Infrared spectroscopy, as a routine analytical tool, became popular in the mid to late 1950's. Since that time, it has blossomed, it has nearly died, and now it is experiencing a major rebirth in the form of Fourier transform infrared spectroscopy (FT-IR). The move away from traditional sequential scanning, dispersive instruments was relatively fast once moderately priced FT-IR instruments became available. Although sequential scanning instruments still provide a very cost effective solution for routine infrared analysis, their use in research and analytical problem solving applications is now minimal compared to FT-IR.

Near infrared (NIR) spectroscopy has been in a rapid development phase for the last 20 years! While it has become more widely recognised as a valid analytical technique, there is still a body of opinion which believes that it is a black box technique without a solid theoretical background. This paper reviews the progress of NIR analysis, the reasons why it is theoretically complex but simple in practice and the most recent development in instrumentation which makes it even more "user friendly".

The NPL hemispherical reflectometer/transmissometer can be used to measure a complete set of spectral radiometric properties from 2.5 to 56 micrometres for any kind of sample, whether or not it has a regular component of reflectance or transmittance. A rotatable hemispherical copper mirror focusses radiation from a cylindrical source on to the sample. Systematic errors are caused by radiometric non-linearity from interreflection effects and by temperature changes of the source arising from interreflection and from the necessary purging air current. A custom-built power supply is operated in a constant-load-resistance mode to maintain the source at a constant temperature to within 0.1 K. This allows simplification of the correction procedures, but the use of standard reference samples is still required for calibration service work. A new method of use for absolute measurements employs a supplementary mirror for additional measurements to let the monochromator optics view the source directly, without and with the sample present. The ratio of these readings then provides a simple correction factor for the reflectometer value, allowing immediate calculation of absolute spectral reflectance. Experimental tests show that the method is valid from 2.5 to 36 micrometres with a grating spectrophotometer, but becomes impracticable at longer wavelengths. A Fourier transform instrument has been modified radically to accommodate the hemisphere reflectometer for further tests.

This paper reports on the development of a small, robust, battery operated near infra-red (NIR) reflectance device, designed for rapid on-farm measurement of the moisture content of forage crops without prior sample preparation. It has potential application to other agricultural or food materials. The instrument is based on two light emitting diodes (LEDs), a germanium detector and a control CMOS single chip microcomputer. The meter has been calibrated to give a direct read out of moisture content for 4 common grass varieties at 3 stages of development. The accuracy of a single point measurement on a grass sample is approximately ± 6% over a range of 40-80% (wet basis). However, the potential accuracy on a homogeous sample may be as goon as 0.15%.

Mid-infrared laser optoacoustic spectroscopy has been used to detect a variety of gases and vapours. Performance was calibrated using the signal from a known concentration of ethene, and then the method applied to the perfume alcohol geraniol. Detection limits were found to be 1 ppb for ethene and 70 ppb for geraniol on their strongest absorption lines for a few seconds measurement time.

Each process application of IR spectroscopy, and there are a great many, easily qualifies as a dedicated subject, but since time and space does not allow for this, it is clear that this paper will represent more a catalogue of tried and tested techniques rather than a technical incite into their design and operation. It will cover temperature measurement, gas and liquid analysis by transmission techniques and the analysis of solids by reflectance methods. Atmosphere monitoring techniques will be treated as a special case and will touch on the use of laser techniques and other long path length methods of IR analysis.

The well known advantages of process infrared analysers in selectivity and stability have been somewhat offset in practice by the need for sampling systems of varying degrees of complexity. These can add to 100% of the basic analyser cost. Even more significantly the sampling system may not be as reliable as the analyser at the centre of it. It is now possible to minimise sampling requirements and three specified examples were presented. The ideal analyser, whatever technique it uses, would be as easy to apply as a temperature sensor. So this would suggest a probe, immersed in the sample, be it gas or liquid, and an electronic unit to connect the infrared signal to the concentration being measured. It is now possible to come close to this. To do this, there must be a way of doing or making unnecessary each of the functions of the concentrated sampling systems.

Two trends can be discerned in the recent development of methods for quantitative analysis using infrared spectroscopy. The first is the adoption of methods which use whole regions or complete spectra rather than a limited number of discrete frequencies. This can be attributed to the availability of the necessary computing power at low cost. Fourier transform spectrometers provide a complete spectrum with no time penalty relative to a measurement at a single frequency. This too has encouraged the development of these methods. The second trend has been the move away from reliance on Beer's Law. Statistical approaches, well established in other fields, can perform well in situations where there is no linear, or other simple relationship between absorbance and the value of the property of interest. These methods have encouraged the application of IR spectroscopy to the quantitative analysis of increasingly complex mixtures, and to the determination of properties other than component concentrations. This paper will deal with the move away from methods based explicitly on Beer's Law. This change is probably more significant than the extension of traditional approaches from discrete frequency to complete spectra. The mathematics will not be treated in any detail. It is possible to identify the advantages and limitations of the various approaches by examining the underlying ideas rather than the detail of the algorithms.

The specificity and sensitivity of infrared spectroscopy for the characterisation of organic compounds have long been recognised, and with continuing improvements in instrumental performance and in sampling procedures it is now possible to obtain good quality spectra from rangecebiological materials and from coals, both of which were previously regarded asrather intractable. The structural complexity of these materials, and of some synthetic polymers, leads to the occurrence of strongly overlapped peaks, whose half widths are usually appreciably greater than the instrumental resolution employed, thus preventing their separation. Furthermore, these three types of material, being of high molecular weight, are not readily soluble, must be examined in the condensed phase, where intermolecular interaction adds to the band broadening and the consequential overlap. This overlap reduces the value of the spectra for structural diagnosis and quantitative measurements.

A major driving force in the development of automated analytical systems is the production of reliable, relevant information within the timescale of a manufacturing operation. The overall time constraints can be met only if these systems are transferred from the analytical laboratory to the production area. Separative techniques do not lend themselves to this type of development. However, the performance of modern spectrophotometers, from the uv to mid ir, allows qualitative and quantitative information to be extracted from spectral data using powerful computational techniques, without prior separation of the components.

Photoacoustic and photothermal techniques have found wide application in infra-red spectroscopy over the last few years. Samples are illuminated with amplitude modulated radiation which produces excited spectroscopic states. These excited states relax by radiationless de-excitation pathways producing a modulated heat source within the sample. The modulated thermal signal can be recovered by a variety of photoacoustic and photothermal probes. Since the intensity of the thermal signal is a function of the amount of energy absorbed by the sample, the use of tunable radiation sources allows spectroscopy to be carried out. In addition, information can be obtained about the excited state behaviour of the system, and its physical characteristics. Techniques are applicable to a wide range of sample states and types. This paper reviews recent literature in the area, concentrating on the main current trends.

A brief description of the principles of Fourier transform mid infrared photoacoustic spectroscopy is given. The factors affecting the quantitative interpretation of the spectra and discussed with particular reference to the effects of particle size. It is concluded that if suitable care is taken photoacoustic spectroscopy can yield useful quantitative results comparable to other methods.

In scanned photothermal radiometric imaging a sample is periodically heated using an intensity modulated laser beam. Absorption of the light energy generates a thermal wave which diffuses into the bulk of the sample. The resulting changes in surface temperature are measured using a focused infrared detector. One of the principal advantages of thermal waves is that they provide a convenient means for performing non-contacting subsurface imaging. Often a very tightly focused laser spot size is used as the heat source, which allows excellent surface resolution. In this case, however, the physics of image formation and the factors limiting the subsurface resolution need to be carefully considered. This paper presents a simple but rigorous Fourier transform analysis which enables one to calculate the temperature distribution throughout a material with a finite sized source and detection area. We plot results for various detector/source dimensions and develop criteria to determine whether attenuation or spreading of the thermal wave limits the spatial resolution. In addition the analysis is particularly useful for determining the range of validity of the simple one dimensional model which is often used to interpret experimental results.

Thermal wave inspection is introduced with particular reference to the measurement and testing of surface coatings. Measurements are presented which demonstrate the effectiveness of the technique for assessing the thickness of plasma sprayed coatings and for detecting and imaging sub-surface defects. A new miniaturised testing system employing a high power semiconductor laser and a high collection efficiency infrared detector is described.

The decoration and protection quality of coatings on polymers is of considerable interest for industrial applications. However, at present there is no non-destructive (NDE) method to monitor the quality of these coatings during the manufacturing process or while they are in use. As an approach for such a method we use photothermal analysis where the propagation and reflection of optically generated thermal waves is investigated. We found that one can monitor the drying process, the effect of surface temperature treatment, and coating thickness (accuracy + 2 μm in 50 μm thickness). The information obtained with this remote NDE method is adequate for most industrial applications, eg car manufacturing.

Carbon fibre reinforced plastics (CFRP) are of increasing importance with respect to industrial applications. There is an urgent demand for the nondestructive characterisation and evaluation of these materials. In the search for new non contacting methods we investigated the applicability of photothermal wave analysis where the propagation of laser generated thermal waves is evaluated. With remote photothermal infrared radiometry we could monitor the curing and ageing process of CFRP-prepregs. Fibre content and orientation could be measured. Defects (cracks in matrix, fibre-break, insertions of various materials) were located and characterised. Our results can well compete with those obtained using conventional NDT-methods.

Early attempts to extend staring-mode sensing into the thermal infrared spectrum failed, because the resulting imagery was dominated by fixed pattern noise. The source of this noise was modulation of the infrared background by sensor response non-uniformities. in 1973, use of internal photoemission from Schottky silicide arrays was proposed as a means of achieving the photoresponse uniformity necessary to obtain useful thermal imaging capability. Since that time, there has been a steady evolution in silicide sensor technology. Current silicide cameras have sensitivity comparable with the best scanning systems. These cameras are based upon the largest infrared arrays now available. This paper will describe recent advances in silicide sensors and project future technology trends.

For the last two decades, ground attack and strike aircraft have possessed an impressive capability to reach and destroy targets by day and in favourable weather conditions. Operations in poor weather and darkness have been denied until recently when developments of thermal imaging combined with Night Vision Goggles (NVG) have opened up new operational concepts. Most aircraft have been severely limited at low level by the pilot's ability to acquire visually targets at a reasonable distance such that a first pass attack is possible. Besides the basic problems of terrain screening the pilot is faced with the effects of daylight haze and mist combined with the smoke/dust of a battlefield scenario.

The two most common spectral bands used in the thermal infrared are the 3-5 μm and 8-12 μm bands. For years, substantial controversy has occurred over which of these spectral bands has the greater merit. A qualitative approach to address this question is presented.