Calibration of dispersive spectral instruments has always been a challenge for researchers. While plotting a spectrum along a pair of axes, the x-axis usually signifies wavenumbers or wavelength, while the y-axis represents intensity. Normally the x-axis is calibrated by measuring the positions of two or more mercury emission lines and interpolating between them. The precision with which intermediate wavelengths may be classified into the following categories:
- Not specified,
- Only as good as the interpolation routine, and
- Generally much worse than the user expects.
IntelliCal Wavelength Calibration Routine
In 2010, Princeton Instruments introduced 64- bit LightField data acquisition software featuring IntelliCal, a patent pending wavelength calibration routine based on the Rietveld refinement algorithm used in x-ray spectroscopy. By using an emission line source and a table of lines from the NIST spectral database, IntelliCal solves for the wavelength of every pixel simultaneously across the complete focal plane of a 26 mm CCD mounted on a dispersive spectrograph. The wavelength accuracy of the calibration at every pixel is stored with the spectral data file, thus ensuring consistent data validation and traceability. By comparing the results to a table of emission lines as shown in Figure 1, the accuracy of the IntelliCal wavelength calibration routine is typically 4 to 10 times higher than that of a traditional interpolative method.
Figure 1. Comparison of wavelength calibration accuracy: IntelliCal versus a conventional interpolative method.
Wavelength inaccuracy is a serious issue. A shift in wavelength by even just 1 pixel in a Raman spectrum cannot be recognized in the difference spectrum. Due to this the instrument may have to be calibrated again and the whole experiment repeated.
Figure 2 shows a true difference Raman spectrum (displayed in red) and a spurious difference spectrum obtained from a shift of a single pixel (displayed in blue).
Figure 2. Real difference Raman spectrum (red) and spurious difference spectrum obtained from a 1 pixel shift (blue).
Actually, differentiating between the two seems impossible, hence it is essential to have a superior calibration routine or one may end up publishing spurious data. Wavelength accuracy is critical in areas such as forensic, and hazmat identification, where a spectral fingerprint serves to recognize the unknown material.
Search-match algorithms compare a measured spectrum to spectral libraries using pattern recognition techniques. An accurate knowledge of the Raman shift of a given band, or the exact emission wavelength of a line in a LIBS spectrum, makes sure the match is successful, while inaccurate spectra increase the risk of both false positives and negatives.
Intensity Calibration Fundamentals
Intensity calibration is highly challenging. The spectrometer data is never the true spectrum, but is modified by every optical element along the light path, which includes mirrors, lenses, , the diffraction grating and the detector. Each of these elements has its own spectral response.
Figure 3 shows the responses of a typical holographic diffraction grating (green curve) and a backthinned CCD camera (black curve).
Figure 3. Spectral responses of a typical holographic diffraction grating (green curve) and a back-thinned CCD camera (black curve). The response of each optical element along the light path in a given experimental setup modifies the recorded spectrum.
A spectrum recorded using these components will include their instrumental artifacts. Typical instrumental intensity artifacts are shown in Figure 4. These spectra show photoluminescence from zinc oxide on a high-resolution spectrometer. The red curve is the true spectrum; the blue and green curves show spurious peaks resulting from discontinuous changes in grating diffraction efficiency as the grating is rotated around its axis.
Figure 4. Photoluminescence from zinc oxide recorded on a high-resolution spectrometer. True spectrum (red curve) versus spurious peaks arising from discontinuous changes in grating diffraction efficiency as the wavelength is scanned across the focal plane (blue and green curves).
The spectra presented in Figure 5 are of an incandescent lamp. The black curve shows the theoretical blackbody spectrum; the three colored curves show the measured data with a sharp feature that arises from a Wood’s anomaly.
Figure 5. Theoretical blackbody spectrum (black curve) versus measured data (colored curves) for an incandescent lamp.
Figure 6 shows Raman spectra of stearic acid, a common component of pharmaceutical tablets. The spectrum shown in red represents raw data, while the spectrum shown in orange has been corrected. The excitation wavelength is 785 nm; the C-H stretching peaks at 2900 cm-1 are at 1015 nm, where the detector’s quantum efficiency is quickly decreasing to zero. Intensity calibration restores the optimal peak-height ratio while flattening the baseline, making it much easier to quantify the fraction of stearic acid in a mixture.
Figure 6. Raman spectra of stearic acid: raw data (red) versus corrected data (orange).
Intensity Calibration Using IntelliCal
In 2011, Princeton Instruments released LightField 4.0, an upgrade to the innovative data acquisition package, with new support for PI-MAX 3 intensified CCD cameras. IntelliCal 2.0 was also launched that includes both the fully automated wavelength and intensity calibration routines. The intensity calibration engine is a multi-LED, USB- powered, light source with emission from 400 to 1100 nm. Each highly stable source is calibrated separately against a standard and the spectrum is recorded in the device’s firmware. The entrance slit is illuminated with the IntensiCal source and the spectrum is recorded as shown in the left side of Figure 7. The uncorrected spectrum displayed in Figure 7 shows a spectral stitching artifact at 610 nm and sonsiderable etaloning at 830 nm that arises from interference fringes on the back -illuminated sensor utilized to collect the data. In the corrected spectrum shown at the right side of Figure 7, the stitching and etaloning artifacts have disappeared. The relative peak heights have also been corrected. For this particular example, the IntensiCal light source was utilized to correct its own spectrum, but the routine as such is general. A Raman, photoluminescence, fluorescence, absorption, or LIBS spectrum can be corrected as easily as that of the light source itself.
Figure 7. Uncorrected spectrum (left) versus intensity-corrected spectrum (right).
Princeton Instruments has once again elevated the science of dispersive spectral instrumentation by adding unprecedented usability to the list of key performance criteria. Spectroscopists need not waste their precious laboratory time creating their own correction routines, they can use automated IntelliCal to gain 100% confidence in both axes of their recorded spectra.
This information has been sourced, reviewed and adapted from materials provided by Teledyne Princeton Instruments.
For more information on this source, please visit Teledyne Princeton Instruments.