWHM) is predicted to be smaller by a few percent than the Lorentzian width <5vl = (tcT2) 1 (FWHM in wavenumber units) for the same T2 (45). The numbers for the example DMSO in Table 1 (last three lines) give some experimental... 

The SNR for particular measurement is rigorously defined as the inverse of the relative standard deviation of the measured value. For example, the SNR for the peak intensity of a Raman band is the average peak height, S (usually above the baseline), divided by the standard deviation of the peak height (Oy). As with any determination of standard deviation, the accuracy of the SNR improves with the number of measurements averaged ... 

Figure 4.1. Definition of S as the mean signal above baseline for the case of a Raman signal on a non-negligible background (curve A). Curve B is the difference of two successive spectra similar to curve A. The correct SNR is S/ay, determined at the peak intensity. <7, is the standard deviation of spectrum B at the peak of interest, divided by /2.

Figure 5.11. Spectra of polyacrylic acid homopolymer (Carbopol 934p, also carbimer resin, CAS registry 9003-01-4) unpacked powder. Spectrum A is an open powder obtained with the conditions of Figure 5.9, except 100 mW at sample and averaging of forty 3.5 sec integrations. A multipoint baseline was subtracted to yield spectrum A. Spectrum B obtained with a Bruker IFS 66 FT-Raman, germanium detector, 325 mW at sample, 1024 scans, 30 min total acquisition time, 4 cm" resolution.

The first line of Table 10.5 indicates short-term repeatability. The second line indicates the reproducibility of the calibration procedure, based on six recalibrations over a 10-day period. Long-term drift is indicated by the third line, over a period of time selected by the user, with recalibration. These results may be used to verify specifications claimed by the manufacturer and to provide a baseline for future measurements. Table 10.5 provides an assessment of the Raman shift accuracy of a particular instrument, obtained using samples and procedures that should be similar to the intended application. 

FIGURE 3.9 (a) Raman spectrum (A. = 514.5nm) of the carbyne deposit shown in Figure 3.7. Top spectrum as-measured, bottom spectrum baseline subtracted and deconvoluted. (b) UV-Raman spectrum (X = 325nm) recorded at different surface points of the carbyne deposit shown in Figure 3.7. The inset shows the down-shifting of the diamond peak position from 1332 to 1328 cm with decreasing crystallite size. 

Figu re 2.2 (a) Raw spectra of a Raman emulsion layer image (b) Spectra after de-noising by principal component analysis (PCA) (c) Spectra after de-noising and baseline correction by asymmetric least squares. 

Raman spectroscopy, compared to IR spectroscopy, is considered to be easy spectroscopy. Often, it is possible to identify spectral features that are unique to each chemical component in the sample, and to use them for univariate analysis. When the number of spectra in a Raman map is small and the chemical composition of the sample is simple, then a univariate analysis is usually sufficient. The typical analysis strategy would be to pretreat the spectra (baseline correction, nor-malizahon, etc.), explore to identify any unique spectral features, and to create intensity maps of those features as Raman images. When differences between spectral species are reflected in peak shifts or band broadenings, the peak positions or bandwidths can also be mapped to create Raman images. 

The data generated from a NIR or Raman spectrum do not immediately provide the concentrations of the species at any time, so there is no predictive capability. Construction of a calibmtion set requires an independent measure of the property, e.g. by HPLC or by NIR of known mixtures of the components. Two such methods are principal-component regression (PCR) and partial least squares (PLS). As soon as quantitative analysis is considered, the question of noise and reproducibility of the data set becomes important. It is therefore necessary to treat the mw data to remove the drift in baseline etc. which will occur over a long period of spectml acquisition. 

There have been many studies of the most effective way to address a common feature in NIR and Raman spectra of reacting systems, viz. the change in spectral baseline. Figure 3.48(a) shows a typical output from a NIR diffuse-reflectance spectral measurement of 12 kinds of ethylene-vinyl acetate (EVA) copolymers differing in vinyl acetate content (Shimoyama et al, 1998). 

---