Capítulo 1: Marco teórico
1.1. El adolescente
1.1.3 Formación académica- escolar
1.1.3.1. La orientación profesional
One of the major obvious differences between dispersive and FT-NIR spectrophotometers is the fact that FT instruments can achieve high optical resolution without compromising signal-to-noise ratio (Figure 5.5 and Figure 5.6). An inherent advantage of the FT is that the resolution does not directly depend on mechanical light-limiting devices such as a slit. In a dispersive instrument each resolution element is a defined sliver of the total available light. The levels of signal-to-noise on the dispersive instrument detectors are limited by the light level thus the higher resolution carries the price of worse signal-to-noise ratio. High optical resolution with high signal-to-noise is important for quantitative analysis.
Dispersive
FT-NIR
7000 7500
7800 0.00 0.05 0.10 0.15 0.20 0.25 0.30 0.35 0.40
Absorbance
cm–1
FIGURE 5.5 Comparison of spectra from a dispersive and FT-NIR spectrophotometer.
0.15 0.14 0.13
H2O 0.12
0.11 0.10 0.09
Absorbance
0.08 0.07 0.06 0.05 0.04 0.03 0.02 0.01
5500 5400
cm−1
5300 5200
FIGURE 5.6 High-resolution water-vapor lines (first overtone).
The FT instrumentation has the proportional advantage in the mid-IR, where the whole experiment is almost always light limited. In the NIR it was argued in the past that there is just so much light that the advantages of the FT cannot be realized. It is true that the modulated and collected radiation from a quartz halogen source (10 to 50 W) can overwhelm NIR detector, heating the detector and causing nonlinearities. In actual sampling situations however, the light levels are reduced drastically, a pharmaceutical tablet for example can reduce the light levels by a factor of a million, so the light-starved detector performs much better in an FT, due to the multiplex advantage.
(b) (a)
10,750
log(1/R)log(1/R)
10,700 10,650 10,600 10,550 10,500 cm−1
32 cm−1 16 cm−1
8 cm−1 4 cm−1 2 cm−1 32 cm−1 64 cm−1
16 cm−1 8 cm−1 4 cm−1 2 cm−1
10,450 10,400 10,350 10,300 7400 7350 7300 7250 7200 7150
cm−1
7100 7050 7000 6950
FIGURE 5.7 NIR spectrum of talc: (a) effect of resolution on band shape, (b) effect of resolution on signal-to-noise.
When evaluating FT-NIR it is important to consider real applications for the comparisons. It is obvious that much higher resolution can be achieved using an FT instrument. The higher resolution allows, for example, the ability to check the wavelength calibration using the vibrational–rotational water vapor lines [3]. Water-vapor lines in a real application could be a source of error if not completely resolved, because of the changing humidity levels in the laboratory environment. Also, many materials, have very sharp bands, which would not be resolved at 10 to 20 nm resolution of typical dispersive instruments. The better-resolved sharper bands give better chances for a selective analysis because of the increased peak height and the more characteristic spectral shape. Figure 5.7 shows a portion of the NIR spectrum of talc, where the smaller side lobe is not even resolved as a shoulder with 32 cm−1resolution (equivalent to 6 nm at 7200 cm−1). There are many chemic-als, for example liquid hydrocarbons, some pharmaceutical materichemic-als, vitamins, and others, where the improved resolution does improve quantitative results. As is evident from Figure 5.7b, the improved resolution also reduces the signal-to-noise ratio. This is a minor peak with low response collected with one scan (∼(1/2) s) to show the trade-off of resolution and signal-to-noise in a FT-NIR spectrophotometer. The top trace shows the sharpest feature, but it is also the noisiest.
In order to increase resolution, FT-NIR spectrophotometer instruments move the scanning mirror longer, thus allowing less signal averaging per unit time. In addition, the higher resolution results in more data points; thus the wavelength element is better defined. The resolution advantage of
FT allows scanning with the lowest resolution in many real applications thus collecting the highest signal-to-noise spectrum, while the optical resolution is still far superior to dispersive instruments.
A vitamin mixture was also prepared to study the effect of resolution on quantification. The actual concentrations of component were:l-ascorbic acid (vitamin C) with target value of 250 mg (varied in the calibration set from 2 to 55% of total); thiamine (B1) target 100 mg (varied 9–30% of total);
nicotinamide (niacin) 20 mg target (2–6% of total); riboflavin (B2) 30 mg (3–7% of total); pyridoxine (B6) target 5 mg (0.2–2% of total); filler, added to bring total weight to 500 mg (0.5–60% of total);
and cellulose and hydroxypropyl methyl cellulose. The spectra were collected in diffuse reflectance mode using an integrating sphere. For most of the components, higher resolution did not show any improvement in the quantification (as measured by comparing correlation coefficients). The minor component pyridoxine did show an improvement (Figure 5.8). Resolution effects can also help in transmission experiments. For example, determining an active compared to a placebo (Figure 5.9).
The multiplex advantage of the FT-NIR spectrophotometer can be readily seen in energy-limited experiments such as transmission tablet analysis. A typical layout for this experiment is seen in Figure 5.10.
FIGURE 5.8 Correlation coefficient R2 as a function of resolution for l-ascorbic acid a, thiamine b, nicotinamide c, riboflavin d, and pyridoxine e.
4200
FIGURE 5.9 Spectra of an active pharmaceutical formulation compared to a placebo.
Detector
Flexible light shield
Pharmaceutical tablet
Optical window
Near-infrared illumination
FIGURE 5.10 Schematic of transmission experiment for tablet analysis.
A typical example of a sample that requires transmission is soft-gel pharmaceutical formula-tions. For example, a vitamin E soft gel gives very little spectral information in reflection because of the thickness of the gelatin coating. Transmission through the vitamin shows detailed spectral information. Comparing this spectrum with the spectra of the oil extracted from the sample and measured using a fiber-optic dip probe shows the power of this technique (Figure 5.11). Studies have also been performed looking at changes of soft-gel formulation with moisture (Figure 5.12).
5.4 STANDARDS
The key assumption of most chemometric methods is that all of the spectral data have the same x-axis alignment. Even slight shifts in peak positions among spectra can create a significant cause of variance that may drastically reduce the reliability of the method. These issues are particularly important to the pharmaceutical industry. While the wavelength precision must be validated on any instrument, the requirement for consistent wavelength registration across a group of instruments is a major concern for many NIR analytical methods [4,5].
Because NIR is typically a low-resolution technique, the importance of accurate wavelengths or wavenumbers was not emphasized. The proposed USP method for wavelength accuracy defines three peaks with tolerances of±1 nm at 2000, 1600, 1200 nm. This translates to allowable errors of 5.0, 7.8, and 13.9 cm−1. A typical FT-NIR spectrophotometer has a precision of better than 0.01 cm−1. In Figure 5.13, a three-week monitoring of the water-vapor peak at 7299 shows a standard deviation of 0.0027 cm−1(0.0005 nm). This precision has required careful evaluations of the standards that are used for NIR.
However, even in FT-NIR spectroscopy the actual wavelength accuracy depends to a small degree on the optical design of the instrument. The measured wavenumber v= v{1 − (1/2)α2} where α is
Reflection
Extracted oil (Dip-probe)
Transmission
10,000 8,000 6,000
cm–1
FIGURE 5.11 Comparing vitamin E spectra of reflection, transmission, and extracted oil from inside the capsule.
% Moisture (loss on drying)
Soft-gel moisture
14 12 10 8 6 4 2 0
0 5 10 15
R2 = 0.9975 RMSECV = 0.210
FIGURE 5.12 NIR-derived moisture values compared to actual values.
related to the etendue or aperture. The differences are a linear shift and the observed differences are typically< 0.1 cm−1deviation [3]. To correct for the differences, the spectral features of a known standard are measured and the laser frequency adjusted to shift the spectrum. In practice, this is rarely done because the observed differences are small.
Bomem has reported using toluene as test of both wave number and absorbance repeatability and reduced differences to less than±0.1% [5].
Many FT-NIR spectrophotometers are designed to have internal standards for calibration proce-dures. The most common internal standard is a polystyrene sheet (Figure 5.14) used in transmission.
Polystyrene has also been used in reflection.
A sheet of polystyrene was compared to SRM 1921 (the NIST mid-IR standard) and shown that a thicker sample of polystyrene plastic that has been validated with the SRM 1921 standard would make an excellent reference material for verifying wavelength accuracy in a medium resolution FT-NIR spectrophotometer [6]. Traceable standards such as the SRM 2035 and SRM 1920× are also used and designed to fit in the sample position.
cm–1
7299.045
7299.035
7299.025 7299.055 7299.065 7299.075
Accuracy tests FIGURE 5.13 Trend line of position of water-vapor peak.
SRM 1920a
SRM 2035
KTA 1920×
Polystyrene
10,000 8,000
cm–1
6,000
FIGURE 5.14 NIR standards.
5.4.1 ABSCISSA
Photometric linearity is an important area as well. Photometric qualification is based on a set of transmission or reflectance standards with known values. In transmission, filters with known trans-mittance values are used. In reflectance, Labsphere™ Spectralon gray standards with reflectance values from 0.99 to 0.02 are used.
5.5 CONCLUSION
In the last decade, FT-NIR spectrophotometers have made inroads into the traditional NIR applica-tions areas: particularly the pharmaceutical, petrochemical, and chemical markets. The advantages of FT provide the ability to produce accurate, reproducible spectra for identification, and quanti-fication of difficult samples. This makes FT-NIR an invaluable tool for quality control and quality assurance applications.
REFERENCES
1. P. R. Griffiths and J. A. de Haseth, Fourier Transform Infrared Spectrometry, Wiley, New York, 1986.
2. R. J. Bell, Introductory Fourier Transform Spectroscopy, Academic, New York, 1972.
3. S. R. Lowry, J. Hyatt and W. J. McCarthy, Applied Spectroscopy (submitted for publication).
4. United States Pharmacopeia, Near-Infrared Spectrophotometry, Pharmacopeial Forum, 24: 1998.
5. H. Bujis, The Promise of FT-NIR: Universal Calibrations.