Scientists dream of being able to probe variations in the chemical composition and structure of materials without any form of sample preparation. Raman spectroscopy, a noncontact, nondestructive analysis tool that yields information on the chemical, vibrational, crystal and electronic structure of materials at the submicron scale, promises this. The primary hurdle has been keeping samples in optical focus during imaging measurement. Most commercial Raman systems use an objective lens to focus laser light onto a sample and to collect the Raman scattered light. One of the key strengths of Raman is that it does not require any sample preparation; provided the laser can be shone on the sample, data can be collected from it. This allows Raman measurements to be conducted on both liquids and solids, on macroscopic and microscopic objects, through glass windows and into reactors. In the last decade, advancements in instrument and laser technology have significantly increased the speed of Raman spectroscopy, allowing large area Raman imaging (or mapping) to become routine. These measurements are conducted by moving the sample using a motorized stage and taking an array of Raman spectra at regular intervals. An image is then generated by applying a metric to these spectra, for instance the intensity of a Raman band, and then turning the data into a false color image. Figure 1. (a) Sample in focus, (b) focused above sample and (c) focused into the sample. In all cases, the sample volume probed and the Raman intensity will vary. Courtesy of Ranishaw. Rapid Raman imaging is now used in a wide range of applications, with people imaging everything from 2D materials to teeth. This has created a new challenge: To collect accurate Raman images, it is vitally important to maintain the focus of the laser on the sample. This focus dictates the illumination of the sample with the Raman laser and the collection of Raman scattered light. During Raman imaging the sample is scanned under the microscope and, if the sample has any variation in height, it will move in and out of focus (Figure 1). Any deviation from the true focus will decrease the amount of Raman light collected from the sample, increasing the required measurement time. As focus is lost, the volume from which the Raman scattered light is collected changes, potentially resulting in a loss of spatial resolution. In a worst-case scenario, the sample can be so far from focus that no Raman signal is collected and all the information is lost (Figure 2). Figure 2. Raman images illustrating the intensity of the graphene 2D band collected from a sample of graphene on copper. (a) Due to the change in height across the sample, only a strip at the center of the image is in focus. Further from focus, the image becomes blurred, obscuring the morphology. Finally, no Raman signal is returned, resulting in complete loss of information. (b) The same area collected using LiveTrack. Here the sample height is adjusted during measurement to ensure that all Raman spectra are collected in focus, allowing accurate Raman information to be collected and enabling the full morphology to be seen. Courtesy of Renishaw. Removing ‘tilt’ is an arduous task Ideally, all samples would be flat, but this is rare. Even the simplest sample, such as a silicon wafer, may have some inherent tilt when placed under the objective. While insignificant over small length scales, over tens of mm this will mean the sample will go out of focus when using a high magnification objective. In this case, the theoretical solution is simple; remove the tilt by adjusting the sample so it lies perfectly perpendicular to the microscope objective. In practice this can be very challenging depending on the precision required. For instance, removing a tilt of a micron per centimeter is an arduous task. Samples are often tilted, rough or have complicated geometries, which makes keeping them in focus when collecting Raman images difficult. Optical images of different examples of these types of samples are shown in Figure 3. Figure 3. Optical images of (a) a tilted sample (graphene on a metal foil), (b) a rough sample (electrode from lithium ion battery) and (c) a sample with complicated surface geometry (a crystalline powder, l-cysteine). Courtesy of Renishaw. One option to image these samples is to replace the objective lens with one that has a larger depth of field, allowing the focus to be maintained over a greater range of sample heights. This typically means reducing the numerical aperture (NA) of the lens, but this has its own drawbacks. The measured Raman intensity is proportional to the NA2, ensuring measurement times will be longer. The lateral spatial resolution is inversely proportional to the NA and the axial spatial resolution NA2, so the collection volume will be larger and the resolving power will be worse. In effect, changing the objective lens will increase measurement times and provide lower resolution data, but will remove any artifacts caused by poor focus. Really difficult samples such as minerals or pharmaceutical tablets need to be mounted, sectioned and polished to ensure they are suitably flat. Any modification of the sample is extremely undesirable, as there is always a risk this preparation can affect the chemical composition of the sample. Worse still, some samples may be priceless and as such cannot be modified. As neither of the discussed solutions is ideal, Raman spectrometer vendors have produced proprietary systems to maintain sample focus when imaging. These can be broadly separated into two categories: • Prescan methods that determine the sample topography before Raman data are collected. During the Raman measurement the surface is fed into the software to adjust the height of the sample at each point. If the sample changes during measurements, the collected surface will not correspond to the true surface and focus may be lost. • Active methods in which a feedback technique is used to determine if the sample is in focus. Corrections are made either immediately before or during the Raman measurements at each point. More detailed discussion of these techniques is made in the accompanying table. Adjusting sample stage height Renishaw PLC has recently released an innovative surface tracking technology called LiveTrack that takes improvements in autofocus technology from other fields and applies them to Raman spectroscopy. The technology employs a closed loop system, using optical feedback that continuously adjusts the sample stage height to maintain perfect focus on the sample during Raman measurements and when viewing the sample optically. Being able to browse the sample optically and not worry about manually maintaining focus makes it easy to locate and define regions of interest for Raman imaging. LiveTrack itself runs in parallel with Raman measurements and does not contribute to the measurement time. The feedback is directional and is sufficiently fast that it is compatible with high-speed Raman imaging, allowing Raman imaging to be applied to a wide range of samples that would not have been practical in the past. One such example is the use of Raman spectroscopy for assessing the distribution of the active components in pharmaceutical tablets. Figure 4 shows data from an over-the-counter analgesic tablet snapped in two so that the cross section can be measured. Snapping the tablet has resulted in a very rough surface, with height changes on the order of 1 mm. The measurements shown (Figure 4) would be impossible without the new technology and it would be necessary to modify the sample, in this case polish it flat, risking contamination or chemical modification. Figure 4a shows the distribution of acetaminophen, aspirin and caffeine within the tablet overlaid on the determined surface of the tablet. The topography information, shown separately in Figure 4b, is obtained by recording the height of the stage at each Raman measurement point, effectively using LiveTrack as an optical profilometer. Figure 4. Raman image of an over-the-counter analgesic tablet (a) illustrating the distribution of acetaminophen (blue), aspirin (red) and caffeine (green) within the tablet. Topography image (b) of tablet collected by LiveTrack. Courtesy of Ranishaw. Another use comes with the analysis of graphene, which has demonstrated huge potential in a wide range of technologies, from electronics to composites. However, for this potential to be realized, improvements must be made to enable the growth of high-quality, large-area material. Industrial production of graphene is already underway with some companies growing material over large areas (more than a square meter) on copper foils. It is hard to maintain quality over these expanses; the material can become highly defective and may consist of multilayer regions. Raman spectroscopy is the go-to technique for analyzing graphene as it can quickly provide comprehensive information on quality and the number of graphene layers. Unfortunately, graphene on copper is hard to analyze over large areas as the foils are never flat; they undulate, causing the graphene to go in and out of focus during Raman measurement. Figure 5 shows a large-area measurement of such a sample. Here, the surface height variation is about 150 µm, orders of magnitude higher than the depth of the field of the objective used (0.4 µm). The Raman image illustrates the change in width of the 2D band, which can be linked to the number of graphene layers present, in this case varying from single to multilayer. Figure 5. Raman image (a) of large area of graphene on copper, illustrating the width of the 2D band, which allows empirical determination of the number of graphene layers present, (b) topography image of the sample, and (c) Raman image overlaid on topography (Z-scale exaggerated to illustrate the undulating surface). Courtesy of Ranishaw. Improvements in focusing technology have removed some of the intrinsic limitations of rapid Raman imaging, allowing it to be applied to a much wider range of samples while avoiding any sample preparation. This opens up a variety of new applications, in particular those linked to quality control applications. Advanced focus tracking adds an additional level of automation to Raman systems, allowing even novice users to collect highly accurate Raman images. Meet the author Tim Batten is senior application scientist at Renishaw PLC. He has more than 10 years’ experience in Raman spectroscopy, and specializes in the application of Raman spectroscopy to material systems, carbons, 2D materials, semiconductors, etc.; email: tim.batten@renishaw.com