Neil Everall

Subsurface probing in diffusely scattering media using spatially offset Raman spectroscopy

We describe a simple methodology for the effective retrieval of Raman spectra of subsurface layers in diffusely scattering media. The technique is based on the collection of Raman scattered light from surface regions that are laterally offset away from the excitation laser spot on the sample. The Raman spectra obtained in this way exhibit a variation in relative spectral intensities of the surface and subsurface layers of the sample being investigated. The data set is processed using a multivariate data analysis to yield pure Raman spectra of the individual sample layers, providing a method for the effective elimination of surface Raman scatter. The methodology is applicable to the retrieval of pure Raman spectra from depths well in excess of those accessible with conventional confocal microscopy. In this first feasibility study we have differentiated between surface and subsurface Raman signals within a diffusely scattering sample composed of two layers: trans-stilbene powder beneath a 1 mm thick over-layer of PMMA (poly(methyl methacrylate)) powder. The improvement in contrast of the subsurface trans-stilbene layer without numerical processing was 19 times. The potential applications include biomedical subsurface probing of specific tissues through different overlying tissues such as assessment of bone quality through skin, providing an effective noninvasive means of screening for bone degeneration, other skeletal disease diagnosis, and dermatology studies, as well as materials and catalyst research.

Modeling and Measuring the Effect of Refraction on the Depth Resolution of Confocal Raman Microscopy

A simple ray-tracing analysis has been used to predict the effect of refraction at the sample/air interface, on the depth resolution of confocal Raman microscopy. This analysis applies to the “z-scanning”, or “optical sectioning”, approach to obtaining depth profiles, where the laser beam is incident normal to the sample surface, and spectra are recorded sequentially as the focus is moved deeper into the material. It is shown that when a “dry” metallurgical objective (the most common configuration for commercial Raman microscopes) is used, both the position and the depth of focus increase dramatically as the beam is focused deeper into the sample. It quickly becomes impossible to obtain “pure” spectra of thin layers that are buried more than a few micrometers below the air interface. Equations are presented which model the intensity response expected when focusing through a coating into a substrate. The model requires knowledge of the sample refractive index, the numerical aperture, and the laser beam intensity distribution at the limiting aperture, of the objective. Given these values, one can predict the substrate Raman intensity as a function of the nominal focal point within the sample. For a 36 μm coating on a thick substrate, we predict that even for a perfectly sharp interface (<<1 μm), substrate bands rise slowly (over an apparent distance of 10 μm or more), and are strong when the focus is apparently only ∼ 18 μm below the air/coating interface. This prediction was confirmed through experimental observation. The model was also used to analyze literature data that had been interpreted previously as showing interfacial diffusion in polymer laminates. The model correctly reproduced the main features of the observed data without invoking interfacial penetration—the optical aberrations alone accounted for almost all the observed broadening and the fact that the apparent thickness of the buried layer is also distorted. It was concluded that, with the use of this illumination geometry, it is very difficult to detect or quantify interfacial broadening unless it occurs on a very large scale indeed (tens of micrometers). It is concluded that ‘optical sectioning’ cannot be recommended for quantitative depth profiling at significant depths using metallurgical objectives. The optimum practical solution is to cut a cross section and map laterally across the sample, thereby utilizing and maintaining the excellent (lateral) resolution of the Raman microprobe. An alternative solution is to use an immersion objective to minimize refraction at the sample surface.

Confocal Raman Microscopy: Why the Depth Resolution and Spatial Accuracy Can Be Much Worse Than You Think

This paper describes continuing studies on the effect of refraction on the depth resolution and spatial accuracy of confocal Raman microscopy. Previous work showed how the apparent position of interfaces, and their breadth, was grossly in error when “optical sectioning” was carried out with a high-power metallurgical objective (the usual configuration for commercial confocal Raman instruments). Simple equations were presented which model these effects given only the numerical aperture of the objective and sample refractive index. This paper extends these studies to the measurement of the position and thickness of buried structures and shows how the refraction theory provides a good basis for interpreting Raman intensity-depth profiles that have been acquired by optical sectioning of complex structures. To summarize the magnitude of the problem, it is typical for the apparent thickness and positional depth of a buried layer to be about half of the true values—hence, correcting for refraction is critical when interpreting intensity-depth profiles. In order to minimize the effects it is possible to use a specialized objective (e.g., an oil immersion objective) to reduce refraction at the sample surface. Data are presented which show that the apparent position and thickness of structures obtained with such an objective are much closer to the actual values, even when the sample index is not perfectly matched by the coupling fluid. The use of immersion objectives is highly recommended if depth profiling by optical sectioning is to be attempted.

Numerical Simulations of Subsurface Probing in Diffusely Scattering Media Using Spatially Offset Raman Spectroscopy

We present the first elementary model predicting how Raman intensities vary for a range of experimental variables for spatially offset Raman spectroscopy (SORS), a recently proposed technique for the effective retrieval of Raman spectra of subsurface layers in diffusely scattering media. The model was able to reproduce the key observations made from the first SORS experiments, namely the dependence of Raman signal intensities on the spatial offset between the illumination and collection points and the relative contributions to the overall spectrum from the top layer and sub-layer. The application of the SORS concept to a three-layer system is also discussed. The model also clearly indicates that an annular geometry, rather than a point-collection geometry, which was used in the earlier experiments, would yield much improved data.

Picosecond time-resolved Raman spectroscopy of solids: Capabilities and limitations for fluorescence rejection and the influence of diffuse reflectance

It has been known for many years that it should be possible to discriminate between Raman and fluorescence phenomena on the basis of their differing temporal responses. However, it is only relatively recently that optical technology has advanced sufficiently to achieve the necessary combination of high repetition rate and picosecond laser pulses, coupled with “gateable” multichannel detectors with matched repetition rates and short on-times. Both electronic and optical gating technologies have been shown to significantly improve the Raman spectra of highly fluorescent solutions. However, the performance of such systems with solid materials has not been reported in detail. To partially redress this imbalance, this article describes the ps-time-resolved Raman spectroscopy of solid films and powders. Excellent temporal resolution and fluorescence rejection was obtained with homogeneous films, but with powders, multiple scattering has the potential to significantly blur the time resolution. For example, after incidence of a 1-ps pulse on a powdered sample of trans-stilbene, the Rayleigh signal was spread over 100 ps in time and the Raman signal persisted for more than 300 ps. Simple models are presented that predict these temporal responses on the assumption that photons randomly “diffuse” through the powder, scattering at particle boundaries and sometimes reemerging to be detected at a later time. These dynamics imply that fluorescence rejection with bulk powders might be less effective than with homogeneous solids as the broadened Raman signal would be incompletely captured within the short detector “on” period. The fluorescence would be rejected, but so would the Raman signal (to some extent), giving a poor signal-to-noise ratio. This long-term signal persistence could also complicate the interpretation of pump-probe spectroscopy studies. However, further work is needed to assess the practical implications of these findings.

Photon Migration in Raman Spectroscopy

Monte Carlo simulation has been applied to study time-resolved Raman and Tyndall photon migration in opaque samples under isotropic and forward scattering conditions. For isotropic scattering, Raman and Tyndall intensities are predicted to decay according to t(1–n) and t−n, respectively, where the value of n depends on the ratio of the optical collection aperture to the mean scattering length. The simulation correctly reproduced the analytical results of n = 3/2 and n = 5/2 for a point source in infinite and semi-infinite media, respectively. In addition the model can be used to relate the time at which a Raman photon exits the sample to the mean depth at which it was generated. This could provide a useful tool for depth profiling the chemical composition of turbid systems, and hence be a useful addition to the established array of photon-migration techniques. The model was applied to analyze experimentally observed Raman and Tyndall decay profiles from powdered trans-stilbene. The transport mean free path (lt) was calculated to be ∼400 μm, which was significantly larger than the particle sizes present in the sample (∼10–100 μm). This implies that the particles were highly forward scattering, as would be expected for this size range. When highly anisotropic scattering was introduced into the model a much more reasonable scattering length (ls ∼ 40 μm) was obtained. Finally, a simple analytical model was developed that gives the correct relationship between the Raman and Tyndall decay exponents for isotropic scattering. To the best of our knowledge this work represents the first detailed study of Raman photon migration under time-resolved conditions.

Depth profiling in diffusely scattering media using Raman spectroscopy and picosecond Kerr gating.

We demonstrate how pulsed laser Raman excitation (∼1 ps) followed by fast optical Kerr gating (∼4 ps) can be used to effectively separate Raman signals originating from different depths in heterogeneous diffusely scattering media. The diffuse scattering slows down photon propagation through turbid samples enabling higher depth resolution than would be obtained for a given instrumental time resolution in an optically transparent medium. Two types of experiments on two-layer systems demonstrate the ability to differentiate between surface and sub-surface Raman signals. A Raman spectrum was obtained of stilbene powder buried beneath a 1 mm over-layer of PMMA (poly(methyl methacrylate)) powder. The signal contrasts of the lower stilbene layer and upper PMMA layer were improved by factors ≥5 and ≥180, respectively, by rejecting the Raman component of the counterpart layer. The ability to select the Raman signal of a thin top surface layer in preference to those from an underlying diffusely scattering substrate was demonstrated using a 100 μm thick optically transparent film of PET (poly(ethylene terephthalate)) on top of stilbene powder. The gating resulted in the suppression of the underlying stilbene Raman signal by a factor of 1200. The experiments were performed in back-scattering geometry using 400 nm excitation wavelength. The experimental technique should be well suited to biomedical applications such as disease diagnosis.

Confocal Raman microscopy: performance, pitfalls, and best practice.

C onfocal Raman microscopy is an extremely useful technique that permits nondestructive, spatially resolved measurements deep within transparent samples simply by focusing the laser beam at the point of interest. Moving the laser focus allows generation of one-dimensional (1D) depth profiles, and 2D and 3D (volumetric) images. However, in order to correctly interpret the data, it is important to understand exactly where the laser beam is focused and to know the volumetric resolution of the probe beam. These are actually non-trivial questions. The objective of this article is to summarize the critical factors that determine the spatial accuracy, resolution, and sensitivity of confocal Raman microscopy and to highlight the precautions that should be taken to collect high quality, quantitative data. No attempt is made to review the applications of Raman microscopy; these are simply too diverse, spanning topics from art conservation to medical diagnosis. However, the same basic principles must be adhered to, irrespective of the application, if reliable conclusions are to be drawn. Two main topics are considered. The first is the need for properly corrected objectives for depth profiling beneath the surface of transparent samples. If this is not done, the confocal profile will have an incorrect depth scale, degraded depth resolution, and reduced spectral intensity and signal-to-noise ratio (S/N). Even if modeling is used to account for the aberrations and to compute corrected profiles, degraded depth resolution and S/N still occur, which limits the performance. The second key issue is that even with a corrected objective operating with the best attainable resolution, the axial point spread function, which determines the depth resolution, has quite broad wings, so weak signals can be detected from regions quite distant (tens of micrometers) from the point of tightest focus. With thick transparent samples, the integrated signal from these out-of-focus domains can be significant or even dominant, resulting in unusual and counterintuitive observations. This effect is noticeable both for confocal profiling and for lateral scanning over cross-sections; in short, one cannot simply assume that data is acquired with a volumetric resolution of ~1 lm. The effect is especially important when one needs to chemically interpret the spectra rather than just view an image or a profile, since this leads to contamination of spectra with spurious bands. Finally, it is important to note that while some of the effects discussed here seem strange when they are first encountered, most have been known since the early days of confocal micro-spectroscopy. Consequently, few of the results discussed here would necessarily surprise a skilled microscopist. However, it is clear from the literature over the last decade or so that many Raman microscopists (the author included) are gradually re-learning these lessons and, as a result, significant advances have been made in the acquisition and interpretation of confocal Raman data. It therefore seems appropriate and timely to summarize these learning points in a review article.

Optimizing Depth Resolution in Confocal Raman Microscopy: A Comparison of Metallurgical, Dry Corrected, and Oil Immersion Objectives

Spherical aberration is probably the most important factor limiting the practical performance of a confocal Raman microscope. This paper suggests some simple samples that can be readily fabricated in any laboratory to test the performance of a confocal Raman microscope under realistic operating conditions (i.e., a deeply buried interface, rather than the often-selected alternative of a bare silicon wafer or a thin film in air). The samples chosen were silicon wafers buried beneath transparent polymeric or glass overlayers, and a polymer laminate buried beneath a cover glass. These samples were used to compare the performance of three types of objectives (metallurgical, oil immersion, and dry corrected) in terms of depth resolution and signal throughput. The oil immersion objective gave the best depth resolution and intensity, followed by a dry corrected (60×, 0.9 numerical aperture) objective. The 100× metallurgical objective was the worst choice, with degradations of ∼5× and 8× in the depth resolution and signal from a silicon wafer, comparing a bare wafer with one buried under a 150 μm cover glass. In particular, the high signal level obtained makes the immersion objective an attractive choice. Results from the buried laminate were even more impressive; a 30× improvement in spectral contrast was obtained using the oil immersion objective to analyze a thin (19 μm) coating on a PET substrate, buried beneath a 150 μm cover glass, compared with the metallurgical objective.

FT-IR imaging of polymers: an industrial appraisal

Chemical structure and conformation and physical property anisotropy at the microscopic level can have a major influence on the macroscopic performance characteristics of polymer products. Images based on infrared spectral differences and changes generated using FT-IR microscopy techniques are becoming increasingly used to highlight both chemical structure variations and morphology gradients within polymer articles. This paper illustrates potential industrial applications to polymer characterisation and analysis using mid-infrared FT-IR microscopy systems fitted with focal plane array (FPA) detectors.