Considerations in Choosing an FT-IR Imaging System

FT-IR imaging is a powerful tool for materials characterization.

By

FT-IR microscopy has been a powerful technique for the analysis of small sample areas for almost two decades. Application areas include the identification of impurities, defects, and inhomogenieties down to 10  10 μm in size in a wide range of materials from polymers to pharmaceuticals. However, in many cases a complete view of the distribution of a molecule or functional group across the sample is required to check for changes in composition. This requires many individual spectra to be acquired to build up an image which is time-consuming using a traditional FT-IR microscope.

To generate these chemical “pictures” of a sample, FT-IR imaging systems with multi-element detectors have been developed to allow rapid collection of spectra from many points across a sample area (Figure 1). This has accelerated the adoption of FT-IR imaging and it is now seen as an essential tool for research and development (R&D) and troubleshooting laboratories — assuring product performance and reducing development time for many products. New application areas are constantly emerging as the technology develops, including biomedical applications and studies of pharmaceutical formulations.


Figure 1: An IR image of the distribution of a
carbonyl band at 1740 cm-1 in a multilayered
polymer cross-section. This highlights two adhesive
layers, one of which is slightly distorted. The area
shown is approximately 0.5 x 0.4 mm and consists
of 5300 data points. Data collection time for the
whole image was less than two minutes.
 

CHOICE OF DETECTOR

The first FT-IR imaging systems employed large focal plane array detectors which were originally designed for military surveillance applications. These systems suffered from poor reliability, low sensitivity, and a reduced spectral range which limited their usefulness for routine operation in the laboratory. New approaches have included a patented linear mercury cadmium telluride (MCT) array detector with 16 individual liquid nitrogen-cooled elements in which the sample image is swept across the detector in a precise linear pattern, and by synchronizing the movement of the stage with the interferometer of the FT-IR spectrometer, up to 170 high-quality, full-range (7800–700 cm-1) spectra per second can be collected, which equates to over 10,000 per minute.

One significant advantage of this approach is that by moving the sample across the detector, images of variable sizes and aspect ratios can be measured rather than a fixed size which is dictated by the layout of the detector array. This clearly saves time in many analyses as only the area of immediate interest needs to be sampled.

Another important consideration is the spectral range of the array detector. Many commercially available arrays show very poor sensitivity below 1000 cm-1 and cannot effectively collect data in this region. This restricts the ability to characterize materials with inorganic content such as filled polymers or paint chips. Analysis of the latter is particularly important in identifying trace evidence in forensic science.

VARIABLE IMAGING RESOLUTIONS

Many imaging systems can measure down to a pixel size of around 10 micro m. However, a range of pixel sizes should be available to match the resolution of the measurement to the expected size of the areas of chemical difference in the sample. This may be determined by particle size or physical composition. For example, many polymer laminates have layers of only a few microns in thickness so a very high resolution is required to reveal the individual layers in an IR image. Conversely, samples such as pharmaceutical tablets may have relatively large particle sizes in some formulations. In many studies of excipient distribution, collecting data at very high-resolution is unnecessarily time-consuming and a lower resolution would be more appropriate. In addition, some systems now have the ability to survey the sample rapidly at a resolution of the order of 50 microns. This is extremely useful as a first-pass screen to quickly identify any areas of inhomogeniety across the sample. If necessary, the user can quickly re-measure areas of interest at higher resolution to characterize fine structure.


Figure 2: Visible and IR Reconstructed Images from Embedded Laminate

ATTENUATED TOTAL REFLECTANCE IMAGING

As mentioned above, polymer laminates often have layers which are a few microns thick, presenting particular challenges for analysts. Even with very careful preparation of very thin microtomed cross-sections, it is difficult to maintain sample integrity throughout the sampling process. Also, using traditional transmission spectroscopy, it is impossible to obtain clean spectra of these layers as the spatial resolution becomes diffraction-limited.

ATR (attenuated total reflectance) imaging can overcome some of these limitations and resolve detail which is difficult, if not impossible, to observe using conventional IR microscopes and imaging systems. As a reflectance technique, the sample need not be cut into thin cross-sections — samples are typically mounted in an embedding resin and polished to a flat surface. Also, the amount of sample interrogated by the IR beam is relatively low — around 1-2μm using germanium ATR crystals. Images tend to be sharper and spectra show fewer artefacts due to beam divergence,1 and interference fringes. Additionally, ATR imaging provides data with higher spatial resolution than transmission imaging.2 A resolution of less than 4 μm is achievable with ATR, whereas the physical diffraction limit for transmission work is typically three to four times this in the mid-infrared fingerprint region.

To obtain good ATR images, the sample must be brought into close, uniform contact with the crystal across the entire measurement area. To achieve this, the laminate needs to have a flat surface at the point of contact and be adequately supported to avoid distortion under pressure. The polished resin sample block is placed directly on an anvil on the ATR accessory and the anvil is raised to bring the sample into firm contact with the crystal.

APPLICATION EXAMPLE: ATR IMAGING OF A POLYMER LAMINATE PACKAGING MATERIAL

Figure 2a shows the visible image of a packaging material section embedded in epoxy resin. Despite the surface scratches, it was possible to generate good ATR images because the sample was slightly compliant, allowing the crystal tip to be pushed into the sample. Also, appropriate data treatment can minimize the effects of the slightly varying contact across the image. Figure 2b shows the reconstructed IR image where the scratches are no longer apparent.

To obtain the reconstructed image, the image spectra were derivatized, offset-corrected, and subject to principal components analysis (PCA). This technique sorts the image spectra into an independent set of sub-spectra (principal components) from which the image spectra can be reconstructed. For example, if there are five layers present in an image of say 1,000 spectra, then five subspectra would be sufficient to describe all the 1,000 image spectra. In practice, more than five spectra are usually required, due to the presence of impurities and other spectral contributions like variations in baseline and atmospheric absorptions. The amounts of the principal components in the original image spectra, or scores, are calculated at each pixel and the resulting score images are extremely useful in enhancing IR image contrast. For this sample, the first three PC score images and corresponding spectra are shown in Figure 3. The second and third images show the major laminate layers, a polyethylene and polyamide, with intermediate layers around 6 μm in thickness, sandwiched by the PE and PA, as revealed in the PC4 image. Here the layers are readily identified by examination of the underlying raw image pixel spectra. The first score image, PC1, is due to the embedding medium and is not shown here.


Figure 3: Score Images and Spectra from Principal Components 2, 3, and 4


Figure 4: Score Image and Spectra for Principal Component 5


Figure 5: Score Image and Spectra for Principal Component 6

In addition, the features due to the minor variance are shown in the PC5 and PC6 score images. The PC5 image shows a feature 3-4 μm thick close to the outer edge of the sample. Using software, it is possible to view the raw spectra underlying these features, tracking through the feature at steps of 1.56 μm. This shows that there are unique carbonyl features in the spectra of the layer which are not present on either side, thus confirming the distinct chemical composition of the layer (Figure 4). This can be compared with the situation in Figure 5 where the PC score image also reveals a layer. However, no distinct chemical feature appears on examination of the spectra, rather a relatively smooth gradation in spectral intensity from one material to the next. This indicates a physical boundary, such as a ridge, rather than a distinct chemical constituent.

FULL AUTOMATION AND OTHER CONSIDERATIONS

The above application example shows how a powerful, fully-featured software package is important in order to extract the maximum information from each analysis. The ability to manipulate large image files is important and it is essential to have access to powerful processing tools such as PCA in order to enhance image contrast and separate chemical and physical differences across a sample. It is equally important to choose a system in which the software allows very rapid and simple set-up of analyses. For example, simple features such as the ability to define the imaging area directly on the visible image of the sample (see Figure 2a) are very convenient and save valuable time.

It is now possible to automate virtually all of the main functions of the imaging system to provide simpler, more reproducible operation by eliminating manual adjustments. These include provision of fully automated stage movement including autofocusing, automatic aperture adjustments in single-point mode, and automatic optimization of sample illumination in visible viewing mode. Additionally, many images can be collected and stored in an unattended operation from one sample or from many samples. The imaging areas are defined in advance by inspection of the visible image, the imaging parameters set-up, and the system can collect data unattended for upwards of 24 hours, facilitated by a large volume auto-feed liquid Nitrogen Dewar to keep the detector cooled.

SUMMARY

FT-IR Imaging is now able to meet many of the diverse challenges of materials characterization applications. Automation and powerful software features are increasing productivity and accelerating adoption in virtually every field of study. Traditional limitations such as sensitivity, reliability, and spectral range have been overcome by new detector designs. Modern systems can now operate at a variety of spatial resolutions, allowing the operator to optimize the analysis according to the sample properties and the desired measurement time. ATR Imaging has provided a solution to the established problem of sample preparation and has allowed very small features, of the order of a few microns, to be characterized at a very high spatial resolution.

References

  1. A J Sommer. “Mid-Infrared Transmission Microspectroscopy,” in handbook of Vibrational Spectroscopy, Vol 2, 1369, Wiley, 2002.
  2. A Canas, R Carter, R Hoult, J Sellors, S Williams. “Spatial resolution in mid-IR ATR Imaging: Measurement and Meaning,” FACCS Conference, 2006.

David Clark is the Consumables and Channel Development Manager, Materials Characterization Business, PerkinElmer Life and Analytical Sciences, Seer Green, UK. He can be reached at David.R.Clark@perkinelmer.com.

 

Categories: Laboratory Technology

Published In

Management's Role in Laboratory Automation Magazine Issue Cover
Management's Role in Laboratory Automation

Published: January 1, 2008

Cover Story