Raman microscopy has been attractive because of its ability to characterize materials on a spatial scale commensurate with optical microscopy. Typically the lateral spatial resolution is quoted as determined by the Airy disc[1] which is 1.22λ/NA where λ is the wavelength of the illuminating light, and NA is the numerical aperture which is equal to nsinθ, where n is the index of refraction of the medium (1.0 in the case of air) and  is the angle subtended by the optics. However, the Airy disc description cannot be correct for a Raman microscope. The Airy disc assumes uniform illumination of the focusing optic, and the laser profile is anything but. In addition, in some instruments the Gaussian laser profile is not well matched to the aperture of the focusing objective. At any rate, this article is going to concentrate on the depth resolution of the Raman microscope. Optical calculations for depth resolution of an optical microscope state that the it is proportional to λ/(NA) 2 . The essential point to recognize is that the spatial resolution of any Raman microscope depends on the detection optics as well as the focusing optics. How effectively does the optical system collect the Raman signal excited in the laser focal spot, and reject the signal from the surrounding volume that is illuminated by the laser but not in focus?Then the essential question becomes how to evaluate the depth resolution experimentally. Historically people have used a piece of a polished Si wafer for these tests. But this material was originally chosen more for the repeatability of any measurement of its signal rather than its appropriateness for answering the questions of depth resolution. The problem with using silicon is that when performing a depth profile, as the sample surface is moved away from the focal plane, the laser-illuminated area is increased; the final signal is a convolution of the losses because the laser illuminated area is not passed efficiently through the confocal hole, and the increase in signal because of the increased excitation volume. In addition, the depth of penetration of the laser into the crystal is not necessarily negligible. At 633nm, it will penetrate 3µm, at 785nm 12 µm. In trying to determine a better way to determine the confocal properties of a Raman system microscopic polymer beads were selected. With such a sample, when the sample is defocused, the Raman volume cannot be larger than the volume of the bead. Comparison of depth profiles of 2µm and 0.5µm beads and silicon will be shown to provide insight into the confocal behavior of the Raman microscope. These measurements are done using the 532nm excitation wavelength whose depth of penetration into silicon is about 0.7µm. This avoids the complications of volume effects. Figure 1 shows depth profiles of the 2µm and 0.5µm spheres of polystyrene recorded while varying the confocal hole. As the confocal hole is increased, the signal strength increased because light from more of the bead volume is transmitted. However, the full 360

Raman microscopy is a hybrid technique between Raman spectroscopy and microscopy. Raman microscopes using optical microscopes were introduced in the 1970's, and take advantage of diffraction-limited optics to provide molecular and crystalline information. The spatial resolution of a Raman microscope is reported to be the order of the laser wavelength [1, 2]. A Raman microscope equipped with a automated stage records Raman maps using the point mapping method, which is, so far, the most popular method to record Raman maps. Successful data processing of Raman spectra within a Raman map results in Raman images that represent spatial/volumetric distribution of different chemical components within the sample. Since Raman images are distillations of spectral information, the spectral quality (e.g. signal-to-noise ratio) is directly correlated to the image quality [3]. Also, the spectral differences between two point measurements (chemical contrast) are directly correlated to the resolving power between two points, which influences the spatial resolution [4]. As a result, the spatial resolution in Raman Imaging must be determined differently from that of optical (?) microscopy.

Raman imaging takes advantage of the excellent sensitivity and specificity of the Raman technique, especially when high spatial resolution is available by employing a confocal microscope. One largely overlooked aspect of point mapping technology is its flexibility. The area, shape and the spacing between spatial data points are all customizable to fit the characteristics of the sample and the goal of the project without losing the spectral resolution or signal intensity. A pharmaceutical over-the-counter analgesic tablet was selected and measured to demonstrate the impacts of the optimized experimental conditions on the image, spectra and information content.The LabRam ARAMIS (Horiba Jobin Yvon, Edison, NJ, USA) was used to measure all samples. The excitation light was 632.8 nm line of the He/Ne laser. The confocal hole was opened to a diameter of 300 µm and the slit to a width of 150 µm. The 1200 gr/mm grating was used to measure the spectral range of 263.5 -1151.2 cm -1 . All data were measured using LabSpec 5 (Horiba Jobin Yvon, Edison, NJ, USA), and processed with LabSpec5 and ISys 4.0 (Spectral Dimensions, Inc., Olney, MD, USA).The target sample (a commercial tablet of over-the-counter analgesic) contains three ingredients of ~ 10 %, ~ 40 % and ~ 50 %, respectively. Based on the size and weight of the sample, the chemical domains (areas where a single chemical component concentration is significantly higher than surrounding areas as to be identifiable as a 'domain' of the specified component) were estimated to be of the order of 100 µm in diameter.To identify the region of interest where spatial heterogeneity is significant, a coarse map of 200 × 200 µm 2 at 25 µm increments between measurement points was obtained. The area measured in this map is being considered as the total sampling area and the locations of subsequently measured maps are subsets of this map. At each spatial point, a spectrum was obtained by measuring and averaging two accumulations of 1 s exposure signal. The total data collection time for this measurement was approximately 4 min. Raman chemical images were created from spectral lines of each species. The representative spectra of each species are shown in Figure 1c. It is noteworthy that, while each spectrum is unique, it is difficult to find a 'clean' band that does not overlap with spectral features of other species. Multivariate analysis results will be employed to overcome this obstacle.For each species, a set of cursors was selected between which integrated intensities were calculated for each spatial point spectrum [271.5 -315.0 (red), 767.3 -802.8 (green), and 1034.0 -1060.9 (blue)]. By subtracting a linear baseline defined by the cursors, the color-coded intensity correlated to the selected species. The composite chemical image is shown in Figure 1a. The rectangular insets indicate two areas that show particularly high spatial heterogeneity. These areas were measured again with the increasing increments (2.5, 1 and 0.5 µm) for further study. The data were processed and visualize...

. Raman microscopy and fluorescence lifetime microscopy are two complimentary techniques used as contrast methods for a biologically interesting cellular image. Often enough, the optical phenomena of fluorescence and Raman interfere with each other depending on the molecular and optical properties of a cellular sample. Instead of avoiding and Raman signals in fluorescence microscopy and avoiding fluorescence signals in Raman microscopy, the two are used as complimentary techniques to extract molecular information from a cellular sample. Exciting this particular sample with NIR light and collecting the image using a Raman microscope gives information about molecular structures around the different parts of the cell. Different contrast information in a cell can be used by fluorescence microscopy. With excitation in the visible wavelength range, the fluorescence signal from the same sample is enhanced and the image shows differences in molecular fluorescence by the cellular structures, lending different variations in the image. In addition, the fluorescence lifetime image is taken using the same fluorescence microscope on the same sample, supporting a third method of differentiation in the microscope image.All Raman images were measured on an XploRa Raman microscope with a 785 laser for excitation. All fluorescence images were measured with a DynaMyc fluorescence lifetime microscope with a 450 nm, 50 MHz pulsed laser diode and both a fluorescence camera and a photomultiplier tube (PMT). Dichroic filters were selected for emission collection on both microscopes. Difficulties in using both methods here include high amounts of scattered light, and liquid media. While highly scattering samples is optically preferred for Raman signals, this phenomenon can cause artifacts in fluorescence signals. On the reverse side, Raman signals are usually very weak when using liquid media due to the lack of scattering signal. By utilizing optimal filters and excitation sources, these two issues can be overcome by optimization of the spectroscopic and microscopy instrumentation and methodologies presented here.With fast scanning of the lifetime decays using time-correlated single photon counting, both fluorescence intensity and lifetime images, are easy for looking at a single image with different contrast methods. Raman microscopy lends the third level of contrast with information on molecular structure and environment with a cell.

Hyperspectral Raman imaging follows the recent trend in the total characterization of the sample using more than one technology. Typically performed with a confocal Raman microscope with an automated mapping stage, the sample is first visualized using high performance microscopic capability. Then the region of interest is identified for molecular analysis. The area is mapped, most often with point mapping (taking Raman measurements from each of points on the predetermined grids) producing a Raman map. Raman spectra within the Raman map is then analyzed to provide chemical identifications, crystal phases, degree of crystallinity, orientations and order of molecules and polymorphic forms. Spectral analysis results produce Raman images highlighting locations of each chemical compound in correlation to the microscopic analysis results. This way, both spatial and spectral composition of the sample is characterized in one analysis step.One particular advantage of hyperspectral Raman imaging in comparison to typical fluorescence imaging is that it requires virtually no sample preparation and non-invasive. There is no need to stain the sample to highlight the target molecules. The selection rules of Raman spectroscopy allow only weak water bands in the finger printing region, even in-vivo and in-situ measurements are straightforward [1-2].The next goal was to achieve the fast enough speed that allows hyperspectral Raman imaging to be practical. Improvements were made continuously including in detectors (increased sensitivity to reduce the per spectrum measurement time), stages (faster moving stages with better stability and precision) and electronics. The most recent developments (e.g. SWIFT™, HORIBA Jobin Yvon) finally have reached the level of speed that allow to record over 10,000 full range Raman spectra in a few minutes.The most striking advancement is to map the entire area of a large sample (millimeters or centimeters) with a small spot laser (a micrometer or less). The areas between measurement points were skipped and undetected in point mapping unless the step size matched the spot size, which led to impractically large map with little information added. A simple approach would be increasing the spot size to match the sample scale using a low numerical aperture (N.A.) objective. However, low N.A. weakens the system sensitivity, slowing the measurement. DuoScan™ (HORIBA Jobin Yvon) was developed to maintain the best of two worlds. The pixel sizes are adjusted (1 × 1 -300 × 300 µm 2 ) from the software without changing the objective, matching the spatial resolution the sample requires while maintaining the high N.A. objective lens for high sensitivity for spectral measurements.One particular advantage of DuoScan™ is for large samples with small elements. For example, carbon nanotubes (CNTs) doped on a Si substrate is not easy to locate due to the large area to survey