Brian R. Stallard

Near-IR Reflectance Spectroscopy for the Determination of Motor Oil Contamination in Sandy Loam:

We have investigated the application of near-IR reflectance spectroscopy to the determination of motor oil contamination in sandy loam. Although the present work is concerned with a specific case of contamination, we discuss the possibility of applying the method to other organic contaminants and other types of soil. The spectral region considered was 1600–1900 nm, which contains the first overtone of the CH stretch. Using a commercial Fourier transform spectrometer together with cross-validated partial least-squares data analysis, the one-sigma precision for the determination of motor oil in sandy loam was 0.17 wt % (0.13 to 0.26 wt % at the 95% confidence level). The largest contribution to the precision of the determination was sampling error, or inhomogeneity in each sample. Given the precision limit imposed by the sampling error, we found that the performance of the spectrometer could be lowered without affecting the overall precision. In a modeling exercise, adequate performance was obtained with a spectrometer having only seven spectral channels with a spectral resolution of 10 nm and a spectral noise level of 10−3 absorbance units. A design for an inexpensive miniature instrument is presented.

Generation of Synthetic Background Spectra by Filtering the Sample Interferogram in FT-IR

The calculation of an absorbance spectrum depends on the measurement of a blank, or background spectrum. In many cases, such as the determination of atmospheric constituents with the use of open-path Fourier transform infrared spectroscopy (FT-IR) or the determination of water vapor in a gaseous sample, it is very difficult to obtain a good background spectrum. The difficulty is due to the fact that it is nearly impossible in these situations to measure a spectrum with no analyte features present. We present a method of generating a background spectrum based on filtering the analyte features from the sample spectrum. When the filtering method is used, the accuracy of the results obtained is found to be dependent upon the analyte peak width, peak height, and type of filter employed. Guidelines for the use of this background generation technique for quantitative determinations are presented.

A two dimensional array of optical interference filters produced by lithographic alterations of the index of refraction

We describe a new concept for producing, on a single substrate, a 2D array of optical interference filters where the pass-band of each element can be independently specified. The interference filter is formed by optically contacting two dielectric mirrors so that the top quarter-wave films of the two mirrors form a Fabry-Perot cavity having a half-wave thickness. In the new device, we propose to etch an array of subwavelength patterns into the top surface of one of the mirrors before forming the cavity. The patterns must have a pitch shorter than the operational wavelength in order to eliminate diffraction. By changing the index of refraction of the half-wave layer, or the optical thickness of the cavity, the patterning is used to shift the pass-band and form an array of interference filters. One approach to producing the array is to change the fill factor of the pattern. Once the filter array is produced it may be mated to a 2D detector array to form a miniature spectrophotometer.

NEAR-IR VERSUS MID-IR : SEPARABILITY OF THREE CLASSES OF ORGANIC COMPOUNDS

Recently there has been a surge of interest in spectroscopic sensors operating in the near-IR, although it is recognized that the mid-IR contains more spectral information. The general question addressed in this paper is, How much specificity is lost in choosing the near-IR over the mid-IR for sensor applications? The example considered is the separability among three classes of organic compounds: alkanes, alcohols, and ketones/aldehydes. We use spectra from two sources: the Hummel polymer library (mid-IR) and the library of Buback and Vögele (near-IR). This is the first paper on class separability to make use of this new near-IR library, available in digital form only since July 1995. Five spectral regions are considered: region 5, 10,500 to 6300 cm−1; region 4, 7200 to 5200 cm−1; region 3, 5500 to 3800 cm−1; region 2, 3900 to 2500 cm−1; and region 1, 2500 to 500 cm−1. Class separability is explored both qualitatively and quantitatively with the use of principal component scatter plots, linear discriminant analysis, Bhattacharyya distances, and other methods. We find that the separability is greatest in region 1 and least in region 2, with the three near-IR regions being intermediate. Furthermore, we find that, in the near-IR, there is sufficient class separability to ensure that organic compounds of one class can be determined in the midst of interference from the other classes.

Use of diffracted light from latent images to improve lithography control

As the microelectronics industry strives to achieve smaller device design geometries, control of linewidth, or critical dimension (CD), becomes increasingly important. Currently, CD uniformity is controlled by exposing large numbers of samples for a fixed exposure time which is determined in advance by calibration techniques. This type of control does not accommodate variations in optical properties of the wafers that may occur during manufacturing. In this work, a relationship is demonstrated between the intensity of light diffracted from a latent image consisting of a periodic pattern in the undeveloped photoresist and the amount of energy absorbed by the resist material (the exposure dose). This relationship is used to simulate exposure dose control of photoresist on surfaces which have different optical properties chosen to represent surfaces typical of those found in operating process lines. Samples include a variety of photoresist materials and substrates with a wide variety of optical properties. The optical properties of the substrates were deliberately varied to determine the effect of these properties on CD (in the presence and absence of an exposure monitor) during lithography. It was observed that linewidth uniformity of the developed photoresist can be greatly improved when the intensity of diffracted light from the latent image is used to control the exposure dose. Diffraction from the latent image grating structures was modeled using rigorous coupled wave analysis. The modeling is used to predict the diffraction from a latent image as a function of the substrate optical properties and the parameters of the latent image (i.e., linewidth, sidewall angle). Good agreement is obtained between theoretical and experimental observations. Conversely, the inverse problem is solved in which the parameters of the diffracting structure (the latent image) are determined from a measurement of the diffracted power. Therefore, the diffracted power can be monitored for the purpose of determining when the latent image will produce the proper CD upon development.

Semi-autonomous registration of satellite imagery using feature fitting

Interband coregistration of multispectral satellite imagery is essential to exploiting the spectral information inherent in these data. A semi-automatic image registration method has been developed for Multispectral Thermal Imager (MTI) data. This registration method, based on feature fitting within the image, is applicable to the 14 MTI spectral bands that contain ground information; these spectral bands range from 0.45 to 10.7micrometers . The feature fitting registration method requires selection of an appropriate scene feature in the image, usually a crossroad or other feature with moderately high contrast to compute the required shift in x and y for each band. This paper describes the algorithm and provides examples of images registered using this method. Preliminary results show that for MTI image registration, feature fitting yields better results than cross-correlation. Results also show that this algorithm works well for a broad variety of scenes; this algorithm has been applied to images with scene content ranging from desert images with very little structure to heavily forested images. This method has been developed in support of the MTI mission, but may easily be extended for use on image data collected by other multispectral sensors.

Quantifying Multivariate Classification Performance - the Problem of Overfitting

We have been studying the use of spectral imagery to locate targets in spectrally interfering backgrounds. In making performance estimates for various sensors it has become evident that some calculations are unreliable because of overfitting. Hence, we began a thorough study of the problem of overfitting in multivariate classification. In this paper we present some model based results describing the problem. From the model we know the ideal covariance matrix, the ideal discriminant vector, and the ideal classification performance. We then investigate how experimental conditions such as noise, number of bands, and number of samples cause discrepancies from the ideal results. We also suggest ways to discover and alleviate overfitting.

Anomaly detection using simulated MTI data cubes derived form HYDICE data

In this work we quantify the separability between specific materials and the natural background by applying receiver operating curve (ROC) analysis to the residual errors from a linear unmixing. We apply the ROC analysis to quantify performance of the multi-spectral thermal imager (MTI). We describe the MTI imager and simulate its data by filtering HYDICE hyperspectral imagery both spatially and spectrally and by introducing atmospheric effects corresponding to the MIT satellite altitude. We compare and contrast the individual effects on performance of spectral resolution, spatial resolution, atmospheric corrections, and varying atmospheric conditions.

Using optical absorption to measure the state of charge of lead-acid batteries

Proposed here is a method of measuring the state of charge of lead-acid batteries based on the assumption that the addition of sulfuric acid to water will reduce the optical absorption at certain absorption peaks of pure water. This reduction is expected to result from the accompanying drop in the concentration of water molecules. Measurements of the absorption in the vicinity of 0.97, 1.20, and 1.45 ?m indicate that this is indeed the case, although deviations from a linear dependence on water concentration are noted. Two ways of implementing this phenomenon in a lead-acid battery are presented; one involves an absorption cell in a flooded battery, while the second involves an optical fiber woven into an absorbed-glass-mat battery. In the second case, the absorptive electrolyte serves as the cladding of the optical fiber and introduces attenuated total internal reflection into what would otherwise be perfectly guided modes.

Near-Infrared Spectroscopy with a Dispersive Waveguide Device

Miniature, low-cost sensors are in demand for a variety of applications in industry, medicine, and environmental sciences. As a first step in developing such a sensor, we have etched a grating into a GaAs rib waveguide to serve as a wavelength-dispersive element. The device was fabricated with the techniques of metal-organic chemical vapor deposition, electron-beam lithography, optical lithography, and reactive ion-beam etching. While full integration is the eventual goal of this work, for the present, a functional spectrometer was constructed with the addition of a discrete source, sample cell, lenses, and detector. The waveguide spectrometer has a spectral resolution of 7.5 nm and a spectral dispersion of 0.11°/ nm. As presently configured, it functions in the spectral range of 1500 to 1600 nm. A demonstration of the analytical capability of the waveguide spectrometer is presented. The problem posed is the determination of diethanol amine in an ethanol solution (about 10 to 100 g/L). This procedure involves the detection of the first overtone of the NH stretch at 1545 nm in a moderately absorbing solvent background. The standard error of prediction for the determination was 5.4 g/L.